ThezmC-

o(,(j)

N)§(o

N)§

§zm

@mTo

New England]ournal of Medicine

Established in 1812 as The NEIV ENGLAND JOURNAL OF MEDICINE AND SURGERY

VOLUN'IE 33I \o\rE\tBER 10. 199.t NUN,IBER 19

Original ArticlesLow Serum Thyrotropin Concentrations as

a Risk Factor for Atrial Fibrillationin Older Persons . L249

C.'f. SlrvIx .c.xo Orrrlns

Bone Marrow Transplants from HLA-Identi-cal Siblings as Compared with Chemo.therapy for Children with AcuteLymphoblastic Leukemia in aSecond Remission ....., . 1253

AJ. Bannur .rxo ()rnens

Accumulation of Nuclear p53 and TumorProgression in Bladder Cancer .. 1259

D. llsnrc ,c..'o OrrrEps

Case Records ofMassachusetts General

A 77-Year-Old Woman with Fever,and Pain in the Head and Legs

\I. H.r;us l:o E."l'. HrpI-cv-\\'Hvrs

theHospitalSweats,

.. L293

EditorialSubclinical Hyperthyroidism - Just a Low

Serum Thyrotropin Concentration, orSomethingMore?. ....... 1302

R.D. Urrr;rn

Sounding BoardFor Every Dollar Spent - The Cost-Savings

Argument for Prenatal Care .. ... 1303

J. Hurrrxcrox .lxo F.A. Coxxrlr

Images in clinical Medicine correspondence

coronary Arteriovenous Fistula on coronary .^^. iii}i::ii[lï:î"Jrlîï;;; : : : :: ::: : : :: li33Angiography " " ' l2ti5 Acetaminophen Poisoning and Liver Function. . . . 1310

K. Fr.Jrse .rso \\'. Srtnrr-rx Mineral Balance in Postmenopausal WomenTreated with Potassium Bicarbonate . 1312

Pulmonary-FunctionTesting .... 1313Special Article Risksof sâlmeterol?....... .... l3l4

Cancer Therapy Meets p53 ...... 1314State Practice Environments and the Supply of Postmortem Viâfifty of H..-u. Immunodeficiency

Physician Assistants, Nurse Practitioners, Virus - Implications for the Teaching of, anà Certified Nurse-Midwives . .. 1266 Anatomy .. . 1315

E.S. Sr.xscr:xsxI, S. Srrsorr. C. Brzrr.l'II.E. Srrrror.., esii F. \It'rr..rr Occasional Notes

The Train Is Leaving the Station .... f 316

Review Articles B']' \\'IrLr'+,s

Book Reviews . .. .. . l3l7Medical Progress: Severe Adverse Cutaneous ' - -' )

/. Reactioris to Drugs .....( tZZÿ Books Received .. . . . 1320

/ .J.C. Rorrrrr' :so R.S. S'runx '\-r' Notices . . . . . . 1323

^--^. T---^r^-*:-- Correctioni Mechanisms of Disease: TransformingI GrowthFactoiBirrri.ro"Fibros"is..... 1286 MoreontheGUsTotrial"' "" 1323

I fV..n. Bororn rro \.A. \oerr Information for Authors .. ' 1324

iI Owned, published, and OCopyrighted, 1994, by the Massachusetts Medical SocietyI (Jwned, ruDllsneq' anq euuPrrr5rt

ithe\Iassachusetts\1edicalSocietrirrrdprirlttclinErrqIand\ bv St. Ii.cs (Àndoicr) Lrd.. \\'cst Pornçav. .\ndovcr. SPI[) 3SF. U.K

\ ,' , . L Rcgistered as a ne\\.spapcr at the Post Olice lSS\ 0028-+793'r tt-/ -l-rl

t o__ rÿ.

3

TIre Europeaî

Euro interactive folrrmTrans =

for continuing

Med medical educationLive by satellite to anexpanding network ofhospitals in 2O Europeancountries

22 November Improving survival in patients with coronary heart disease:the role of lipid management.A special progralzrrne based on tbe Results presentatîon of tbe 45Study at a bot-line session of tbe American Heart Associationmeeting in Dallas on 17 Nouember 1994

23 November Quafity of üfe, quality of care in oncology

29 November Therapeutic strategies in osteoporosis

6 December Optic nerve health

L3 December The difficult-to-manage glaucoma patient

Earlier 1994 prograrnmes available on VIIS cassette include:

Euthanasia: the example ofThe Nethedands . Diagfrosis and treatment of biliary strictures .AIDS patients: emerging trends in fungal infections . The Menopause:the consequences of

hormonal change . Secondary prevention of CHD in women . New guidelines for the delayingof progression of renal disease . Innovations in the treatment of schizophrenia

. ESC Guidelines on the prevention of CHD in clinical practice

Bor 1995, EuroTransMed invites additional sponsorship for prograflunes on:Minimally invasive surgefy; Cancer (in association with the European School of Oncology

and the EORTC); Medical ethics; Genetics; Endocrine and metabolic diseases;

Medical imaging and technology; Gastroenterology.

Further inforrnation frorn:EuroTransMed, The London Television Centre, Upper Ground, London SE1 9LT, UK.

(tel: +44 7 1 7 37 8458; fax: +44 7 1 261 37 38)EuroTransMed Foundation, c/o Department of Surgery, University Hospital, PO Box 9600,

2300 RC Leiden,The Netherlands. Tel[Fax:+3171 219779

Officiol iournol of rhe

lnlernotionol Epidemiologicol lnternotionql Journol ofAss'c"'n Epidemiology

Editor: Peter O D Phorooh, Deporiment of Public Heolth, Universiÿ of Liverpool, Liverpool, L K

The International Journal of Epidemiology continuesto be the leading journal in its field, and is essentialreading for anyone who needs to keep up-to-date withthe very latest developments throughout the world.

The joumal has for many years encouraged communica-tion among those engaged in the research, teaching and

application of epidemiology of both communicable and

non-communicable disease. Research into healthservices and medical care is also covered, as are newmethods for the analysis of data used by those whopractice social and preventative medicine.

Volume 23, number 1

Space-Time Clustering in lnsulin-Dependent Diabetes Mellitus (IDDM)in South-East Sweden. Ull Samuelsson, Calle Johansson,John Carstensen and Johnny Ludvigsson.

DNA Probes as Epidemiological Tools for Surveillance ol Plasmodiumfalciparum Malaria in Thailand. Bobert H Barker Jr,

Trairal Banchongaksorn, Jeanne M Gourual, Illhnnapa Suwonkerd,Kamolwan Bimwungtlagoon, Nitaya Srittong and Dyann F Wirlh.

Volume 23, number 2

Verbal Auiopsies Tor Adult Deaths: lssues in their Development

and Validation. 0aniel Chandramohan, Gillian H Maude,Laura C Rodriques and Richard J Hayes.

Childhood and Adolescent Passive Smoking and the Risk of Female

Lung Cancer. Fu-Lin Wang, Edgar John Love, Ning Liu andXu-Dong Dai.

Why lnfant Very Low Birthweight Rates have Failed to Decline in theUnited States Vital Statistics. Samuel Sepkowitz.

Prevalence and Causes ol Visual lmpairment in ltaly.Allredo Nicolosi, Paolo E Marighi, Paola Fizzardi, Alberto Osellaand Stelano Miglior.

A Mathematical Model for the Prediction of the lmpact of HIV lnfection0n Tuberculosis. Michael Schulzer, M P Radhamani,Stelan Grzybowski, Edwin Mak and J Mark Fitzgerald.

Volume 23, numbet 3

Differential ETTects oT Tar Content, Type of Tobacco and Use oT a Filteron Lung Cancer Risk in Male Cigarette Smokers. Simone Benhamou,Ellen Benhamou, Ariane Auquier and Roberl Flamant.

Diarrhoea - Defining the Episode. Saul S Mottis, Simon N Cousens,

Claudio F Lanala and Betty R Xilkwood.

Sex Differences in Measles Mortality: A World Revier,r,'

Michel Garenne.

Volume 23, nurflber 4

Forced Movements of Population and Health Hazards in TropicalAfrica. R mansell Prothero.

Myocardial lnfarction Case-Fatality Gradient in Three French Regions:ïhe lnfluence 01 Acute Coronary Care. Phillipe Amouyel, DominiqueArveiler, Jean-Piere Cambou, Michèle M0ntaye, Jean-BenardRuidavets, Annie Bingham, Paul Schafler and Jaccyues-LucienRichard.

Determinants of Stunting and Recovery from Stunting in NorthwestUganda. Venanzio Vella, Andnw Tomkins, Amandlo Borghesi,Giovanni Batlisla Migliori and Vincent Yooman 0nyem.

Volume 23, number 5

Stratospheric 0zone and Health. Bruce K Armslrong.

Short-ïerm Etlects of Air Pollution on Daily Mortalily in Athens: A

Time-Series Analysis. G Touloumi, S J Pocock, K l(atsouyanni andD Trichopoulos.

W-@tuil@@dù.É-d+3iiM+È.i e6éi

Modelling AIDS Reduction Strategies. D A Kault.

Drtferential Response to Early Nutrition Supplementation: Long-Term

Effects on Height at Adolescence. M T Ruel, J Rivera, J-P Habichtand R Martorell.

Mortality Following Radiation Treatment for lnfertility of Hormonal0rigin or Amenorrhea. E Ron, J D Boice Jr, S Harnbuqer andM Stovall.

Changing Patterns in the Epidemiology and Hospilal Care o1 Peptic

Ulcer. P Primalesla, M J Goldacrc and V Seagroatt.

Determinants of Excellent Health: Ditferent from the Determinants oflll-Health? J P Mackenbach, J van den Bos, I M AJoung, H van deMheen and K Stronks.

IIIIIIII'IIT IIIIIl^.; Please send me subscriplion details and rr free sample copyI of the lnternational Journal of Epidemic,logy.

III

(Ee4t2)

IIII etease return to: Journals Marketing Depl,, O:dord University press,

I Walton Street, Oxford OX2 6DP, UK. Tel: +44 (O) A65 56767. Fax: +44 (O) 865 267773. ln N. America, please return to: Journals

I Marketing Dept., Oxford University press, 2001 Evans Road, Cary,I NC 27513, USA. Iel: 919 677 0977 Fax: gtg 677 1714

#:: ilhs

.A-*'t,5-G.

:}'r*4.

\. ri't,{l"r

rlit\'l'1ii

Tlo get more information frmay \rye respectfully ask you to

\ilhile the technology that wentinto the new SONOS 2500 is quitecomplex, the thinking behind it issimple: the more information youhave on yornr patient's condition,the more co,nfidence you'll havein your diagnosis.

By listening to cardiologists andsonographers from around theworld, we have created an imag-ing machine with the capabilitiesmost important to you-capabil-ities that enable the 2500 to giveyou more diagnostic information

than any other ultrasound system.

It is the only system that offersour breakthrough Acoustic Quan-tification technology, giüng youreal-time areas, volumes and beat-to-beat ejection fractiors faster and

I

o

om the machine on the left,Iook at the machine on the right.

easier than ever before. And it is thefirst system with HP's exclusiveDynamic Elevation Beamformingand Extended Definition ColorFlow for superior image quality.

PIus, the fi,rll potential of the 2500

is unleashed by our new digitaluser interface. With responsivetouch screens, it puts all modesand controls right at your finger-tips to give you instant access topatient information. So you workmore efficiently.

For more information, call yourIocal sales representative. Æter all,giving you more information iswhat the SONOS 2500 is all about.

Vî'3FY-ITJ'

Penicillins Third-generationcephalosporins

4.5 g/q t h (piperacillin sdium ftazûaclam sodium)

Fast, strong, broadagainst serious infection

Lederle (lnternational).\ Dir: ol Anrerican Cvanamid\\irrne \l 07-170

Side Effects: \,lost corrnon in clinicai trials utr.'diiirrhea. nausea. and herdiicher a: seel rith other iniectablc perririllin:.

Ful) prescribing infornrar:on i-. avajlablc l-.r conlacting rour local Lcderlc/ \nterican Cvanamid ollice-/aCÊT.rl or Âf,rencan Cyê-iir d Coîpary

@1g-qJ. Leder e

The New England

Jolrrnal of Medicine@Copyright. 1994, by the Massachusetts Medical Society

\iolume 331 \OVE\IBER 10. 199+ \umber l9

LOW SERUM THYROTROPIN CONCENTRATIONS AS A RISK FACTOR FOR ATRIALFIBRILLATION IN OLDER PERSONS

Ct-.c,nr T. S-lr.r'rx. \{.D.. Axoxlrv Gr:,I-lrn., \I.D.. Prrrr-rp A. \\rolr. \I.D., Ar-saR.rJ. Brlexcar<, NI.A..Ennol B.rrrr. Pn.D.. P..rrrrr.r B.lcH.rrrc:H, 8.A., Pr,rE'n \\'.F. \\'rr-sox, \I.D..

Elrr.r-rr.|. BrxJ.l,urx. I,I.D.. exo Rer.pn B. D'--\cosrrro. Pn.D.

Abstract Background. Low serum thyrotropin concen-trations are a sensitive indicator of hyperthyroidism butcan also occur in persons who have no clinical manifesta-tions of the disorder. We studied whether low serum thy-rotropin concentrations in clinically euthyroid older per-sons are a risk factor for subsequent atrial fibrillation.

Methods. We studied 2007 persons (814 men and1193 women) 60 years of age or older who did not haveatrial fibrillation in order to determine the frequency ofthis arrhythmia during a 1O-year follow-up period. Thesubjects were classiïied according to their serum thyrot-ropin concentrations: those with low values (<0.1 mUper liter; 61 subjects); those with slightly low values(>0.1 to 0.4 mU per liter; 187 subjects);those with nor-mal values (>0.4 to 5.0 mU per liter; 1576 subjects);and those with high values (>5.0 mU per liter; 183 sub-jects).

Results. During the 1O-year follow-up period, atrial fi-brillation occurred in 13 persons with low initial values forserum thyrotropin, 23 with slightly low values, 133 with

ATRIAL fibrillation is a well-knorvn manilestationl_\ of hr perthr roidism. ,\mong older people. inu'hom atrial fibrillation is common.r-r hvpertht,roid-ism is relatitelv Llncommon, hou'ever.'''; Thus. al-though hl'perthvroidism is a risk factor lor atrial fi-briliation, most oldcr people tvith atrial fibrillationdo not have h,vperthvroidism.s!' Subclinical h,vper-thvroidism. defined as a lou, serum thvrotropin con-centration in an asvmptomatic p"..on r,vith normalserum thr.roid hormor-re concentrations. is more com-mon amons older persons than overt hvperthyroid-ism. For example, altcr patients taking thr-roid hor-mone havc been excluded, Iow serum thl,rotropinconcentrations have been lound in 0.9 to 1.9 percent of

From thc Medical and Medical Research Seruices. Boston Vcterans AffairsMedical Center. Boston (C.T.S.. A.G.. 8.8., P.B.): the Section oT PrevenriveMedicine and Epidemiology-. Evans Memorial Depanment of Clinical Research(P.A.W.). and thc Departments of Neurology (P.A.W.). l\'lathematics (A.J.B..R.B.D.), and Medicine (C.T.S.. E.J.B.). Boston University, Boston; and thcFramingham Heart Study. Framingham. Mass. (P.A.W.. P.W.F.W.). Addressreprint requests to Dr. Sawin at the Boston Veterans Altàirs Medical Center.150 S. Huntington Ave.. Boston. MA 02130.

Supponed in pan b) the N{etlical Research Service. Depanment of Veterans

Affairs: bl grants from the National lnstitute of Neurological Disorders and

Srroke (2-R0l-NS'17950-12). the National Heart. Lung. and Blood Institute(R0l-HL.l042l-04). and Boots Pharmaceuticals: and b1' a contracl (NOl-HC-38038) with the National Heart. Lung. and Blood Institute.

normal values, and 23 with high values. The cumulativeincidence of atrial fibrillation at 10 years was 28 percentamong the subjects with low serum thyrotropin values(<0.1 mU per liter), as compared with 11 percent amongthose with normal values; the age-adjusted incidence ofatrial fibrillation was 28 per 1000 person-years amongthose with low values and 10 per 1000 person-yearsamong those with normal values (P : 0.005). After adjust-ment for other known risk factors, the relative risk of atrialfibrillation in elderly subjects with low serum thyrotropinconcentrations, as compared with those with normal con-centrations, was 3.1 (95 percent confidence interval, 1.7 to5.5; P<0.001). The 1O-year incidence of atrial fibrillation inthe groups with slightly low and high serum thyrotropinvalues was not significantly different from that in the groupwith normal values.

Conclusions. Among people 60 years of age or older,a low serum thyrotropin concentration is associated with athreefold higher risk that atrial fibrillation will develop in thesubsequent decade. (N Engl J Med 1994;331:1249-52.)

older pcrsons, Ièu' ol r,r,hom rvere judged to have clini-cal irvperthr,roidism. ro- I 5

\\'hether people rvith subclinical hr.perthvroidismhave an increased risk of atrial fibrillation is unknorvn.Because atrial fibrillatiorr is an independent risk factorfrrr stroke and can dccrease cardiac output, it is impor-tant to identifv anv factors that preclispose paticnts tohave tiris arrhythmia. To assess this risk u.e examinedprospectivelv the incidence of atrial fibrillation in re-lation to serum thvrotropin concentrations over l0vears among older people (more than 60 vears ol age)r,vho rvere participating in the Framinsham HeartStudv.

Mrrrroos\\'c studied all the members of the original cohort ol the Fra-

mingham Heart Studr rrho rvere 60 r'ears ol'aee or oldcr at thc timeol' the l5th biennial exanrination in 1978 through 1980. 'flrosenho h:rd atrial Iibrillation ar rhar rirne or had a historv of the:rrrhvthmia uere excludccl. :rs \!ere t\\'o subjects krund to hivc clini-cal hvpt'rthvroidism at tliat cxamination. Subject,. rvho did not re-turn lbr anv follorr'-up esaminations altcr the l5th L'xarninJti()n \\ crr:also excluded. \\'e revieuecl the records of thc remainine 2007 pcr-sons (Bl4 mcn ancl Il93 nomcn) to dcternriue uhether thev rveretakinq a thvroid hornrone prcpararion ar the rime of the lith esanri-natiorr and to idenril\ atrial fiLrrillation occurring during the nest I0lears. The 60 subjects rçith a histon ol lrr perthvroidism beforc the

I 250

I 5th examination and the I I 5 subjects rlho rlcre taking a thr roidhormone preparation at that time nere includecl.'l'hosc rvith a his-ton' ol hvperthvroidism had been treated uith surgeri' (21) sub-jects), antithvroid druss (15 subjects). radioiodine (13 subjects), oriodine (ir sulr.jccts): had not bcen treated (l sutrject)l or thcir treat-mcnt \\'ils unknortn (3 subjccts). À total of l()*B ol the men andwomen stuclied at the l5th examination returned lbr the 2Oth exami-nation: 6+8 persons had dicd in the interim. and the other 3l I hadrcturned f<rr onc or more of the intcrvening exarminations.

Serum sample s u cre collccted during the l 5th examination andstored at -20'(1. Serum thvrotropin cr,ncentrations rrere mt'asureclin 1990 and l99l uith ;r chemoluminescence assav (London l)iag-nostics. Eden Prairic. \linn.: this assav is norl madc br' \icholsInstitutc Diagnostics, San.Juan Capistrarlo. Clalif.): the sensitiritrof the assav rvas 0.005 mU per liter, and the interassav coefËcicnt o1'

variation *as 5 perccnt at I mU pcr liter and I I pcrcent at 0.0,1 mUper liter. Srrum thvrotropin conccntrations liad been measurcd inl98l through 1983 u,ith an older radioimmunoassar-.r'i'fhe nrcan of'

the serum thvrotropin concentrations that u'ere greater than 5 mUper liter and less th;in or equal to l0 mU per liter uas 6.8 mU pcrliter rvhen measured bv the older assav and 6.7 mU per liter bv thenewer zrssa).; lor valucs abovc 1 0 rnU per lite r, the meirns rqerc I 7.9mU per liter and 16.8 mU per liter. respectivelv.'l'he correlationcoe{licient of the t\\'o assat's lor values above 5 mU per liter rvas

0.91. These results inclicate the stabilitv of serum thvrotrupin cun-centrations mcasured in lrozen serum samples. Iror the neu er assar',the normal range ol'serum thyrotropin values in vounger adults lvasfrom >0.4 to 5.0 mU per liter.

Serum thvroxine concentrations \rcre nleasured in I98l throughl9B3 b,v radioimmunoassav (Diagnostic l)roducts. Los Aneeles): thcnormal range rças -1.7 to 12.0 pg per deciliter (60 to 15-l nmol perliter). Sixtr,-fir'c persons (52 rvomen and l3 men) tere takins estro-gen at (he time ol the I5th examinationl amclng thosc takins estro-gen rvho had normal serum thyrotropin tuncentrations and rlere nottaking a thyroid hormone preparation. the mean serum thvroxineconcentration rvas 7.6 pg per deciliter (98 nmol per liter). indicatinethat the estrogen therapv had little ellcct on the serum thvroxineconcentration. r\ll assavs rvere done singly, sasspl those on samplesrvith values outside the normal rangc. rlhich rçere reassaved in du-plicate.

In our analvscs. the subjects rçere divided irrto lour qroups a(-cording to their serum thvrotropin values: those *ith values S0.lmU per liter (lo* r'alues): those rvith values )0. I to 0..1 mU pe r liter(stightlv lorl values): those n'ith valucs )0.{ to 5.0 mL per liter(normal values): and those with valucs >5.0 mU pcr liter (hiqhvalues). fhese categories rçere based on the normal ranqe :rnd thelikelihood that a serum thvrotropin valuc of 0. I rnU per liter or less

is indicative ol'hvperthvroidism. \\'ith this thrrotropin assa\,. mostpatients rrith clinicallv evident htperthvroidism had serum thvrot-ropin concentrations l;clorv 0.05 mL per liter :rnd nonc had ir con-(enlratiun aborc U.l rrrt per lit, r.l;

Atrial fibrillation s'as diasnosed bv electrocardiosraphv pcr-formed at the biennial cxaminations and during anv inrerveninghospitalizations during the l0 r'ears of fbllorr-up. altcr revierv of therecords bv tno cardiologists; the date of onset rvas considered to bethe date of thc first electrocardiographic documentation ol atrialfibrillation. 'l'he incidence of'atrial fibrillation during the lO-r'earlollou-up period. adjusted bv the direct method lbr the age distribu-tion ol'the lull cohort at the l5th biennial examinari«rn. rlas calcu-lated for each uroup accordins to the Kaplan-\Ieicr techr.rique.lsProportional-hazards analvsisrl'rr,as used to test the association of'the serum th1'rotropin groups rlith the iucidence of atrial fibrillationafter adjustment for age.

A multivariate model uas also computed to control Iàr otherknoun risk factors lor atrial fibrillation rhar \r'erc present at the I5rhexamination (smoking, diabetes mellitus. hvpertension. left ventric-ular hlpertrophr'. mvocardial inl:rrction, conscstive heart failure.and hcart murmur). The rcsults are expressed as rates per 1000person-\,ears of lollou-up or as relative risks. uith 95 perccnt conli-dence inten'als, lor the group in question. rvith thc uroup lithnormal serum thvrotropin conccntrations as the relerence group.2"Calculations rlere perlornred lor all subjects toqether and also aftcrthe exclusion of those takins thvroid hormone prcparations at thel5th examination and those t'ith a histon'of'hvperthvroidism be-

fore l97B through l9B0. .\ll P values :rre b:rsed on a two-tailedanalvsis.

THE, \E,\\' E\GL:\\D JoUR\,\L OF \IEDIC]I\E \or'. 10. 1994

Rrsur-rs

lncidence ol Atrial Fibrillation

During the 1O-vear follorv-up period, atrial fibrilla-tion developed in 192 pcrsons. The overall age-adjust-ed rate was l2 per 1000 person-vcars ('fable 1).

lnitial Serum Thyrotropin Concentrations and lncidenceof Atrial Fibrillation

Among tl.re subjects \r.ith lo$' serum thvrotropinconcentrations (<0.1 mU per liter) in 1978 through1980, the cumulative incidence ol atrial fibrillationafter l0 vears was 28 percent (Fig. i)l the comparablerates in the other groups \\'ere l6 percent in the eroupwith slightlv lon,serum thvrotropin values, I I percentin the group \r'ith normal values, and l5 percent in thegroup with high values. The age-adjusted rate of atrialfibrillation in the group lr'ith lo\r' serum thvrotropinconcentrations was significantlv higher than that inthe group with normal concentrations (P : 0.005)('I'able l). There was a nonsignificant trend in thesame direction for the slightll'-lou'-serum-thvrotropinand hieh-scrum-th) rotropin sroups.

\Vhen the results \,vere adjusted for age and sex andfor the presence of other knou,'n risk lactors ibr atrialfibrillation, including smokins. diabetes mellitus, hv-pertension, left ventricular hypertrophv, myocardialinfarction, congestive heart failure, and cardiac mur-mur, the relative risk of new atrial fibrillation in thosein the lor'-serum-thvrotropin group \{as 3.1 (95 per-cent conlidence interval, 1.7 to 5.5), significantl,v diÊferent (P<0.001) from that in the normal-serum-th)'-rotropin group (Table 2). Aftcr adjustments. thesubjects rvho had slightl,v lou' serum thvrotropin con-centrations also had a somelvhat higher risk thanthose n,ith normal concentrations (reiative risk, 1.6;P:0.05). Excluding the subjects receivins thvroidhormone therapv at thc i5th examination had onlv alimited e{Iect on the relative risk in anv of the groupswith abnormal scrum th1'rotropin concentrations (Ta-ble 2). Furthermore, the exclusion of those u'ho hadhad hyperthvroidism had no cllect rvhen the low-serum-thvrotropin group \ÿas compared rvith the nor-mal-serum-thyrotropin group (relatir.'e risk. 2.4; 95percent confidence interval. 1.3 to 4.8: P : 0.007).

Relation of Serum Thyroxine to Serum Thyrotropinand Subsequent Atrial Fibrillation

Subjects in the lor,r'-serum-thvrotropin group had asignificantl,v higher mean (tSD) serum thvroxineconcentration (8.9-t2.4,ug per deciliter [1 l5+31 nmolper literl) than those in the normal-serum-thvrotropingroup (7.311.7 ttg per deciliter 194122 nmol per li-tcrl; P : 0.001) at the I5th examination. Amons the25 persons in the lorv-serum-th\irotropin group who\vere not taking thvroid hormonc, the serum tht roxineconcentration was rvithin the normal ranqe in 21.There u.as no relation betrveen the serum tht'roxineconcentration altd the subsequent occurrencc of atrialfibrillation in the stud,v group as a u'hole (P:0.71rn'ith adjustment lbr age : P : 0.60 with acljustmenr forage and risk factors). After adjustment for the serumthvroxine concentration. the relative risk of' atrial fi-

Vol. lllll No. l1)

brillation in thc subjects ingroup was 3.0 (95 percent5.5: P<0.0{Jl).

Clinical Hyperthyroidism andin Subiects in Whom Atrial

,\mong the l3 persons in thc lol'-serum-thyrotropingroup \{'ho had atrial llbrillation during the 10-,vearfollorv-up period. onlr'2 had clinical hvperthvroidismduring the same period. In onl,v one of tl.rem, rvho hada recurrencc ol Graves' hy'perthyroidism that had pre-viouslv been successfully treated, did the hyperthy-roidism appear at the same time as atrial fibrillation.This person rvas also the onh' I of the 13 rvho had a

serum thvroxine concentration above I0 trrg pcr decili-ter (129 nmol per liter) at the l5th cxamination; hcrserum thvroxine concentration at this time lvas I4.4pg per decilitcr (185 nmol per liter) . Thc other personu'ho later had hypcrthvroidism had a serum th1'roxineconcentration of 10.0 pg per dccilitcr (129 nmol perliter) at the 15th examination. In fbur other subjects,tu.o in the lou,-serum-thvrotropin eroup and onc eachin the slightly-low- and normal-serum-thvrotropin

Table 1 . Serum Thyrotropin Concentrations in 1978 through 1980and lncidence of Atrial Fibrillation (AF) in the Next l0 Years.

RarF oF AF§-o. slrH (PER lüX)

AF PERSoN-YEARS)* P Val-uEi

)c

l6

lt

l5t2

*The age-adjusled incidence per 1000 [E6o.-veaA of fbllow-up.iFor the comptison wilh the rate in the group sith the nomal serum thyrotropin values.

groups, h1'perthvroidism developed after the l5th ex-amination. but thev did not have atrial fibrillationduring tlie follou-up period.

Overall, onh' 2 ol the 192 subjects rvho had atrialfibrillation also had spontaneous h1'perthvroidism;both ol them had lou' serum thl,rotropin concen-

I 251

Years

Figure 1. Cumulative lncidence ot Atrial Fibrillation among Sub-jects 60 Years of Age or Older, According to Serum Thyrotropin

Values at Base Line.

Low serum thyrotropin values were defined as <0.1 mU per liter;slightly low values, >0. 1 1o 0.4 mU per liter; normal, >0.4 to 5.0

mU per liter; and high, >5.0 mU per liter.

46 were takins thyroid hormone at the time of the15th examination. No further data on serum thyrot-ropin concentrations were available lor l0 of the46. Among the remaining 36 subjects, l9 had serumthyrotropin concentrations of 0.1 mU per liter orIower on at least one subsequent occasion and 6 oth-ers had concentrations of 0.2 mU per liter or lowerat least once; I9 also had at least one subsequent se-rum thyrotropin concentration within the normalrange.

DrscussroN

The sensitive assays for serum thyrotropin2l nowavailable make it possible to distinguish normal fromsubnormal values and to identiÿ desrees of suppres-sion of thvrotropin secretion. Persons with low serumconcentrations of thyrotropin but no clinical mani-festations of hyperthvroidism can be lollowed forthe manifestations (such as atrial fibrillation) thatare usuallv associated with overt hyperthyroidism.In our stud,v we found that a low serum thyrotro-pin concentration (<0.1 mU per liter) in persons60 1,ears of age or older was an independent risk factorlor atrial fibrillation.

Lo\\.SERUII THYROTROPI\ AND ATRIAL FIBRILLATIo\ I\ oLDER PERSONS

thc lou'-serum-tl-rvrotropinconfider-rce intcrval, 1.7 to

Thyroid Hormone TreatmentFibrillation Developed

ù30co'E 25(g

s20TL

E15

b10G)o-CCoEô0C

SERU\I THrRorRoPlN No. ar S[xVALUE Rrsx (!l/F)

Low (<0.1 mu/liter) 61 8/51 l3Slightly low (>0.1 «r 187 84/103 23

0.4 mU/litcr)

Nomal (>0.4 to 1576 680/896 133

5.0 mu/liter)

High (>5.0 mu/liter) 183 42tl4l 23

All subjects 2W1 814/1193 192

0.005

0.1 I

0.08

trations at the 15th examination.l-our of the 57 subjects rvho be-gan taking thvroid hormone dur-ir.rg the follou,-up period also hadatrial fibrillation, 3 of uhom u'ereamong the 28 subjects in the high-serum-thvrotropin group u'ho be-gan taking thvroid therapv duringthis peririd.

Low Serum ThyrotropinConcentrations in Subjects withoutAtrial Fibrillation

Fortv-six subjects in tl're low-serum-thvrotroPin groull did nothavc either atriai fibrillation oro\.ert hvpcrthvroidism during the

lollou'-up period. 'Ihirtr ol' the

Table 2. Serum Thyrotropin Concentrations in '1978 through 1980 and Relative Risk olAtrial Fibrillation, with and without the Exclusion of Subjects Beceiving Thyroid

Hormone Therapy.*

SERLNl THYRoTRoPT\

Low (<0. I mU/liter)

Slightly lou, (>0.1 to0..+ mu/litcr)

Nomal (>0..+ to 5.0mU/liter)

High (>5.0 mu/liter) I8l 4).it1t 1..1

rO q-1 lr

,{r.r SLBTE( r s

REI.A] IVE

\O, ÀT 5E\ RISK

RrsK (\1 F) (957 CI)*

ExcLUDrNc SuBJEcrsRr( Et!rNc TH\ RolD

HoRMoNE

RELATI! È

NO, ÂT S[\ RISK

P VAI,L! RI5K (M F) (95% CI)* P VAI-UE

61 8/5.3 3.1 <0.001 25 5/20 3.8 <0.001(l.7-5.s) (1.7-8.3)

187 8,1/103 1.6 0.05 168 80i88 t.6 0.04il .0-2.5) 0.0_2.5)

1576 680i896 1.0 1530 6741856 t.0

0. 12 169 40/129 L6 0.06(t.0-2.4)

*Adjulted br other risk tialors for atflal fibrillaircn l\eù the Rc\uli\ section)ralues *crc the refercnce rroup. CI denotes c()nfidence interval.

Thc subjects with norm!l serum th)- rotropin

I 2t2

In previous studies the prevalence of atrial fibril-lation at the time of the diagnosis of o\-ert h)'perth\'-roidism ranged lrom 2 to 30 percent2r-5' the Preva-lence is higher among patients more than 60 I'ears olage than amoug vounger patients.J3-l8 Ferv stuclics,hou'ever. har,'e examined the relation betu'een lou'se-rum thvrotropin concentrations and the subsequentdeveiopment of atrial fibrillation. In one stud_v, basedon serum samples obtained lrom a central referencelaboratory', atrial fibrillation developed in 3 of 32 sub-jects r.t'ith subclinical hyperthyroidism during twoyears of follow-up, as compared with none of 35 withnormal serum thyrotropin concentrations.?7 In an-other studl'. there rvas an increased risk of unspccifiedischemic heart disease in hospitalized paticnts rl'hohad been taking thvroxine. but this risk u'as not relat-ecl to tl-re serum th\-rotropin concentration and rvassignificant or-rlr, lor patients vounger than 65 r'ears ofage.29 Our finding that a lou' se rum tht rotropin colr-centration u-as a risk làctor lor atrial fibrillation u'asbased on ciata from a large. unselected. communitt'-based populatior.r ol older pcrsons lollorved for up to10 r'ears.

The mean serum thvroxine concentration rr,as high-er among the subjects r,r,ith low serum thYrotropin con-centrations than among those rvith normal serum thv-rotropin concentrations. although in mosl the valuesn,ere u'ithin the normal range. 'Ihere lvas no relationbetween the serum thvroxine concentration and thelater development of atrial fibrillation

- an observa-

tion that is probablv related to the large variation inserum thvroxine concentrations in the seneral popu-lation and to the variation in cardiac sensitivitv tothvroxine.

Approximatelv l0 to l5 percent of patients u'ithovert hvperthl-roidism r'r'ho have atrial fibrillationhave an arterial embolic event.:r(r"ii Hence. the identifi-catioll ol risk factors lor atrial fibrillation is important.A lou' serum thvrotropir.r concentratiorl

- that is. sub-

clinical h1'perthl'roidism -

appears to l:e one suchfactor. l'hvroid secretion need not increase much, if atall. for atrial fibriliation to occur. Onlv tu'o ol thesubjects in this studv t'ho had lon,serum thvrotropinconcentrations and subsequentlv had arial fibrillationalso had overt hvperthvroidism during tire follou'-upperiod. Exclusion from the anah'sis of those rvith a

historl' of hvpertln,roidism or those taking thvroidhormonc at the l5th examination had little effèct onthe results. Overall, u'hen atrial frbrillation occurs.there is onlv rarelt' either concurrent or subsequento\rert hvperthvroidism; it is more commonll' associat-ed rvith subclinical hvperthl'roidism. which can occureither spontaneouslv or in association with thvroidhormone therapv.

Older persons rvith lorv serum concentrations olthvrotropin should be follou'ed for the development olovert hvpcrthvroidism and atrial fibrillation.'r: \\'heth-er antith\.roid treatment can prevent atrial fibrillationin such persons is not knor,r'n. Among those t'ho arereceiving thvroid hormone and have iolr'- serum tht'-rotropin concentratiorls. the risk of arial fibrillationcan be lessened bv avoiding excessivell'high doses.

l'HE \E\\'E\GL.\\D IOUR\.{t. OF \IEl)I(llNE

il.

ll

r3

\or'. 10. 1991

RErrnrNcBs

Kanncl WB. Abbott RD. Savage DD. N{cNamara PM. Epidemiologic tèa-

tures oi chronic atrial fibrillation: the Framingham Study. N Engl J Med

1982:306:1018-22.Wolf PA. Kannel WB. McGee DL. Mceks SL. Bhaucha NE, McNamara

PM. Duration of atnal librillation and imminence o[ stroke: the Framingham

Study. Stroke 198-1:1.1:66.1-7.

Petersen P. Godtfredscn J. Atrial flbrillation - a review of course and

prognosis. Acta N'[ed Scand 198,1:2t6:5-9.Wolf PA. Abbon RD. Kannel WB. Atrial fibnllation: a major contributor tostroke in the elderly: thc Framingham Study. Arch Intem Mcd 1987;147:l 56 I -'1.

Furszyfer J. Kurland LT. McConahey WM. Elveback LR. Graves' disease

in Olmstead County, Minnesota, 1935 through 1967. Mayo Clin ProcI 970:45:636-44.Phillips DI. Barker DJ. Rees Smith B. Didcote S, Morgan D. The geo-

graphical distribution of thyrotoxicosis in Enqland according to the presence

or absence of TS H-reccptor antibodies. Clin Endocrinol (Oxf ) I 985:23:183-7

Lundgren E. Borup Christenscn S. Dccreusing incitlenc!'ofth)'rotoxicù\i\ inan endenric goitre inJantl area of Srveden. CIin Endocrinol (Oxf) 1990:33:I3.3-tt.

Fagcrberg B. Lindstedt G. Stromblad SO. et al. Thyrotoxic atrial fibrilla-tion: an undcrdiagnoscd or overdiagnosed condition? Clin Chem 1990;36:620-1.Siebers N'[J. Drinka PJ. Vergauu'en C. H1'perth;"roidism as a causc of atriallibrillation in long-term care. Arch [ntem Mcd 1992:152:2063-4.E-egensen R. Petersen K. Lundberg P-A. Nystrom E. Lindstedt G. Screen-ing for thyroid disease in a primar-"- care unit with a thyroid stimulatinghormone assal' uith a low detection limit. BIt'lJ l98tt:297:1586-92.Bagchi N, Broun TR. Parish RF. Thlroid d1'sfunction in adults over age

55 vears: a sturll in an urban US communitl. Arch Intern Med I990:150:785-7 .

Parle JV. Frankll n JA. Cross KW. Jones SC. Sheppard MC. Prevalcnce and

follou'up of abnormal th)rotrophin (TSH) concentrations in the elderly- inthe United Kingdom. Clin Enclocrinol tOxf ) l99l:-l-1:77-83.Sauin CT. Gcller A. Kaplan MI\l. Bacharach P. Wilson PW. HershmanJN'I. Lou, serum thlrtx«rpin (th)roid-stimulating hormone) in older personsuirhout hlpcrthlroidism. Arch lntcm \'Ied l99l:151:16-5-8.Sundbeck G. Jagcnburg R. Johansson P-\'t. Etlcn S. Lindstedt G. Clinicalsi-cnilicance of lou serum thyrotropin conccntration bl chemiluninomet-ric assav in 85,1ear-old uomen and mcn. Arch Intem Mcd l99l:l5l:5.19-56.Friedman D. Reed RL. l\looradian AD. The prevalence of ovemedicationwith levoth,r-roxine in ambulatory elderly patients. Age 1992;15:9-13.Pekrl AE. Hershman JM. Parlo* AF. A sensitive and precise radioimmu-noassal tbr human thy'roid-stimulating bomrone. J Clin Endocrinol MetabI 975:4 t:676-84.Ross DS. Daniels GH. Gouveia D. The use and limitations of a chemilu-minescent thyrotropin assay as a single thyroid tunction test in an out-patientendocrinc' clinic. J Clin Endocrinol Mctab 1990:71:764-9.Kaplan EL. Meier P. r-onparametric estirnation fiom incomplete obsen'a-tions. J Am Star Assoc l95tJ:53:.157-81.Cor DR. Regression models and life-tables. J R Stat Soc [B] t972;3.1:187-220.§{iscellaneous topics. ln: Kalbfleisch JD. Prentice RL. The statisticâl analv-sis of failure time data. New York: John Wilei'. 1980:199-201.Weeks I. Stur,sess M. Siddle K. Jones NlK. Woodhcad JS. A high scnsiti\-it\ imnrunochemiluùrinometric assav for human thvrotrophin. Clin Endo-crinol (Oxf ) 198-1:10:.1ti9-9-5.White PD. Aub JC. The clectrocardiogram in thyroid diseasc. Arch lntemMcrt l9l8:22:766-9.Sandler G. Wilson GM. The nature and prognosis of hean disease in thyro-toxicosis: a revieu of l-50 patienrs treated sith rrrI.

Q J Nled 1959:52:3,17-69.Peterscn P. Hansen lNl. Strokc in thyrotoxicosis uith atrial Ilbrillation.Stroke 1988:19:15-8.Nordyke RA. Gilben FI Jr. Harada AS. Graves' disease: inlluence ofagc onclinical Iindings. Arch Intem Med 1988:148:626-31.Daly JG. Greenuood R\{. Himsuonh RL. Th1'rotoxic atrial fibrillation.BMJ 1982:285:157.1.

Tenerz A. Forberg R. Jansson R. Is r more active attitude wamanted inpatients nith subclinical thyrotoxicosis'l J Infem Med 1990:228:229-33.Tibaldi JM. Barzel US. Albin J. Surks M. Th1'rotoxicosis in the very old.Am I Med 1986:81:619-21.Leese GP. Jung RT. Guthrie C. Waugh N. Browning MC. N{orbiditl inpaticnts on r.-th1 mrine: a comparison of rhose with a nonnal TSH to thoseuith a suppressed TSH. CIin Entlocrinol (Oxl) I992:17:500-3.Sraffurth JS. Gibberd MC. Ng Tang Fui S. Arterial embolism in thyroroxi-cosis uith atrial librillation. BMJ 1977;2:688-910.Presti CF. Hart RG. Thyrotoxicosis. atrial fibrillation, and embolism. revis-itcd. Arn Hean J lt)89:l l7:q7o-7Singer DE. Ranrlornized trials of warfàrin tbr atrial fibrillation. N Engl J

Med l99l:317;l-15t-3.

9

IO

l5

t6

l.l

t7

t8

lg

:l

25.

26.

27.

28.

29.

30.

ll.

32.

Vol, 33l \o. l!) tsO\E }I,\RRO\T TR.\NSPLAN'IS VERSUS CHE\IO'THERAPY FOR C]HILDRE,\ \\'ITH .{i,I, I 253

BONE MARROW TRANSPLANTS FROM HLA-IDENTICAL SIBLINGS AS COMPARED WITHCHEMOTHERAPY FOR CHILDREN WITH ACUTE LYMPHOBLASTIC LEUKEMIA IN A

SECOND REMISSION

,\. Joux Benrr:,rr, \I.D.. N'Ienv \L HonowIrz, II.D., Bna» H. PoLrocx, Pn.D., Nlr,r-Jrr Zuexc, PH.D.,Nlonuurn NI. Bonrrx, N[.D.,* GroncE R. BucHaNax, N'I.D., Br<uca NI. Cenrrrre, N{.D.,

Juorrn Ocns. N{.D.. JoHx Gn,c,HeM-PoLE, À,{.D., Purlrp A. Ror'r,r-rxcs, }I.8., B.S..

Arrxrp A. Rrlrrr. Pu.D.. Jon:'r P. KreIN, PH.D., Joxsru.lx J. SursrEn, Pu.D.,K.q,rni-rex A. Sosocrxsrr, M.S., arv» RoeEnr Pnrr,n Gar-r,, \I.D., Pu.D.

tation in the transplant recipient. A total of 255 matchedpairs were studied.

Results. The mean (*SE) probability of a relapse atfive years was signilicantly lower among the transplantrecipients than among the chemotherapy recipients (45+4percent vs. 8013 percent, P<0.001). At five years theprobability of leukemia-free survival was higher aftertransplantation than after chemotherapy (4013 percentvs. 17+3 percent, P<0.001). The relative benefit of trans-plantation as compared with chemotherapy was similar inchildren with prognostic factors indicating a high or low riskof relapse (the duration of the first remission, age, leuko-cyte count at the time of the diagnosis, and phenotype ofthe leukemic cells).

Conclusions. For children with acute lymphoblasticleukemia in a second remission, bone marrow transplantsfrom HLA-identical siblings result in fewer relapses andlonger leukemia-free survival than does chemotherapy.(N Engl J Med 1994;331:1253-8.)

to 20 percent.+-rr The main determinant of the out-come of chemotherapy is the duration of the first re-mission: a child with a brief first remission fares worsethan one rvith a long first remission. Other variablesreported to predict the outcome of chemotherapy areage, the leukocyte count at the time of the diagnosis,and the phenotvpe of the leukemic cells. Resistance tochemotherapy is the main cause ol treatment failure.

Bone marrow transplantation from an HLA-identi-cal sibling is an alternative treatment for childrenin a second remission.r2-r6 Transplantation has result-ed in ieukemia-free survival at fi\'e years in 22 to 64percent of patients in large series.rT-2+ The durationof the first remission is also a primarv determinantof the outcome of bone marrow transplantation.re-2rUnlike the results 'n'ith chemotherap_v, treatment-related mortalin' and resistant leukemic cells contrib-ute equalll' to the lailure ol treatment rvith bone mar-row allogralts.

\\rhether a child r,vith acute lvmphoblastic leukemiain a second remission n.ho has an HlA-identical sib-ling should receive chemotherapy or a bone marrolvtransplant is a matter ol intense debate.l5 16'21 2r To ourknolvledge, no randomized trials have addressed thisquestion. because the relativel,v lolv incidence of acutelymphoblastic leukemia and the limited number o{'HLA-matched donors make accrual of patients di{fi-cult. even for multicenter cooperative eroups.26 Otherreports comparing chemotl.rerapv and transplantationare either inconclusive or contradiclo.u.l:r.22.25-2u {r, uprevious comparison of data from the InternationalBone \'Iarrorv 'Iransplant Registrv (IBMTR) anddata from trials rvith chemotherapv, we suggested thatbone marrorv transplantation in patients in a second

Abstract Background. lt is unclear how best to treatchildren with acute lymphoblastic leukemia who are in asecond remission after a bone marrow relapse. For thosewith HLA-identical siblings, the question of whether to per-form a bone marrow transplantation or to continue chemo-therapy has not been answered.

Methods. We compared the results of treatment withmarrow transplants from HlA-identical siblings in 376children, as reported to the lnternational Bone MarrowTransplant Registry, with the results of chemotherapyin 540 children treated by the Pediatric Oncology Group.A preliminary analysis identified variables associatedwith treatment failure in both groups. We selected co-horls by matching these variables. A possible bias associ-ated with differences in the interval between remissionand treatment was controlled for by choosing matchedpairs in which the duration of the second remission inthe chemotherapy recipient was at least as long asthe time between the second remission and transplan-

1--IURREN-I resimens of intensive chemotherapy\-,,{ produce rcmissions in almost all children withacute lvmphoblastic leukemia. Additional treatmentu'ith consolidation and maintenance chemotherapvcures up to 70 percent of these children, but in about25 percent of them the disease recurs in the bonemarrorv.l-:l \Iost of these children have a second re-mission u,ith chemotherapl,', but the choice of subse-quent treatment is conroversial. One approach is

more chemotherapv. In large pubiished studies, che-motherapv resulted in leukemia-free survival at fivevears in B to 76 percent ofpatients but generaily in 10

From rhc National Hean. Lung. and Blood lnstitute. Bethesda, Md. (A.J.B.):the lntemational Bone Manow Transplant Registr)'. Health Policy lnstitute(NI.M.H., M.-J.2.. M.M.B., P.A.R.. K.A.S.). and the Departments of Pediat-rics (B.M.C.)and Biostatistics (J.P.K.). Medicai College of Wisconsin, Milwau-kee: the Pediatric Oncology Group Statistical Office (8.H.P., J.J.S.) and the

College ot lledrcine. Universitl'of Florida (J.G.-P.). Gaincsville; the Universityof Texas Southwestem Medical Center. Dallas (G.R.B.): St. Jude's Children'sHospital. Memphis. Tenn. (J.O.); Case westem Reserve University School of\{edicine. Cleveland (A.A.R.): anrl Salick Health Care. Los Angeles (R.P.G.).Address reprint rcquests to Dr. Horowitz at the Intemational Bonc Manow Trans-plant Registr). Medical College of Wisconsin. 8701 W. Waterto§n Plank Rd.,Ntilwaukee. Wl 53226.

Supported by grants (POl-CA-40053, Ul0-CA-29139. CA-33625, CA-29281,CA-32053. CA-31566. and CA-30969) from thc Nâtional Cancer lnstitute. theNarronal Hean. Lung. and Blood Institutc. und the ){ational lnstitutc of Allergyand Infectious Diseases and bl grants from the Alpha Therapeutic Corpora-tion. Armour Pharmaceutical Company. Lynde and Harry Bradley Foundation.

Bristol-N'1]ers. Buroughs Wellcome Companl. Charles E. Culpeper Founda-

tion. Elcanor Nallor Dana Charitable Trust. Eppley' Foundation for Research.Hoechst Rousscl Phamaceuticals. Immunex Corporation. Kettering FamilyFoundation, Robert J. and Helen C. Kleberg Foundation. Eli Lilly and CompanyFounclation, Nada and Herben P. Mahler Charities, Marion Menell Dow, Am-brore Moncll Foundation. Samucl Robcrts Noble Foundation, Ortho Biotech

Corporation, John Oster Famill' Foundation. Jane and Lloyd Pettit Foundation.

RGK Foundation. Roerig Dir ision of Pfizer Pharmaceuticals. Sandoz Research

Instirute. Stackncr Famill Foundation. Star Foundation. Joan and Jack Stein

Charities. Suiss Cancer Lcaguc. and W)clh-Ayerst Rescarch

"Deccased.

l 254

remission after a bone marrow relapse resulted in a

higher probabilitv of leukemia-free survival in chil-dren with a short first remission (<18 months) but notin those rvith a long first remission.2e Hor,r'ever, prog-nostic and treatment information that mieht have ac-counted for the dillerence !ÿas not available for thepatients receiving chemotherapv.

In this studv rve performed a matched-pair analvsisto compare the results of treatment with bone marrowtransplants from HLA-identical siblings in 255 chil-dren, as reported to the IBNITR, with the results ofchemotherapy in 255 similar children enrolled in trialsconducted by the Pediatric Oncologv Group.

MrrHo»s

Patients Treated with Transplantation

The ItsIITR collects data (iom over 250 transplantation centersworldu idc rhar report informatiou on consecutive patients receir inqallogeneic or identical-tu'in bone marrou' ransplants. Participantsaccount for about trvo thirds of all actite transplantation teams.l"The registn'. rrhich includes information on ,10 to 50 percent ol allallogeneic bone marrorv transplantations since 1970. is the lareestdata base lor transplantations in patients rlith acute l1'mphoblasticleukemia. !-or this studv. the cohort of transplant recipients rtasdrawn from a population of patients 18 l ears of aee or vounger u,horeceived transplants from HI-A-identical siblinss between 19U3 and1991, while ther.uere in a second remission after a bone marro\{'relapse. Children uith an isolated extramedullan, relapsc. Ivm-phoma rrith a leukemic transformation, or Dortn's svndrome uereexcluded. The selection o[ candidates for transplantation variedaccording to the policies of the transplantation teams. \Iost trans-plantations u'ere performed in children rçithout serious concurrentillnesses. The interval betlveen the second remission and transplan-tation varied from I to 222 neeks (median. I0). For this paticntpopulation, the results of transplantation reported to the IB\ITRare similar to those reported bv most centers.l:lI

Patients Treated with Chemotherapy

The Pediatric Oncologl' Group is a multicenrer clinical-trialseroup including over 80 institutions in the Unitcd States. Canada,and Europe. 'fhe cohort of patients lvith acute lvmphoblastic leuke-mia who were treated with chemotherapv tras dral n from a popula-tion of patients rvho ltere l8 vears of age or vounqer and had had asecond remission after treatment ficr a first bonc marro\r relapse.Thev rçere treated betu'een April l9B3 and IIav l99l in studr'8303.8304,8710, or 8862. f)etails of the treatments are reported else-u'herell3r and are also available from the group's statistical center.Study 8303 excluded patients who had a relapse more than sixmonths after maintenance chemotherapy had been discontinued(late rclapse). Studr' 830,] included onlv patients \ÿirh a late relapse.Studv B7l0 included patients u'ith either an earlv or a lare relapsebut escluded those rvith T-cell acute lvmphoblastic leukemia orDorvn's svndrome. Stud_v 8862 included parienrs u,ith 'f-cell acutelvmphoblastic leukemia or T-cell non-Hodgkin's lvmphoma but ex-cluded those x-ith Dorvn's svndrome. The patients trith T-cell non-Hodgkin's l,vmphoma uere excluded from this anall.sis. The resultsof these trials are tt'pical of the published results of'studies in theLnited States and other countries in n,hich children rçith acutelymphoblastic leukemia are treated after a first bone marrolr. re-lapse. l'3

Statistical Analysis

The initial studv group consisted of 376 children in the IB\tTRcohort and 5.10 in thc Pediatric Oncology Group cohort. Two poten-tial sources of bias ltere considered in comparing these groups.First. patient- and disease-related variables associated rvith the out-come of chemotherapv nright have influcnced the selection ofpatients for bone marroll transplantation. This could have resultedin an imbalance in prognostic factors betueen the groups. Second,

\or'. 10. I991

Table 'l . Variables Signiticantly Associated with Treatment Fail-ure (Relapse or Death) among Children with Acute LymphoblasticLeukemia in a Second Remission Receiving either Continued

Chemotherapy or Bone Marrow Transplants fromHLA-ldentical Siblings. *

VARIABLE

Age >10 yr

T-cell phenotype

læukocyte count at diagnosis<50.000rmm150.001- 100.000/mmr> t00.000immr

Duration of nrst remission (mo)<1818.01-36>36

CHEIIoTHERAPY TR{\SPI-ANTAnoN(N = 5.10) (N = 376)

RELATIIE P REr-ÂIIIE P

RISX VTL|E RISX \'ÂI UË

r-S l.5l 0.003

NS 2.16 <0.001

t.00+0,9: NS1.58 0.002

1.88 <0.001 2.01 <0.0012.85 <0.00t 1.42 0.05I.00+ L00+

*Sex and the yeil of the diagnosis were not signi,icanrl! associated with th. risk of trea(menrtàilure. NS denotes not significant (P>0. I ).

iRefercûce group.

the IB\ITR cohort included onlv children rrho *ere in remissionlong enough to rcceive a transplant, uhereas the Pediatric Oncol-opçr'Group cohort included all children er.rrolled in the group's stucl-ies rrho had a second remission. Thc transplantation grclup there-fore ercluded children u'ho died earll or had an earh secoudrelapse. 1'his could have introduced a bias favorinq transplan-tation.3z lll

\\'e used the l'ollorving statistical methods to address these issues.First, using a Cox proportional-hazards regression.r+ rçe identifiedpatient- and disease-related variables associated uith treatmentfailure (relapse or death) in each group. 'fhe variables \\'e tested inthe model were age. sex. thc leukocvte count at the tirnc of thediagnosis. the phenotl'pe. the vear of the diagnosis, and the dura-tion of the first remission. \\'e then sclected pairs of chemotherapr'and transplant recipients by' matching members liom tlte trlo co-horts for all variables associated u'ith the outcome o[ either therapr'(P<0.1).'fhese variables *ere the age at the time of thc secondremission (0 to 2. 3 to 10. or I I to l8 r'ears), the leukocvte count atthe time of the diagnosis (<50.000, 50.001 to 100.000. or >100.000cells per cubic millimeter). the T-cell phenotlpe (r'es or no). and theduration of the lirst remission (rvithin six months). II'more than onepatient in the chemotherapv group rras eligible for matchins rvith apatient in the transplantation eroup. rve selected the chemotherapvrecipient rçith a first remission closest in duration to thar of thetransplant recipient. In addition. the chemotherapv recipient ineach pair was selected liom among the children uith a second re-mission at least as lons as the interval benveen the second remissionand transplantation lbr the transplant recipient. \\e identified 255pairs uho met these criteria.

-I'he primarv end points rve analvzed uere the duration of sur-vival uithout a relapse of leukemia (leukemia-free surr-ival). thetime to treatment-relaled mortalitr'(death in continuous completeremission), and the time to a relapse. The probabilities of leukemia-free survival. treatment-related mortalitv. and relapse rverc cal-culated *'ith Kaplan-\leier methods and compared bv pairedlog-rank test.si For thc analvses of leukernia-free survival. treat-ment rras considered to have lailed at the time of' a relapse atany site or at the time of de:rth lrom an1'cause: data on patientsn,ho rvere alive and in continuous complete remission rvere censoredonh' at the time of the Iast follorr-up visit. For analvses oftreatment-related mortalitv. failure sas defined as death duringa continuous complete remission: data ltere censored at the timeof a relapse or. among patients uith continuous remissions, arthe time of the Iast lollorv-up visit. For anall'ses ol relapse. fail-ure was defined as the recurrence ol acute lvmphoblasric leuke-mia at anv site: data ryere censored at the time of a death or thelast follrrv-up visit during a continuous remission. \\'e used aCox proportional-hazards regression model to test {br inrerac-tions between treatment and age. sexr the leukocvte count at

THE \E\\' E,NGLA\D.JOI,,iR\AL OF \tEDICI\T]

Vol. 331 No. 19

the time of the diagnosis, the leukcmia-cell phenonpe. and theduration of the first remission.

RnsurrsCharacteristics of the Patients

Table I shorvs the variables associated with treat-ment failure in the unmatched groups of children r.r,ithacute lymphoblastic leukemia in a second remissionu,ho received either chemotherapy or bone marrowtransplants. Among the 540 patients who receivedchemotherapv, an increasecl risk of treatment failurewas associated with a leukocyte count )100,000 percubic millimeter at the time of the diagnosis and by afirst remission that n'as (36 months long. Among the

Table 2. Characteristics of the Matched Chemotherapy andTransplantation Cohorts.x

BONE }IARRO\\.TRANSPI,:\\.IS VERSUS CHE\IOI.HERAPY FOR CHILDREN \VITH ,\LL I 255

o

= 1.0

aar 0.8o)

E 0.6ota=_3 o.4

o

È0,-o(g

3 o.o0-

Transplantation 376

Chemotherapy 540

12 24 36 48 60

Months after Treatnr ent

192 113 87 65 48

187 105 72 50 35

38

24

CHÂRACTERIsnc

Male sex -

no. (%,)

Age at diagnosis - yr

MedianRange

Age group at diagnosis -

no. (7c)

0-2 yr3- l0 yrll-18 yr

Yeu of diagnosistMedianRange

T-cell phenotype - no. (7r)

YesNo

Leukocyte count at diagnosis -

per mmrMedianRange

Leukocyte count at diagnosis -

no. (lcl<50,0ü)/mml50.001- 100,000/mnrr>100,ü)0rmml

Duration of first remission - moMedianRange

Duration of first remission - no. (%)<18 mo18.01-36 mo>36 mo

Weeks from second remission to trans-plantation

MedianRan-{e

*Because of rounding. not all percenlages total 100.

iP : 0.9 by paired t-tesl.

376 children u'ho received marro\\, transplants, an in-creased risk of treatment failure was associated withan age ) l0 vears, the presence of a T-cell phenotype,and a first remission lasting (36 months. 1'able 2

shows the characteristics of the matched cohorts.which were very similar.

Outcome

Fisure I shows the actuarial probabilities of leuke-mia-free survi\.al in the unmatched transplantationand chemotherapy cohorts. The mean (aSE) prob-ability of leukemia-free survival at frve vears was3613 percent in the transplantation group and 16+2percent in the chemotherapy group (P<0.001).

Figure 1. Actuarial Probability of Leukemia-free Survival in Un-matched Cohorts of Children Receiving Chemotherapy or Under-

going Transplantation.The numbers below the tigure indicate the nurnbers of

children at risk.

For the 255 matched pairs from the tv,'o cohorts,the probability of leukemia-free survivai at five vearswas significantly higher after transplantration thanafter chemotherapl' (40-+- 3 percent vs. l7t3 percent,P<0.001). \,loreover, the risk of a relapse n'as signit-cantly lower after transplantation than aft,er chemo-therapy (+5'r+ percent vs. B0i3 percent, P<0.001)(Fig. 2 and 3 and Table 3). The probabilit.', of treat-ment-related death within five vears was l4t4 per-cent with chemotherapy and 27 !3 percent 'with bonemarro\r' transplantation (P<0.001 ).

Table 3 shorvs the probabilities of leul:emia-freesurvival and relapse in subgroupsr of children ac-cording to the prognostic factors listed in Table 1.

12 24 36 ,48 60 72

Months after Treatment

CHEMOTHERAPY(N = 255)

163 (64)

6.00.4-18.1

l6 (6)163 (64)

76 (30)

l 983r975- 1990

2l (8)

234 (92)

10.0000-600.000

214 (84)24 (9)

t7 (7)

25<l-105

9l (36)88 (35)

76 (30)

TRANSPLANTATION

(N = 25s)

r63 (6+)

7.00.5- 18.4

r6 (6)163 (6.+)

76 (30)

1983

1971 -t989

2l (8))?A rO)r

9000r000-456,000

214 (84)24 (9)t7 (71

26< l- 104

89 (35)90 (3s)76 (30r,

t0< t- I40

=U)o 0.8o(U

E 0.60)Y

_9 0.4

o

È o.zb(U

€ o.oo-

Transplantation 255 133 81 60 ,47

Chemotherapy 255 105 58 39 29

37 28

19 15

Figure 2. Actuarial Probability of Leukemia-free Survival inMatched Cohorts of Children Receiving Chenlotherapy or Under-

going Transplantation.The numbers below the figure indicate lhe numbers ot

children at risk.

I 256

A Cox prop()rtional-hazards regression model fittedu'ith first-ord er interaction variable s did not show sig-nificant interactions between the treatment and anv ofthe proenost: c variables studied - that is, the relatir.ebenefit of tnansplantation as compared rvith chemo-therapl' rvas similar in all groups. The probability ofIeukemia-free survival at fi\,e years lvas higher aftertransplantat,ron than after chemotherapy both in chil-dren whose l'irst remission had lasted 36 months or less(35X4 percrrnt vs. l0-r3 percent for 179 pairs) (Fig.4A) and in those u'ith a first remission that exceeded36 months (:t3!7 percent vs. 32 16 percent) (f ig. aB) .

Comparisons of pairs of children with a first remissionlonger than 48 months (36 pairs) or lonser than 60months (lB pairs) vielded results that rvere similar tothose for chiLldren with a first remission that excecded36 months.

DrscussroN

Our stuc.y provides evidence that treatment withbone marrolv transplants lrom HLA-identical siblingsresults in a statistically greater likelihood of leukemia-lree sun,ivzrl at five years than does chemotherapy inchildren u'rth a bone marrorv relapse after a secondremission of acute lymphoblastic leukemia. The rea-son lor this diflerence was the lower risk of relapseafter transplantation, which outweighed the higherrisk of trc:atment-related mortality associated lviththis treatm ent. The outcome after transplantation wassuperior to the outcome aftcr chemot]rerapy in sub-groups of <:hildren with lavorable or unfavorable prog-nostic lactrrrs (duration ol first remission, < 36 monthsor )36 monthsl leukocvte count at the time of thediagnosis, < 100,000 per cubic millimeter or > I 00,000per cubic rnillimet er; age, { I0 years or } I 0 vears; andphenotvpr:, T-cell or non-T-cell acute lymphoblasticleukemia). Since the numbers of matched pairs that

0122436486072Months after Treatment

\or'. 10, 1994

Table 3. Probabilities of Relapse and Leukemia-free Survival FiveYears after Chemotherapy or Bone Marrow lransplantation.

No. oF LEtrxEMt^-rREEPArRs RELAPSE SuRvrval

CHEMO- TRANSPLAN. CHEMO- TRANSPLAN.

]HERÀPY TATION THERAPY TÀTION

nkan ( =SE) Perceùt

255 8013 45:t4 l7+3 40+3

.fHE NEW ENGLANDJOURNAL OF \IEDICINE

All patientsAge (yr)

<10>10

SexMaleFemalc

Leukocyte count at

diagnosis< 100.0ül/rnmr 238> 100.0001mmr 11

T-cell phenotypeYesNo

Duration of first remission (mo)

<18r8.01--36>36

4315 I8a351!'7 t2!4

441430r6

49 l: 5 19 t4 31 .!:4

3716 l3+4 46*6

45+4 l7 !3 39t:437:! 13 916 5l rr 17

50r:12 lt+7 32*1044!4 17 .L3 41!4

63*6 l1:t4 29+539+7 7+3 4l+630a7 32*6 53+7

17976

r6392

781:486a5

79+183 t5

80+-19l j6

861880* 3

2t

91 83i488 91 1476 66!6

Figure 3. Actuarial Probability of a Relapse in Matched Cohorts olChildren Rec.eiving Chemotherapy or Undergoing

Transplantation.

The numbers b,elow the tigure indicaie the numbers ofchildren at risk.

could be evaluated in some of these subgroups weresmall (21 pairs with'I'-cell acute lvmphoblastic leuke-mia and 17 with leukocyte counts over 100,000 percubic millimeter), the relative benefits of chemothera-py and transplantation remain uncertain in patientswith these characteristics.

This study indicates that leukemia-free survival islonger after transplantation than after chemothera-pi, but does not completel,v answer the question ofthe best treatment strategy for children with acutelvmphoblastic leukemia in a second remission. Ouranalysis does not consider a third option: to reservetransplantation lor children who have a second re-lapse after receiving chemotherapy for their first re-lapse. Horvever, the poor outcome of chemotherapy inchildren rvhose first remission lasts lor 36 months orless (lcukemia-free survival at five years, ( l0 percent)is an argument for early transplantation if an HLA-identical sibling is available. For chilclren whosefirst remissir>n is longer than 36 months and in whomchemotherapy results ir.r a better outcome (leukemia-free survivai, about 30 percent). it mav be reasonableto deler transplantation until a subsequent relapseoccurs.

Because there are no randomized trials compar-ing chemotherapy and transplantation, we used amatched-pair design to control both for known prog-nostic lactors in childhood acute lymphoblastic leuke-mia and for a time-to-treatment bias. Although thetwo cohorts were matchcd lor these factors, there mayhave been unknown factors that differed between thecohorts. Consequentlv, our flndings should be inter-preted cautiously.

Our conclusions applv only to the chemotherapyand transplantation resimens used for the patientsrve studied and only to the transplantation ol graftslrom HlA-identical siblings. Other chemotherapy

'1.0

3 o.ao_(!6tr 0.6

E o.4(§

o ^^-! u,zLL

0.0

ïransplanlation 255

Chemotherapy 255

'133 81

105 58

60 47

39 29

37 28

19 15

Vol. lllll No. 19

Transplantation 179

Chemotherapy 179

Transplantation 76

Chemotherapy 76

24 36 48 60

Months after Treatment

46 37 32 24

22 16 14 10

BO\E }I.-\RRO\\'TRÀ\SPLANTS VERSUS CHE}IOTHE,RAPY I'OR (]HILDRE\ \\'ITH ALL

12 24 36 48 60 72

Months after ïreatment

t%1

Champlin R. Gale RP. Acute lymphoblastic leukemia: recent advanccs in

biology and therapy. Blood 1989:73:2051-66.Rivera G. Murphy SB. Aur RJ. Vcrzosa MS. Dahl GV, Mauer AM. Recur-rent childhood lymphocytic leukcmia: clinical and cytokinetic studies ofcytosine arabinoside and methotrexate for maintenancc of second hemato-Iogic remission. Cancer I978:42:2521-8.Ekert H. Ellis MW. Waters KD. Matthews RN. Poor outlook fbr child-hood acute lymphoblastic leukacmia with relapse. Med J Aust 1979.2:224-6.

Chessells JM, Combleet M. Combination chemotherapy tbr bone marrowrelapse in childhood lynlphoblastic leukaemia (ALL). Med Pcdiatr OncolI 979:6:359-65.Kimbali JC, Herson J. Sullivan MP. Favorablc response to maintenancetherapy of second or subsequent remissions in childhood acute lvmphocyticIeukcmia. Cancer 1980:.15: 1093-7.

Amadori S. Spiriti MA. Mebni G. Pacilli L. Papa G. Mandelli F. Combina-tion chemothcrupy tbr marrou' relapse in childrcn and adolesccnts with acutelymphocytic leukaernia. Scand J Haenratol l98l;26:292-6.Baum E. Nachman J. Ramsay N. et al. Prolonged second remissions inchildhood acute lymphocltic leukemia: a report from the Childrens CanccrStudy Group. Med Pediatr Oncol 1983:l l:l-7.Hcnze C. Fenglcr R. Hartmann R. ct al. Six-year cxperiencc with a comprc-hensive approach to thc treatment of recunent childhood acute lvmphobias-tic leukcmia (ALL-REZ BFM 85): a relapse study of the BFM group. Blood199]t:78:1t66-72.Sadowitz PD. Smith SD. Shuster J. Wharam MD. Buchanan GR. RiveraGK. Treatment of late bone manow relapse in children with acute lympho-blastic leukemia: a Pediatric Oncology Group study. Blood 1993:ttl:602-9.Ramsay NK. Kersey JH. lndications fbr bone manow transplantation inacute lynrphoblastic leukemia. Blood 1990:75:815-8.Bàrett AJ. Bone mmow transplantation for acute lymphoblastic leukae-mia. In: Treleavan J, Barett AJ. eds. Bone marow transplantation in prac-tice. Edinburgh, Scotland: Churchill Livingstone. 1992:33-41.Sanders JE. Thomas ED. Buckncr CD. Done]- K. Manou, transplantationtbr children with acute lymphoblastic leukcmia in second remission. Blood1987:10:324-6.Chessells JM. Rogers DW. Lciper AD, et al. Bone-mmow transplantationhas a limited role in prolonging second mmow remission in childhoodlymphoblastic leukaemia. Lancct 1986:l: 1239-41.Pinkel D. Allogeneic bone mmow transplantation in children with acute

leukemia: a practice whose time has gone. Leukemia 1989:3:212-4.Coccia PF, Strandjord SE. Warkentin PI, et al. High-dose cytosine arabino-side and fractionated total-body inadiation: an improved preparative regi-men for bone marow transplantation of children with acute lymphoblasticleukemia in second remission. Blood 1988:71:888-93.Banen AJ. Horowitz MM. Gale RP. et al. Marow transplantation fbr acutelymphoblastic leukemia: factors aftècting relapse and survival. Blood1989:74:862-7 I .

Bmett AJ, Joshi R, Kendra JR. et al. Prediction and prevention of relapse ofacute lymphoblastic leukaemia after bone manow transplantation. Br J Hac-matol 1986:64:179-86.Herzig RH, Bortin MM. Bffiett AJ. et al. Bone-marow transplantation inhigh-risk acute lymphoblastic leukaemia in first and second remission. Lan-cet 1987:l:786-9.Brochstein JA. Keman NA, Groshcn S, et al. Allogeneic bone mmowtransplantation after hyperfiactionated total-body irradiation and cyclophos-phamide in children with acute leukemia. N Engl J Med 1987:317:t618-24.Dopfer R, Henze G, Bender-Gotze C, et al. Allogeneic bone marow trans-plantation for childhood acute lymphoblastic leukemia in second remissionafter intensive primary and relapse therapy according to the BFM- and

CoAll--protocols: results of the German Cooperative Study. Blood l99l;78:2780-4.Wingard JR, Piantadosi S. Santos GW. et al. Allogeneic bone mmowtransplantation for patients with high-risk acute Iymphoblastic leukemia.J Clin Oncol 1990;8:820-30.Poynton CH. Bamett AJ. Bone manou transplantation in childhoodlymphoblastic leukaemia. Lancet 1986;2:395-6.Johnson FL. Thomas ED. Clark BS. Chard RL. Hartmann JR. Storb R. Acompæison of manow transplantation with chcmotherapy for children withacute lymphoblastic leukemia in second or subsequent remission. N Engl J

Med 198 I :305:846-5 I .

Darbyshire PJ. Pinkerton CR. Stevcns RF. Oakhill A. Treatment of acute

Iymphoblastic leukaemia after relapse. Lancet 1990;335:733.Buchanan GR. Boyett J. Rivera GK. Comparison of bone manow transplan-tation and intensive chemotherapy for children uith acute lymphoblasticleukemia in second manow remission: a Pediatric Oncology Group (pOG)study. Bk)od l9tl6:68:Suppl I :2 I 9a. abstracr.Hanis R. Feig S. Coccia P. et al. ALL in second renrission

- a CCSG srudyol maintenance chemotherapy vs ntamow transplantation, [n: Gale RP.Champlin R. eds. Pro_srcss in bonc marow transplantation. Ne§ York: AlanR. Liss. I987:91-5.

o

=ao 0.80)

(!E 0.6c)-Y=_9 0.4

È0,-o(d

3 o.oÈ 72

18

7

t085

53

lt

l6

t7

(s

= 1.0

ao 0.8o)

(U

E 0.6(l)L=_9 0.4

oÈ o.zô(ü

Ëoo0-

18.

19

z0

21.

22.

23

T2

l3

15.

15 13 10

15 9 8

Figure 4. Actuarial Probability of Leukemia-free Survival inMatched Cohorts of Children Receiving Chemotherapy or Under-

going Transplantation, According to the Duralion of the FirstRemission.

The numbers below the figure indicate the numbers olchildren at risk.

and transplantation regimens and autografts or trans-plants lrom donors other than HlA-identical sib-lings u'ere not considered and may have dillerent out-comes. However, the methods used for this study canreadily be applied to comparisons of these otherapproaches.

This article is dedicatecl to tht' memon of Dr. \'Iortimer \{.Bortin. rçho diedJulr'25. I99+. Dr. Bortin uas a pioneer o1'bonemarro\\' transplzrntatiolr, participatillg in olre of' the first successfirltransplantations in humans. He helped fbund the InternationalBone \Iarron Transplant Registrr in 1970 and sen ed as its scientil:it director lur o\cr lft rerrs.

RErrnnNcrs

l. Rircra GK. Raimondi SC, Hancock ML. et al. lmproved outcome in child-hood acute limphoblastic leukacmia with reinforced early treatment and

rotâtional cunbination chemothcrapy. Lancet 1991:337:61-6.

2. ChessellsJN{. Treatmentofchildhoodacute llmphoblastic leukaemia: pres-

ent issues and future prospects. Blood Rev 1992:6:193-203.

48 35 23

at Jo tJ

25

26

27

Remission >36 mo

I 258

Butturini A. Rivera GK. Bortin MM. Gale RP. Which treâtment for child-hood acute iymphoblastic leukaemia in second remissionl Lancet 1987:l:425-32.Horow itz MM. Bortin MNt. The role of regisiries in evaluating the results ofbone manow transplantation. ln: Treleavan J. Banctt AJ. eds. Bone manowtransplantation in practicc. F-dinburph. Scotland: Churchill l-ivinsstone.1992:367 -77 .

Buchanan GR, Rivera GK. Bo1'ett JM. Chauvenet AR. Crist WM. ViettiTJ. Reinduction therâp) in 297 children sith acute lvmphoblastic leukemiain first bone manou'rclapse: a Pediatric Oncolo-ey Group Study. BloodI 988:72: I 286-92.

fHE \E\\' E\Gr.,\\D.JC)L R\.\L OF \rEt)lcI\E,

l3

l4

35

\or'. 10. 199{

Begg CB. McGlave PB. Bennen JM, Cassileth PA. Oken MM. A criticalcomparison of allogeneic bone manou transplantation and conventionalchemotherapl as treatment fbr acute nonlymphocytic leukemia. J Clin On-col 198,1:2:369-78.

Horowitz MM. Messerer D. Hoclzer D. et al. Chemotherapy compared withbone marow transplantation fbr atlults with acute lymphoblastic leukemia inlirst remission. Ann lntern N{ed l99l:l t5:13-8.Cox DR. Regression models and Iifc-tablcs. J R Stat Soc [B| 1972:34:187-220.Andersen PK. Borgan O. Gill RD. Keiding N. Statistical models bascd oncounting processes. Neu' York: Springer-Verlag. 1993:386-8.

ll

Yol. 331 \o. l!) \(:CL \IUL.-\TIO\ OF \UCILE'\R p53 À\D TL\IOR PROGRESSIO\ I\ BLÀDDER C:.\\CER l2i9

ACCUMULATION OF NUCLEAR p53 AND TUMOR PROGRESSION IN BLADDER CANCER

Drvro Esnro. \I.D., Doxar-o Enrelrex, \I.D.. Suses GnosHEN. PH.D., Jonx A. Fna,Ert.rx. \LD..Jonx P. Srrrx. \'I.D., Su-Cnru CsEx. \I.S., PErr,n \\:. Nrcnor-s. N[.D., Doxer.o G. Srrxrnn, \1.D.,

Pr,rr,n A. JoxEs, Pu.D., ,A.No RrcHero.]. Corr,, NI.D.

Abstract Background. We have previously demon-strated a strong association between nuclear accumula-tion of p53 protein, as determined by immunohistochemi-cal analysis, and mutations in the p53 gene. The purposeof this study was to determine the relation between nucle-ar accumulation of p53 and tumor progression in transi-tional-cell carcinoma of the bladder.

Methods. Histologic specimens of transitional-cell car-cinoma of the bladder (stages Pa, noninvasive disease, toP4, disease with direct extension into adjacent organs orstructures) trom 243 patients who were treated by radicalcystectomy were examined for the immunohistochemicaldetection of p53 protein. Nuclear p53 reactivity was thenanalyzed in relation to time to recurrence and overall sur-vival.

Results, The detection of nuclear p53 was signif icantlyassociated with an increased risk of recurrence of bladdercancer (P<0.001) and with decreased overall survival(P<0.001). ln patients with cancer confined to the bladder,the rates of recurrence for stage Pl, P2, and P3a tumors

U'IATIO)iS of.the p53 gene are the most com-mon genetic defect in human tumors.l The p53

gene functions as a tumor-suppressor gene and morespecificallr,' as a cell-ct'cle regulator.2 Levels of p53protein increase in response to damage to DNA, ar-resting the cell c,vcle and allou,ing time lor the repairof D\A.

\'lutations o[ the p53 gene occur in a high pcrcent-age of invasive transitional-ccll carcinomas of thebladder3 and appcar to be an earlv event in the forma-tion o1'carcinoma in situ.+ They are much less frequentin noninvasive papillarv tumors.-'o \Iutations of thep53 gene and nuclear accumulation of p53 protein areassociated rvith the srade and stage ol bladder canceriand mav be important in the multistep progression ofbladder cancer.T'!''I'he immunohistochemical detec-tion ol the protein, u,irich exploits the difference in iilespan betrveen mutated and rvild-tvpe p53. has beenshorvn to correlate strongll' u.ith mutations of the p53gene in bladder cancer.'

The primarr. lactors used in determining the clinicaltreatment of patients u'ith earlr'-stage bladder cancerare the depth of tumor invasion, the tumor grade, andthe presence or absence of lvmpl'r-node metastases.Radical cvstectom)' is a eenerally accepted treatmentfor patients rvith invasive cancer con{ined to thebladder, particularlv lhen there is evidence of mus-

From thc Departments of Urologl'(D. Esrig. D. Elmajian. J.A.F., J.P.S..D.G.S.. P.A.J, ), Preventive Medicine (S.G., S,-C.C. ), and Pathology (P.W.N. ,

R.J.C.), University of Southern California School of Medicine and KennethNonis Jr. Comprehensivc Cancer Center. Los Anseles. Address reprint requertsto Dr. Cote at the Depannrcnt of Pathôlog)'. University of Southem CalitbmiaSchool ol- Medicine. Kenncth r'.oris Jr. Cancer Center, l.l4l Eastlakc Ave.. LosAngeles. CA 90033.

Supponed in part by grants (R35 CA,19758 and P30 CA1,1089) from the Na-

tional Crncer lnstitute, b) the American Foundation for Urologic Disease-National Kidney Foundation Fellowship. and by the Firestein-Genz Cancer Re-

search Fund.

that had no detectable nuclear p53 reactivity at five yearswere 7, 12, and 11 percent, respectively, as comparedwith 62, 56, and 80 percent, respectively, for tumors thathad p53 immunoreactivity. Similar results were obtainedwhen the presence or absence of p53 in the nuclei of thetumor cells was studied in relation to overall survival. ln amultivariable analysis stratified according to grade, patho-logical stage, and lymph-node status, nuclear p53 statuswas an independent predictor (and in cancer confined tothe bladder, the only independent predictor) of recurrenceand overall survival (P<0.001).

Conclusions. ln patients with transitional-cell carcino-ma confined to the bladder, an accumulation of p53 in thetumor-cell nuclei detected by immunohistochemical meth-ods predicts a significantly increased risk of recurrenceand death, independently of tumor grade, stage, andlymph-node status. Patients with transitional-cell carcino-ma confined to the bladder that demonstrates nuclear p53reactivity should be considered for protocols of adjuvanttreatment. (N EnglJ Med 1994;331:1259-64.)

cle invasion.rt) Surgical treatment is curative in a sub-stantial proportion of such patients, but in a largenumber incurable metastatic disease appears aftersurgcr)'. Adjuvant therapl' (including chemotherapr.and radiation therap-v) is under investigation in thissetting, and sholvs encouraging results.rr'r2 It ma1'thus be important to distinguish patients for rvhomsurgery alone is adequate and who can avoid the mor-bidity and expense of adjuvant therapv from patientswho might benefit from both surgerv and additionaltreatment.

In the present studv rve sought to determine rvheth-er a nuclear accumulation of p53 is an important fac-tor in predicting the clinical behavior ol bladdercancer. Previous work has suggested that alterationsof p53 can predict disease progression in superficial-ly inr.asive bladder cancer.l'l Horvever, a studv ofpatients rvith bladder cancer u,ho received a varietvof treatments suggested that nuclear accumulation ofp53 is not a predictor of disease progression that isindependent of clinical stage and the rate of pro-liferation of the tumor cells.t+ We demonstrate, in243 patients rvith bladder cancer treated br. radicalcvstectom,v, that nuclear accumulation of p53 identi-fies transitional-cell carcinomas with a propensitl' forprogression that is independent of tumor grade andstage.

MBrrroos

Population of Patients

\\'e studied 2.[3 patients rlho undcr§enr radical cr.stectontr.. pel-vic-lvmph-node dissection. and urinarv divcrsiorr irt the Kennetlr\orrisJr. Comprehcnsite Cancer Center. To assess the associationbet$'een nuclear accumulation ol p53 and clinic;rl outconrc. rr e iu-cluded all patients rvith transitional-cell carcinoma treatecl br radi-cill c)stectomv fiom April l9B3 through June lgBT lirr uhom fbl-lon-up data and tumor samples lrom the cvstectomv specimens

l 260

(preserved in archival parafân-embedded tissue blocks) were avail-able. An additional 33 patients who underwent cystectomy bet\4r'een

July 1987 and December l9BB were studied in order to increase thenumber of patients with cancer confined to the bladder. The studyalso included 53 patients described previously whose tumors under-v/ent both molecular and immunohistochemical analysis for p53.sThus, ol the entire group of 243 patients, 190 had not previouslybeen evaluated for nuclear accumulation of p53. Patients with pureadenocarcinoma, squamous-cell carcinoma, or small-cell carcinomaof the bladder were excluded. The median age of all patients was 63years (range,49 to B3); 77 percent were men, and 23 percent werewomen.

Cllnical and Pathological Evaluation

The indications for radical cystectomy included invasion of mus-cle or prostate stroma by the tumor; high-grade, superficially inva-sive tumors associated with carcinoma in situ; carcinoma in siturefractory to intravesical chemotherapy or immunotherapy; andrecurrent multifocal disease alter conservative therapy. None ofthe patients received pelvic irradiation or systemic chemotherapybefore surgery. All the specimens included in this study were transi-tional-cell carcinomasl a minority demonstrated glandular or squa-mous differentiation. The histologic grading was performed ac-cording to the method of Bergkvist et al.,rs and the pathologicalstaging \^/as done according to the tumor-node-metastasis classifi-cation.lG All tumors were reevaluated for histologic grade by twoof the investigators. Among the 243 tumors, 11 were classifiedhistologically as grade 2, 157 as grade 3, and 75 as grade 4. Five

Nov. 10, 1994

tumors \/ere classified as stage Pa (noninvasive papillary tumors),I I as stage Pis (carcinoma in situ), 50 as stage Pl (superficial inva-sion),32 as stage P2 (superÊcial invasion into muscularis propria),35 as stage P3a (deep invasion into muscularis propria),68 as stage

P3b (invasion into perivesicular fat), and 42 as stage P4 (directextension into adjacent organs or structures). Sixty-six Patientswere found on pathological examination to have metastatic diseasein the pelvic lymph nodes. None of the patients had distantmetastatic disease at the time of cystectomy. Forty-one patientsreceived systemic chemotherapy after surgery, either cisplatin,cyclophosphamide, and doxorubicin or methotrexate, vinblastine,doxorubicin, and cisplatin. The median follow-up for the 243 pa-tients was 6.0 years, with Bl percent having at least 3 years offollow-up. The patients were seen at three-month intervals duringthe first postoperative year, every four months during the secondyear, and every year thereafter. Follow-up consisted of a biochemi-cal profile, chest radiography, and physical examination. A com-puted tomographic scan or bone scan was performed to confirmsuspected recurrences of disease. Data on recurrences of transition-al-cell carcinoma and causes of death were obtained from office andhospital records.

Monoclonal Antibodies and lmmunohistochemical Analysis

Five-micrometer sections of archival formalin-fixed, parafEn-embedded tissue were placed on slides coated with poly-lJysine(Sigma, St. Louis). The immunohistochemical procedure was per-formed as described elsewhere.5 In briet after deparalfinization andblocking of endogenous peroxidase, the antipS3 mouse monoclonal

THE NEW ENGLANDJOURNAL OF MEDICINE

A

Figure 1. lmmunohistochemical Detection ol ps3 ttü:"il$# Xffiflj,J lrÂiïËional-Cell

Carcinomas of the Bladder with the Anti-psg

No detectable nuclear staining is seen in Panel A, only a few cells (1 to g percent) with nuclear reactivity are seen in Panel B,heterogeneous nuclear reactivity is seen in 10 lo 49 percent of tumor cells in Panel C, and intense homogeneous nuclear reactivity isseen in 50 to 100 percent of tumor cells in Panel D (all panels x 105). Tumors with 0 to g percent nuclear staining were considered p53-

negative, whereas tumors with 10 to 100 percent nuclear staining were considered ps3-positive.

\:ol. 331 \o. I9 ACICIU\IULATION OF )iLCLEAR p53 AND TU\IOR PROGRL,SSION I\ BLADDER CA\CER I26l

VARIÂBI-E

Gradet23

4

Stage§

Metastases absentPa or PisPIP2P3a

P3b

P4Metastases present

Table 1. Association of p53 lmmunoreactivity with Grade andPathological Stage of Bladder Cancer and Presence or Absenceof Lymph-Node Meiastases in "lg0 Patients Not Previously Test-

ed for p53 Alterations.

No. oFPaTrENrs NUCLEAR p53 REAcflvlTt! P V{LU[

NECAI I\ E TSITIVE

number lpercent)

9 7(78) 2(22)122 79 (65) 43 (35)59 33 (56) 26 (411

15 12 (tt0) 3 (20)39 30 (77) 9 (23)29 l9 (66) l0 (34)20 1l(55) 9(45)36 2l (58) t5 (42)9 7(78) 2(22)

42 19 (45) 23 (55) 0.0031

xNegatiÿe indicates that less than l0 percenr of tunror cell nuclei demonstrated p53 immunoreaclivil). ând positive lhat I0 percent or morc of tumor-cell nuclei demonstrâted p53immünoreactisit) .

iAccording to the method of Bergkvi5l ls

iDerived tiom Peason's chi-square test for a tso-b!{wo table comparing grade 2 ând 3

tumors with grade 4 tumors. l'

§Pathological staging was performed according to the tumor-node-metastasis sysrcm.l6N'letastases refers to lymph-nodc metastases identilied during the pathological eÿaluation.

lBased on the likelihood-ratio tesr for logistic regression.li

antibod)' PAblB0l (IgGl class. l:10 dilution, Bioqencx, Sarr Ra-mon, Calif.) uas incubated overnight rr.ith tissue at .1"C. P.\blB0lrecosnizes the p53 protein at a dcnaturation-resistant epitope corre-sponding to amino acids 32 through 79. l'issues \\'ere thcn incubat-ed uith a biotinllatcd horse antimouse secondarv antibodr'(\rectorLabs, Burlingame, Calil). and reactivitv rças visualized rrith anavidin-biotin-immunoperoxidase svstem (\'ector) using diamino-benzidine (0.03 percent) as the clrromosen and hematoxvlin as thecounterstain. Samples of bladder c:rrcinoma rlith knorvn p53 muta-tions and documented accumulations ol'p53 protein bv inrmuno-histochemical analrsis lere used as positive controls. Normal uro-thelium and nonepithelial cells (hmphootes, stromirl cells, andendothelial cells), used as internal neqativc controls. demonstratcdno immunoreactivitr'. Onlv nuclear localization of immunoreactir'-itv lvas evaluated. 'I'he extent of nuclear reactit'itt rtas classified infour catccories: no nuclear reactivitvi a leiv focallr positive nuclei( I to 9 percent of tumor cells): heterogeneous nuclear reactivit\' ( I 0to 49 percent of tumor cells): and intense homogeneous nuclearreactivit) (50 to 100 percent of tumor cells) (l:ie. 1). C)nh'samplesdemonstrating ar least I 0 percent nuclcar reactivitv rvere consideredto be p53-positive (to have an alteration in p53). \\'e based thiscriterion on our demonstration ol' a slrong correlation of mu tarionsin the p53 gene rvith the accumulation of p53 protein in l0 percentor more of the tumor-cell nuclei.r The immunohistochemical analr'-sis was performed rçithout knorçledee ol the tumor stase or theresults ol clinical follorr-up.

Statistical Analysis

Survir.al was calculated lrom cvstectomv to the clate ol death orthe date of thc last follorv-up (either a clinical r,isit or a discussionrvith the patient's referring phr,sician)l deaths duc to anv cause wereconsidered to represent treatment làilures. 'fime to recurrence s,ascalculated lrom cvstectomY to the date of thc first document-ed clinical recurrence or the last folk»r.up: data r:n patients rÿhodied free ol disease beltlre anv recurrence u'ere censored at the timeof death. One patient u'as knorvn to have clied of disease, but thedate of recurrence was unkno\\n and therelore a datc nriduav be-

rueen the last clinic visit;rnd the date of'dcath $'as used. For thepurpose of the statistical analvsis. nuclear accumulations of'p53rvere classifred as either positive (accumulation in I0 percent ormorc uf tumor cells) or neqatirc.

Contingencv tables. Pearson's chi-square test. aud loeistic regrrs-

sionlt u'ere usecl to evalu:rte the association of p53 accumulationnitli l;mph-node st.ttus. pathological stage of the priman tumor.and histologic grade. Kaplan-\Ieier plotsls and the loe-rank testll'l'cre used to evaluate the association of these three standard clinicalprognostic variables, as rvell as the expression of' p53. rrith the timeto recurrence and u'ith sun-ival. Greenu'ood's I'ormularq rças used toestimate thc standard errors of the Kaplan-\Ieier estimates of theprobabilitv of survilal or recurrence. To determine rvhether anaccumulation of p53 prolided prognostic inlormation betond thatprovided bv the three standard clinical variables. a stratified loÉi-

rank test uas used. Àll P values reported are nro-sided.The statistical analvses rrere performed for the entire cohort of

2.13 patients and rsere perficrmed separatch lor the 190 patients norpreviouslv tcsted for accumulation of p53.

RnsurrsOf the 190 bladder tumors not previouslv examined

for accumulation of p53. I 12 had no nuclear reactivityand 7 had a ferv locallv positive tumor nuclei but lessthan l0 percent. These ll9 tumors were consideredp53-negative. In 54 of the remaining 7l tumors, p53protein was detected in l0 to 49 percent of tumor-cellnuclei. and in 17 tumors p53 was seen in 50 to 100percent of tumor-cell nuclei. These 7l bladder cancers'rvere considered p53-positive. In all 2+3 patients, 142tumors were classified as p53-negative and 101 as p53-positive.

Association of p53 Nuclear Reactivity with Tumor Grade,Pathological Stage, and Lymph-Node Status

Analysis of the 190 tumors not previously tested forp53 revealed that nuclear p53 was identified more fre-quently in grade 4 tumors than in grade 2 or 3 tumors)but this association !!'as not statisticallv significant(P:0.20) (Table l). Bvcontrast, thepresenceof nu-clear p53 was significanti)' associated rvith pathologi-cal stase (p : 0.003); 32 percent ol the patients withnegative lvmph nodes, as compared with 55 percent ofthe patients u'ith positive lymph nodes, had nuclearaccumulation of p53 (P : 0.008). In patients with noevidence of lymph-node metastases, the presence ofnuclear p53 lvas associated with greater depth of inva-sion; 22 percent of the patients with superficial disease(stage Pa, Pis, or Pl), as compared with 38 percent ofthe patients with disease invasive of muscle (staees P2through P4) , had nuclear p53 in the tumor cells(P : 0.044) . Similar results were obtained for the en-tire cohort of 243 patients.

Association ol p53 lmmunoreactivity with Recurrence andOverall Survival

In the group of 190 patients not previously testedfor p53 alterations, nuclear accumulation of p53 wassignificantll' associated with an increased probabilityof tumor recurrence (P<0.001) and a decreased prob-ability of survival (P<0.001). Identical resuks wereobtained for the entire cohort of 243 parients; Figures2 and 3 sholv curves for recurrence and survival for allpatients.

Among the patients u,ith no evidence ol lvmph-node involvement, nuclear accumulation of p53 lvassienificantly associated with recurrence in those withdisease confined to the bladder (srage Pi, P2. or P3a).Horvever, this association \r'as not statistically sienifi-cant amons the patients with extravesical extension of

0.20+

o)'Ê'E.trooÈGô9O

à*E(r(!_ooo-

I 262

Lymph nodes negative

Confined to bladderPa or Pis 15

Pl 39P2 29P3a 20

Not conlined to bladderP3b

P:t

33*27 0.48 75!13 6'7 *2762+17 0.002 9315 78a l.l56* t7 0.007 79+ l0 30* l580* 17 0.01 I 6.1a l5 22!14

of the bladder treated bv cvstecto-mv, the nuclear accum;l;tion ofp53 correlated rvith a significantlvincreased risk of recurrence and de-creased overall survival. The linkbetlveen nuclear p53 and prognosisrvas independent ol tumor grade,pathologicai stage. and l"'mph-nodestatus.'fhe strongest associationbetueen p53 immunoreactivitv intumor-cell nuclei and tumor pro-gression lvas observed when thedisease rr.as confined to the blad-der. 'Ihe presence ol nuclear p53was stronglv associated n,ith an in-creased risk of recurrence amongpatients rlith Pl. P2. or P3a diseaseand with decreased overall survivalamons those u,ith Pl or P2 disease.It rtas the onlv independent predic-

THE \E\\' E,\GLA\D.JOUR\AT, OF \IEDICI\E \ov. 10. 1994

tumor (stage P3b or P4) ('fable 2). The five-r'ear re-currence rates for patients r.vith stage Pl. P2. and P3a

tumors that rvere p53-negative rvere 7. 12. and I I per-cent, respectivelr-. In contrast, the fir.'e-l-ear recurrencerates were 62, 56, and B0 percent for stage Pl. P2, andP3a tumors that n'ere p53-positir-e (Table 2). In allsubgroups of tumors confined to the bladder, the re-currence rates at five vears for pS3-negative and p53-positive tumors di{Iered significantlv. Signi{icant diÊferences in estimated fir'e-\'ear survival rates \1'ere alsofound among patients with tumors conlined to thebladder (except stage P3a tumors) (Table 2). Identi-cal results rvere obtained ficr the entire cohort ol 243patients. Figures 4 and 5 shon' the Kaplan-\Ieiercurves for recurrence and survival in all 243 patients,stratified according to the depth of tumor invasion.

In a multivariable anah'sis stratified according tograde, patholoeical stage, and presence or absence oflymph-node metastases nuclear p53 status u,as an in-dependent predictor of the recurrence ol bladder can-cer and of o'n'erall survival (P<0.001). \\'hen patientsrvith ivmph-node-negative cancers confined to thebladder were stratified accordine to p53 status, patho-logical stage (i.e., depth of inr,asion) lvas not an inde-pendent predictor of either recurrence (P = 0.62) orsurvival (P : 0.23).

Forty-one of the 243 patients received postoperariveadjuvant svstemic chemotherapr'. \Vhen these pa-tients were removed from the analysis. the presence ofnuclear p53 remained highlv associated rvith recur-rence and death. 'Ihis anah'sis demonstrates the pre-dictive value of nuclear p53 reactir,'itv in a cohort ofpatients uniformlv treated rvith radical cvsrectomyand pelvic-lvmph-node dissection but no adjuvanttherapy.

DrscussroN

\\'e found that the immunohistochemical detectionof p53 protein in the nuclei of tumor cells can pro-vide prognostic information in patients with bladdercancer. In patients u'ith transitional-cell carcinoma

c')

.=Èf,ao.=-o(§_ooo-

0 1 2 3 4 5 6 7 8 9 10

Years since Cystectomy

Figure 2. Probability of Bemaining Relapse-free in 243 Patientswith Bladder Cancer and either ps3-Positive or ps3-Negative

Tumors.

ldentical results were obtained for the subgroup of 190 patientsnot previously lested for p53 alterations. Each tick mark repre-sents a patient who had not had a recurrence of disease at the

time of the last follow-up.

0 1 2 3 4 5 6 7 I I 10

Years since Cystectomy

Figure 3. Probability of Survival in Patients with Bladder Cancerand either ps3-Positive or ps3-Negative Tumors.

Each tick mark represents a patient who was alive at the time ofthe last follow-up.

Table 2. Estimated Rates of Recurrence and survival at Five years in 1g0 patientswith Bladder Cancer, According to p53 Status of Tumors and Pathological Stage.*

No. oFCRoLp AND SrAoE PÀT|ENTS RAIES oF RECURRENCE Relrs or SunvrrlL

p53- p53- P ps3- p53- PNEGA'IIVE rcSITIVE VALUE NECATIVE rc§ITtVE VAI.UE

18+127+5

l2+8Il+ll

0.'730.00.10.0230.20

36o

591 ll 73*t2 0.23 52+11 27+l 0.1843+t9 100 0.20 57t19 0 0.12

69* 12 9l t6 0.054 32:t I I l3+7 0.19Lymph nodes positive 42

*Rates of recurcnce and sun i\'{l &c based on Kaplan- Meier esiimaresril plus-minus values üe estimates of the standardcrror. calculaled u ith Greenwood s lbmula.l9 Tumors considered lo bc p5-l-negative had less than l0 percent of tumor-cellnuclei demonsralrng p53 immunoreactivity. and those considered to be p53-positive had l0 percent or nx)re of tumor-cellnuclei demonstratrn! f5l immunoreactrvl§.

pS3-positive (n = 101)

pS3-negative (n = 34)

ps3-positive (n = '1 1)

Vol. 331 \o. 19 ,\CCU\ILL.\'I'IO\ ()F \tICLEAR p5l) ,\\D'tU\lOR PROGRI'ISSIO\ I\ BL.\DDER CANCIER l2€i:l

0)

; 1.00a(§

P 0.75

=cF o.soCc)(ro o.2s.=

(§

€ o.ooÈ

0.s0

o.25

0.000 I 2 3 4 5 6 7 I I 10

Years since Cystectomy

Figure 4. Probability of Remaining Relapse{ree According toPathological Stage in Patients with Organ-Confined Bladder Can-

cer but No Regional Lymph-Node Metastases.All patients with P1, P2, or P3a dlsease from the entire cohort of243 patients were included in this analysis. ldentical results wereobtained tor the patients with P1, P2, or P3a disease in the sub-group of 190 patients not previously tested for p53 alterations.

tor o[ disease proeression in a multivariable compari-son of p53 status, pathologicai stagc. and histologicgracle.

The product of â mutated p53 gene t:an be a nleta-bolicallv stable protein rvith a lons half-lifè.ri' In con-trast, rvild-npe p53 has a short htrll'-litè. orih'6 to 30

minutes, and thus it dot's lrot scncrally accrrmulate inhigh enough levcls to be clctected l>r' standard immu-

nohistochemical methocls. A long-lived mutated p53protein, in contrast. is usuallr" detectable by such tech-niques. \\'e have previouslv demonstrated a signifi-cant association ol p53 mutati«;r'rs *'ith the immuno-histochemical detection of p53 protein in tumor-cellnuclei.:' Immunohistochemical techniques identiÊedmost cases ir-r u'hich thc cells ]rad mutated p53 genes.but 15 to 20 percent of tumors 'rvith demonstrablemutatiolls of the p53 gene were ncgative bv immuno-histochemical analvsis.r \Ioreovcr. a substantial pro-portion of tumors rvith nuclear accumulation ol p53lacked evider.rce of p53 gene mutation. Alterationsthat rer.rder the p53 protein clctectable bv immunohis-tochemical analr-sis mav not require a mutatiorl in thep53 gcne. For example. celh-rlar oncoselle productsthat bind to and inactivate n'ild-tvpe p53 protein, suchas À,IDNI2,rr mzrv prolong the half-lile ol'p53, allowingit to accumulate in the nucleus. Although this phe-nonlenon has not been shou'n directlv, recent studiesindicate th.rt cells that overexpress NIf)NI2 can alsooverexprcss p53 protein in thc absence of a p53 genemutation.r: N{oreover. the level of u'ilcl-tvpe p53 pro-tein can incrcase in response to D\A damage.::r Thus,increases in the ler.el ol the p53 protein and mutationof the p53 sene occur together in most cases but can beseparate events. Although u,e have shorvn that thepresence of p53 alterations as detected bv immuno-histochcmical methods is clinicallr- rele'u'ant, it mavalso be important to assess mutations of the p53 genedirectlv as a prognostic indicator in bladder cancer.

'lhe most important fincling of this studv is that inpatients u'ith transitional-cell carcinoma confined tothe bladdcr (stages P1, P2. and P3a). and no evidenceof lvmph-node metastases. p53 nuclear reactivit)-iclentifies tumors rvith a tendencv to progress. In thesecases. the detection ol nuclear p53 is more stronglvassociated lvith tumor recurrence and decreased sur-vival than is the depth ol invasion or histoloeic grade.Patients u'ith p53-negative tumors had relativell. lou,rates of recurrence. r.vhereas metastases developed inthe majoritt' of patients rvith p53-positive tumors, re-gardless of depth ol invasion. Although depth ol inva-sion is clcarlv associated n'ith the proeression of dis-ease in patients rvith bladder cancer, this ma1' reflectthe proportion of tumors in each stage that have p53alterations.

This studv mav in{luence the selectir-rn ol patientsfor ad.juvant treatment of blaclder cancer. r'r'hich cur-renth'ciepends on the depth of tumor invasion and thedetection of regional ivmph-node metastases. The re-sults of t\\'o recent studies are encouraging," '' but therole ol ad.jur.ant chemotherapv after c)'stectom)', par-ticularlv in patients rvith disease con{rnecl to the blad-der. remains unresoh'ed. Patients rvith p53-negativetumors confined to the bladder havc a lou. rate ofdisease prouression (even lvhen there is deep muscleinr,asion) and mav not recluire adjuvant treatmentafter radical cvstectom\,. In c()ntrrISt, paticnts \\'ithp53-positir.,e cancer conhned to the bladder (includingthose rtith onh. superficiallv invasive tumors) have apoor prognosis. 'fhev mav bencfit lrom adjuvant trear-

p53-negative (n = 20)

P = 0.018

pS3-positive (n = 12)

ps3-positive (n = 11)

(,)c.=

fao.=o(§ooÈ

l 26+

1.00

0.75

0.s0

0.25

0.00

1.00

0.75

0.50

o.25

0.00

1.00

0.75

0.50

0.25

0.o0012345678910

Years since Cystectomy

Figure 5. Probability of Survival According to Pathological Stagein Patients with Organ-Confined Bladder Cancer but No Regional

Lymph-Node Metastases.

ment. This studv and the studv bv Sarkis et al.r:r arealso relevant to the controversial issue of surgicaltreatment of superficially invasive bladder cancer. Pa-tients with superliciallv invasive p53-positive tumorsIiave a high rate of disease progression and mav there-fore benefit from earlv radical c),stectomv. N{oreover,our study indicates that patients rvith p53-positive tu-mors are at high risk of progression dcspite radical

THE NE\\' ENGI,A\D JOURNAL OF \IEDICJINE Nor'. 10. 199{

cvstectom-v and mav therelrrre benefit from adjuvanttreatment. Thus, the immunohistochemical detectionof p53 mav identif,v patients u'ith cancer confined tothe bladder rvho could benefit lrom radical surgervar.rd adjuvant therap\'. Furthermore, the absence ofdetectabie p53 mav be an indication for conser\iativetl.rerapv. er,en in the presence of localll' advanceddisease.

\\'e are in<lebted to f)rs. Garl Licskovskr and Stuart BovdIbr providing sonrc of the paticnts included in this studr'. andto I)rs. Ron \atale and Cllive 'far lor for tht'ir critical rer,ieit ofthis rvork.

RrprnnNcrs

l. Hollstein M. Sidranskl'D. Voqelstein B, Hanis CC. p53 Mutations inhuman cancers. Sciencc l99l:253:.19-53.

2. Lane DP. Cancer: p53, guardian of the -senome. Nature 1992:358:15-6.3. Sidransky D. Von Eschenbach A. Tsai YC. et al. Idcntification of p53 gcne

mutations in bladder cancers and urine samples. Science l99l:252:706-9.

.1. Spruck CH Ill. Ohneseit PF. Gonzalez-Zulueta M, et al. Two molecularpathwa]-s «) transitional cell carcinoma of the bladder. Cancer Rcs 1994:

54:784-8.5. Esrig D. Spruck CH llt. Nichols PW. et al. p53 Nuclear protein accumula-

tion conclates with mutations in the p5l gene. tumor,erade. and stage in

bladder canccr. Am J Pathol 1993:143:1389-97.6. Sarkis AS. Zhang Z. Cordon-Cardo C. et al. p53 Nuclear overexpression

and disease progression in Ta bladder carcinoma. Int J Oncol 1993:3:355-60.

7. Sidransky- D. Messing E. Molecular gcnetics and biochemical mechanismsin bladder cancer; oncogenes. tumor suppressor genes. and growth factors.Urol Cf in North Am 199219:629-39.Cordon-Cardo C, Wartinger D. Petrylak D. et al. Altered expression ofretinoblastoma gene product: prognostic indicator in bladder cancer. J NatlCancer Inst 1992:ti4: I25l-6.Logothctis CJ, Xu HJ. Ro JY. et al. Altcred expression of retinoblastomaprotcin and known prognostic variables in localll,advanced bladder cancer.

J Natl Cancer In« 1992;84:1256-61.Skinner DG. Manâgement of invasive bladder cancer: a meticulous pelvicnode dissection can make a diffèrence. J Urol 1982:128:34-6.Skinner DG, Daniels JR. Russell CA, et al. The role of adjuvant chemother-apy followin-e cystectomy for invasive bladder cancer: a prospective com-parative trial. J Uro! t99l:145:,:159-67.Stijckle M. Meyenburg W. We llek S. et al. Advanced bladder cancer (stages

pT3b, pT4a. pNl and pN2): improved survival after radical cystectomy and

3 adjuvant cyclcs of chemotherapy: resu lts of a controlled prospective study.J Urol 1992:t48:302-7.Sarkis AS. Dalbagni G. Cordon-Cardo C. et al. Nuclear overexpression ofp53 protcin in transitional cell bladder carcinoma: a marker for diseaseprogression. J Natl Cancer lnst 1993:85:53-9.

Lipponen PK. Over-expression of p53 nuclear oncoprotein in transitional-cell bladder cancer and its prognostic value. Int J Cancer I993:53:365-70.

Bcrgkvist A, Ljungqvist A. Moberger G. Classificalion of bladder tumoursbased on the cellular pattern. Acta Chir Scand 1965:130:371-8.Hermanek P. Sobin LH. eds. TNM classilication of malignant tumors. 4thed. New York: Springer-Verlag. 1987:133-4.Fienberg SE. The analysis of cross-classified categorical data. Cambridge.Mass.: MIT Press. 1977:9-86.Kaplan EL. Meier P. Nonparametric estimation from incomplete observa-tions. J Am Stat Assoc 1958:53:457-81.N'liller RG Jr. Suruival analysis. New York: John Wiley. l98l:44-102.Finlay CA, Hinds PW, Tan TH. Eliyahu D, Oren M. Levine MJ. Actiratingmutations for transformation by p53 produce a gene protJuct that foms an

hsc70-p53 complex with an altered half-life. Mol Cell Biol 1988:8:531-9.Oliner JD, Kinzler KW. Meltzer PS. George DL. Vogelstein B. Amplifica-tion of a gene encoding a p53-associated protein in human strcomas. NatureI 992:358:80-3.Cordon-Cardo C. Latres E, Drobjnak M. et al. Molecular abnormalities ofmdrn2 and p53 genes in adult soti tissue sarcomas. Cancer Res 1994:.54:794-9.Kastan MB, Onyekwere O. Sidransky D. Vogelstein B, Craig RW. Panici-pation of p53 protein in the cellular response to DNA damage. Cancer Res

l99l:51:6304-l I.

II

12

21

23

l3

14.

15.

16.

t7.

18.

19.

20.

22

pS3-negative (n = 14)

ps3-positive (n = 12)

\irl. ll3l \o. l9 III.\GE]S I\ CLI\ IC],\L IIEDI(JI\E l 265

Images in Clinical Medicine

Kru En<;r.r, M.D., Eorr<>n

Coronarl Arteriouenous F'istula on Coronary Angiograph2

A large coronary arteriovenous fistula can be seen originating in the left coronary artery, coursing over the lateral and posterior walls ofthe left ventricle, and eventually emptying into the right atrium. The solid arrow indicates the origin of the fistula (not well visualized), andthe open arrow indicates a normal-sized left anterior descending coronary artery, which is of much smaller caliber than the greatly dilatedfistula. Patients with such fistulas often have loud continuous cardiac murmurs and may have congestive heart {ailure, myocardialischemia (due to a coronary-artery "steal" phenomenon), and pulmonary hypertension. Coronary arteriovenous listulas typically arise as

congenital anomalies. With obliteration of the fistula by suturing, the prognosis is usually excellenl.

KENrcnr Fuyrse, II.D.University of Texas HealthScience Center at Houston

Houston, TX 77030

\\trnnEs Suenlrax, \I.D.Beth Israel Medical Center

)Jen'York. \Y 10003

THE \E\\' E\Gr..\\D.IOUR\'\1. Or \IEDIC:l-\E \or . 1t). l9!t{

SPECIAL ARTICLE

STATE PRACTICE ENVIRONMENTS AND THE SUPPLY OF PHYSICIAN ASSISTANTS, NURSEPRACTITIONERS, AND CERTIFIED NURSE-MIDWIVES

Eou'eno S. Sprsorxsrr, lvl.P.H., Srr,pHaxIr, SeNsol,r, trI.P.H., NI.P.P., Canol BezclL, \{.D., N[.P.H.,Nlanr-,q, E. Seluox, Sc.D., R.N., axn Frlznc;cn Nlurr-ex, M.D.

Abstract Background. Most proposals to increase ac-cess to primary care in the United States emphasize in-creasing the proporlion of generalist physicians. Anotherapproach is to increase the number of physician assist-ants, nurse practitioners, and certilied nurse-midwives.

Methods. We analyzed variations in the regulation ofnurse practilioners, physician assistants, and certifiednurse-midwives in all 50 states and the District of Colum-bia. Using a 100-point scoring system, we assigned nu-merical values to specific characteristics of the practiceenvironment in each state for each group of practitioners,awarding a maximum of 20 points for legal status, 40points for reimbursement for services, and 40 points forthe authority to write prescriptions. We calculated coeffi-cients for the correlation of summary measures of thesevalues within states with estimates of the supply of practi-tioners per 100,000 population.

Results. There was wide variation among states inboth practice-environment scores and practitioner-to-pop-ulation ratios for allthree groups of practitioners. We found

\ 7[OS'I proposals to increase access to primarr'IVI care in the United States emphasize increasingthe proportion of generalist phl,sicians.r-i Another ap-proach is to increase the number of other practitioners

- specificallv, physician assistants, nurse practition-ers, and certified nurse-midwives.'t!'Within their areasof competerrcv, and lvith appropriate training and su-pervision. these practitioners may pror.ide medicalcare similar in qualit,v to that of phvsiciar-rs and at lesscost.r''ri' 'fhese practitioners ma_v be especiallv valu-able in areas rvhere there are shortages of primarycare phvsicians. Yet state legislation and regulationmav discourage or prevent them from seeking emplol'-ment, even rvhen jobs would otherwise be ar.ailable.To understand the relation betweerr statcs' practiccenvironments and the suppl,v of these practitioners,ll'e analyzed variation in the regulation of nurse prac-titioners. physician assistants, and certified nurse-midwives in all 50 states and the District of Columbia(n,hich, for the purposes ol this analvsis. rve consid-ered a state).

Although the education, Iicensure. and reeulationof nurse practitioners and physician assistants diffêr.manv have similar job descriptions.r'j 'fhe1, diaenoseillness, perform physical examinations, order and in-

From the Bureau of Health Professions. Health Resources and Services Ad-ministration, Depanment of Health and Human Services. Rockville. Md. Ad-dress reprint requests to Mr. Sekscenski at the Bureau of Health Prolessions.Health Resources and Services Administration. Rm. 8-47. 560O Parklawn Dr..Rockville. MD 20857.

The vieus cxpressed in this article are srictly those ofthc authors. No ofhcialendorsement b! the Department ol Health and Human Services or any of its

components is intended or should be infened.

positive correlations within states between the supply ofphysician assistants, nurse practitloners, and certifiednurse-midwives and the practice-environment score forthe state (Spearman rank-correlation coefficients, 0.63lP<0.0011,0.41 [P : 0.003], and 0.51 [P<0.001], respec-tively). Positive associations were also found in the statesbetween the supply of generalist physicians and the sup-ply of physician assistants (r : 0.54, P<0.001) and nursepractitioners (r = 0.35, P : 0.014). Nevertheless, in the17 states with the greatest shortages of primary care phy-sicians, favorable practice-environment scores were stillassociated with higher practitioner-to-population ratios forphysician assistants (r = 0.68, P : 0.003), nurse practi-tioners (r : 0.54, P : 0.026), and certified nurse-mid-wives (r:0.42, P = 0.09).

Conclusions. State regulation of physician assistants,nurse practitioners, and certified nurse-midwives varieswidely. Favorable practice environments are stronglyassociated with a larger supply of these practitioners.(N EnglJ Med 1994;331:1266-71.)

terpret laboratory tests, establish and carrv out treat-ment plans, suture wounds, and provide preventivehealth services. Each profession is about 25 r'ears oldin the United States. Phl.sician assistants are saiariedemplovees nho bv law must work under the supervi-sion of a ph,vsician. Of the 22,300 phl.sician assistantspracticing in 1992, 44 percent rvorked in primarv carespecialties. and another B percent lvere in emergenc).medicine. The majoritv were educated in trvo-yeartraining prosrams. About 34 percent of phvsician as-sistants rvorked in rural areas.l7

In some states, nurse practitioners can establish in-dependent practices and be reimbursed directlv fortheir services. Because the states have no commondefinition of nurse practitioners. estimates of theirnumber varv widelv. 1-hrough 1992. about 42.600emploved registered llurses had receir,ed lormal train-ing as nurse practitioners bevond their professionaleducation as nursesrs: estimatcs of the number prac-ticing as rlurse practitioners ransed lrom 21,900re to27,200.2" 'lhe majoritv were educatcd in certificateprosrams a'"'erasins about one vear in length; 4 ofeverl' l0 l'rad rnaster's degrees. About three quartersrvere in primarr, care. Eighteerl percent of nurse prac-titioners lvorked outside metropolitan areas in 1992.re

Certified nurse-midwives are registered nurses u'ithadvanced education in the pror,ision of prenatal, peri-natal, postpartum, neu'born, and routine gvnecolog-ic care. About fJl pcrcent hacl master's degrees in1991.'rr Sincc 1971. national certification as a nurse-nridrvifc has rccluired eraduation ll'om an accreditedprogram lbr nurse-midr,vifelv and the passine of an

Vol. 331 \o. l9

examination administered by the American Collegc of\urse-Nlidu'ir-es. In 1992, 43 states recosnized certi-fied nurse-midwives in their statutes or regulations.22About half of all states allou'ed direct reimburse-ment ltir the services of a certified nurse-midwife. In1992, betn een 3500 and 4300 certified nurse-midwit,esu'ere eligible to practice in private offices, communityhealth centers, free-standing birthing centers, and oth-er health care settinss2'r (and unpublished data) . Cer-tified nurse-midwives attended 4.1 percent of all delir,-eries in the United States in l99l.i+ Betu-een 1l and 22perccnt practiced in rural areas.l32*

Mr:rHoos

Practicc environnrents in the states u'ere assesscd bv rerierringjournal articles and lcgislation and bv consultinq rvith researchers.legal scholars. anrl 1;rolessional organizations. In all jurisdictions.infbrrn:rtion las sorrght aborrt conditions in lgg2. Spccilic criteriaare shoun in 'I'able l.

A 100-point scoring svstem \\'as corlstructed for each group; a

maximum ol-20 points rras allocated if practitioners had legal sta-tus as professionals. '10 points i[ rcimbursement lor their srrvicesrvas requircd. and.l0 points il thcv had the authoritv to \lrite pre-scriptions. \Irrrc leigltt uas gir-en to the second and third categoriesbecausc the simple recognition ol prolessional iclentitv cntailcd inthe confèrring o['lce:rl status alonc uas consiclered less important;houevcr. rvhen equ:ll rveiqht rvas given to each o{'the three rnajurcatesories. the results did not change substantiallr'. Points rrereallocated llithin cach categorv and then totaled..\ score of 100represcntecl the most lavorable enVironmcnt. and a score of 0 theleast favorablc. The assessnrent *as perlormed consistenth lirr allthe states in a given discipline. but the actual criteria lbr the disci-plincs varied bccause ol' prolcssional and resulatorv di{Ierences.Thus, comparison of scort's benveen states is more appropriatervithin a discipline than between disciplincs.

Thc practice environments Iôr phvsician assistants \iere quan-tified primarilv on tlle basis ol' inlormation lrom the AmericanAcademv of Phvsician Assistants!; and other published studies.tr'26'fhe practice environrnents làr nursc practitioncrs rrere quantificdon the basis of inlbrmation lrom published studies.rili The prac-tice ertvironnrcnts [or certified nurse-midu ives were quantifiedon the basis ol information f-rom tlie .\merican Ciollege of \ urse-\{idrvives2328 and a survev bl the Ollice of the Inspector Generalof' the Department of' Health and Human Services.lr Supplemen-tal information on all threc groups ol practitioners *as obtainedfrom the 1993 annual report of the Phvsician Parment ReviervCommission.+

Estimates of the supph of nonphr.sician practitioners in each

PR-\C]-|IC]E E\VIRO\\{ENTS.\ND THE SUPPLY OF NO}iPHYSICIAN PR.\CTII'IO\ERS 1267

state h'ere obtained lrom various sources (Table 2). The estimatesof 27,200 practicing nurse practitioners and '1300 certified nurse-midrvives u'ere the onll available estimates that providecl state-specific figures. The estimates of the number of phvsician assist-ants do not include federal emplolees. The supph of sencralistphvsicians was cstimated as the total number of nonfederirl allo-pathic phvsicians activeh'involved in patient care u'ho designatedthemselves as being in eeneral practice. lamilv practice, generalinternal medicine, or general pediatrics in the 1992 Area ResourceFile.rs Practice-environment scores and estimates of the supplvof practitioners u'ere calculated independentll'. Data from theBureau of the Census on state populations in 1992 were obtainedfrom the Ànrerican \ledical Àssociation.'j'r Estimates ol' the percent-age ol each state's population that was Iiving in areas designated as

having a shortage of primarv care u'ere obtained lrom the Bureau ofPrimar.v Health Care of the Department of Health and HumanService s.'ll

Becausc the practice-erl'ironment scores \\'e developed hadnon-normal distributions. the\' \r'ere analt zed lvith nonparamet-ric methods. States rçere ranked according to their practice-enri-ronment scores and the number of practitioners per 100.000population (practitioner-to-population ratios). Rank-correlationcoe{Hcients n'ere derivrd lbr pairs of individual practitioner-to-population ratios and practice-en\ ironment scores. according to themethod of Spearman.32

\\'e calculated partial correlation coe{Êcients that compared thesupplv of nonphr.sician practitioners rvith that ol generalist phlsi-cians. rvith control lbr the state population.3s AII P values are basedon t\r'o-tailed tests.

Rrsurrs

There was lvide variation in both state practice-environment scores and practitioner-to-population ra-tios for all the groups of practitioners (Table 2, Fig. I ).For phvsician assistants, the practice-en\rironmentscores ranged from a high of 100 in the state of Wash-ington to 0 in N{ississippi. Twenty states had scores of90 or higher; l4 had scores below 50. Practitioner-to-population ratios

"'aried from a high of 24.6 phvsician

assistants per 100,000 population in Maine to a lo!r,of0.2 in Mississippi. Twenty-one states had I0 or morephvsician assistants lor every 100,000 people, whereas13 states had 5 or fewer.

Practice-environment scores lor nurse practitionersranged lrom 100 in Oregon to l4 in Ohio and Illinois.Twelve states scored 86 or abol'e; l9 had scores belo$'50. The ratios of the number ol nurse practitioners

Table 1. Scoring System Used to Quantiÿ the Practice Environment in States in Regard to Physician Assistants, NursePractitioners, and Certified Nurse-Midwives.

ScoRrNc CArEcoRt

Legal status(20 points)

Rcimburscment(40 poinrs)

Authority toprcscribe(.+0 points)

PHYsrctAN AssrsrÀNTs

License recognition. 5 points: scope

of practice regulations, 0-5 pointsmore: practice under physician'sindirect superuision. additional0 l0 points.

Mandated payment. 30 points: pay-mcnt for services under indirectsuperuision, l0 points more unlesspayment was less than that paid tophysician. then percentage mülti-

' plied by I0, for 0-10 points.Any authority to write prescriptions.

20 points:0-20 more points based

on absence of specilic restrictions.(Limited auth0ritl'to order medi-cations in inpatient settings. 0- l0poinrs.)

NURsE PRAcTlrloNERs

Licensc or title recognition. 6 points:scope of practice defined by boardof nursing alone. 7 more points: norequired superuision by physi-cian, another 7 points.

Mandated palment. 20 points: :ervicescovered. another 0- 10 points;percentâge of physician feespaid by Medicaid (times l0).0 l0 points more.

Continuum of points bascd on levelof independence: no authority.0 points: full authority withoutphysician oversight. 40 points.

CERTTFTI:D NriRsF-MrDwrvFs

License or title recognition, 10 points;regulation by board of nursing alone,additional l0 points.

Mandated payment, 20 points: typesof matemity. perinatal, or family-planning services covered. addi-tional 0-20 points.

Three categories: no authority.0 points; limited or restrictedauthùrity. 20 poinrs: tull authority.40 points.

I 268 THE NE\\- E\GL.\\D JOUR)iAL Ot- \lEDICINE \or'. 10. 199,1

Practice-enr,ironment scores for certified nurse-midwives ranged from 100 in Nlinnesota to 25 in Indi-ana. Six states had scores of 90 or higher; 13 hadscores of 50 or less. The number of certified nurse-midu,ives per 100,000 population was low in all thestates, ranging from a high of 6.4 in Alaska to a low of0.1 in Nebraska. Six states had practitioner-to-popula-tion ratios of 4 or above, and 14 had fe"r'er thanI certified nurse-midwife per 100,000 population.

\\'ith a felv exceptions, states that had favorablepractice-environment scores for one group ol practi-tioners also had favorable scores for the other tlvogroups. States r,vith more favorable practice-environ-ment scores were clustered in the West and North-west: several states r,r,,ith less lavorable scores were inthe Southeast.

Among states ll,ith generally unfavorable practiceenvironments, the lack o[ authority to write prescrip-tions rvas an important contributor to lou, scores for alleroups. For example, 16 of the l7 states with the low-est practice-environment scores lor physician assist-ants prohibited these practitioners from writing pre-scriptions, as did l1 of thc 17 states rvith the lowestscores lor certified nurse-midu'ivcs and 9 of the l7rvith the lowest scores for nurse practitioners.

Reimbursement n,as an important factor in thepractice-environment scores lor nurse practitioners,but it was somelvhat less important for physicianassistants, who, as salaried employees, are not re-imbursed directly for their services, and for certi-fied nurse-midr,vives. Of the l7 states r,r,ith the leastlavorable practice environments for nurse practition-ers, 4 had a score of 0 on the reimbursement scale,and none scored higher than 20 out o1'a possibie 40points.

Correlations among Groups of Practitioners

We found significant positil'e correlations for allthree groups of practitioners betneen favorable statepractice-environment scores and higher practitioner-to-population ratios (Table 3). Positive associationsn'ere also found in the states betrveen the supplyof physician assistants and the practice-environmentscore of nurse practitioners and between the sup-piv of nurse practitioners and the practice-environ-ment score of physician assistants (Table 3). Thissuggests that in most instances a greater suppl,v ofpractitioners in one group \ÿas not associated rvithbarriers to practice for the other. \Ve examined thepossibilitv that the supph' of physician assistants,nurse practitioners, and certified nurse-midwives ina particular state depends on educational opportuni-ties for these practitioners. \\''e analvzed the sup-ply of each group of practitioners in relation tothe number of accredited schools in the states in1992. The results \\'ere inconclusive (data not shorvn).Although several states rvith schools had a higher-than-average suppll' of practitioners of the disciplinein question, no overall correlation was found betweenthe state-specific supply of practitioners and thenumber of accredited schools for any group.

to members of the population also varied w-idel1,,

lrom 37.2 per 100,000 in the District of Columbiato 2.7 per 100,000 in Nebraska. Trventv-six stateshad ratios of more than 10, including 10 above 20.In contrast, five states had ratios of 5 or fewer.

Table 2. Practitioner-to-Population Ratios and Practice-Environ-ment Scores for Physician Assistants, Nurse Practitioners, and

Certified Nurse-Midwives According to State, 1992,*

STATE

AlabamaAlaskaArizonaArkansasCalifomiaColoradoConnecticutDelawareDistrict of ColumbiaFloridaGeorgiaHawaiiItlahoIllinoisIndianalowaKansasKentuckyLouisianaMaineMarylandMassachusettsMichiganMinnesotaMississippiMissouriMontanaNebraskaNevadaNew HampshireNeu'JerseyNew MexicoNew YorkNorth CarolinaNorth DakotaOhioOklahomaOregonPennsylvaniaRhode IslandSouth CarolinaSouth DakotaTennesseeTexasUtahVemontVirginiaWashingtonWest VirginiaWisconsinWyoming

MeanMedianStandard deviation

CERTIFIEDPHYsrcrAN NURSE PRACT! NURSE-

AssrsrANTSi uorrnsi lrIrD\\1\ ES§

RATIO SCORE RÂTIO SCORE RATIO SCORE

2.720.8

8.91.2

5.1I 1.015.37.75.98.29.04.65.12.1

9.3I0.66.22.2

24.6l3.98.9

10.4

7.t0.22.0

13.0

6.2t2.72.3

10.4

39 6.390 30.599 23.354 28.958 12.980 16.287 20. l55 20.492 37.348 t4.959 7.838 9.489 8.r59 '1.O

5t 1.499 8,387 1.442 6.031 3.894 21.549 I 1.883 18.3

89 .1.5

83 9.70 4.8

39 9.698 16.093 2.798 8.295 22.931 5.091 17.098 ll.792 7.?87 t2.351 7.846 7.199 21.386 10.8

93 19.7

37 9.1

94 l0.l42 7.2'77 6.593 10.786 t7.342 15.,1

r00 2t.496 7.795 1.597 l 1.0

72.8 t2.786.0 10.1

25.3 7.4

33 0.9 3293 6.4 8486 2.8 7648 0.4 3530 t.7 8059 3.2 5058 3.2 9360 2.t 6053 4.7 6068 2.3 9832 2.7 't0

27 1.8 1246 0.1t 5414 1.4 31

34 0.4 2573 0.s 5552 0.,1 6878 1.4 6820 0.4 3712 2.6 9093 3.r 6968 3.5 5145 t.3 7068 2.3 10072 1.1 5963 0.5 2198 t.2 9846 0.1 5073 0.8 3095 4.0 7065 2.1 5462 4.6 1893 2.3 6743 t.2 9098 l.l 55t4 0.9 6040 0.5 54r00 4. l 8066 1.5 3450 2.5 844t 2.6 5965 l.l 7027 0.9 5642 t.0 5491 3.2 7368 5.1 51

38 r.3 4790 2.8 7089 1.2 4767 1.0 6294 0.7 80

60.2 2.O 62.162.0 t.4 60,023.E 1.4 19.2

5.0s.3

I 1.5

10.5

2.1

r 9.55.14.89.1

12.74.6

I r.4

t2t5l84

l0t2t2

I8.95.3

*Practitioneuo-population ratios for crch group ofpractitioners are based on estimates of rhe

total suppl.v of pnctitioners per 100.0m population in each state.

+Estimates of the supply ofpracticing nonfederal physician assistânts were obtained from theAmerican Academ)- of Ph!sician Assistants.lT with extrapolations for nonmemtrers. Federaleûrployees ilc not include<i.

iEstimates of the suppl) of practicing nurse practitioners were obtained from the studl_ b)Morgan.lo

§Estimates of the supply of praclcing ce(illcd nurse-midsives wcre calculated b)'lheDir ision ofNursing.,f lhe Bureau of Health Professions on the ba\is ofdah liom the AmericanCollege of Nurse-Midrr rres.:l

Vol. 3lll \o. I9 PR'\C1'ICE EN\rIRO\\'{E\TS AND THE SUPPLY OF NO\PHYSIIIIA\ PRACI'IITIO\ERS I 269

90 1000

15

15

10

0

co(u

=o-oo-ooo-ooo(u

-ooo

.(U

.9o_cIL

C.E(!

=o-(LoOoooa<l)co.F(.)(d

ILq)ofz

co(!

=o_o(Looooooo

=pq)U)

fzro0)

.EL0)()

0+0

r = 0.626

P<0.001

aa

' ttt

t. i.

aa3 .. '

aata'.

To examinc the potential effect of competition be-trveen phr-sicians and either phvsician assistants ornurse practitioners, we comparcd the supph' ol eachgroup of practitioners u'ith that of seneralist phvsi-cians in the state. (Ccrtified nurse-midwi\res were ex-cluded from this analvsis because of their relativelysmall numbers; federallr, emplo,ved ph'n'sicians u,erealso excluded.) \\'e found sisnificant positive rela-tions betu'een the supph' ol each group of practi-tioners and the suppiv of allopathic generalist phvsi-cians (rvith residents in graduate medical trainingexcluded) after controlling for state population (Table4) . In most states, a sreater number of phvsician as-sistants and nurse practitioners did not appear to beassociated \vith a lesser supplv of generalist phvsi-cians.

\Ve found a positive correlation in the states be-tlveen the supplv of phvsician assistants and thenumber of resident phvsicians in graduate medicaltraining, suggesting that a disproportionate numberof phvsician assistants mav be emploved in teach-ing hospitals (Table 4). No comparable relation l'asapparent behveen the supplv of nurse practition-ers and the number of resident phvsicians in tl.re

states.Some studies suggest that competition betr,veen

phr,'sicians and nonphvsicians has triggereci the cre-ation of barriers to practice lor nonphr-sician practi-tioners in some states.li 3+ \\'e therefore compared thesuppl,v of generalist phvsicians in the states rvith thepractice-environment scores for nurse practitionersand physician assistants. No associations were loundbe§veen the practice-environment scores for eithergroup of practitioners and the supplv of generalistphvsicians. u,hether resident phvsicians n'ere includedor excluded from the anaivsis.

Areas with Shortages of Primary Care Physicians

An adequate suppll' of phvsician assistants, nursepractitioners. and certified nurse-midu'ives mav beparticularlv important in areas lacking sulHcient num-bers of primarv care phvsicians. \Ve repeated some ofour analyses for the I 7 states rvith the hiehest propor-tions of people living in areas desisnated as having ashortage of primary care in 1992 (Table 3). Theseproportions ranged lrom I l.6 percent of people inNorth Carolina to 25.0 percent in \orth Dakota.Three quarters ol these areas u'ith priman' care short-ages rÿere rural. Nonphl'sician practitioners are notcounted in the lormula used b1'the federal so\iernmentto characterize these areas.3r

In the l7 states^ lavorable practice-en\.ironmenrscores for phl,sician assistants and nurse practitionerswere associated rvith practitioner-to-population ratiossignificantlv above the national average (Table 3). Forcertified nurse-midlvives there was a similar trend(P : 0.09).'fhe nine states with practice-environmenrscores ol 90 or hieher for phvsician assistants had anaverase ratio of 13.4 phr.sician assisrants per i00,000people. as compared rvith LB in the states n'ith scoresof 40 or less. The four states rvith practice-environ-

'10 20 30 40 50 60 70 80 90 100

Practice-Envi ronment Score

10 20 30 40 50 60 70 B0

Practice-Environment Score

10 20 30 40 50 60 80 90 100

Practice-Environment Scorec

Figure 'l . Correlation of the Praclitioner-to-Population Ratio withthe Practice-Environment Score for Three Groups of Practitioners

in Each of the 50 States and the District of Columbia, 1992.

The numbers of practicing, nonfederally employed physician as-sistants (Panel A), nurse practitioners (Panel B), and licensed,certified nurse-midwives (Panel C) per 100,000 population areshown, with Spearman rank-correlation coetficients and P values

for the correlations.

r =0.412P<0.003

o'attt ' '

lr

r = 0.505

P<0.001

.t

j.'.' '

1270

ment scores of 90 or higher for nursc practitioners hadan average of 17.5 nurse practitioners per 100,000people, as compared with 6.4 in the five states rvithscores of 40 or less. For certified nurse-midu'ives, theratios were 1.2 per 100,000 for the two states withscores of 90 or higher. and 0.8 per 100,000 for the fourstates with scores of 40 or less.

DrscussroN

Favorable state practice environments for phv-sician assistants, nursc practitioners, and certifiednurse-midwives were strongl-y associated with a great-er supplv of these practitioners. States r,r,ith less fa-vorable practice environments had ferver such practi-tioners for everv 100,000 people. In general. practiceenvironments rvithin a statc rvere consistentlv favor-able or unlavorable for all thr-ee groups. Inabilitvor limited ability to write prescriptions was a ma-jor factor in lorvering practice-en\-ironment scoresfor ail three groups. Reimbursement issues n'ere im-portant in lorvering the scores for nurse practition-ers, but thev u'cre ol iesser importance tor ph,vsi-cian assistants and certified nurse-midlvives.

Table 3. Rank-Correlation Coefficients and P Values lor Practi-tioner-to-Population Ralios and Practice-Environment Scores forAll States and for the 17 Stales with the Largest Proportions of

People Living in Areas with a Shortage of Primary CarePhysicians. *

]'HE \E\\' E\GL,\ND.JOUR\AL OF \IEDICTI)iE

GRouPs CoUPARED

Table 4. Partial Conelation Coefficients and P Values lor theComparison of Estimates of the Supply of Physician Assistants

and Nurse Practitioners with That of Generalist Physicians andResidents in the 50 States and the District of Columbia, with

Control for State Population, 1992.*

\or'. 10. 199-l

No\RlstDENT(;ENERAI-rsr ALL Ar LoPÀTHtc RÈst-PH\\t(r\\s= DE\r PH\st(l\\(i

r PvaruE r PVALUE

*Spearman rank-conelation cæfhcients are shos n. Shonage arcas were determined on thebasis of the percentaSe of each state's population that was Iiving in areas with a designatedshona3e ol healrh professionals in Septenrber l 992. according to thc Bureau of PrimaN Health

Car.- ll The I 7 srart'. u irh the highest lere!'ntagÈ\ of p.ople residing in such ârr'r\ §ere NonhDakota (25.0 percentt. Mississippi (24.3). South Dakota (22..1), New I'lerico (21.9). Idaho(19.6). W)oming 119.5). Louisiana (19.1). Wcsl Virginia (19.1). South Carolina (18.,1).

Alabama ( 16.8). Ala\ka ( i.1 0). \'lontana ( l3 5). Arkansat ( 13.5). Oeorsia (13.5). the Districrof Columbia {D.C.) (12-0). Illinois (l 1.9). and No(h Carolina ( I 1.6).

ilndicate\ thât the ralio ibr rhc lirsr gr()up lisred i\ compared u ith lhe score for the second

sroup.

ilndicates that rhc ..ore lbr the first group li\tcd is comparecl s ilh lhe score for the rccondgroup

Physicians and physician assistânts 0.539 <0.001 0.583 <0.001

Physicians and nurse practitioners 0.3.17 0.014 -0.033 NS

*First-order panial corelation coefficienls wilh population held constant a

denotcs not signifi canl-

ilncludcs all allopathic physicians in general practice. lamil) praclicc. general intemalmedicine. or general petiiatrics. excluding federal enrployees, accordiûg to lhe Bureau ofHealth Professions.:9

ilncludcs all allopathic re\idents and fellows in all spccialties in 199f. according to the

Association of American Medical Colleges.lr)

Our findings do not support the hvpothesis that a

larger suppll,' of gene raiist phvsicians in a state is asso-

ciated r,r,ith a Iess Iàvorable practicc en"'ironment fornonphvsician practitioners. Indeed, rve lound that thesupplies ol' generalist ph1'sicians. phl'sician assistants.and nurse practitioners rvithin states were positivelvassociated. Horvever, states u,ith documented short-ages of primarv care phr.sicians that had environ-ments favorablc to ph1'sician assistants and nursepractitioners had more such practitioners than the na-tional average.

Factors other than those rve identified affect thepractice environment lor nonphvsicians at the stateIevel. For example, acceptance as professionals bvphvsicians (including the extension of hospital admit-ting privileges and professional collaboration), inclu-sion in the terms of private and corporate healthinsurance policies, ability to obtain malpractice insur-ance, and acceptance bv the public are probablv im-portant determinants of the supplv ol practitioners atthe communitr,, regional, and state lcvels. Ir.r addition,because our analysis applied to onl,v one short period,u'e could not determine u,hether the greater supplv ofnonphvsician practitioners preceded the removal ofbarriers to practice! or the rcverse.

Our stud1. demonstrates that regulation bv thestates of phvsician assistants. nurse practitioners, andcertified nurse-midn,ives varies u'idelr.. Thcse findingsmav help state legislators and regulators reduce spe-

cific barriers to practice and thus make these practi-tioners more available to paticnts.

t.

2.

3.

4.

RrrnnnNcrs

Budetti PP. Achicving a unifom tèderal priman care policy: opponunitiespresented b1' national health reform. JAMA 1993:269:49ti-501.Kindig DA, Culticc JM. Mullan F. The clusive generalist physician: can wereach a 507r goal? JAMA 1993:270:1069-73.

Rivo ML, Satcher D. Improving access to health care throu-uh ph1'sician

workforce retbrm: directions for the 2 I st century. JAMA I 993;270: I 074-8.Annual repon to Congress I993. washington. D.C.: Physician Payment

Review Commission, 1993.

Starfield B. Primary care: concept, evaluation, and policy. New York: Ox-ford University Press. 1992.Josiah Macy, Jr., Foundation. Report of the Josiah Macy. Jr.. Foundation:for July l, 1991 through Junc 30, t992. New Yorkr The Foundation.t992.Mundin-qer MO. Advanccd-practice nursing - good medicine Iilr physi-cians? N Engl J Med 1994:330:21 l-4.

MEASLRES CovPARED

Ratios and scores within group

Physician assistantsNurse practitionersCertilicd nurse-midwives

First-practitioner ratio and second-practitioner scorei

Physician assistants and nursepractitioners

Physician assistants and ceniliednurse-midwivcs

Nurse practitioners and ph;-sicianassistants

Nurse practitioners and cenifiednurse-midwives

Certified nurse-midwives andphysician assistants

First-practitioner score and second-prâctitioner scorei

Physician assistants and nursepractitioners

Nurse practitioners and ccrtifiednurse-midwiles

Physician assistants and certiliednurse-midwives

ALr STÂrES

AND D,C,(N = 5l)

COEFFI.

CIENI P VALUE

0.626 <0.0010.412 0.0030.505 <0.001

0.452 0.001

0.502 <0.001

0..+16 0.002

0.411 0.002

0.226 0. I l0

0.570 <0.001

0.3s7 0.010

0..132 0.002

SiloRTÀcE AREAs(N = 17)

coEFft-CIENT P vALUE

0.68I 0.0030.518 0.0260.424 0.090

0.513 0.035

0.623 0.008

0.493 0.044

0.142 0,076

0.2 l9 0.398

0.593 0.0r2

0.5 r2 0.036

0.602 0.011

5

6

Vol. 331 No. 19 PRACTICE E\\'IRON\,IEN]'S A\D'I'HE SUPPLY OF NONPHYSICI,\\ PRACI'ITIONE,RS t2i I

10.

The National Health Scrvice Corps White Paper: proposed strategies fbrfulfilling primary care professional needs: pan II: nurse practitioners. physi-cian assistants. and certilied nurse-midwivcs. Rockville. Md.: NationalHealth Service Corps. 1991.Primary cae workforce 2000: federal health policl'strategies submittcd toHillary Rodham Clinton and the Presidcnt's Task Force on National HealthReform. Washington, D.C.: Pew Health Professions Commission, 1993.

Brown SA, Grimes DE. Nursc practitioners and certified nurse-midwives: ameta-analysis of studies on nurse primary cre roles. Washington. D.C.:American Nurses, 1993.

Schaft GE, Cawley J. The physician assistants in a changing health careenvironment. Rockville. Md.: Aspen Publications, 1987.Clawson DK, Osteryeis M. eds. The roles of physiciân assistants and nursepractitioners in primary care. Washington. D.C.: Association of AcademicHealth Centers. 1993.Scupholme A, DeJoseph J, Strobino DM, Paine LL. Nurse-midwifery cueto vulnerâble populations: phase l: demographic characteristics of the Na-tional CNM Sample. J Nurse Midwifery 1992;37:341-8.Nichols LM. Estimating costs of underusing advanccd practicc nurses. NursEcon 1992:10:343-51.Safriet BJ. Health care dollars and regulatory sense: thc role of advancedpractice nursing. Yale J Reg 19929:417-87.Fowkcs V. Meeting the needs of the underserved: thc roles oi physicranassistants and nurse practitioners. In: Clawson DK. Osterweis M, eds. Theroles of physician assistants and nurse practitioners in primæy care. Wash-ington. D.C.: Association of Acadcmic Health Centers. 1993:69-8;1.General census data on physician assistants. Alexandria, Va.: AmericanAcademy of Physician Assistants. 1993.Bureau of Health Protèssions. 1992 Sample suruey of registered nurses.Rockville. Md.: Depanment of Health and Human Seruices. 1993.Idem. Suruey of cenified nurse practitioners and clinical nursc specialists:December 1992. Rockville, Md.: Department of Health and Human Serv-ices.1994.

Morgan wA. Using State Board of Nursing data to estimate the number ofnurse practitioners in thc United States. Nursc Pract 1993:lE(2):65-7.1.Ofllce of the lnspector Gcneral. A survey o[ certilied nurse-midwires.Washington. D.C.: Dcpartment of Health and Human Sen,ices. 1992.Barickman C. Bidgood-Wilson M. Ackerly S. Nurse-midwifery today: a

lcgislative update. J Nurse Midwifèry 199237:207.National Commission on Nurse-Midwifery Education. Education of nurse-midwives: a strategy for achieving afTordable. high-quality maternity care.Washington. D.C.: American College of Nurse-Midwives, 1993.National Center tbr Health Statistics. Advance repon of flnal natality staris-tics, 1991. Mon Vital Stat Rep 1993;42(3):Suppl:l-48.Gara N. Physician assistant state laws and regulations. 6th ed. Alexandria.Va.: American Academy of Physician Assistants. 1993.Willis J. Baniers to physician assistant practice in primary care and ruralmedically underserved settings. J Am Acad Physician Assist 1993:6:418-22.Pearson LJ. 1992-93 update: how each state stands on legislative issuesaffecting advanced nursing practice. Nurse Pract 1993;18(l):23-38.Nurse-midwifery today: a handbook of state legislation. Washington, D.C.:American College of Nurse-Midwives. Political and Economic AffairsCommittee- 1992.

Bureau of Health Prolèssions. Area resource lile. Rockville. Md.: Depart-ment of Hcalth and Human Seruices. September 1993.Roback C, Randolph L. Seidman B. Physician charactcristics and distribu-tion in the U.S.: 1993 edition. Chicago: American Medical Assocration.1993.

Bureau of Primary Health Care. Health professions shortage areas. Wash-ington, D.C.: Department of Health and Human Services. 1993.

Gibbons JD. Nonparametric methods for quantitative analysis. New York:Holt. Rinehan & Winston. 1976.

Klcinbaum DG. Kupper LL. Applied regression analysis aml other multivar-iablc methods. North Scituate, Mass.: Duxbury Press. 197E.Ginzberg E. Ostow M. Physician supply strategy: the case of the South.Health Aff (Millwood) 1992:l l(2):193-7.

20.

21.

22.

23.

25.

26.

27.

28.

29.

30.

ll

t2

13.

14.

15.

16.

17

t8

l9

31.

32.

33.

34.

THE \L,\\' E\GL.\\D.JOLR\.\L OI \IEDICTI\E

REVIEW ARTICLES

\or'. l(). 1994

MEDICAL PROGRESS

SEVERE ADVERSE CUTANEOUSREACTIONS TO DRUGS

Jr,ex Crauoe Rotlnet', II.D.,A\D RoBERT S. Srens. \I.D.

/t LTHOU(iH the ratc o{'acute severe advcrse cuta-.É\ ,r"orrs reactions to medications is lou'. these reac-tions can aflèct anvone rvho takcs medications and canrcsult in death or disabilin'.r Even a small number ofcases associated rvith a particular drug mal' alter therecommendations lor its use.z-l Prompt differentiationof severe adversc cutaneous reactions Irom less seriousskin disorders mav be difficult. Rapicl recognition ofsevcre reactions is cssential. Prompt rvithdra',val of theoflending drus is often the most important action tominimize morbiditv.

Adr,erse cutaneous reactior-rs to drugs are frequent.affecting 2 to 3 pcrcent of hospitalized patients.i \Ianvcommonlv used drugs l'rave reaction rates above I per-cent.j Fortunatelr'. most adverse cutaneous reactionsare no1 severe. and [èrt, are làtal.

Complications o[ drug thcrapv are the most com-mon tvpe of adverse event ill hospitalized paticnts.accounting for I9 percent o[ such events." Cutaneousor allergic reactions to druss are responsible for ap-proximatelr' 3 percent of all disabline injurics duringhospitalization.'; The reported percentage of cutane-ous drug reactior.ls that pht'sicians diagnose as poten-tiallv serious varies greatlv but is probablv about2 percent.TB \\te estimate that about I ol everv 1000hospitalized patients has a serious cutancous drug re-action. Each vear thousands ol- outpatients have cuta-neous reactions that mav rcsult in substantial morbid-in. or death unless promptll, recognized and treated.Not all serious adverse reactions to drugs lvith aprominent cutaneous component develop rapidlv. Forcxample. the distinctive cutancous chanses of eosino-philia-mvalgia svndrome ciruse great morbiditv butusuallv occur alter prolonged exposure.l'

In this article, '"r'e shall emphasize the clinical rec-osnition, epidemiologv. pathophvsiologv. ar.rd treat-mer-rt of acute. serious cutancous adverse reactiorls.

From the Depanmcnt of Dematology. Henri Mondor Hospital. University ofParis Xll. Creteii. France (J.C.R.). and Beth lsrael Hospital. Hanud MedicalSchool. Boston (R.S.S. ). Addrcss reprinl rcquests to Dr. Stem at the Departmentof Dematologr. Bcth Israel Hospital. 330 Brookline Ave.. Bosron. MA 02215.

Supportcd in pan b! a grant from TNSERIV1 (9O-O812).

'fable I presents the kcr' clinical features ol the re-actions u'c shall discuss.

RBcocNrrroN

Drug eruptions are most often morbilliform or ex-anthematous (Fig. 1).i ? Thev usuallv fade in a fervda1's but nla\' lr'orscn. In rare instances in which noalternative therapy is available, a drug may be contin-ued in spite o{' a morbilliform eruption. Unfortu-natelv, a morbilliform eruption is often the initialpresentation of more serious reactions including toxiccpidermal necrolvsis, hvpersensitivitv svndrome, andserum sickness. 'Iable 2 lists clinical features thatshould alert the ph1'sician that a reaction is serious.\\'hen a drug reaction is suspected, the presencc olurticaria. blisters, mucosal involvement. facial edema.ulcers, palpable or extensive purpura, fever, or lvm-phadenopathv almost ahvavs necessitates discontin-uation of the drug.

Several algorithms have been proposed lor the as-sessment of adr.erse drug reactions,l0-r''l but none havepror,'ed to be both sensitive and specific. The follorvingcriteria and 'Iables I and 3 provide euidelines for for-mulating a diflerential diagnosis. First, alternativecauses should be excluded. especiallv inlections, sincemanv inlectious illnesses are difficult to distinguishclinicallv from the adverse effects of drugs used totreat infections. Second, the interval between the in-troduction o[ a drus and the onset of a reaction shouldbe examined. f hird. anv imprclvement after drugu.ithdrawal should be noted. Fourth, the phvsicianshould determine rvhether similar reactions have beenassociated with the same compound. Fifth, any reac-tion on readministration ol the drug should be noted,

A skin biopsv is often critical for an accurate diag-nosis. but biopsv does not help in establishing rvhetherthe disease is drug-induced. In vivo tests include read-ministration of the drue (rechallenge) and skin tests.Reactions after rechallenge mav be worse. Rechal-lenge should not be performed after a serious reaction.

Skin tests and in vitro tests (such as the radioaller-sosorbent test) help diaenose IgE-mediated type Ihypersensitivitv reactions, especially to penicillin.r3 Inother t),pes of eruptions, skin testing has lorv sensitiv-it,v and specificitv.rtln vitro testing ol cellular proliÊerativc responses to drugs is usually' not helpful.15Although still investigational, in vitro studies of en-hanced toxic effects of drugs or drug metabolites oncells mav someday aid in the diagnosis and under-standing of the pathogenesis of some tvpes ol reac-tions. l6'17

SrrvnNs-JoHNsoN SvN»nolrn aN» ToxrcEprornuer Nrcnorvsrs

Stevens-Johnson svndrome and toxic epidermal ne-crolvsis are two relatcd mucocutaneous disordcrs rvith

Vol.33l No. 19 MEDICAL PROGRESS

Table 1. Clinical Features of Selected Severe Cutaneous Reactions Often lnduced by Drugs.

Erythema then purpura and necrosis, espe-cially of fatty areas

Urticaria or swelling of central part of face

t273

PERCENT PERcaNrFREQUENT sloNs

^ND SyMmoMs DRUG-INDUCED FATAL

10-3070 of cases involve fever, 50 <5lesions of the respiratory tract*md gastrointestinal tmct

Neârly all caæs involve fever, >80 30"acute skin failure,"*t leu-kopenia, lesions of the respi-ratory tract* and gastro-intestinal tract

Pain in affected reas 100 >10

Respiratory distress,. cardio- >50§ l-6vascular collapse*

DrÀGNosts Mucos^L LEsloNs

Stevens-Johnsonsyndrome Erosionsusuallyat >2 sites

Toxic epidemal necrolysis Erosions usuallyat >2 sites

Hypersensitivitysyndrome Infrequenl

Small-vessel vasculitis Infrequent

TvBc^L SKrN LEsloNs

Small blisters on dusky purpuric macules oratypical argets, rarc areas of confluence, detachment of <107o of body-surface æea

Individual lesions like those seen in Sievens-Johnson syndrome, confl uent erythema,outer layer of epidemis sepüatesreadily from basal layer with lateralpressure,t ltrge sheets of necroticepidemis, total detachment of>30% of body-surface area

Severe exanthematous rash (may become 30-507o of cases involve fever,purpuric), exfoliative dematitis lymphadenopathy, hepatitis,*

nephritis,* crditis,* eosino-philia, atypical lymphocytes

Palpable purpura, most often on the legs; 30-507o of cases involve thenodules; ulcerations; urticaria gâstrointestinal tract,* neu-

ritis, fever, glomerulo-nephritis*

Morbilliform lesions, sometimes with urticilia Fever, arthralgias

>90

Serum sickness or reactionsrcsembling serumsickness

Anticoagulant-inducednecrcsis

Angioedema

Absent

Infrequent

Often involved

<5

*Poæntial caüæ of death.

fSystemic conrequenccs of widespread injury to the skin, as seen with thcmal bums.

iNikolsky's sign.

§The flgure refers to the percenlage among hospitaliæd patients; a much smaller percentage of all cases æ drug-induced.

high rates of morbidity and morraliry (Table l).t'ts'tsAlthough the nosology and specific diagnostic criteriafor these disorders remain controversial, we believecertain clinical features help define these conditions2o(Tables I and 3).

Clinica! Features

ln 1922, Stevens and Johnson described childrenwith febrile erosive stomatitis, severe ocular involve-ment, and a disseminated cutaneous eruption of dis-crete dark-red macules, sometimes with a necroticcenter. This became known as Stevens-Johnson syn-drome.2l In 1956, Lyell introduced the term "toxicepidermal necrolysis" to describe patients with exten-sive loss of epidermis due to necrosis that leaves theskin surface looking scalded.22 In severe cases, Ste-vens-Johnson syndrome can include extensive areasof epidermal necrolysis. In most cases of toxic epidermal necrolysis, the discrete red macules typicallyseen with Stevens-Johnson syndrome occur aroundlarger necrolytic areas. The similarities between thehistopathological findings and the drugs responsiblesuggest that these two conditions are part of a sin-gle spectrum.ta'te'23'24 The term Stevens-Johnson syn-drome is also frequently used as a synonym for erythe-ma multiforme major, resulting in confusion. In ouropini<ln, the two are different conditions that are usu-ally clinically distinguishable.20 Patients with erythe-ma multiforme major have typical target lesions, pre-dominantly on the extremities (Fig. 2). Erythemamultiforme major usually occurs after infections, espe-cially herpes simplex and mycoplasma, and has a

benign course.25 Patients with widely distributed pur-puric macules and blisters (Fig. 3) and prominent in-volvement of the trunk and face (Fig. a) are likely tohave Stevens-Johnson syndrome, which is usuallydrug-induced.

Patients may present with a clinical picture of Ste-vens-Johnson syndrome that evolves to one of toxicepidermal necrolysis within a few days. Fever and in-fluenza-like symptoms unexplained by infectious ill-

Figure 1. A Morbillilorm Drug Eruption with Numerous Erÿhema-tous Macules and Papules That Vary in Size and Are

Symmetrically Distributed.

Most lesions are faint, but some may be slightly inliltrated andresemble urticaria. This exanthematous eruption often starts onthe trunk, as in this patient. lt may also begin on areas subjected

to pressure. The rash may become confluent.

127 I 'rHE NEW ENGLAND JOI-IRNAL OF. \{EDICINE Nov. 10. l9!)4

Table 2. Clinical and Laboratory FindingsThat Should Alert Clinicians That a Drug-

lnduced Cutaneous EruptionMay Be Serious.

Clinical findings

CutaneousConfluent erythemaFacial edema or central facial involvementSkin painPalpable purpuraSkin necrosisBlisters or epidemal detachmentPositive Nikolsky's sign*Mucous-membrane erosionsUrticariaSwelling of tongue

GeneralHigh fever (temperature >40"C)Enlarged lymph nodes

Arthraigias or arthritisShonness of breath, whcezing. hypotension

Laborâtory results

Eosinophil count ) 1000/mmlLymphocytosis with atypical lymphocytesAbnomal results on liver-function tests

+The outer layer of the epidemis spuates readily from thebasal layer with lateral pressure.

ness often precede the mucocutaneous lesions of thesetwo conditions by one to three days. Burning andpain occur. Initiallv, these eruptions are symmetri-cally distributed on the face and upper trunk, areasthat usually remain the most severely allècted.!'r Therash spreads rapidly and is usuallv maximal withinfour days, sometimes within hours. The initial skin

lesions are usuallv poorly defined macules rvith darkerpurpuric centers that coalesce (Fig. 5).

Although precise diagnostic boundaries betweenthe two disorders have not been established, cases

with Iimited areas of epidermal detachment are usual-ly labeled Stevens-Johnson syndrome and those withextensive detachment toxic epidermal necrolysis. Weclassi§' cases with detachment of less than l0 percentol the epidermis as Stevens-Johnson syndrome andthose with more than 30 percent as toxic epidermalnecrolysis.20 In cases with detachment of l0 to 30percent of the epidermis we consider the two syn-dromes to overlap.2o In toxic epidermal necrolysis,sheet-like loss of epidermis and raised flaccid blis-ters, rvhich spread rt'ith pressure, often occur, andNikolsk,v's sign (i.e., dislodgment of epidermis bylateral pressure) is positive on ervthematous areas.With trauma, full-thickness epidermal detachment(Fig. 6) yields exposed, red, sometimes oozing dermis.In other areas, pale necrotic epidermis may remain(Fig. 7).

Al:out 90 percent o[ patients with each disorderhave mucosal lesions, including painful erosions andcrusts on anv surface (Fig. B).r'r Impaired alimenta-tion, photophobia, and painful micturition often re-sult. The epithelium of the trachea, bronchi, or gastro-intestinal tract may be involved.2o-2t Often overlooked,these lesions may cause substantial morbidity. About85 percent of patients have conjunctival lesions.rs'2s'30

These range from hyperemia to extensive pseudo-membrane lormation.2e-:rr Synechiae between eyelids

Table 3. Factors to Consider in Diagnosing Severe Cutaneous Adverse Reactions and Their Causes.

DRucs lt osr OmEN

DIAcNosts RESTNSIBLE

Stevens-Johnson syn- *

drome

Toxic epidermal ne- *

crclysisHypersensitivity syn- Anticonvulsants, sulfona-

drome mides. allopurinol

Drug-induced vas- ïculitis

Serum sickness or re- Intravenous pro-

actionsresembling teins,p-lactamscrum sickncss antibiotics

Anticoagulant-induced Warfarinnecrosis

HeparinAngiædema Penicillin, cephalosporins,

contrast medium. glafe-nine,i drugs used inanesthesia

Nonsteroidal antiinflam-matory drugs

ACE inhibitors

5- I 0 daysA few minutes lnsect stings, foods

to a few houni

I -7 days

<4 wk

Tvprcar lNrelver rrouBEcrNNrNc oË DRUG

THÊRAPY To ONsET

oF REÂcrtoN

l-3 wk

t-3 wk

2-6 wk

I-3 wk

8- 14 days

3-5 days

ALTERNATTvE CAUSES NoT RELÂt ED

To DRUos

Postinfectious erythema multiformemajor (especially in the case ofinfection with herpes simplexor mycoplasma)

Cutaneous lymphoma

Infection, rheumatic diseases, lym-phomas

lnfection

Disseminated intravascultr coagu-lopathy. septicemia

HELPFUL TESTS

Skin biopsy with immuno-fluorescence testing

Skin biopsy with immuno-fluorescence testing

Skin biopsy. blood count,eosinophil count, liver-function tests

Antinuclear antibodies.rheumatoid faclor

C-3 and C4 complement

Protein C defrciency

Platelet countSpecific IgE antibodies for

penicillin allergy

*See Table .1 for a conrplete list.iSee Table 5 for a complete list.

iGlafenine is no lonfer mekered.

).11.1)!( : \1. l,li( )( ; tti.\\

;;.,:.

\i.{1. . {-+lÈ

Figure 2. Typical Target Les ons of Erythema Multiforrne Major.This case was classifled as erylhema multiforme major becauseof associated mucous-membrane lesions. These tarqel lesionsinclude three zones: an erythematous or dusky small cèntrai pap-ule that may blister. a raised edematous middle ring. and an

erylhematous outer ring.

ancl corrjun<'tivir olicn oct rrr'. Kerirtitis :rrrd r.orrrt'alerosions arc l('ss l)'eqtrt'rrt. Ircver is usllallv lriqhr.r intrixic t PiclcrrnlLl rrt'crolvsis (icntpcnrtrrrt.. )38"(11 thlrnin Stt'r'cns-Joirrrson svnrlrr,rrrc. lrrrrl astlrcrriir, skinpairr. :irrd anrii'tv aIe olit'rr ('\tr('nt(,.

'l'ht' cor-rrltlic'irtions ol loric t:piclt'rrnlrl nt't lolvsisarrrl t rlcnsivc tlrcrr-nal [rurns lrre -rinrilirr'.

'f lrt. sr vt.ritris prol;orrior)ill l() the extt'rrt ol skin rrtt'r'osis. -\lrLssir,.trans('pidcrrnal Iluicl losst's (l'i to I litt'r's clailv in atlrrltsn itlr lrall thcir lrorlr -srrrl'ucc irrca i1l\ ()l\'(,(l 1 o« cLrr rvithass()( iatecl elt't'ttolvtc inilrirllrnct-.1" 1'r't'rt'nal :rz()t(.rni.ris cornnror-t. Bat tt'rilrl colorriziitir,n ril' the skin lln(l (lc-cl'(':ls('(l intrnrrrrc rt'sporrsivt'nrss irrtlr.usc tlrt' likr.ii-Iroorl ol s('psis. -\ hr pt'rclrtrrlrolic stirtt'. s()nictillls \\ itl)irrlribition ol in:ulirr s(,.r'('ti()n or irrsulin lcsist.rrrt t . irc()rrrnr()n. Ditlirst intelstitilrl pnr'rinrorritis. r'hiclr nr.r.It'ltri to thc irrlrrlt rcspinrtolr (iistr( ss sr rrrlronrt,. sorrrr'-1inrc's rlcvelop,s.

l,r t'rt i1 t]rt' tlilrqrtosis o1'Slcle tts .Jolrnsotr 5\'nrlr'()ln('or torir cpiclt'r rn:rl nt r:r't.,lr sis is c lirrir':illr t,r'irlt.rrt. lskirr biopsr' ht'l1rs conlirnr tlrt' cliaqnrrsis. tlrus rrsrrirllr,cxcludinq ïtullorrs clist'iist's n()t r('lat('(l to clrug thclapr.I',arlr on. tht't't' is lirll-thit ktress cltirlcn-n:rl lrr.( r'()si-s

irn(l (l('ta( irm( rrt. \r itlr irn rrtrlt sliqhrlr lrltcrt'ci un(lcr'-lr irrq cit't mis l l iq-. 6t. 1'ltt' trst' ol' ft'ozt,n st.c tions lrllon,sir ral)i(l clilrq-nosis. Irnnr rrnoflrror('s( ('nc(' strrclit's orrlr'lrt'JP t'rcltrclc otirt'r'llrllotrs rli-scast's. .\ntrnia arrtl lr ni-pltr,1;t'rrirr nlt- li t'r1rrcnt. lrtrt cosinol-llrili;r is l'lrrr.. \ r.rr-1r'olrcrri:t sllqg(,sts a por)f llloqnr,sis. :

'l'lrc rt,qrouth ol epidclnris mar'llcqin uiti-,iri ilavsbLrt rrsualh ttrkt's :rlrout tlrltc u'r'cks. iltt'tr pic:rl l,,,,,lilro1' tlrc }rospit:riizlrtior-r.l .\r,.as sub jcct t() l)rasslrr(. un(l

1tt'r'iolilicial ar('lts ofi('n hr'aI Iast. ()r'trlar seclrrr'lirr. .r1-

lèr't ulrout 35 pcrct'rtt ot lrutients ulro sun,irc torit:cpirlCrnral nt't |olt si-. :tnrl e sn]ltllt't l)crcciltirq( ()f

tir,st' u iti'r St,'r't'ns-.Johnsorr s\ rldroln('.li -\ Siii,glcn-

Figure 3. Dusky or Purpuric Macules Typrcal of Stevens-JohnsonSyndrome.

These lesions may develop an overlying blister. They do not havethe three zones of typical target lesions (shown in Fig.2) and

generally are irregularly shaped and vary rn size.

i,t-Figure 4. Widespread Lesions Characteristic of Stevens-John-

son syndrome.The lesions are most heavily concentrated on the trunk and proxi-mal extremities. The darker areas are sites of epidermal necrosis.

. ' ':§.. .'*:læ*S8§

Figure 5. Purpuric Macules Typical ot Stevens-Johnson Syn-drome.

The macules may coalesce 1o torm blisters.

,"

1276 THE NEW ENGLANDJOURNAL OF IVIEDICINE Nov. 10, 1994

like sicca syndrome with a deficiency of mucin intears, inturned eyelashes, proliferation of squamousmetaplasia, and neovascularization of conjunctivaand cornea, symblepharon, punctate keratitis, andcorneal scarring may develop.30 Persistent photopho-bia, burning eyes, visual impairment, and even blind-ness may result. Other possible sequelae include scar-ring, irregular pigmentation, eruptive nevi, persistenterosions of the mucous membranes, phimosis, vaginalsynechiae, and abnormal regrowth of nails.rB

Differential Diagnosis

Skin disorders involving desquamation, exfoliation,or blistering are sometimes misdiagnosed as Stevens-Johnson syndrome or toxic epidermal necrolysis. Ex-foliative dermatitis is characterized by generalized

Figure 6. Cross Section of Epidermis and Upper Dermis of Nor-mal Human Skin.

The number 1 denotes stratum corneum, 2 stratum granulosum,3 stratum spinosum, 4 basal cells, and 5 dermis. ln toxic epider-mal necrolysis, the necrosis of cells from the basal layer andstratum spinosum results in detachment ol the epidermis lrom thedermis. Exfoliative dermatitis is characterized by increased thick-ness of the stratum corneum. ln staphylococcal scalded skin syn-drome and exanthematous pustulosis, detachment occurs be-tween the stratum granulosum and the stratum corneum or within

the stratum granulosum (hematoxylin and eosin, x400).

Figure 7. Necrolysis of Skin in Toxic Epidermal Necrolysis.

Varying degrees of erythema are seen. The wrinkled areas repre'sent full-thickness necrosis of the epidermis. This dead skin will

be lost, resulting in superficial skin ulcers.

Figure 8. Ulcerations and Erythema of the Oral Mucous Mem-branes and Lips Caused by Toxic Epidermal Necrolysis.

These findings can also be seen with erythema multiforme major,Stevens-Johnson syndrome, and primary bullous diseases such

as pemphigus vulgaris.

erythema and scaling (Fig. 9).3+ When the scales sepa-rate in large sheets, especially on the palms and soles,desquamation may be clinically confused with full-thickness epidermal detachment (Fig. l0).

In infants, staphylococcal scalded skin syndromemay resemble toxic epidermal necrolysis. Specificstaphylococcal exotoxins cause extensive subcornealseparation of the stratum corneum (Fig. 6 and I l).35Acute exanthematous pustulosis is drug-induced andresembles pustular psoriasis.36 The subcorneal asepticpustules are usually distinctive and may coalesce toproduce extensive superficial exfoliation (Fig. l2).The mucous membranes are infrequently involved.Subcorneal skin separation (Fig. 6) and the absence ofnecrosis in both conditions facilitate their pathologicaland clinical diagnosis.

Paraneoplastic pemphigus of acute onset may beconfused with toxic epidermal necrolysis.3T Direct im-

Vol. 331 No. l9 MEDICAL PROGRESS 1277

Figure 9. Exfoliative Dermatitis.

There is widespread, scaling, brawny erythema and desqua-mation.

munofluorescence microscopy can be used to distin-guish these disorders. Thermal burns, phototoxic re-actions, and pressure blisters occurring in comatosepatients may resemble toxic epidermal necrolysis,even on pathological analysis. The pattern of the blis-ters and the clinical history facilitate proper diag-nosis.

Epidemiologic Features

Although infrequent, toxic epidermal necrolysis andStevens-Johnson syndrome occur in all ages, all races,and both sexes, with an incidence ranging from 0.4 to1.2 and 1.2 to 6 per million person-years, respective-ly. t,3,ls,:a-'«r

Most cases of toxic epidermal necrolysis are drug-induced. Fewer than 5 percent of patients report nodrug use.3'le A strong association with specific medica-tions is observed in about 80 percent of the cases.

Other occasional reported causes include chemicals,

mycoplasma pneumonia, viral infections, and immu-nizatlon.+t'a2 That there is a less frequent clear-cut re-Iation of drugs to Stevens-Johnson syndrome (in

about 50 percent of cases) probably reflects the com-mon confusion between this syndrome (Fig. 3 and 4)and erythema multiforme major (Fig. 2).

Drug-induced Stevens-Johnson syndrome andtoxic epidermal necrolysis typically begin one tothree weeks after the initiation of therapy but occurmore rapidly with rechallenge.rB More than 100 differ-ent compounds have been implicated in both syn-dromes.3, I B' te'33 3a'4o,42'43 Table 4 lists frequently impli-cated drugs. For all drugs, the reported reaction ratesare relatively low. The drugs with the highest estimat-ed incidence include co-trimoxazole (trimethoprim-sulfamethoxazole; I to 3 reactions per 100,000 us-ers),3'3e'a0 a long-acting combination of sulfadoxineand pyrimethamine (Fansidar-R; l0 reactions per100,000 users),4'23'aa and carbamazepine (14 reactionsper 100,000 users).45 These estimates, which werebased on retrospective series or spontaneous reports,may substantially underestimate the true incidence.

Patients often have underlying diseases. A role forinfection as a cofactor has been postulated, but there islittle supporting evidence.a3 Conditions that alter im-munologic function, including systemic lupus erythe-matosus, may increase risk.a6 The HLA phenotypeBl2 is associated with a threefold increase in risk.aT

Toxic epidermal necrolysis has been described inan animal model of cutaneous acute graft-versus-host disease (GVHD).48 Toxic epidermal necrolysishas developed in humans a few weeks after bonemarrow transplantation.4e'so In transplant recipientscutaneous necrolysis is most often related to acuteGVHD, but in some cases it is drug-induced.50,sr Ocu-lar lesions are rare in acute GVHD and frequentin drug-induced toxic epidermal necrolysis.ts'50'5rWhether drug-induced or related to acute GVHD,epidermal necrolysis after bone marrow transplanta-tion suggests a very poor prognosis.4e-sl

Figure '10. Superficial Blisters of the Feet Consisting of Sheets olthe Upper Epidermis in a Patient with a Severe Morbillilorm

Eruption.

This kind of desquamation, especially on the palms and soles,should not be conlused wilh the true full-thickness necrosis of

toxic epidermal necrolysis and Stevens-Johnson syndrome.

THE NEW ENGLANDJOURNAL OF MEDICINE Nov. 10,1994

Patients with the acquired immunodeficiency syn-drome have a higher incidence of many drug-inducedskin rashes, including Stevens-Johnson syndrome andtoxic epidermal necrolysis, with a combined incidenceof I per 1000 person-years.s2-ss Sulfonamides are themost frequently implicated agent. The risk of reac-tions to sulfonamides is l0 to 100 times higher amongpersons infected with the human immunodeficiencyvirus (HIV) than among other persons. This high riskreflects more frequent drug use and greater suscepti-bility.5a'55

Pathophysiology

Patients with Stevens-Johnson syndrome or toxicepidermal necrolysis induced by sulfonamides or anti-convulsant agents often have an alteration in thedetoxification of reactive drug metabolites.56's7 T'herecurrence of Stevens-Johnson syndrome and tox-ic epidermal necrolysis within 48 hours of rechal-lenge (although the initial reaction occurs about 14

days after treatment is begun) argues against a di-

Figure 11. Staphylococcal Scalded Skin Syndrome.

The patient has ÿpical periorilicial efihema and crusting, aSwell as superiicial peeling and erosions of the upper epider-mis with indistinct underlying erythema. The clinical picture

is similar to that seen with a superticial thermalscalding.

Figure '12. Acute Exanthematous Pustulosis.

Small pustules are seen on erythematous skin. Gonfluent pus-tules may produce superficial erosions and sometimes blisters.

rect toxic effect and is more consistent with immu-nologic mechanisms.rs

The immunopathologic pattern of early lesionssuggests a cell-mediated cytotoxic reaction againstepidermal cells.ss-6r The epidermis is infiltrated byactivated lymphocytes, mainly CD8 cells, and macro-phages.s-6r An immune reaction against drug-reactivemetabolites produced in excess may be responsible.Because infiltrating cells are present in only mod-erate numbers, it is unlikely that these cells are theprincipal cause of epidermal necrosis. Cytokines, re-leased by activated mononuclear cells and keratino-cytes, may contribute to local cell death, fever, andmalaise.

Prognosls and Treatment

Mortality rates are below 5 percent for Stevens-Johnson syndrome but about 30 percent for toxic epi-dermal necrolysis.rs'le Sepsis is the principal cause ofdeath. More extensive epidermal detachment, in-creased age, increased blood urea nitrogen concentra-tions, and visceral involvement indicate a poorer prog-nosis. The prognosis does not appear to be affected bythe type and dose of the responsible drug or the pres-ence of HIV infection.

The physician is responsible for the early recogni-tion of the reaction, the withdrawal of all poten-tially responsible drugs, and the initiation of intrave-nous-fluid replacement. Although some drugs areclearly more often responsible than others (Table4), all drugs, especially those introduced within onemonth of the reaction, should be considered suspect.Patients with widespread skin involvement shouldbe transferred to an intensive care unit or burnunit. During transfer, pain control, fluid replacement,aseptic handling, and avoidance of any adhesivematerial are important. The main principles of thera-py are the same as for thermal burns, including ag-gressive fluid replacement, nutritional support, andantibacterial treatment.62,63

\/ol. 3:ll \o. 19

Table 4. Drugs Associated with Stevens-Johnson Syndrome and Toxic Epidermal

Necrolysis.

DRUCS MosrFRteutNr I-Y AssmlATED*

Sullàdoxine

Sulfadiazine

Sulfasalazine

Co-trimoxazole

Hydantoins

Carbamazepine

Barbilurates

Benoxaprofeni

Phenylbutazone

Isoxicami

Piroxicam

Chlomezanone

Allopurinol

Amithiozone

Aminopenicillins

DRUcs ALsoASSoCIATED

Cephalosporins

Fluoroquinolones

Vancomycin

Rrfampin

Ethambutol

Fenbufen

Tenoxicam

Tiaprofenic acid

Diclolenac

Sulindac

Ibuprofen

Ketoprofen

Naproxen

Thiabenduole

+Together these drugs account for approximatcly two thirds ofthe cases attributed to drugs in lage series in France, Gemany.and the Uni(ed States. l l 19 ll ll

iThis drug is no longer markete<.l.

Many interventions meant to halt the progressionof toxic epidermal necrolysis have been tried, eachin a few patients. A positive result, usually defined as

one that halts the spread of necrolysis, has typical-ly been noted after several previous "ineffective"treatments. However, in untreated patients, the aver-age duration of progression is less than four days.Therefore, the results ol these uncontrolled studiescannot be interpreted. Short courses of corticoster-oids early in the disease have been advocated,Ga buttheir e{fectiveness has never been demonstrated incontrolled trials. Toxic epidermal necrolysis can de-velop in patients who are receiving hish-dose corti-costeroids.3'65 Retrospective studies demonstrate nobenefit ol corticosteroids or higher rates of morbidityand mortality in corticosteroid-treated patients.66-68We recommend against their use. Case reports claim-ing that plasmapheresis, cyclosporine, cyclophospha-mide, and monoclonal antibodies directed againstcytokines are helpful should be resarded with skepti-cism. s9,6e.70

Because these disorders progress so rapidly, manycases have evolved lully before the patients are hospi-talized, thus limiting the practical value of such treat-ments. Therefore, therapies that reduce morbidity as-sociated with skin loss or accelerate regrowth of theskin are the most promising.

ff yprnsrNsr:rrvrry SyNonoun

A variety of hypersensitivity responses are responsi-ble for most cutaneous reactions to drugs. The term"hypersensitivity syndrome" refers to a specific severeidiosyncratic reaction. The syndrome typically in-cludes skin rash and fever, often with hepatitis, ar-thralgias, l,vmphadenopathy, or hematologic abnor-

I 279

malities (Tables I, 2, and 3). Perhaps because of itsrelatively late onset, slow evolution, and clinical simi-larity to many infectious illnesses, the diagnosis ofhypersensitivity syndrome may be delayed.

The aromatic antiepileptic agents (phenytoin, car-bamazepine, and phenobarbital) - with an estimatedincidence of I reaction per 5000 patients and per-haps a higher rate among black patienl5 - 2nd strl-fonamides are the most frequent causes of hyper-sensitivity syndrome.56'57'7r-76 Other drugs, especiallyallopurinol, gold salts, dapsone, and sorbinil, are alsoassociated with the syndrome.TT-80 Hypersensitivitysyndrome may be di{ficult to distinguish from serumsickness or drug-induced vasculitis. Laboratory find-ings often help distinguish these clinicall,v similar con-ditions from each other and from infectious diseases(Table 2).

The hypersensitivity syndrome t,vpicallv developstwo to six weeks after a drug is first used, later thanmost other serious skin reactions (Table 3). With anti-epileptic drugs, fever and rash are the most frequentpresenting symptoms (in B7 percent of cases). Lt'm-phadenopathy (in about 75 percent) is frequent andusually due to benign lymphoid h,vperplasia.ri Atvpi-cal lymphoid hyperplasia and pseudolymphoma occa-sionally occur.sr Some ol these cases resolve with with-drawal of the drug, but in some cases lymphomaeventually develops.B2 Hepatitis (51 percent); intersti-tial nephritis (l I percent); hematologic abnormalities,especially eosinophilia (30 percent); and mononucleo-sis-like atypical lymphocvtosis are also common.l' In-volvement of the heart, lung, thyroid, and brain is lessfrequent.rT'83 Severe cases of hepatitis may be life-threatening.Br

A genetically determined inability to detoxify thetoxic arene oxide metabolic products of anticonvul-sant agents has been observed in patients with thehypersensitivity syndrome, but the syndrome also oc-curs in patients without this abnormality.rT85 Cellsfrom the parents of affected patients have a degree ofin vitro sensitivity to these toxic metabolites that isintermediate between that of a{fected patients andthat of controls.rT Positive tests have been noted inmultiple family members.s6 Cross-sensitivity betweenthe various aromatic antiepileptic drugs is well docu-mented, making it dilficult to select alternative anti-convulsant therapy.BT'88

Rashes of all types are reported with carbamaze-pine or phenytoin therapy.T3Be Most o[ these rash-es are morbilliform (Fig. l) and will abate even ifthe drug is continued. Unfortunately, the hypersen-sitivity syndrome often initially presents as a mor-billiform eruption indistinguishable from less seriousreactions (Fig. 1). The reaction may become indu-rated and infiltrated (Fig. l3). Any cutaneous reac-tion associated with aromatic anticonvulsant agentsthat includes facial swelling, exfoliative dermati-tis (Fig. 9), fever, lymphadenopathy, eosinophilia, ar-thritis, hepatitis, or bullous or purpuric skin lesions

NIEDICAI, PROGRESS

Vol. 331 \o. l9 X,{[,DICAI. PROGRESS 1279

Table 4. Drugs Associated with Stevens-Johnson Syndrome and Toxic Epidermal

Necrolysis.

DRUGs It{osrI_REquri\TLY Ass(rATLr)'

Sultadoxine

Sulfàdiazine

Sulfasalazine

Co-trimoxazole

Hydantoins

Carbamazepine

Barbiturates

Benoxaprofenf

Phenylbutazone

Isoxicami

Piroxicam

Chlomezanone

Allopurinol

Amithiozone

Aminopenicillins

DRrr(is At soAss6tATED

Cephalosporins

Fluoroquinolones

Vancomycin

Rifampin

Ethambutol

Fenbufen

Tenoxicam

Tiaprofenic acid

Diclofenac

Sulindac

Ibuprofen

Ketoprolen

Naproxen

Thiabendazole

*Togcther these drugs account for approximately two thirds ofthe cases attributed to drugs in large series in France. Gemany,and the United state(.l.l'le'll 4l

ïThis drug is no lonqer marketed.

Many interventions nleant to halt the progressionof toxic epidermal necrolysis have been tried, eachin a few patients. A positive result, usually defined asone that halts the spread of necrolysis, has typical-ly been noted after several previous "ineffective"treatments. However, in untreated patients, the aver-age duration of progression is less than four days.Therefore, the results of these uncontrolled studiescannot be interpreted. Short courses of corticoster-oids early in the disease have been advocated,Ga buttheir effectiveness has never been demonstrated incontrolled trials. Toxic epidermal necrolysis can de-velop in patients who are receiving high-dose corti-costeroids.3'6'" Retrospective studies demonstrate nobenefit of corticosteroids or higher rates ol morbidityand mortality in corticosteroid-treated patients.66-68We recommend against their use. Case reports claim-ing that plasmapheresis, cyclosporine, cyclophospha-mide, and monoclonal antibodies directed againstcytokines are helpful should be regarded with skepti-cism.se,6s.70

Because these disorders progress so rapidly, manycases have evolved fully belore the patients are hospi-talized, thus limiting the practical value of such treat-ments. Therefore, therapies that reduce morbidity as-sociated with skin loss or accelerate regrowth of theskin are the most promising.

HyprnsrNsITrvrrY SyN»notr.r

A variety of hypersensitivity responses are responsi-ble for most cutaneous reactions to drugs. The term"hypersensitivity syndrome" refers to a specific severeidiosyncratic reaction. The syndrome typically in-cludes skin rash and fever, often with hepatitis, ar-thralgias, lymphadenopathy, or hematologic abnor-

malities (Tables l, 2, and 3). Perhaps because of itsrelatively late onset, slow evolution, and clinical simi-larity to many infectious illnesses, the diagnosis ofhypersensitivity syndrome may be delayed.

The aromatic antiepileptic agents (phenytoin, car-bamazepine, and phenobarbital) - with an estimatedincidence of I reaction per 5000 patients and per-haps a higher rate among black patients - and sul-fonamides are the most lrequent causes of hyper-sensitivity syndrome.s6'i7'7r-76 Other drugs, especiallyallopurinol, gold salts, dapsone, and sorbinil, are alsoassociated with the syndrome.TT-80 Hypersensitivitysyndrome may be difficult to distinguish lrom serumsickness or drug-induced vasculitis. Laboratory find-ings often help distinguish these clinicalll' similar con-ditions lrom each other and lrom infectious diseases(Table 2).

The hypersensitivity svndrome tvpically developstwo to six weeks alter a drug is first used, later thanmost other serious skin reactions (Table 3). With anti-epileptic drugs, fever and rash are the most frequentpresenting symptoms (in B7 percent ol cases). l,ym-phadenopathy (in about 75 percent) is lrequent andusuall,v due to benign lymphoid hvperplasia.lT Atypi-cal lymphoid hyperplasia and pseudolymphoma occa-sionally occur.sr Some of these cases resolve with with-drawal of the drug, but in some cases lymphomaeventually develops.B2 Hepatitis (51 percent); intersti-tial nephritis ( I I percent); hematologic abnormalities,especially eosinophilia (30 percent); and mononucleo-sis-like atypical lymphocytosis are also common.rT In-volvement of the heart, Iung, thyroid, and brain is lessfrequent.ri'83 Severe cases of hepatitis may be life-threatening.B+

A genetically determined inability to detoxiÿ thetoxic arene oxide metabolic products ol anticonvul-sant agents has been observed in patients with thehypersensitivity syndrome, but the syndrome also oc-curs in patients without this abnormality.ri'B' Cellsfrom the parents of ailected patients have a desree ofin vitro sensitivity to these toxic metabolites that isintermediate between that of affected patients andthat of controls.rT Positive tests have been noted inmultiple family members.BG Cross-sensitivity betweenthe various aromatic antiepileptic drugs is well docu-mented, making it di{ficult to select alternative anti-convulsant therapy.B''B*

Rashes o[ all types are reported with carbamaze-pine or phenytoin therapy.T3'Bs lvlost of these rash-es are morbillilbrm (Fig. l) and will abate even ilthe drug is continued. Unfortunately, the hypersen-sitivity syndrome often initially presents as a mor-billiform eruption indistinguishable from less seriousreactions (Fig. l). The reaction may become indu-rated and infiltrated (Fig. l3). Any cutaneous reac-tion associated with aromatic anticonvulsant agentsthat includes facial swelling, exfoliative dermati-tis (Fig. 9), fever, lymphadenopathy, eosinophilia, ar-thritis, hepatitis, or bullous or purpuric skin lesions

I 280

Figure 13. lnfiltrative Papules Coalescing inlo Plaques.

These papules are ÿpical of the more advanced eruptions seen inthe hypersensitivity syndrome associated with aromatic antiepi-leptic drugs. The histologic appearance of these induraled conllu-ent papules and plaques is often similar to that of early stages of

cutaneous T-cell lymphoma.

or begins more than two weeks after therapy is ini-tiated is especially worrisome.

Sulfonamide-induced hypersensitivity syndromeand that induced by antiepileptic agents are clinicallyindistinguishable.16'57 Slow N-acetylation of sulfon-amide and increased susceptibility of patients' Ieu-kocytes in vitro to toxic hydroxylamine metabolitesare associated with greater susceptibility, but only asmall percentage of people who acetylate sulfona-mides slowly have reactions to these drugs.t6,sz,so

Recovery is usually total, but rash and hepatitismay persist for weeks. Treatment with corticosteroidshas been widely advocated, but controlled studies arelacking.er We have observed dramatic improvementsin symptoms and laboratory measurements in patientsgiven systemic corticosteroids (>0.5 mg per kilogramof body weight). Relapses of rash and hepatitis mayoccur as corticosteroids are tapered. Transient hypo-thyroidism may also develop.s2

Vescur,rrrs aND SERUM Srcxxess

Vasculitis characterized by inflammation and ne-crosis of blood-vessel walls has many causes.s3 Drug-induced vasculitis typically involves small vessels andis a subtype of hypersensitivity vasculitis,sa which alsoincludes cutaneous leukocytoclastic vasculitis and se-rum sickness.sa

In 1905, von Pirquet and Schick described serumsickness in children treated with horse serum contain-ing diphtheria antitoxin.es More recently, serum sick-ness has been noted in patients treated with horseantithymocyte globulins or human diploid-cell rabiesvaccine.es'e7 Serum sickness is a type III hypersensi-tivity reaction mediated by the deposition of immunecomplexes in small vessels, activation of complement,and recruitment of granulocytes. Drug-induced vas-culitis is believed to result from antibodies directedagainst drug-related haptens, but this has not been

THE NETV ENGLANDJOURNAL OF NTEDICI\E Nov. 10, 1994

proved.s8 Alternative proposed mechanisms includedirect drug toxicity against vessel walls, autoantibod-ies reacting with endothelial cells, and cell-mediatedcytotoxic reactions against vessels.e3'es-l0r

Clinical Presentation

Serum sickness has distinctive skin findings. Typi-cally, erythema first occurs on the sides of the fin-gers, toes, and hands, before a more widespreaderuption that is most often morbilliform (in two thirdsof patients), sometimes with urticaria,es-ei lJrticariais seldom seen alone. About half the cases of se-

rum sickness have visceral involvement. Rash, fever,constitutional symptoms, arthralgia, and arthritis arethe most frequent clinical findings.ss'e6

The clinical hallmark of cutaneous vasculitis is pal-pable purpuric papules, classically located on the low-er extremities, although any site may be involved (Fig.l4).e3'r02'103 Hemorrhagic blisters, urticaria, ulcers,nodules, Raynaud's disease, and digital necrosis mayalso occur. The same vasculitic process may also affectthe kidney, liver, gastrointestinal tract, or neryoussystem and can be life-threatening. Histologically,small dermal vessels exhibit fibrinoid necrosis, infil-tration by polymorphonuclear leukocytes, and nucleardust.lo3 The results of direct immunofluorescence areoften positive, with deposits of IgM and C3 comple-ment on capillary walls.ro3

In serum sickness, serum C3 and C4 complementlevels are markedly decreased.e6 Serum sickness be-gins B to l4 days after the initial exposure to a foreignprotein. Other kinds of drug-induced vasculitis typi-cally develop 7 to 2l days after a new drug is begun,but the interval can be longer.ee When otherwise unex-plained palpable purpura develops in a patient, anydrug the patient is taking, especially those inroducedwithin the preceding two months, should be consid-ered suspect. Withdrawing the drug usually leads torapid resolution. Systemic corticosteroids may benefitsome patients.

Diflerential Diagnosls

Drug-induced hypersensitivity vasculitis may bedifficult to distinguish from other types of vasculitis.Schônlein-Henoch purpura usually occurs in youngerpatients, with characteristic large purpuric cutaneouslesions, often on the buttocks. Renal and gastroin-testinal involvement is common. IgA is depositedin vessels.roa Cryoglobulinemia-associated vasculitishas a chronic or recurrent course. Polyarteritis no-dosa and Wegener's granulomatosis sometimes beginas a palpable purpura.to5 Most patients with Weg-ener's granulomatosis have autoantibodies to neutro-phil cytoplasmic antigens,106 a feature that is usuallyabsent in drug-induced vasculitis. Infection and colla-gen vascular disorders can also induce vasculitis.e3Excluding infection as a cause is often the great-est challenge. Drugs cause about l0 percent of casesof acute cutaneous vasculitis. 102,103

Only a small fraction of drug reactions take the

Vol. 331 No. 19 MEDICAL PROGRESS

Figure 14. Drug-lnduced Vasculitis Presenting as Palpable Pur-puric Papules and Plaques, Occasionally with Overlying Small

Blisters, Especially on the Lower Extremities.

form of vasculitis.Ts Propylthiouracil may induce aclinically distinctive vasculitis initially involving theface and ear lobes, with erythema and later purpu-ra.r07,r08 Antinuclear antibodies and antineutrophilcytoplasmic antibodies may be produced.roe,rr0

Table 5 lists drugs that are often implicated in caus-ing vasculitis. Recently reported drugs associated withvasculitis include the retinoids and quinolones andagents used in immunotherapy.t I t-t ts

Reactions resembling serum sickness (rash, fever,and arthralgias) occur in about I of 2000 childrengiven celaclor and have also been reported with mino-cycline, penicillins, propranolol, streptokinase, andother drugs.r14-ll8 Since reduced concentrations of se-rum complement are not generally noted, most suchcases probably do not represent rue serum sickness.

ANrrconcureNr-INoucr» SxrN Nncnosrs

A rare and devastating effect of warfarin therapy isskin necrosis, a consequence of occlusive thrombi invessels of the skin and subcutaneous tissue.lls Typi-cally, warfarin-induced skin necrosis begins three tofive days after therapy is initiated. The use of higherinitial doses, obesity, and female sex appear to in-crease the risk.r20 Red, painful plaques evolve to ne-

crosis (Fig. l5), with hemorrhagic blisters or necroticscars, frequently in areas with large quantities ol adi-pose tissue, including the breasts, hips, and buttocks.Acral involvement is infrequent.

People with hereditary deficiency of protein C, anatural anticoagulant protein, are at highest risk, evenif they are heterozygotes and thus have no history ofrecurrent thrombosis.lls-122 In these persons, warfaringreatly depresses protein C levels before decreasingother vitamin K-dependent coagulation factors, in-ducing a transient hypercoagulable state and throm-bus formation.rls Rapid recognition of painful, redplaques in fatty areas is the key to diagnosis. Ther-apy includes discontinuing warfarin, administerinevitamin K to reverse the effect of warfarin, giving hep-arin as an anticoagulant, and administering mono-clonal antibody-purified protein C concentrate.r23Necrotic tissues may require surgical débridementand grafting. If not rapidly treated, this condition maybe fatal. It develops in I in 10,000 patients receivingwarfarin, a prevalence that is about 2 percent of theestimated prevalence of protein C deficiency.lls,l2aSince most persons with protein C deficiency toleratewarfarin, other factors must play a part. Protein S orantithrombin III deficiency also confers an increasedrisk.r25

Heparin can also cause thrombosis and necrosis inthe skin and other organs.r26 The mechanisms of hepa-rin-induced and warfarin-induced necrosis are almostcertainly different. Heparin can induce vessel throm-bosis with fibrin thrombi at injection sites and distantskin sites and in other organs.l2T'r2B Localized reac-tions at injection sites are frequent, but devastatingwidespread reactions are not. Heparin-induced plate-let aggregation may be responsible for widespread re-actions. These lesions need to be dilferentiated fromother cutaneous reactions to heparin at injection sites,which are most likely immunologic.l26'r27 Neither pro-

Table 5. Agents Most Otten Associated withVasculitis, Serum Sickness, and Reactions

Resembling Serum Sickness.

Vasculitis

AllopurinolPenicillinAminopenicillinsSulfonamidesThiazidesPyrazolonesHydantoinsPropylthiouracil

Raynaud's disease or digital necrosis

Beta-blockersErgot alkaloidsBleomycin

Serum sickness

Serum preparationsVaccines

Reactions resembling serum sickness

Beta-blockersStreptokinasep-Lactam antibiotics

r 282

q.",

Figure 1 5. Warfarin-lnduced Necrosis.ln this woman, painful erythema and induration of the breasts

were followed by necrosis of these tatty areas.

tein C nor protein S piavs a part. ln heparin-inducednecrosis, levels of fibrinogen and fibrin-split productsare usualll' normal. but platelet counts are often de-pressed.r26 Evidence of primar,v vasculitis is iacking.Heparin-induced thrombocytopenia and thrombosismav be an immune-complex disorder.rrlr In additionto discontinuation of the drug, treatment rvith lvar-farin or antiplatelet drugs is useful.126

ANcrororuaImmediate-hypersensitivitv reactions can produce

a range of cutaneous findings from simple urticaria toangioedema or anaphvlaxis. The mechanism andtreatment of IgE-mediated immediate-hvpersensitir-ity reactions inciuding anaphylaxis, which are mostoften induced bv insect stings and food, have beenreviewed recentlv.l30'lrlr Many drug-induced cases ofangioedema are nor mediated b,v IgE. \\'e shail brieflydiscuss newer drugs that cause angioedema or ana-phylaxis.

Antibiotics (especiallv the penicillins), anesthetics,and radiocontrast aeents are the most common caus-es of serious IgE-mediated. drug-induced immediateht'persensitivity.r:o'tst Ansioedema occurs in aboutI per I0,000 courses of penicillin and leads to death inI to 5 per 100.000 courses. In persons receiving long-term penicillin prophvlaxis for rheumatic fever. therisk of angioedema persists during treatment.r32

Other frequently used drugs, including angiotensin-converting-enzvme (ACE) inhibitors, nonsteroidalantiinflammatory drugs, radiocontrast agents, opi-ates, and curare, cause angioedema that is not IgE-mediated. ACE inhibitors induce the majority ofcases of ansioedema that lead to hospitalization.l33'l3rThe observed incidence of drug-related angioedemahas increased in parallel w'ith the increased use ofACE inhibitors, especiallv ionger-acting ACE inhibi-tors.13r-136 Angioedema occurs in 2 to l0 per 10,000

THE \E\\' E\GLÀ\D JOUR);AL OF \TEDICI\E \ov. I0. 199.1

new users of ACE inhibitors - a rate that is probablyhigher than that associated with penicillins.r3i Therisk is highest during the first three weeks of thera-py.137 These reactions ma-v be due to the inhibition ofkinin metabolism.l3s Hemodialysis'o,ith high-flux di-alysis membranes, which may increase the productionof bradvkinin, greatlv increases the risk of anaphviac-toid reactions associated with ACE inhibitors.l3e-r+lReactions occur in up to 35 percent of patients treatedin this manner.l+1

CoNcr-usroNs

Adverse reactions to drugs most often affect theskin, but onlv a small fraction are life-threatening orlead to disabling sequelae. Because of the low frequen-cy of such severe reactions (usually less than I reac-tion per 5000 exposed patients), they are unlikely tobe detected in premarketing clinical trials. Only ifclinicians recognize and report severe reactions toregulatory authorities and manufacturers can ne'ô'drugs associated rvith a high risk of such reactionsbe identified, relabeled, or r,','ithdrawn from the mar-ket. r42.143

For many severe cutaneous reactions to drugs, in-cluding toxic epidermai necrolvsis, Stevens-Johnsonsyndrome, vasculitis, and serum sickness, medical in-tervention is limited to the earlv recognition of thesymptoms and the withdrawal of the offending drug.Even for other reactions that mav benefit from thera-py, early recognition of the symptoms and promptrvithdrau'al ol suspect drugs are usuallv the most im-portant steps. Therefore, clinicians should carefullyevaluate the signs and svmptoms of all adverse cuta-neous reactions thought to be due to drugs and imme-diately discontinue all drugs that are not essential,especialll'when the signs or svmptoms associated withmore severe reactions are present (Table 2). After re-cover)', patients should be advised to avoid the drugthought to be responsible for the reaction and allchemically related compounds. Patients with toxicepidermal necrolysis and hypersensitivitl, svndromeshould alert their first-degree relatives to their ele',,at-ed risk of such reactions to the same drugs.

RrrrnnNcr,s

l. Chan HL. Stern RS, Arndt KA. et al. The incidence of erythema multi-forme, Stevens-Johnson syndrome, and toxic epidemal necrolysis: a popu-lation-based study with panicular reference to reactions caused by drugs

among oùtpatients. Arch Demratol 1990:126:43-7 .

2. Cone KA. Spielberg TE. Adverse drug reaction processing in the UnitedStates and its dependence on physician reporting: zomepirac (Zomax) as acase in point. Ann Emerg Med 1988:17:1,15-9.

3. Roujeau JC, Guillaume JC, Febre JP. Penso D. Flechet ML, Gine JP.Toxic epidemal necrolysis (Lyell syndrome): incidence and drug etiologyin France. 1981-1985. Arch Dermatol 1990 12637-42.

4. Miller KD, Lobel HO, Satriale RF. Kuritsky JN, Stem R, Campbell CC.Severe cutaneous reactions among American trâvelers üsing pyrirretha-mine-sulfadoxine (Fansidar) for malaria prophylaxis. Am J Trop Med Hyg1986:35:45 1 -8.

5. Bigby M, Jick S. Jick H, Amdt K. Drug-induced cutaneous reactions: a

report from the Boston Collaborative Drug Surueillance Program on15,438 consecutive inpatients. 1975 to 1982. JAMA 1986;256:3358-63.

6. Leape LL, Brennan TA, Laird N, et al. The nature of adverse events inhospitalized patients: results of the Harvard Medical Practice Study ll.N Engl J Med 1991t324:37'1-84.

Vol. 331 No. li)

Alanko K. Stubb S. Kauppinen K. Cutaneous drug reactions: clinical typesand causative agents: a nve-year suruey of in-patients (1981-1985). ActaDem Venereol (Stockh) 1989:69:223-6.lves TJ. Bentz EJ. Gwyther Rt. Dematologic adverse drug reactions in afamily medicine setling. Arch Fam Med 1992;1:241-5.Varga J, Vitto J. Jimenez SA. The cause and pathogenesis of the eosino-philia-myalgia syndrome. Ann Intem Med 1992;ll6:140-7.Kramer MS. Leventhal JM, Hutchinson TA, Feinstein AR. An algorithmtbr the operational assessment of adverse drug reactions. L Background,description, and instructions for use. JAMA 1979:242:623-32.Tumer WM. The Fmd and Drug Administration algorithm: special work-shop

- regulatory. Drug InfJ 1984:18:259-66.Louik C. Lacouture PG. Mitchcll AA. ct al. A study of adverse reactionalgorithms in a drug suneillance program. Clin Phmacol Ther 1985;38:183-7.Ressler C, Mendelson LM. Skin test lbr diagnosis of penicillin allergy -cunent status. Ann Allergy 1987l,59:16'7-10.Bruynzeel D. van Ketel W. Skin tests in the diagnosis of maculo-papulardrug eruptions. Scmin Dematol 1987l.6:l19-24.Shea NH. Diagnosing cutaneous adverse reactions to drugs. Arch Dema-tol I990:126:94-7.Rieder MJ. Uetrecht J. Shear NH. Cannon M. Miller M, Spielberg SP.Diagnosis of sulfonamide hypersensitivity reactions by in-vitro "rechal-lenge" with hydroxylamine metabolites. Ann lntem Med l989ll l0:286-9.Shear NH. Spielbcrg SP. Anticonvulsant hypereensitivity syndrome: invitro assessment of risk. J CIin Invest 1988;82:1826-32.Revuz J, Penso D. Roujeau JC. et al. Toxic epidermal necrolysis: clinicalfindings and prognosis factors in 87 patients. Arch Demtatol 1987l.123:I 160-5.

Schôpf E. Stühmer A, Rzany B. Victor N, Zentgraf R. Kapp JF. Toxicepidemal necrolysis and Stevens-Johnson syndrome: an epidemiologicstudy from West Gemany. Arch Dermatol l99l:127:839-42.Bastuji-Garin S, Rzany B. Stem RS, Shear NH, Naldi L. Roujeau JC.Clinical classifrcation of cases of toxic epidermal necrolysis, Stevens-John-son syndrome. and crythema multifome. Arch Dermatol 1993;,129:92-6.Stevens AM. Johnson FC. A new cruptive tèver associated with stomatitisand ophthalmia: report of two cases in childrcn. Am J Dis Child1922.24:526-33.Lyell A. Toxic epidemal necrolysis: an eruption resembling scalding oftbe skin. Br J Dematol 1956:6ll:355-61.Bergoend H. Lôffler A. Amu R, Maleville J. Réactions cutanées surue-nues au cours de la prophylaxie de masse de la méningite céÉbro-spinalepar un sulfamide long-retard (à propos de 997 cas). Ann Dermarol Syphi-ligr (Ptris) 1968;95:481 -90.Ruiz-Maldonado R. Acute disseminated epidermal necrosis types l, 2, and3: study of sixty cases. J Am Acâd Dcmatol 1985;13:623-35.Huff JC. Weston WL, Tonnesen MG. Erythema multiforme: a criticalreview of characteristics. diagnostic criteria, and causes. I Am Acad Der-matol 1983:8:763-75.Roupe C. Ahlmen M, Fagerberg B. Suurkula M. Toxic epidemal necroly-sis with extensive mucosal erosions of the gastrointestinal and respiratorytracts. Int Arch Allergy Appl lmmunol 1986180:145-51.Timsit JF, Mion C. Rouyer N, [æ Gulluche Y, Cusin H. Bronchopulmo-nary distress associated with toxic epidermal necrolysis. Intensive CareMed 1992:18:42-4.Chosidow O. Delchier J-C. Chaumette MT. et al. Intestinal involyemenr indrug-induced toxic epidermal necrolysis. Lancet l99l :337:928.Wilkins J, Morison L, White CR Jr. Oculæutaneous manifestations of theerythema multiforme/Stevens-Johnson/toxic epidermal necrolysis spec-trum. Dermatol Clin 1992; l0:571-82.Binaghi M. Kosos M, Saiag P. Roujeau JC, Coscas G. Atteinte oculaire aucours du syndrome de Lyell: incidence. évolution, pronostic. Ophtalmolo-gie l9E8;2:l2t-2.Prendiville JS. Hebert AA. Greenwald MJ. Esterly NB. Management ofStevens-Johnson syndrome and toxic epidemal necrolysis in children.J Pediatr 1989:l l5:881-7.Westly ED. Wechsler HL. Toxic epidermal necrolysis: granulocytic leuko-penia as a prognostic indicator. Arch Dematol 1984;12O:721-6.Jordan MH, Lewis MS, Jeng JG, Rces JM. Treâtment of toxic epidemalnecrolysis by bum units: another market or another threat'l J Bum CileRehabil l99l ;12:579-81.Marks J. Ervthroderma and its management. Clin Exp Dermatol 198217:4t5-22.Melish ME. Glasgow LA. The staphylææcal scalded-skin syndrome:development of an extrnrimental model. N Engl J Mcd 1970;282:l I l4-9.Roujeau JC. Bioulac-Sage P, Bourseau C, et al. Acute generalized exan-thematous pustulosis: analysis of 63 cases. Arch Dermatol 199|.127:I 333-8.Mutasim DF, Pelc NJ, Anhalt GJ. Paraneoplastic pemphigus. DematolClin l993lll:473-81.Boniger LE. Strandberg l, Westerholm B. Drug-induced febrile mucocuta-neous syndrome with a survey of the literature. Acta Med ScandI 975:198:229-33.

I 283

39. Naldi L, Læati F. Marchesi I, Cainelli T. Incidence of toxic epidemalnecrolysis in Italy. Arch Dermatol 1990; I 26: I 103-4.

40. Strom BL. Cmon JL. Halpem AC, et al. A population-bæed study ofStevens-Johnson syndrome: incidence and antecedenl drug exposures.Arch Dematol l99l : I 27:831-8.

41. House RA, Jakubovic H, Wong L. Holness DL. Work-related toxic epider-mal necrolysis? J Occup Med 199234:135-9.

42. Nethcrcoû JR, Choi BC. Erythema multiforme (Stevens-Johnson syn-drome)

- chart review of 123 hospitalized patients. Dematologica

1985:l7l:383-96.43. Stem RS, Chan HL. Usefulness of case repon litemture in detemining

drugs responsible for toxic epidemal necrolysis. J Am Acad Dermatol1989;21:317-22.

,14. Hellgren U, Rombo L, Berg B, Cillson J, Wilholm BE. Adverse reactionsto sulphadoxine-pyrimethamine in Swedish travellers: implications for pro-phylaxis. BMJ 1981 ;295:365-6.

45. Askmark H, Wiholm B-8. Epidemiology of adversc reactions to crbamz-epine as seen in a spontâneous reporting system. Acta Neurol Scand1990:81:l3l-40.

46. Burge SM. Dawber RPR. Stevens-Johnson syndrome and toxic epidemalnecrolysis in a patient with systemic lupus erythematosus. J Am AcadDermatol 1985;13:665-6.

47. Roujeau JC, Huynh Tn-, Bracq C, Guillaume JC, Revuz J, Touraine R.Genetic susceptibility to toxic epidemal necrolysis. Arch Dermarol1987;t23:l l7l-3.

48. Billingham RE, Streilein JW. Toxic epidemal necrolysis and homologousdisease in hamsters. Arch Dematol 1968:98:528-39.

49. Peck GL, Elias PM, Graw RG Jr. Graft-versus-host reaction and toxicepidermal necrolysis. Lancet 1972;2:l l5l-3.

50. Villada G, Roujeau JC. Cordonnier C, et al. Toxic epidemal necrolysisafter bone mmow transplantation: study of nine cases. J Am Acad Derma-tol 1990:23:870-5.

5 I . McDonald BJ, Singer JW, Bianco JA. Toxic epidermal necrolysis possiblylinked to aztreonam in bone marow transplant patients. Ann Phmacother1992:26:34-5.

52. Cmpman SA, Johnson RA, Platt R, Stern RS. Cutaneous disease and drugreactions in HIV infection. N Engl J Med 1993t328:1670-4.

53. Porteous DM. BergerTC. Severe cutaneous drug reactions (Stevens-John-son syndrome and toxic epidemal necrolysis) in humm immunodeficiencyvirus infection. Arch Dematol l99l :127 t1 4O-1.

54. Saiag P. Caumes E. Chosidow O, Revuz J, Roujeau JC. Drug-inducedtoxic epidermal necrolysis (Lyell syndrome) in patients infected with thehuman immunodeficiency virus. J Am Acad Dematol 1992,26:567-74.

55. Rzany B, Mockenhaupt M, Stocker U, Hamouda O. Schopf E. Incidenceof Stevens-Johnson syndrome and toxic epidermal necrolysis in patientswith the acquired immunodeficiency syndrome in Germany. Arch Derma-tol 1993;129:1059.

56. Spielberg SP, Gordon GB. Blake DA, Goldstein DA. Herlong HF. Predis-position to phenytoin hepatotoxicity assessed in vitro. N Engl J Medl98l'.305:722-7 .

57. Shear NH, Spielberg SP. Grant DIr{, Tang BK, Kalow W. Differences inmetabolism of sulfonamides predisposing to idiosyncratic toxicity. Annlntem Med 1986:105:179-84.

58. Merot Y. Gravallese E, Guillen FJ, Murphy GF. Lymphocyte subsets andLangerhans' cells in toxic epidemal necrolysis: report of a case. ArchDermatol 1986: I 22:455-8.

59. Heng MC, Allen SG. Efficacy of cyclophosphamide in toxic epidemalnecrolysis: clinical and pathophysiologic aspects. J Am Acad Dematoll99l:25:778-86.

60. Villada G. Roujeau JC, Clerici T, Bourgault I, Revuz J. lmmunopathologyof toxic epidermal necrolysis: keratinocytes, HLA-DR expression, Langer-hans cells, and mononucler cells: an immunopathologic study of fivecases. Arch Dematol 1992;128:50-3.

61. Coneia O, Delgado L. Ramos JP, Resende C, Toninha JA. CutaneousT-cell recruitment in toxic epidemal necrolysis: funher evidence ofCD8+ lymphocyte involvement. Arch Dematol 1993;129:466-8.

62. Heimbach DM, Engrav LH. Mæin JA, Hamar TJ. Grube BJ. Toxicepidemal necrolysis: a step forwrd in treatment. JAMA 1987;257:2171-5. [Enatum, JAMA 1987;258:1894.]

63. Demling RH. Bums. N Engl J Med 1985;313:1389-98.64. Sheretz EF. Jegasothy BV. Lazms GS. Phenyroin hypersensitivity reac-

tion presenting with toxic epidermal necrolysis and severe hepatitis: reponofa patient treated with corticosteroid "pulse therapy." J Am Acad Derma-rol 1985:12:178-81.

65. Rzany B, Schmitt H, Schôpf E. Toxic epidermal necrolysis in patientsreceiving glucæonicosteroids. AcIa Dem Venereol (Stockh) l99l:71:t7 t-2.

66. Kim PS, Goldfrb lW, Gaistbrd JC, Slater H. Stevens-Johnson syndromeand toxic epidemal necrolysis: a pathophysiologic review with recommen-dations for a treatment protocol, J Burn Cre Rehabil 1983;4:91-100.

67. Halebian PH. Corder VJ. Madden MR, Finklestein JL. Shires GT. Im-proved bum center survival of patients with toxic epidemal necrolysismanaged without conicosteroids. Ann Surg 1986:204:503- I 2.

ME,DICAL PROGRESS

20

2t

7.

6.

9.

10.

l3

t4

l5

l6

t7

t8

ll

t2

22.

23.

24

25

28.

29.

32.

33.

34.

35.

16.

26.

27

30

3l

31 .

38.

I 281

Weston WL. Oranje AP, Rasmussen JE. et al. Corticosteroids for erythe-ma mültifome? Pediatr Dermatol 1989:6:229-50.Kamanabroo D, Schmitz-Landgraf W, Czmetzki BM. Plasmapheresis insevere drug-induced toxic epidemal necrolysis. Arch Dermatol 1985:l2l:I 548-9.Renfro L, Grant-Kels JM, Daman LA. Drug-induced toxic epidemal ne-crolysis treated with cyclosporin. Int J Dematol 198928:441-4.Pelekanos J, Camfreld P, Camfield C, Gordon K. Allergic rash due toantiepileptic drugs: clinical features and mânagement. Epilepsia l99l;32:554-9.Pellock JM. Carbamazepine side effects in children and adults. Epilepsia1987;28:Suppl 3:S64-S70.Rapp RP, Norton JA, Young B, Tibbs PA. Cutaneous reactions in head-injured patients receiving phcnytoin for seizure prophylaxis. Neurosurgery1983:13:272-5.Temkin NR, Dikmen SS, Wilensky AJ. Keihm J, Chabal S. Winn HR. Arandomized. double-blind study of phenytoin for the prevention of post-traumatic seizures. N Engl J Med 1990l'323:497-502.Hmda F. Phenytoin hypersensitivity: 38 cases. Neurology 1979129:1480-5.Livingston S, Villamater C, Sakata Y. Pauli LL. Use of crbamazepine inepilepsy: results in 87 patients. JAMA 1967:200:2M-8.Lupton GP, Odom RB. The allopunnol hypersensilivrty syndrome. J AmAcad Dermatol 1979:.l :365-7 4.Sene H, Sany J, Rosenberg F. Les effets secondaires de la chrysothérapie.Ann Med Inteme (Paris) 1916l,127:552-9.Tomecki KJ, Catalano CJ. Dapsone hypersensitivity: the sulfone syndromerevisited. Arch Dematol 1981:l l7:38-9.Spielberg SP. Shear NH, Cannon NI. Hutson NJ, Gunderson K. In-vitroassessment of a hypersensitivity syndrome associated with sorbinil. AnnIntem Med l99l:1 l4:720-4.Rijlaarsdam U, Scheffer E. Meijer CJ, Kruyswijk MR, Willemze R. My-cosis fungoides-like lesions associated with phenytoin and carbamazepinetherapy. J Am Acad Dermàtol 199l|.24:216-20.Rosenthal CJ. Noguera CA, Coppola A, Kapelner SN. Pseudolympho-ma with mycosis fungoides manifestations, hyperesponsiveness to di-phenylhydantoin. and Iymphocyte disregulation. Cancer 1982;49:2305-14.

Ray-Chaudhuri K, Pye IF, Boggild M. Hypersensitivity to carbamazepinepresenting with a leukemoid reaction, eosinophilia, erythroderma, andrenal failure. Neurology 1989:39:436-8.Parker WA, Shearer CA. Phenytoin hepatotoxicity: a case report and re-view. Neurology 1979;.29:17 5-8.Maguire JH, Wenrell G, Rane A. Apparently nomal phenytoin metabo-lism in a patient with phenytoin-induced rash and lymphadenopathy. Br JClin Phamacol 1987 :24:554-1.Gennis MA. Vemuri R, Bums EA, Hill JV, Miller MA, Spielberg SP.Familial occunence of hypersensitivity to phenytoin. Am J Med l99l;91:63 I -4.Beran RG. Cross-reactive skin eruption with both carbamazepine and ox-crbazepine. Epilepsia 1993;34: 163-5.

Pirmohamed M. Graham A, Roberts P. et al. Crbamazepine-hyperrensi-tivity: assessment of clinical and in vitro chemical cross-reactivity withphenytoin and oxcilbazepine. Br J Clin Phmacol l99l:'32:74ï-9.Wilson JT, Hojer B, Tomson G. Rane A, Sjoqvist F. High incidence of a

concentration-dependent skin reaction in children treated with phenyloin.BMJ 1978:l:1583-6.Rieder MJ, Shear NH, Kanee A, Tang BK. Spielberg SP. Prominence ofslow acetylator phenotype among patients with sulfonamide hypersensitiv-ity reactions. Clin Phmacol Ther l99l:49:13-7.Murphy JM, Mashman J, Miller JD, Bell JB. Suppression of carbamaze-pine-induced rash with prednisone. Neurology l99l :41 : 144-5.Gupta A, Eggo MC. Uetrecht JP. et al. Drug-induced hypothyroidism: thethyroid as a ttrget organ in hypersensitivity reactions to anticonvulsantsand sulfonamides. Clin Phrmacol Ther 1992:51:56-67.Fauci AS. Haynes B. Katz P. The spectrum of vasculitis: clinical, patho-logic. immunologic and therâpeutic considerations. Ann lntem Med 1978;89;600-76.Calabrese LH. Michel BA. Bloch DA, et al. The American College ofRheumatology 1990 criteria for th€ classification of hypersensitivity vascu-lilis. Arthritis Rheum 1990:33: I 108-13.von Pirquet C. Schick B. Serum sickness. Baltimore: Williams & Wilkins,195 1.

Biebry L, Yancey KB. Young NS, Frank MM, Lawley TJ. Cutâneousmanifestations of serum sickness in patients receiving ântithymocyte glob-ulin, J Am Acad Dcrmatol 1985:13:41 l-7.Wanington RJ, Martens CJ, Rubin M, Rutherford WJ, Aoki FY. Immuno-logic studies in subjects with a serum sickness-like illness after immuniza-tion with human diploid cell rabies vaccine. J Allergy Clin lmmunolI 987:79:605- I 0.Wintroub BU. Stern R. Cutaneous drug reactions: pathogenesis and clini-cal classif,cation. J Am Acad Dermatol 1985:13:167-79.

THE \E\\' ENGLAND JOURNAL OF \'IEDICINE

to2.

lll.

I12.

I r3.

I 14.

I 15.

Dubost JJ, Souteyrand P, Sauvezie B. Drug-induced vasculitides. Bail-lieres Clin Rheumatol 1991:5:l l9-38.Brasile L, Kremer JM, Clarke JL, Cerilli J. Identification of an autoanti-body to vascular endothelial cell-specific antigens in patients with systemicvasculitis. Am J Med 1989:87:74-80.Falk RI. Jennette JC. Antineutrophil cytoplasmic autoantibodies withspecificity for myeloperoxidase in patients with systemic vasculitis and

idiopathic necrotizing and crescentic glomerulonephritis. N Engl J MedI 988:3 I 8:165 I -7.Ekenstam EAF, Callen JP. Cutaneous leukocytoclastic vasculitis: clinicaland laboratory features of 82 patients seen in private practice. Arch Derma-tol 1984:120:.184-9.

Sanchez NP, Van Hale HM. Su WP. Clinical and histopathologic spectrumof necrotizing vasculitis: repon of findings in l0l cases. Arch Dermatol1985: l2 I :220-4.Van Hale HM, Gibson LE, Schroeter AL. Henæh-Schonlein vasculitis:direct immunofluorescence study of uninvolved skin. J Am Acad Dematol1986: I 5:665-70.Hoffman GS. Ken GS, Leavitt RY. et al. Wegener granulomatosis: ananalysis of 158 patients. Ann Intem Med 1992;l t6:488-98.D'Cruz D. Hughes G. Systemic vasculitis: new tests, new treatments. BMJ1992:3M:269-70.\ryemer MC, Romaldini JH, Bromberg N, Wemer RS, Farah CS. Adveneeffects related to thionamide drugs and their dose regimen. Am J Med Sci1989:297:216-9.Carrasco MD, Riera C, Clotet B, Grifol M, Foz M. Cutaneous vasculitisassociated with propylthiouracil therapy. Arch Intem Med 19871,147:1677 .

Horton RC, Shepprd MC, Emery P. Prophylthiouracil-induced systemiclupus ervthematosus. Lancet 1989:2:568.Dolman KM, Gans RO, Vervaat TJ, et al. Vasculitis and antineutrophilcytoplasmic autoantibodies associated with propylthiouracil therapy. Lan-cet 1993:342:651-2.Dwyer JM, Kenicer K. Thompson BT, et al. Vasculitis and retinoids.Lancet 1989:2:494-6.Huminer D, Cohen JD, Majadla R, Dux S. Hypersensitivity vasculitis dueto ofloxacin. BMJ 1989:299:303.Taylor RJ. Hypersensitivity vasculitis occuning in a patient receiving im-munotherapy. J Allergy Clin Immunol l99l;87:889-90.Heckbert SR. Stryker WS. Coltin K, Manson JE, Platt R. Serum sicknessin children after antibiotic exposure: estimates of occurence and morbidityin a health maintenance organization population. Am J Epidemiol 1990;t32:336-42.Platt R, Dreis MW, Kennedy DL. Kuritsky JN. Serum sickness-like reac-tions to amoxicillin, cefaclor. cephalexin, and trimethoprim-sulfamethoxa-zole. J Infect Dis 1988:158:474-7.Yen MC, Piszczek JE, Mintzer DL. Serum sickness-like syndrome associ-ated with propranolol therapy. Postgrad Med 1983:74:291-4.Puyana J, Urena V, Quirce S. Femandez-Rivas M, Cuevas M. Fmj J.Serum sickness-like syndrome assæiated with minocycline therapy. Aller-gy 1990;45:313-5.Noel J, Rosenbaum LH, Gangadharan V, Stewart J, Galens G. SerumsicknessJike illness and leukocytoclastic vasculitis following intracoro-nary arterial streptokinase. Am Hean J 1987;l l3:395-7.Bauer KA. Coumarin-induced skin necrosis. Arch Dermatol 1993',129:766-8.Comp PC. Coumrin-induced skin necrosis: incidence, mechanisms, man-agement and avoidance. Drug Safety 1993;8:128-35.Teepe RG, Brækmans AW, Vermeer BJ, Nienhuis AM, Loeliger EA.Recunent coumarin-induced skin necrosis in a patient with an acquiredfunctional protein C deficiency. Arch Dematol 19861122:1408-12.Vigano D'Angelo S. Comp PC, Esmon CT. D'Angelo A. Relationshipbetween protein C antigen and anticoagulant activity during oral anti-coagulation and in selected disease states. J Clin Invest 1986',77:416-25.Schramm W. Spannagl IVI, Bauer KA, et al. Treatment of coumrin-induced skin necrosis with a monoclonal antibody purilied protein C con-centrate. Arch Dermatol 1993l.129:7 53-6.Miletich J. Sheman L, Broze G Jr. Absence of thrombosis in subjects withheterozygous protein C deliciency. N Engl J Med 1981l,317:991-6.Grimaudo V. Gueissaz F, Hauen J, Sanai A, Kruithof EK, Bachmann F.Necrosis of skin induced by coumarin in a patient deficient in protein S.

BMJ 1989:298:233-4.MacLean JA. Moscicki R, Bloch KJ. Adverse reactions to heparin. AnnAllergy 1990;65:254-9.Tuneu A, Moreno A. de Moragas JM. Cutaneous reactions secondary toheparin injections. J Am Acad Dematol 1985l.12:1072-7.Kelly RA, Gelfand JA. Pincus SH. Cutaneous necrosis caused by system-ically administered heparin. JAIIIA I 98 I ;246: I 582-3.Kelton JG. Sheridan D, Santos A, et al. Hepæin-induced thrombocytope-

nia: laboratory studies. Blood I 988;72:925-30.Bochner BS, Lichtenstein LM. Anaphylaxis. N Engl J Med 19911324:I 785-90.

Nor'. 10. 1994

68

69

70.

1t.

72.

73.

15.

76.

11

78.

79.

80.

84.

85.

90

9t

92

il8

I 19.

120.

Izt.

122

123

86.

87

88

89.

99.

100.

103.

104

105.

106.

107.

108.

109.

I10.

81

83

93

I 16.

lt7.

t24.

125.

95.

96.

97.

t26.

127.

128.

t29.

98 I 30.

Vol. 331 No. 19

Horan RF, Schneider LC, Sheffer AL. Allergic skin disorders and masto-cytosis. JAMA 1992:268 :2858-68.Intemational Rheumatic Fever Study Group. Allergic rcactions to long-term benzathine penicillin prophylaxis for rheumatic fever. tancet 1991;

337:1308-10.Megerian CA, Amold JE, Berger M. Angioedema: 5 yeus' experience,with a review of the disorder's pres€ntation and treatment. Laryngoscope1992:lO2:256-60.Hedner T, Sarnuelsson O, Lunde H, Lindholm L, Andren L, Witrclm BE.Angieoedema in rclation to treatment with angiotensin converting enzymeinhibitors. BMI 1992:304:941 -6.Thompson T, Frable MA. Drug-induced, life-thrcatening mgioedema rc-visited. Laryngoscope 1993;103:10-2.Bielory L, I,ee SS, Holland CL, Jaker M. I-ong-acting ACE inhibitor-induced mgiædem. Allergy Præ 1992;,13:85-7.Slater EE, Menill DD, Guess HA, et al. Clinical profrle of angioedemaassociated with angiotensin converting-enzyme inhibiüon. JAMA 1988;260:967-70.

r 285

Andemon MW, deShæo RD. Studies of the mechanism of angiotensin-converting enzyme (ACE) inhibitor-associated mgioedema: the effect ofan ACE inhibitor on cutaneous responses to bmdykinin, codeine, and

histamine. J Allergy Clin Immunol 1990;85:856-8.Verresen L, Waer M, Vanrenterghem Y, Michielsen P. Angiotensin-con-verting-enzyme inhibitors and anaphylactoid reactions to high-flux mem-brane dialysis. Lancet 1990;336: I 360-2.Pames EL, Shapiro WB. Anaghy\actoid reactions in hemodirlysis patientsûeated wftfr me ANÔ9 diâIyær. Kidney Int t99I;4O:l l4E-52.

Brunet P, Jaber K, Berland Y, Bz M. Anaphylactoid reactions duringhemodialysis and hemofiltration: role of æsociating AN69 membrane and

mgiotensin l-converting enzyme inhibiton. Am J Kidney Dis 1992;19:

444-7.Kessler DA. Introducing MEDïy'atch: a new approach to rcporting medica-tion and device adverse effects and product problems. JAMA 1993;

269:2765-8.Moorc N, Biour M, Paux G, et al. Adverse drug reaction monitoring: doingit the French way. Lancet 1985;2:1056-8.

l3l

t32

135.

136,

137 .

MEDICAL PROGRESS

139.

138.

t33.

1341«).

14l.

142

143.

Massachusetts Medical SocietyRegistry on Continuing Medical Education

To obtain information on continuing medical education courses in the New England area,call between 9:00 a.m. and 12:00 noon, Monday through Friday, (617) 893-4610 or in

Massachusetts 1-800-322-2303, ext. 1342.

l 286 ]'HE NE\\' ENGL.A,ND JOURNAL OF N,{EDICINE \oi. 10, 1994

remodeling, whether planned (as in embrvogenesisand development) or unplanned (as in carcinogenesisand tissue repair after injurl ).-''

TGF-B is a prototvpical. multifunctional cvtokinethat u'as isolated from platelets and characterizedjustover l0 years aeo.i The name is derived from the ob-servation that TGF-B stimulates normal cells to growin soft asar as though thev had been virally trans-formed. In mammals the cvtokine has three isoforms,TGF-pl, 2, and 3, rvhose biologic properties are near-Iy identical. 'lhe TGF-Bl gene is up-regulated in re-sponse to tissue injury, and TGF-Bl is the isoformmost implicated in fibrosis. In this review the genericterm'|GF-B is used in discussins properties that areprobably shared bv ail three isoformsl the specific iso-form is mentioned. however. i{-it has becn identified orused in a particular studv.

TGf-p1. svnthesized as a 391-amino-acid precursormolecule, is proteolvticallv cleaved to vield peptidefragments and a ll2-amino-acid subunit. ActiveTGF-Bl is a 25-kd dimeric protein composed of tu'osubunits linked by a disulfide bond. TGF-Bl is secret-ed in an inactive (latent) form that requires activa-tion before it can exert a biologic eflect. The latentform of TGF-Bl is a high-molecular-n'eight complexin which TGF-Bi is noncovalentlv bound to anotherdimeric peptide, the latenct'-associated peptide, whichis formed from cleavage lragments of the TGF-BIprecursor. Latent TGF-Bl is sl.ored at the cell sur-face and in the extracellular matrix and is convertedto active TGF-BI at these sites bv an unknown mech-anism.B

TGF-B binds to at least three membrane proreins,referred to as receptor tvpes I. II, and III, that existon virtuallv all cells. The tvpe I and tvpe II receptorsare transmembrane serine-threonine kinases that in-teract with one another and facilitate each other's sig-naling.!)The tvpe III receptor, also called betaglycan,is a membrane-anchored proteoelycan that lias no sig-naling structure but acts to present TGF-B to the oth-er receptors.r0 The effects of TGF-B on the svnthesisand deposition of extracellular matrix are mediatedby the type I receptor. The effects on cell grorvth andprolileration are mediated by the type II receptor. Theregulation of TGF'-Bl secretion and action involvescomplex post-transcriptional events, including mes-senger RNA (mRNA) stabilization, the assembly andactivation of the latent 'IGl'-pl complex. and themodulation of receptor expression.r I

Other cytokines involved rvith TGF-Bl in tissueremodeling alter injurv are platelet-derived growthfactor, basic fibroblast grolvth factor, tumor necro-sis factor, and interleukin-l.r Each cvtokine has dis-tinctive, synergistic roles in tissue repair. as recentstudies involving in vivo gene transfection. gene dis-ruption ("knockout"), and the administration of cyto-kines have sho\&,n.1!-l:r The dominant eflect of platelet-derived grolvth factor is to stimulate cell prolilerationand migration; fibrobiast grorvth factor induces the

MECHANISMS OF DISEASE

FneNxr.rx H. Ers'rrrN, M.D., Editor

TRANSFORMING GROWTH FACTOR PIN TISSUE FIBROSIS

WayNr A. Bon»nn, NI.D.,aNo NaNcy A. Nonrr,. Pn.D.

f)ROGRESSIVE fibrosis in the kidney, liver, lung,I. heart, bone marrow, and skin is both a majorcause of suffering and death and an important con-tributor to the cost of health care. All of this is likely tochange in the future. Advances in cell and cytokinebiology have brought a new understanding of the mo-lecular events underlying tissue fibrosis. It is becom-ing clear that fibrogenesis is not a unique pathologicprocess but is due to excesses in the same biologicevents involved in normal tissue repair.r

A central event in tissue repair is the release ofcytokines in response to injury. Several lines of evi-dence point to transforming grorvth factor B (TGF-B)as a key cytokine that initiates and terminates tissuerepair and whose sustained production underlies thedevelopment of tissue fibrosis.2 'Ihis review will ex-plore the biology of tissue repair and the propertiesand actions of 1'GF-B that make it such a potent fibro-genic molecule. Understanding the actions of TGF-pin fibrosis could lead to the development ol clinicallyuseful antifibrotic asents.

CyroxrNrs rN Trssur IN;unvTissue is made up of organized groups of cells at-

tached to an extracellular matrix and surrounded by anetwork of blood vessels. Tissue homeostasis is main-tained b1' coordinating cell growth and prolilerationwith the production and turnover of the extracellularmatrix. Cells achieve this coordination by constantsignaling to themseives (autocrine activity) and eachother (paracrine activity) by means of polypeptidescalled cytokines (also known as growth factors).3 Cy-tokines difler from conventional hormones in that theyact locall.v, not at a distant site. Ihe action of a cyto-kine can be positive or negative, depending on theinfluence of other cvtokines and the physioiogic stateof the target cell and the surrounding extracellularmatrix. This variability of cytokine action providesceils and tissues with a range of potential responses toany stimulus.a Cytokines regulate all aspects of tissue

From the Division of Ncphrology. University of Utah School of Medicine, 50N. Medical Dr.. Salt Lake City, UT 84132, where reprint requests should beaddressed to Dr. Border.

Suppo(ed by a grant (DK 43609) from the National Institute of Diabetes andDigestive and Kidney Diseases.

Vol. 331 No. 19

formation of ner,v blood vessels (angiogenesis); andtumor necrosis factor and interleukin-l promote in-flammation, cell migration, and proliferation. TGF-Blis unique in its widespread actions that enhance thedeposition ol extraceliular matrix. It also acts as apotent regulator of repair, coordinating or suppress-ing the actions of other cytokines.7,l6

Bror,ocrc AcrroNs or TGF-B rN Trssun RrparnThe healing of a dermal rvound, a paradiem lor

tissue repair in general, is a coordinated sequence ofbiologic events beginning rvith plateiet-induced hemo-stasis, foilowed by an influx of inflammatory cells andfibroblasts, the formation of nell' extracellular matrixand blood vessels (granulation tissue), and the prolif-eration of cells to reconstitute the tissue.r TGF-BIplays an important part in each of these events,lvhich can largely be reproduced in normal tissue bythe administration of TGF-B1.to'r; Platelets containhigh concentrations of TGF-Bl and platelet-derivedgrorvth lactor that are released into the tissue at thesite of injury. Inactive (latent) TGF-Bl, bound local-lv to the extracellular matrix, can aiso be activatedafter injurv. In femtomolar concentrations TGF-Bi isstrongly chemotactic for neutrophils, T cells, mono-cytes, and fibroblasts.r6'18'le \tloving to the site of theinjury, these cells become activated as they encounterhigher (picomolar) concentrations of TGF-B1. NIono-cytes begin secreting fibroblast gro\\'th factor, tumornecrosis factor, and interleukin-1, and fibroblasts in-crease their synthesis of extracellular-matrix pro-teins.l6 TGF-Bl also induces both infiltrating cells and

t2Bl

resident cells to produce more of itself. This autoin-duction amplifies the bioiogic effects of TGF-B I at theinjur,v site and mav have a central role in chronicfibrosis.2r'

At phl.siologic concentrations, TGF-B1 regulatesplatelet-derived growth factor (in smooth-muscle cellsand fibroblasts), fibroblast growth factor (in endothe-Iial cells). and tumor necrosis factor and interleukin-1(in monocvtes) b1'stimulating or inhibiting their pro-duction or modulating their actions to both synchro-nize and control the repair process.2''22 TGF-Bl alsoinhibits the functioning of T cells and B cells and theirproduction ol tumor necrosis factor and interleukin-l.!:rNeonatal mice in which the TGt'-Pl gene has beeninactivated live lor several weeks, until the maternalsupply of TGF-Bl is gone, and then die ol a svstemicautoimmune-like disease in which tissue concentra-tions of tumor necrosis factor and interleukin-l aremarkedlv elevated.l3 r+ TGF-Bl also modulates the cy-totoxicity of macrophages by suppressing the produc-tion of superoxide and nitric oxide.r6'2+

Whereas TGF-Bl can function as either an agonistor an antagonist o{'cel} prolileration and inflamma-tion, it consistentlv and potently acts on cells to inducethe deposition of extracellular matrix.T The accumula-tion of matrix in tissues is the chief pathologic featureof fibrotic diseases. Extracellular matrix is a dvnamicsuperstructure o{' selÊaggregating macromolecules,including fibronectin, collagens, and proteoglycans, towhich cells attach bv means of surface receptors calledintegrins.2i The matrix surrounding cells is continual-l,v degraded by proteases. Figure I illustrates how

lv{ECHANIS\,IS OF DISEASII

Actions of TGF-B

IIIII

l\4atrix proteins

Proteases

Protease inhibitors

lntegrins

ïGF-p

Figure 1. Actions of TGF-B in the Healing of lnjured Tissue.Platelets release TGF-B at the site of tissue injury. To repair the damage, TGF-B then induces the deposition of extracellular matrix bysimultaneously stimulating the production of new matrix proteins (fibronectin, collagens, and proteoglycans), blocking matrix degradationby decreasing the synthesis of proteases and increasing the synthesis of protease inhibitors, and modulating the expression of cell-surface integrins in a manner that enhances cell-matrix interaction and matrix assembly. TGF-B also induces its own production by cells,

thus amplifying its biologic effects.

I 288 THE NE\V ENGLAND.JOUR\AL OF \IEDICINE Nor,. 10, 1994

TGF-Bl causes the deposition of exrracellular matrixby simultaneoush' stimulating cells to increase sever-alfold the synthesis ol most matrix proteins, decreasethe production of matrix-degrading proteases, in-crease the production of inhibitors of these proteases.and modulate the expression of integrins in a mannerthat increases cellular adhesion to the matrix. Thesecomprehensive effects on the extracellular matrix re-flect the diverse biologic properties ol TGF-BI andmav also be part of a negative-feedback loop that nor-mally regulates the expression of TGF-B.26 TGF-Bbinds to proteoglycans in the matrix or near the cellsurface, and such binding may act as a signal to termi-nate the production of TGF-B after tissue repair iscomplete.

ENrreNcuvrBNT oF l{ouNo HrnrrNc nv TGF-B

l'ibrosis represents a pathologic excess of the nor-mal process of tissue repair. Excessive or sustainedproduction of TGF-B I is a key molecular mediator oftissue fibrosis. The topical application of TGF-B ac-celerates wound healing.5'r6 In rats, topical or limitedintravenous administration of recombinant TGF-Blnormalizes r,r,ound healing that is impaired by age orglucocorticoids.2T In humans, TGF-82 has been usedto repair retinal holes. TGF-B therefore has greatpromise as a therapy lor poorly healing wounds.2B

However, the fibrogenic potential of TGF-B is re-vealed with repeated injections of higher doses. Twoweeks of intravenous injections of TGF-Bl producedserious systemic effects in rats, including marked fi-brosis in the kidneys and liver and at the injectionsite.2e Severe cachexia and generalized tissue fibrosisdeveloped in mice given TGF-pl intraperironeallv forl0 davs.ro

The clinical counterpart of these results may be therapid onset of liver and lung fibrosis in patients withadvanced breast cancer who receive high-dose chemo-therapy in preparation for autologous bone marrowtransplantation. In one study, more than 90 percent ofthe patients whose plasma TGF-B concentrationswere 2 SD or more above the normal mean ( l0 ng permilliliter) had liver or lung fibrosis.3r The source of theelevated plasma TGF-B concenrations in these pa-tients is unknown. Measuring TGF-B is expensive andtechnically cumbersome with existing bioassays, but anewly reported bioassav promises to be more sensitiveand specific.32

TGF-p rN FrsRorrc Drsrasrs

Table I lists the animal models and human dis-orders in rvhich TGF-B has been implicated in thepathogenesis of fibrosis. The involvement of TGF-B inkidney, liver, and lung disease has been the most thor-oughly investigated.

Kldney

The intricate architecture and frltratine function ofthe kidney make it particularly vulnerable to the con-

Table 1. TGF-p in Animal Models of Fibrotic Dis-orders and in Human Fibrotic Disorders.

ORGAN AND DTSoRDER* REFERENcE

Animal models

KidneyAcute ATS glomerulonephritis Okuda et al.3rChronic ATS glomerulonephritis Yamamoto et al.r4Anti-GBM elomerulonephritis Coimbra et al.15Habu-venomglomerulonephritis BarnesandAbboudr6Acute and chronic puromycin- Jones et a1.37

induced nephrosisDiabetic nephropathyHIV nephropathyUreteral obstructionAngiotensin-induced nephropathy

LiverSchistosomiasisCarbon tetrachloride-induced

hepatic fibrosisLung

Bleomycin-induced fi brosis

SkinNomal and impaired wound

hcalingArteries

Vascular restenosis

Central nervous systemScming after injury

OtherAcute and chronic arthritisRadiation-induced fibrosisGranulomasPostoperative adhesions

Human disease

KidneyGlomerulonephritisDiabetic nephropathyAllograft rejectionHIV nephropathy

LiverCinhosis

Veno-occlusive diseaseLung

ldiopathic fibrosis

Autoimmune fibrosisSkin

Systemic sclerosis Kulozik et a1.57

Keloids Peltonen et a1.58

Hypenrophic burn scm Ghahary et al.5eEosinophilia-myalgia syndrome Vuga et al.m

AneriesVascular restenosis

Central neryous systemlntraocular fibrosis

OtherRheumatoid anhritisNasal polyposis

Yamamoto et al.rEKopp et al.3eKaneto el al.$Kagami et al.ar

Czaja et al.a2Czaja et al.r:

Westergren-Thoresonet al..$ Khalilet al.{

Spom and Roberts,2ETenell et al.:e

Wolf et al.;5

Logan et al.6

Brandes et al.llBrcellos-Hoff et al.a7Appleton et al.a8Williams et al.ae

Yoshioka et al.50Yamamoto et a1.38

Shihab et al.5lBorder el al.s2

Castilla et al..5r Nagyet al.5o

Anscher et al.rr

Anscher et al.,3lBroekelmann et al.5s

Deguchi56

Nikol et al.6r

Connor et a1.62

Lafyatis et al.6lOhno et al.8

*ATS denoles antith_vmævte serum. GBM glomerula basemenr mem-brâne. and HIV human immunodeliciencv virur

sequences of fibrosis. A model of acute glomerulo-nephritis in rats has provided a unique opportunitvto stud,v the role of TGF-Bl in fibrogenesis, becauseglomeruli can be rapidly isolated and studied invitro throughout the course of disease.33 In theserats a single injection of an antithymocyte seruminjures glomerular mesangial cells. Extracellular ma-trix accumulates in the nephritic glomeruli, reachinga peak level in 14 days, after which the glomeruli

\rol. 331 \o. 19

return to normal. The temporal pattern oî TGF-BIgene expression and the actions ol TGF-BI on theextracellular matrix mirror the pattern of accumu-lation and removal of the pathologic matrix. Forexample, nephritic giomeruli contain man), timesmore TGF-BI mRNA than normal glomeruli, syn-thesize more TGF-BI protein, and produce muchmore fibronectin and proteoglycans.:r3 The plasminprotease system) which is thought to have an impor-tant role in marix degradation. is strikingly sup-pressed owing to a decrease in plasminogen activatorand a substantial increase in the synthesis of plas-minogen activator inhibitor type 1.65 Simultaneously,the synthesis and expression of integrin receptors forfibronectin and collagen increase.oG Platelet-derivedgrowth factor and fibroblast growth factor also medi-ate some biologic events, especially cell proliferation,in these rats.67

'I'hree lines o1' evidence point to a causal relationbetw'een elevated production of TGF-Bl and the accu-mulation of extracellular matrix in the model of glo-merulonephritis. First, the in r.'ivo events that underliethe accumulation of matrix - increased productionof extracellular-matrix proteins, inhibition of proteaseactivity, and increased integrin expression - have allbeen reproduced by incubation of normal glomerulior mesangial cells as well as nonrenal cells withTGF-p1.33 65 66 Second, injecting nephritic rats with anantiserum capable of neutralizing TGF-BI or witha proteoglycan that binds TGF-Bl prevented the in-creased production of matrix proteins by the glomer-uli and blocked the accumulation of matrix.68'6s Third,in r.'ivo transfection of the 7GF-B1 gene into normalrat kidneys led to increased production of TGF-B1 inglomeruli and the rapid development of glomerulo-sclerosis.r2 Identical transfection of the gene for plate-Iet-derived growth factor markedly stimulated theproliferation of glomerular cells, with an associatedslight increase in extracellular matrix. The dillerencesin the biologic activities of TGF-Bl and platelet-derived growth factor, revealed in the gene-transfec-tion experiments, are similar to earlier findings inwhich these cvtokines were delivered in vivo by osmot-ic minipumps.r5

An exciting leature of the glornerulonephritis modelis the recent discovery that animals receir,'ing a secondinjection of the antiserum have persistent elevationsof glomerular TGF-Bl mRNA and TGF-Bl itself.3a

Myofibroblast-like cells appear in the tubulointersti-tium of the kidnev and strongly express TGF-Bl. Thepersistent production of T'GF-Bl in the kidney leadswithin weeks to glomerulosclerosis and tubulointersti-tial fibrosis, a picture closely resembling the histologicfindings in humans with chronic glomerulonephritis.

Elevated concentrations of I'GF-Bl may also be im-portant in the pathogenesis of glomerulosclerosis indiabetic nephropathy. Rats made diabetic with strep-tozocin, a drus that causes insulin deficiency, hadprogressively increasing concentrations of TGF-Bl

mRNA and TGF-Bl in their glomeruli.38 The stimulusthat triggers the expression of TGF-BI in diabetesmay be hyperglycemia or an increase in the activityof the renin-angiotensin system in renal tissue. Inhumans, good control of blood glucose with insulinand the administration of an angiotensin-converting-enzyme inhibitor retard the development of diabeticnephropathy. In diabetic rats, insulin treatment re-duced the increase in the amounts of glomerularTGF-Bl mRNA and the extracellular-matrix proteinsknown to be induced by TGF-BI.38 In cultured ratmesangial cells, both increased glucose and increasedangiotensin II concentrations induced the productionof TGF-BI, which then stimulated the synthesis offibronectin, collagens, and proteoglycans.ai,T0 The ad-ministration of angiotensin II to rats leads to elevatedamounts of glomerular TGF-BI mRNA and type Icollagen mRNA in one week.ar

The relevance of these studies to human glomerulardiseases has recently been demonstrated. In kidney-biopsy specimens from patients with mesangial prolif-erative glomerulonephritis, a disease histologicallysimilar to the model of glomerulonephritis describedabove, glomerular immunostaining for TGF-Bl wasintense, and the intensity correlated closely with theamount of mesangial matrix.5o In the glomeruli ofhumans with diabetic nephropathy. TGF-Bl proteinand matrix proteins induced by TGF-BI were in-creased, as they were in the glomeruli of diabeticrats.38 Glomeruli from patients with renal diseases inwhich fibrosis does not occur and from patients withno renal diseases were negative for TGF-Bl. Recently,elevated amounts of TGF-BI protein were found infibrotic kidneys from patients with human immunode-ficiency virus-associated nephropathy and patientswith chronic allograft rejection.sr'52

Liver

mRNA for type I collagen, the predominant matrixcomponent in injured liver, is increased in cultured rathepatocytes incubated with TGF-Bl.a2 In liver-biopsyspecimens from patients with chronic liver disease, theamount of TGF-BI mRNA correlated closely withthat of type I collagen mRNA.s3 TGF-Bl mRNA con-centrations in the liver are mirrored by serum concen-trations of peptide fragments of type III collagen andthe histologic activitv of the liver disease. In biopsyspecimens from patients with chronic liver disease,TGF-Bl protein was detected by immunostaining inareas of fibrosis, but not in areas of inactive disease orin normal liver.5a As previously discussed, elevatedplasma TGF-B concentrations are highly predictive ofthe development of hepatic fibrosis (veno-occlusivedisease) in the recipients of bone marrotv trans-plants.3r

In two models of hepatic fibrogenesis, one inducedby the administration of carbon tetrachloride and theother by infection with schistosoma, increased concen-trations of TGF-Bl mRNA and TGF-p1 in perisinu-

\IECHANISMS OF DISEASE I 289

Vol. 331 \o. 19

return to normal. The temporal pattern o1 TGF-PIgene expression and the actions of TGF-BI on theextracellular matrix mirror the pattern ol accumu-lation and removal of the pathologic matrix. Forexample, nephritic glomeruli contain many timesmore TGF-BI mRNA than normal glomeruli, syn-thesize more TGF-BI protein, and produce muchmore fibronectin and proteoglycans.33 The plasminprotease system, rvhich is thought to have an impor-tant role in matrix degradation, is strikingly sup-pressed ou'ing to a decrease in plasminogen activatorand a substantial increase in the synthesis of plas-minogen activator inhibitor type l.Ùr Simultaneously,the synthesis and expression of integrin receptors forfibronectin and collagen increase.66 Platelet-derivedgrowth factor and fibroblast growth factor also medi-ate some biologic events, especially cell proliferation,in these rats.67

Three lines o1' evidence point to a causal relationbetw'een elevated production of TGF-Bl and the accu-mulation of extracellular matrix in the model of glo-merulonephritis. First, the in vivo events that underliethe accumulation of matrix - increased productionof extracellular-matrix proteins, inhibition of proteaseactivity, and increased integrin expression - have allbeen reproduced by incubation of normal glomerulior mesangial cells as well as nonrenal cells withTGF-p1.33'65'66 Second, injecting nephritic rats with anantiserum capable of neutralizing TGF-BI or witha proteoglycan that binds TGF-Bl prevented the in-creased production of matrix proteins by the glomer-uli and blocked the accumulation of matrix.68'6s Third,in vivo transfection of the TGF-BI gene into normalrat kidneys led to increased production of TGF-Bl inglomeruli and the rapid development of glomerulo-sclerosis.r2 Identical transfection of the gene for plate-let-derived growth factor markedly stimulated theproliferation of glomerular cells, with an associatedslight increase in extracellular matrix. The differencesin the biologic activities of TGF-BI and platelet-derived growth factor, revealed in the gene-transfec-tion experiments, are similar to earlier findings inwhich these cvtokines were delivered in vivo by osmot-ic minipumps.r5

An exciting feature of the glornerulonephritis modelis the recent discovery that animals receiving a secondinjection of the antiserum have persistent elevationsof glomerular TGF-B1 mRNA and TGF-Bl itself.3+Myofibroblast-like cells appear in the tubulointersti-tium of the kidney and stronely express TGF-B1. Thepersistent production of TGF-Bl in the kidney leadswithin weeks to glomerulosclerosis and rubulointersti-tial fibrosis, a picture closely resembling the histologicfindings in humans with chronic glomerulonephritis.

Elevated concentrations of 1'GF-Bl may also be im-portant in the pathogenesis of glomerulosclerosis indiabetic nephropathy. Rats made diabetic with strep-tozocin, a drug that causes insulin deficiency, hadprogressively increasing concentrations of TGF-Bl

mRNA and TGF-Bl in their glomeruli.38 The stimulusthat triggers the exprcssion ol TGF-Bl in diabetesmay be hyperglycemia or an increase in the activityof the renin-angiotensin system in renal tissue. Inhumans, good control of blood glucose w'ith insulinand the administration of an angiotensin-converting-enzyme inhibitor retard the development of diabeticnephropathy. In diabetic rats, insulin treatment re-duced the increase in the amounts of glomerularTGF-Bl mRNA and the extracellular-matrix proteinsknown to be induced by TGF-B1.38 In cultured ratmesangial cells, both increased glucose and increasedangiotensin II concentrations induced the productionof TGF-BI, which then stimulated the synthesis ofÊbronectin, collagens, and proteoglycans.ar'70 The ad-ministration of angiotensin II to rats ieads to elevatedamounts of glomerular TGF-BI mRNA and type Icollagen mRNA in one week.al

The relevance of these studies to human glomerulardiseases has recently been demonstrated. In kidney-biopsy specimens from patients with mesangial proliÈerative glomerulonephritis, a disease histologicallysimilar to the model of glomerulonephritis describedabove, glomerular immunostaining for TGF-Bl wasintense, and the intensity correlated closely with theamount of mesangial matrix.5o In the glomeruli ofhumans with diabetic nephropathy, TGF-Bl proteinand matrix proteins induced by TGF-BI were in-creased, as they were in the glomeruli of diabeticrats.38 Glomeruli from patients with renal diseases inwhich fibrosis does not occur and from patients withno renal diseases were negative for TGF-BI. Recently,elevated amounts of TGF-Bl protein were found infibrotic kidneys from patients with human immunode-ficiency virus-associated nephropathy and patientswith chronic allograft rejection.5l,52

Liver

mRNA for type I collagen, the predominant matrixcomponent in injured liver, is increased in cultured rathepatocytes incubated with TGF-Bl.a2 In liver-biopsyspecimens from patients with chronic liver disease, theamount of TGF-BI mRNA correlated closely withthat of type I collagen mRNA.s3 TGF-Bl mRNA con-centrations in the liver are mirrored by serum concen-trations of peptide fragments of type III collagen andthe histologic activity of the liver disease. In biopsyspecimens from patients with chronic liver disease,TGF-pl protein was detected by immunostaining inareas of fibrosis, but not in areas of inactive disease orin normal liver.5a As previously discussed, elevatedplasma TGF-B concentrations are highly predictive ofthe development of hepatic fibrosis (veno-occlusivedisease) in the recipients of bone marrow trans-plants.3r

In two models of hepatic fibrogenesis, one inducedby the administration ôf carbon tetrachloride and theother by infection with schistosoma, increased concen-trations of TGF-BI mRNA and TGF-Bl in perisinu-

MECHANISMS OF DISEASE 1 289

I 290

soidal cells paralleled the increased expression of a

collagen sene and increased collagen svnthesis.+2

Lung

In rats u.ith pulmonary fibrosis induced b.v the ad-ministration of bieomvcin. total lung'|GF-Bl content\.vas several times higher than in normai rats. 'Iheincreased production ol TGF-Bl preceded the synthe-sis of collagens, fibronectin, and proteoglvcans.l:' Theprincipal cellular source of TGF-Bl lvas alveolar mac-rophages, in which increased production of TGF-B1couid not be suppressed by high-dose corticosteroidtreatment, a possible explanation for the ineffective-ness ol this treatment in patients rvith idiopathic pul-monary fibrosis.a+

In humans w'ith idiopathic pulmonary fibrosis,TGF-Bl is increased in alveolar walls at the sites atwhich extracellular matrix has accumulated.ri Bron-choalveolar cells obtained by lavage from patients

with autoimmune diseases and lung fibrosis contained

T HE NE\\' ENGL-{-\-D JOURN.{L OF \IEDICINE \or'. J0. 1994

10 times more'fGF-Bl mRNA than similar cells ob-tained from normal subjects or patients with asthma.56

Fibrotic Disorders of Other Organs

'IGf--Bl and collagen are increased in tissue sec-tions from patients with svstemic sclerosis.-'7 keloids,58and hypertrophic scars from burns.5lr TGF-Bi hasbeen implicated in the frbrosis associated rvith eosino-philia.6" Increased amounts of TGI--Bl are also foundin the arteries of rats at the sites ol balloon angioplastyand in vascular lesions associated lvith restenosis inhumat)s,'*:il

OvrnpnooucrroN or TGF-B rN Frsnosrs

In both animals and humans, acute. limited irluryis accompanied br only a transient increase in TGF-Bl, and fibrosis does not occur. \\'ith repeated injury,the increase in TGF-Bl production is sustained. lead-ing to the progressir.'e deposition of extracellular ma-trix and tissue Êbrosis.'ra The manner in rvhich TGF-

lnjury

ÿ!.-..

_ ilq

Healed Tissue

I Protease inhibitor t

Collagens

.§ -/'/'Proteogt'cans...,1

I Repair {5 --;_-'ïd*ætra

Fibronectin

- Protease

Repeated lnjury? Fibrotic Tissue

t

"ryRepair

Figure 2. Overproduction of TGF-P in Fibrogenesis.

ln normal tissue repair, the production of TGF-p and extracellular matrix is terminated by unknown mechanisms as the damaged tissue.

heals. ln patients with chronic disease, repeated tissue injury, a de{ect in TGF-B regulation, or both lead to the continuous production oiTGF-B and extracellular matrix, resulting in tissue fibrosis.

Vol. 331 No. 19 \IEC:HANIS\,{S OF DISEASE l29l

Bl production is normally terminated is unknown. Re-peated injurv. with continued autoinduction of TGI'-Bl, overrides the normal termination signals. creatinga chronic, vicious circle of TGF-Bl overproduction, as

shorvn in l'igure 2.

TGF-B ANracoNrsrs As ANTrFrBRorrc AcrNrsInjecting an antiserum capable of neutralizing the

activitv of TGF-Bl inhibited its action in the kidner',';8skin,Tr lung.72 brain,nt; joint,73 and arterial lvall.+5 Ineach case excessive amounts of extracellular matrixwere not deposited, and tissue repair was normal. Innephritic rats, the injection of the antiserum dramati-callv reduced the synthesis ol matrix proteins and thedcposition of plasminogen activator inhibitor t1'pe I inthe glomeruli and blocked the accumulation of mesan-gial matrix. The collagcn content of dermal rvoundstreated with anti-TGF-Bl rvas substantially reduced,but their tensile strength u'as normal and scar forma-tion w-as minimal. Anti-TGF-Bl reduced fibrous scartissue and inflammation at the site of brain injur1,.In arthritic joints, anti-TGl'-Bl decreased inflamma-tion and bone and synovial destruction. Finally, inrats with carotid-arter,v injurv, the injection of anti-TGF-Bl suppressed the accumulation o1'matrix thatunderlies the development of intimal hvperplasia andrestenosis. These consistent therapeutic successcsmake clear the enormous clinical potential associatedwith decreasing the action of TGF-Bl in vivo.

CoNcrusroNs

As the complexities of TGF-B regulation are unrav-eled, a number of possible therapeutic approaches fordecreasing the action of the cvtokine have arisen thatmay be more suitable than antibodies for use in hu-mans.25 For example, soluble TGF-B tvpe III recep-tors inhibit the binding of TGF-B to its membranereceptors and block its action; soluble type I and IIreceptors, rvhich have higher affinities lor TGF-B,may be even more potent in blocking its action.!'r0Similarl,v, the latenc-v-associated peptide that is re-leased in the process of TGF-B activation mav beused to inhibit the action of TGF-B. Several mem-bers of the superfamilv of retinoid-steroid receptorsmav act as post-transcriptional regulators ficr gcnes ofdifferent isoforms of 'tGF-p.? Manipulation of thisregulation may lead to decreased production of TGF-p. Using TGF'-B antisense oligonucleotides is anotherpossibilitv. A low dietary intake of protein decreasedthe expression of TGF-Bl in rats rl,ith acute glomeru-lonephritis.Tr This {inding mav help explain the bene-ficial ellect of lowlprotein diets in patients rvithvarious kidney diseases. Finally, some proteoglycansbind TGF-B.26 The injection of one of these proteoglv-cans into nephritic rats \\,as as elfective as the injectiono1' anti-TGF-Bl in inhibiting glomerular accumula-tion of matrix.6s

Whetirer any of these approaches will yield an efïèc-tive antifibrotic drug is unknown. Nevertheless, un-derstanding that TGI--B is a key Iàctor in fibrogenesis

offers a target for the development of new therapeuticagents for the many frbrotic conditions associated withthe overproduction of TGF-B.

\\'e are indcbted to Drs. \Iarkus Ketteler and Douglas Brees lortheir usefül comments on the manuscript, and to \Ir. Bruce Goude-lock lor preparinq the fisures.

l.

2.

3.

4.5.

6.

7.

8.

9.

10.

ll.

t2.

13.

14.

15.

16.

t7

t8.

2l

l9

20

RrrrneNcrsDavidson JM. Wound repair. In: Gallin JI, Goldstein lM, Snydeman R.eds. lnflammation: basic principles ând clinical conelates. 2nd ed. NewYork: Raven Press, 1992:809-19.Border WA, Ruoslahti E. Transfoming growth factor-B in disease: the darkside of tissue repair. J Clin lnvest 1992;90:l-7.Spom MB. Roberts AB. Autocrine secretion - l0 yean later, Ann InternMed I992:l l7:408-14.Nathan C, Spom M. Cytokines in context. J Cell Biol 199lll 13:981-6.Roberts AB. Spom MB. Physiological actions and clinical applications oftransfoming growth factor-B (TGF-B). Growth Factors 1993:8:l-9.Idem. Diflerenlial expression of the TGF-B isofoms in embryogenesis sug-gests specific roles in developing and adult tissues. Mol Reprod DevI 992:32:9 I -8.lden. The transforming growth factors-B. In: Spom MB, Roberts AB,eds. Peptide growth factors and their receptors. Vol. 95 of Handbookof experimental pharmacology. New York: Springer-Verlag, 1990:419-72.Flaumenhaft R, Abe M. Mignatti P, Rifkin DB. Basic fibroblast growthfactor-induced activâtion of latent transfoming growth factor B in endothe-lial cells: regulation of plasminogen activator activity. J Cell Biol 1992;I l8:901-9.Ebner R, Chen R-H, Lawler S, Zioncheck T, Derynck R. Detemination oftype I receptor specificity by the type II receptors for TGF-p or activin.Sciencc 1993:262:90O-2.Lopez-Casillas F, Payne HM, Andres JL, Massague J. Betaglycan can act as

a dual modulator of TGF-B access to signaling receptors: mapping of ligandbinding and GAG attachment sites. J Cell Biol 1994:124:557-68.Kim S-J, Puk K, Koeller D. et al. Post-transcriptional regulation of thehuman transforming growth factor-pl gene. J Biol Chem1992.267:13702-7.

Isaka Y, Fujiwra Y. Ueda N, Kaneda Y, Kamada T. Imai E. Glomerulo-sclerosis induced by in vivo transfection of transfoming growth fâctor-B orplateletderived growth factor gene into the rat kidney. J Clin Invest199392:2597 -6Ol .

Kulkami AB, Huh C-G, Becker D, et al. Transfoming growth factor pr nullmutation in mice causes excessive inflammatory response and edy death.Præ Natl Acad Sci U S A 1993:90:770-4.Shull MM. Omsby I, Kier AB, et al. Targeted disruption of the mouse

transfoming growth factor-pl gene results in multifocal inflammatory dis-ease. Nature 1992:359:693-9.Ogawa Y, Ksander GA, Pratt BM, et al. Differences in the biologicalactivities of transfoming growth factor-p and platelet-derived growth factorin riyo. Growth Factors l99l;5:57-68.Roberts AB, Joyce ME, Bolander ME, Spom MB, Transforming growthfactor-beta (TGF-É): a multifunctional effector of both soft and hard tissueregeneration. In: Westemark B, Betsholtz C, Hôkfelt B, eds. Growth fac-tors in health and disease: basic and clinical aspects. Amsterdam: ExcerptaMedica.1990:89-101.Roberts AB, Spom MB, Assoian RK, et al. Translbming growth factortype B: rapid induction of fibrosis and angiogenesis in yiyo and stimulationof collagen tbmation in \'üro. Præ Natl Acad Sci U S A 1986:83:4t67-71.Wahl SM, Hunt DA. Wakefield LM. et al. Transforming growth factor type

B induces monocyte chemotaxis and growth factor production. Proc NatlAcad Sci U S A 1987:84:5788-92.Postlethwaite AE, Keski-Oja J. Moses HL. Kang AH. Stimulation of thechemotactic migration of human libroblasts by transfoming growth factor

B. J Exp Med 1987:165:251-6.Kim S-J, Angel P, Lafyatis R. et al. Autoinduction oftransforming growthfactor Bl is mediated by the AP-l complex. Mol Cell Biot 1990;10:1492-7.

Battegay EJ, Raines EVy', Seifert RA, Bowen-Pope DF, Ross R. TGF-Binduces biomodal proliferation of connective tissue cells via complex con-trol of an autocrine PDGF loop. Cell 1990:63:515-24.Pepper MS, Belin D. Montesano R. Orci L, Vassalli J-D. Transfominggrowth tàctor-beta I modulates basic frbroblast growth factor-induced pro-

teolytic and angiogenic propenies of endothelial cells in vitro. J Cell Biol1990:l 1 l:743-55.Brandes ME, Allen JB. Ogawa Y, Wahl SM. Transfoming growth factorpl suppresses acute and chronic anhritis in experimental animals. J ClinInvest 1991;87:l l0E-13.

22

t292

Vodovotz Y, Bogdan C, Paik J, Xie Q-W, Nathan C. Mechanisms ofsuppression of macrophage nitric oxide release by transforming growth fac-tor p. J Exp Med 1993;178:605-13.Ruoslahti E. Integrins. J Clin Invest l99ll87:l-5.Ruoslahti E, Yamaguchi Y. Proteoglycans as modulatoB of growth factoractiviries. cell 1991 :64:867-9.Beck LS, DeGuzman L. Lee WP. Xu Y, Siegel MW, Amento EP. Onesystemic administration of transfoming growth factor-B I reverses age- orglucoconicoid-impaired wound healing. J Clin Invest 19931'92:2841-9.

Sporn MB, Roberts AB. A major advance in the use of growth factors ro

enhance wound healing. J Clin Invest 19931'92:2565-6.Tenell TG, Working PK, Chow CP, Green JD. Pathology of recombinanthuman transfoming growth factor-Bl in rats and rabbits. Int Rev Exp Pathol199334:43-61.Zugmaier G. Paik S, Wilding G, et al. Transforming growth factor Blinduces cachexia and systemic fibrosis without an antitumor effect in nudemice. Cancer Res l99l:51:3590-4.Anscher MS, Peters WP, Reisenbichler H, Petros WP, Jirtle RL, Trans-forming growth factor B as a predictor of liver and lung fibrosis after autolo-gous bone marrow iransplantation for advanced breast cancer. N Engl J Med1993:328:1592-8.Abe M, Harpel JG, Merz CN, Nunes I. Loskutoff DJ, Rifkin DB. An assay

for transforming growth factor-B using cells ûansfected with a plasminogenactivator inhibitor- I promoter-luciferase construct. Anal Biochem 1994;

216:276-84.Okuda S. Languino LR, Ruoslahti E, Border WA. Elevated expression oftransforming growth factor-B and proteoglycân production in experimentalglomerulonephritis: possible role in expansion ofthe mesangial extracellularmatrix. J Clin Invest 1990;86:453-62. [Eratum. J Clin Invest 1990186:

2t7s.lYmamoto T, Noble NA. Miller DE, Border WA. Sustained expression ofTGF-pl underlies development of progressive kidney fibrosis. Kidney Int1994:45:916-27.Coimbra T, Wiggins R. Noh JW. Menitt S, Phan SH. Transfoming growth

factor-B production in anti-glonrerultr basement membrane disease in therabbit. Am J Pathol l99l:138:223-34.Bames JL. Abboud HE. Temporal expression of autocrine growth factorsconesponds to morphological features of mesangial proliferation in Habusnake venom-induced glomerulonephritis. Am J Pathol 1993;143:1366-76.Jones CL, Buch S, Post M, Mcculloch L. Liu E, Eddy AA. Pathogenesis ofinterstitial f,brosis in chronic purine aminonucleoside nephrosis. Kidney Intl99l:40:1020-3 l.Yamamoto T, Nakamura T, Noble NA. Ruoslahti E, Border WA. Expres-sion of transforming growth factor É is elevated in human and experimentaldiabetic nephropathy. Proc Natl Acad Sci U S A 1993:90:1814-8.Kopp JB. Klotman ME. Adler SH, et al. Progressive glomerulosclerosis and

enhanced renal accumulation of basement membrane components in micetransgenic for human immunodeficiency virus type I genes. Proc Natl AcadSci U S A 1992:89:1577-8t.Kaneto H, Monissey J. Klahr S. Increased expression of TGF-BI mRNA inthe obstructed kidney of rats with unilateral ureteral ligation. Kidney lnt1993.44:313-21.Kagami S, Border WA. Miller DE. Noble NA. Angiotensin II stimulatesextracellular matrix protein synthesis through induclion of lransfominggrowth factor-p expression in rat glomerular mesangial cells. J Clin Invest199493:2431-7.Czaja MJ, WeinerFR, Flanders KC, et al. In vitro and in vivo association ofransfoming growth factor-pl with hepatic librosis. J Cell Biol 1989;108:2471 -82.Westergren-Thorsson G. Hemnâs J, Sâmstrand B, Oldberg Â. HeinegàrdD, Malmstrôm A. Altered expression of small proteoglycans, collagen, and

transfoming growth factor-Bl in developing bleomycin-induced pulmonuyfibrosis in rats. J Clin lnvest 1993:92:632-7.Khalil N, Whitman C, Zuo L, Danielpour D, Greenberg A. Regulation ofalveolar macrophage transfoming growth factor-B secretion by corticoster-oids in bleomycin-induced pulmonuy inflmmation in the rat. J Clin Invest1993t92:1812-8.Wolf YG, Rasmussen LM, Ruoslahti E. Antibodies against transfominggrowth factor-Bl suppress intimal hyperplasia in a rat model. J Clin Invest1994:93:1172-8.Logan A, Berry M. Gotzalez AM, Frautschy SA, Spom MB, Baird A.Eflècts of transfoming growth factor Bl on sca production in the injuredcentral nervous system of the rat. Eur J Neurosci 1994;6:355-63.Barcellos-Hoff MH, Derynck R. Tsang ML-S, Weatherbee JA. Transfom-ing growth factor-B activation in ifiadiated murine mammary gland. J Clinlnvest 1994;93:892-9.Appleton I, Tomlinson A, Colville-Nash PR, Willoughby DA. Temporaland spatial immunolæalization of cytokines in murine chronic granuloma-tous tissue: implications for their role in tissue development and repairprocesses. Lab Invest 1993169:405-14.

THE NE\V ENGI,ANDJOURNAL OF N'TEDICINE

64

Williams RS, Rossi AM, Chegini N, Schultz G. Effect of lransforminggrowth factor p on postoperative adhesion fomration and intact Peritoneum.J Surg Res 1992;52:65-70.Yoshioka K. Takemura T. Murakami K. et al. Transfoming growth factor-

B protein and mRNA in glomeruli in normal and diseased human kidneys.Lab Invest 1993:68:154-63.Shihab F, Yamamoto T. Nast C. et al. Acute and chronic allograft rejecrionin the human kidney corelate with the expression ofTGF-B and extracellu-lar matrix proteins. J Am Soc Nephrol 1993i4:671. abstract.Border W. Yamamoto T. Noble N, Gold L, Nast C, Cohen A. HlV-associ-ated nephropathy is linked to TGF-B and matrix protein expression in humankidney. J Am Soc Nephrol 1993;4:675. abstract.Castilla A, Prieto J, Fausto N. Transfoming growth factors Ê1 and a inchronic liver disease: effects of interferon alfa therapy. N Engl J MedI99l:324:933-40.Nagy P, SchaffZ, Lapis K. Immunohistochemical detection oftransforminggrowth factor-Bl in fibrotic Iiver diseases. Hepatology l99l:'14:269-73.Broekelmann TJ. Limper AH, Colby TV, McDonald JA. Transfominggrowth factor Br is present at sites of extracellular matrix gene expression inhuman pulmonary'librosis. Proc Natl Acad Sci U S A 199l:88:6642-6.Deguchi Y, Spontâneous increase of transforming growth factor B produc-tion by bronchoalveolar mononucletr cells of patients with systemic auto-immune diseases affecting the lung. Ann Rheum Dis 1992:'51:362-5.Kulozik M, Hogg A, Lankat-Buttgereit B, Krieg T. Co-localization of trans-forming growth factor B2 with a I (l) procollagen mRNA in tissue sections ofpatients with systemic sclerosis. J Clin lnvest 199O86:917-22.Peltonen J, Hsiao LL, Jaakkola S, et al. Activation of collagen gene expres-sion in keloids: co-localization of type I and VI collagen and transforming

Srowth factor-Bl mRNA. J lnvest Dematol l99l;97:2.10-8.Ghahary A, Shen YJ, Scon PG, Gong Y, Tredget EE. Enhanced expressionof mRNA for transforming growth tâctor-p, type I and type III procollagenin human post-bum hypertrophic scar tissues. J Lab Clin Med 1993:122:465-73.Varga J, Uitto J, Jimenez SA. The cause and pathogenesis of the eosinophil-ia-myalgia syndrome. Ann Intem Med 1992:l l6:140-7.Nikol S, Isner JM, Pickering JG, Kearney M. Leclerc G. Weir L. Expres-sion of transforming growth factor-p I is increased in human vascular reste-nosis lesions. J Clin Invest 1992:,90:1582-92.Connor TB Jr, Roberts AB. Sporn MB, et al. Corelation of fibrosis andIransfoming growth factor-p type 2 levels in the eye. J Clin Invest 1989;83:

t66t-6.Lafyatis R, Thompson NL, Remmers EF. et al, Translbrming growth factor-p production by synovial tissues from rheumatoid patients and streptococcalcell wall arthritic rats: sttrdies on secretion by synovial fibroblast-like cellsand immunohistologic localization. J Immunol 19891143:1 142-8.

Ohno I, læa RG. Flanders KC, et al. Eosinophils in chronically inflamedhuman upper airuay tissues express transforming growth factor Bl gene(TFGpl). J CIin lnvest 1992:89:1662-8.Tommka S. Border WA, Mrshall BC, Noble NA. Glomerular matrixaccumulation is linked to inhibition of the plasmin protease system. Kidneylnt 1992:,42:1462-9.Kagami S, Border WA, Ruoslahti E, Noble NA. Coordinated expression ofBl integrins and transfoming growth factor-p-induced matrix proteins inglomerulonephritis. Lab lnvest t993;69:68-76.Floege J, Eng E, Young BA, et al. lnfusion of platelet-derived growth factoror basic libroblast growth factor induces selective glomerular mesangial cellproliferation and matrix accumulation in rats. J Clin lnvest 1993:92:2952-62.Border WA, Okuda S, Languino LR, Spom MB, Ruoslahti E. Suppressionof experimental glomerulonephritis by antiserum against transfominggrowth factor Bl. Nature 1990:346:371-4.Border WA. Noble NA. Yamamoto T, et al. Natural inhibitor of transfom-ing growth factor-B protects against scaning in experimental kidney disease.

Nature 1992:360:361-4.Ziyadeh FN, Sharma K, Ericksen M, Wolf G. Stimulation of collagen gene

expression and protein synthesis in murine mesangial cells by high glucoseis mediated by autocrine activation of transforming growth factor-p. J ClinInvest 1994:93:536-42.Shah M, Foreman DM, Ferguson M\vV. Control of scaring in adult wounds

by neutralising antibody to transforming growth factor B. Lancet 1992;339:213-4.Giri SN, Hyde DM, Hollinger MA, Effect of antibody to transforminggrowth factor p on bleomycin induced accumulation of lung collagen inmice. Thorax 1993:48:959-66.Vr'ahl SM, Allen JB, Costa GL, Wong HL, Dasch JR. Reversal of acute and

chronic synovial inflammation by anti-transfoming growth factor p. J ExpMed 1993:177:225-30.Okuda S, Nakamura T, Yamamoto T, Ruoslahti E. Border WA. Dietaryprotein restriction rapidly reduces transfoming growth factor Bl expressionin experimental glomerulonephritis. Proc Natl Acad Sci U S A l99l;88:9765-9.

Nov. 10. 1994

25.26.

27.

50.

5l

52

53.

28

29

30,

ll

32

33

56

57.

58

59

60.

61.

62

63

65.

34

35

36

37

38

39

40

4l

54.

55.

67

68

69,

70.

7t.

42

43

73

45

46

48.

\rol. 3l'|l \o. 19

CASE RECORDSOF THE

MASSACHUSETTS GENERAL HOSPITAL

**0" :'.:îTi::::::l *':*.'

"'Roeanr E. Scur-lr'. Nl.D., Editor

EucrxE .f . \I,ux, trI.f).. ,lssorlale Editor

Wrt-rtelt F. NIcNrEr-v. \,I.D.. ,{ssoczale Editor

Berrv U. \lc\art-v. Assistant Editor

CASE 40-1994

PnrsrNrarroN oF CASE

A 77-vear-old rvoman was admitted to the hospitalbecause of pain in the head and legs. fever, andsweats.

The patient had been r'vell until four lveeks earlier,ivhen she began to experience pain in the neck thatextended over the lon er third of the left side of thehead and lelt mandible. The next morning she had anacute onset of "burning" in the lorver legs, as il thevrvere "on fire." That night she changed her nightgolvnse',.eral times because of sr.l,eats. After several days ofsuch symptoms the patient consulted a dentist; radio-graphs w'ere reported to sho'w'a dental abscess, andcephalexin vr,as prescribed, rvithout improvement inthe pain. ThreeTveeks before admission the patient'sphysician prescribed indomcthacin, rvithout improve-ment. Sixteen da;'s before admission she entered an-other hospital.

She had a history ol hypertension of'more than l0years' duration, which was manased with spironolac-tone-hvdrochlorothiazide; she also used lovastatin be-cause of hvperlipidemia. She had lost 1.5 kg in weightduring the preceding rveek.

The temperature !\ras 37.3"C. the pulse rvas 104,and the respirations rvere 20. The blood pressure was145175 mm Hg. Phvsical examination rvas negative;no rash, l1'mphadenopathy, or temporal-arterv ten-derness rvas detected. The blood chemical findingswere normal. Th1'roid-function tests; tests lor antinu-clear antibodies, rheumatoid factor, hepatitis A anti-body, hepatitis B surface antigen and antibodr', hepa-titis B core antibodl', and hepatitis C antibody; and aserologic test for Borrelia burgdorÿri were negati\.e. Theresults of other laboratory studies are shorvn in TableL Six blood cultures u'ere sterile. A cardiac ultrasono-graphic examination !ÿas negative. Radiographs o[

CASE RECORDS OF]'HE \IASS.ACHUSETTS GE\ER,\L HOSPITAI, I 293

the chest and both ankles and feet n'ere normal. Acomputed tomographic (C'l') scan of the abdomenand pelvis showed no abnormalitv except a small cystat the upper pole of the left kidne,v.

The patient's customary medications were discon-tinued. The temperature rose to 40'C on the secondhospital dav. Marked srvelling of the scalp der.eloped,with tender lumps in the temporal and occipital areasbilateralll,. Ampicillin was given by vein for two days,then ceftriaxone was substituted and continued forone lveek. On the fifth hospital day prednisone (60 mgdailv) lvas begun, and the dose rvas later reduced to 40mg daily for a total course of l0 davs. Fever rvith thetemperature as high as 3B.5oC persisted during theadministration of prednisone, and pain persisted inthe head and lorver lees, althoueh the patient was ableto rvalk rvith the aid of a stationar)' rvalker. The scalpswelling and tender lumps resolved during treatmentwith prednisone, and the patient's appetite improved.On the l Tth hospital da1' she was transferred to thishospital.

The patient was widorved and lived alone in awooded area. She had had a dry' cough for manyyears. She had not smoked tobacco for more than30 years and ingested little alcohol. Nlammogramsobtained nine months earlier were normal. Her medi-cations at the time of the transfer to this hospitalwere prednisone (40 mg dailv) and meperidine (75 mgeverv three hours as required). There was no historyol visual problems, claudication of the jaw, lymphade-nopath,v, dvsphagia, cough, sputum, wheezing, he-moptysis, dvspnea, abdominal pain, nausea, r.omit-ing, diarrhea, dvsuria, drug allergies, swelling ofthe extremities, rash, recent or previous arthritis,Ra,vnaud's phenomenon, tral'el, exposure to farmor wild animals. risk factors for infection u'ith thehuman immunodeficiencv virus (HI\r), exposure totuberculosis, recent insect bites, or ingestion of un-usual or unpasteurized foods.

The temperature lvas 3B.3oC, the pulse rvas 96, andthe respirations were 22. The blood pressure lvas140/80 mm Hg.

Ph,vsical examination lvas negative. No rash, cuta-neous signs of sepsis or vasculitis, or lymphadenopa-thy was found. The head and neck w-ere normal; noscalp or temporal-artery tenderness or lumps were de-tected. The lungs, heart. breasts, and abdomen werenormal. No articular abnormalitv or muscle tender-

Table 1. Laboratory Values at the Other Hospital.

VARIÂBLI

Hemoglobin (8/dl)

Erythrocyte sedimentation rate (mrdhr)

White-cell count (per mmr)

Platelet count (per mmr)

Alkaline phosphatase

Aspanate aminotransferase

7-Glutamyltransferase

VALUE

t2.6

l2l, then 94

29,700-42.700

up to 970.000

Increased

lncreased

lncreased

129-1 THE \E\\- E\GLA\D.JOL R\AI, OF }IEDICI\E \ov. 10. 199{

V{RIABLE

Hematocrit (7o)

Mean corpuscular volume (pmr)Erythræyte sedimentation rate (mtrÿhr)

White-cell count (per mmr)

Differential count (7c)

NeutrophilsLymphocytesMonocytesEosinophils

Platelet count (per mm3)

Table 2. Hematologic Values on Admission. results of serum immunoelectrophoresis and agarose-gel electrophoresis were normal.

Prednisone and meperidine lvere discontinued, andacetaminophen \!-as given. The patient continued tohave intermittent pain in the lower legs but was ableto rvalk n-ith a stationarv walker. The temperaturerose dailv to the ranse of 38.9 to 39.1"C. The bloodpressure did not exceed 140/80 mm Hg. Repeatedphvsical examinations were unchanged. Blood cul-tures $'ere sterile except that one flask grew a co-asulase-negatx,e Staphllococcus aureus; a urine culturevielded moderate enterococci. A urinalvsis was nega-tive; the sediment contained I red cell and 4 whitecells per high-porver field. On the sixth hospital daymicroscopical examination of a biops-v specimen fromthe left temporal arterv sho$,ed intimal fibrosis andIbcal medial calcification. n,ithout evidence of arteri-

Table 4. Results of LiverFunction Tests.

Hosprral SERUM ASPARTATE Sr:ruu ArlNtlt ALKAT-rNE

DAY AMINoTRANSFERASE AMINOTRANSFERASE PHOSPHATASE BILIRUBIN*

ngltll

0.3

0.7

0.5

0.4

VALUE

3t .7

92

90

22.500

78

697

680.000

ness rvas found. Neurologic examination lÿas grosslynegative. The urine \4'as normal except that the sedi-ment contained B white cells and moderate numbersof bacteria per high-pou,er field. 'fhe hematologic andblood chemical findings are presented in Tables 2

and 3. Radiographs of the chest, paranasal sinuses,and mandibles lvere normal. Skin tests with tubercu-lin (purified protein derivative IPPD], 5 TU) and can-dida antigen were negative; a test with mumps antigen\{,as positive. Tests for antineutrophil cvtoplasmicantibodies, hepatitis A antibodv, hepatitis B surfaceantigen and antibod)., and hepatitis C antibodr.u'erenesative. Nlicroscopical examination of an aspiratedspecimen of bone marrow showed normal cellularitywith relative mveloid hvperplasia; iron stores wereabundant; no malignant-tumor cells were seen. The

Table 3. Blood Chemical Values onAdmission.*

I

3

6

8

9

12

146

130

159

144

128

r5r

*To conveft values for bilirubin lo micromoles per liter. multiplt'. by 17.1.

tis: a single small adventitial inflammatorv infiltratethat contained numerous eosinophils $'as considerednonspecific.

On the seventh hospital day indomethacin (25 mgthree times dail,v) was begun, but the temperaturecontinued to rise on most days to 39.1"C, and on tu'odavs it reached 39.6"C. An ultrasonographic examina-tion of the abdomen showed a normal liver and a mildright hvdronephrosis. A consultant in neuromusculardisorders lound that the patient rvas alert and fullvoriented, lvith fluent speech and a normal mental sta-tus. A mild head-nodding tremor \,r'as present. Crani-al-nerve functions \,r'ere intact. N{otor po\ÀIer r,vas 5/5in all major muscle sroups; extension and eversionof the toes \,r,ere graded 4*/5, with decreased tone.Sensation was reduced distal to the midpalms andknees. Coordination was normal. The gait was nar-rou,-based, with an equivocal Romberg sign. The ten-don reflexes lvere *+ except that the anklejerks rvere* bilateralll,; the plantar responses u'ere flexor.

Nerve-conduction studies and electromvographicexamination were performed. NIotor-nerve conduc-tion was normal bilateraliy; no conduction block wasobserved. The sural-nerve action potentials were un-obtainable bilaterallv, and the right ulnar sensory-nerve action potential was of low amplitude. The

90

28

.15

50

40

53

Ulliter

293

149

87

19

60'74

VARIABLL

Urea nitrogen

Creatinine

Protein (g/dl)AlbuminGlobulin

Glucose (mg/dl)

Calcium (mg/dl)

Phosphorus

Iron (pgldl)

Iron-binding câpacity (pgldl)

Bilirubin

Sodium (mmolrliter)

Potassium (mmol/liter)

Chloride (mmol/liter)

Cubon dioxide (mmol/liter)

Magnesium (mmol/liter)

Aspânate aminotransferase (U/liler)

Alanine aminotransferase (U/liter)

Creatine kinase (U/liter)

Alkaline phosphatase (U/liter)

NomalNomal

5.6t.73.9

187

7.7

Nomal

9

t31

Normal

r37

r00

3t.90.65

90

293

1l

146

xTo conven values for glucose to millimoles per liter. multipl!bt" 0.0555 l: ro convefr values for calcium to millimoles per liter.mulliply by 0.250; to conven values for iron and iron-bindingcapacil) to micromoles per liter. mùlliply bl- 0.1791: and toconven values for magnesium lo milliequivâlents per liter. divideby 0.5.

Vol. 331 No. l9

B

Figure 1. CT Scan of the Chest through the Upper (Panel A) andLower (Panel B) Lungs, Demonstrating Bilateral Multifocal

Ground-Glass Opacities Involving All Lobes.

tibial F responses and H reflex lvere prolonged bilater-ally. Needle electroml'ographic examination disclosedspontaneous activity in the flexor hallucis brevis mus-cles on each side. The overall pattern was consistentwith a generalized axonal sensorimotor polyneuropa-thy. 1-he white-cell count graduallv declined to 15,200on the I lth hospital dav; repeated differential countsshowed 0 to 4 percent eosinophils. Nlultiple liver-func-tion tests revealed a trend toward improvement in theaspartate aminotransferase and alanine aminotrans-ferase levels (Table 4) . On the l3th hospital day thepatient coughed more frequently and produced asmall amount of clear sputum. A CT scan of the chest(Fig. l), performed rvithout the intravenous adminis-tration of contrast material, demonstrated bilateralmultifocal patchv and ground-glass opacities in alllobes; atelectasis and pleural thickenine were ob-served at both lung bases; no abnormal lymph nodes

C.{SE RECORDS OFTHE \{ASSACHUSEI'TS GENERAL HOSPI'I'AL

were observed; coronary arterial and mitrai annularcalcifications were present.

A diagnostic procedure r,l'as performed.

DrrrpnBNtteL DrAcNosrs

DR. Mrcnerl Havps*: This 77-year-old womanhad an acute onset ofjaw and neck pain and burningsensations in the lower extremities, with episodes ofdiaphoresis. She r,r'as initiallv treated lbr a dental ab-scess. Examination revealed tachycardia and the de-velopment of scalp swelling, subcutaneous nodules,and findings consistent with a peripheral neuropathv.The pertinent laboratory findings included very ele-

vated white-cell counts, a verv high sedimentationrate, and abnormal results of liver-function tests, anda temporal-arterv biopsy revealed only a small focusof eosinophilic infiltration in the adventitia.

May we review the nerve-conduction and electro-myographic studies, Dr. Cros?

Dn. Drorr,n P. Cnos: ÿIotor and sensory-nerve con-duction studies were performed in the upper and lowerextremities. The amplitude of the compound motorresponses was low in the distribution of both peronealnerves (recorded from the extensor digitorum brevismuscie) and normal in the distribution of the tibialnerves. Conduction velocities were normal, and noevidence of conduction biock was found. Sural sen-sory-nerve action potentials were undetectable bilater-ally; the ulnar sensory-nerve action potentials wereof low amplitude bilaterally. The F responses \^r'ere

slightlv delaved in the tibial distribution bilaterally,and the soleus H reflex was also slightly delayed onboth sides. Needle electrom,vographic studies rn'ere

negative in the thigh and lower muscles. Spontaneousactivitv (fibrillations and positive sharp waves) wasseen in both plantar muscles. Blink-reflex studies werenegative bilaterally. These studies disclosed evidenceof a generalized sensorimotor axonal poiyneuropathy.

Dn. Heves: Dr. Shepard, may we see the radiologicfindings?

Dn. Jo-Axxr O. Snrpano: The x-ray films of thechest were normal. A CT scan of the chest (Fig. 1),with high-resolution imaging, disclosed a diffuse pa-renchymal pulmonary abnormality consisting of mul-tiple patchy and ground-glass opacities in a nonseg-mental distribution throughout all lobes, probablyreflecting an air-space process.

Dx.. Hevrs: Although the clinical picture and theresults of the electrodiagnostic studies were consistentwith a peripheral neuropathy, the onset of the dis-order lvas not typical of an axonal polyneuropathy.which generally presents more insidiously. The sud-den onset and the nature and severity of this patient'spain are consistent with nerve ischemia or infarction.

*Staff physician. Division of Neurology, Department of Medicine, St. Eliza-beth's Hospitali assistant Professor of neurology, Tufts University School ofMedicine.

I 296

If multiple nerves are involved in this manner, thediagnosis of mononeuritis multiplex is usually enter-tained.r The clinical and electrophysiologic hallmarkof that disorder is the involvement of multiple nerves,with other nerves totally una{Iected. If the involve-ment is widespread, the disorder becomes indistin-guishable from a generalized polyneuropathy. CIearly,the findings on electrophysiologic studies in a consid-erable proportion of cases of neuropathy caused by amicrovascular occlusive process are consistent with ageneralized axonal sensorimotor polyneuropathy. Theclinical history must be eiven great weight in thesecases. Ischemic neuropathies present suddenll' withpain and severe disability. There is often pain at thesite of the nerve infarct,2 which is constant and burn-ing. The pain, unlike that associated n,ith a typicalaxonal process, is not preceded by paresthesias or by agradual ascending sensory loss. The common causesof generalized polyneuropathy, such as uremia, "ita-min 8,, deficiency, hypothyroidism, and toxins, canbe readily ruled out in this case, since other clinicaland laboratory features do not support any ol thesediagnoses.

The differential diagnosis o[ painful neuropathy ofsudden onset includes diabetes, sarcoidosis, and vas-culitis.

Diabetes can cause ischemic nerve lesions as well as

a generalized polyneuropathy,3 sometimes with a highsedimentation rate.t The single blood glucose level re-ported in this case was elevated at lB7 mg per deciliter(10.4 mmol per liter). but rve are not told whether thespecimen was obtained while the patient was fastingor during the administration of high-dose corticoster-oids. However, other features of the case. especiallythe skin lesions and high eosinophil count, are notassociated lvith diabetes.

Sarcoidosis is a chronic multisystem disease inwhich noncaseating granulomas in various organs areassociated with derangement of the normal tissuestructure.5'6 The findings are protean and may includemild eosinophilia, an elevated sedimentation rate, pe-ripheral neuropathy or mononeuropathy multiplex,Tand lung involvement affecting the alveoli, small bron-chi, small blood vessels, and occasionally the largevessels. In the so-called acute form of the disease con-stitutional signs, such as fever, anorexia, and weightloss, may be seen.5'6'B We are not told whether theangiotensin-converting eîzyme level was determined;it is elevated in approximatel,v two thirds of cases,6,sbut a normal level does not rule out the diagnosis.Several features of this case, hor.r'ever, are unlike thoseof sarcoidosis. Hilar lymphadenopathy, which is seenin up to 90 percent of cases of sarcoidosis, was absentin this case. The sedimentation rate, r,r'hich ma,v beelevated in sarcoidosis, is not as high as that seenrepeatedly in this case. Hypergammaglobulinemia iscommon in patients with sarcoidosis but was absent inthis patient. Also, although subcutaneous nodules areobserved in sarcoidosis, they usually involve the trunk

THE \E\\'E\GL.{\D JOUR\AL OF \IEDICI\E \ov. 10. 1994

and limbs and are not associated with angioedema,which was present in this case. For these reasons adiagnosis of sarcoidosis is unlikely.

Another disorder that was initially appealing in thiscase is the hypereosinophilic syndrome,r0'12 w'hich ischaracterized by peripheral eosinophilia and eosino-philic infiltration of various organs. The disease pre-sents in several forms, often r,r'ith lung involvementand angioedema. Peripheral neuropathy is seen in 6 to14 percent of the casesr3ra; it is not caused by eosino-phil infiltration or vasculitis but results from abnor-malities seen in nerve-biopsy specimens that are pos-tulated to be secondar,v to injury to the blood-nervebarrier, with a consequent increase in endoneural-fluid pressure.r+ Dystrophic changes in Schwann cellsand dissolution of myelin and axonal damage u'ithwallerian degeneration are seen. Toxins released byeosinophils have been postulated to be the cause. Fe-ver, weight loss, and malaise are also observed withthis disease, as they rvere in the patient under discus-sion. However, this patient did not have the hyper-eosinophilic syndrome. Cardiac involvement, which isprominent in the hypereosinophilic syndrome,r0 wasabsent in this case. Moreover, the diagnosis rests onprolonged peripheral hypereosinophilia and an in-creased number of eosinophilic promyelocytes evidenton bone marro\^r' biopsy, and neither feature was pres-ent in this case.

Did the patient have some form of vasculitis? Vas-culitis is a major consideration in the diferential diag-nosis of ischemic neuropathy15-r7 or any neuropathyassociated with a very elevated sedimentation rate.NIany classifications of vasculitis have been suggest-ed. The simplest ones classify vasculitis according tothe size of the vessels involved.rs-20

The large-vessel arteritis group includes'fakayasu'sarteritis and temporal arteritis. Takal''asu's arteritis,2lwhich causes symptoms related to the aortic arch andthe carotid and renal arteries, is easily ruled out in thiscase. Although the high sedimentation rate and ane-mia and possibly the jar.r'pain led to a temporal-arter-vbiopsy, temporal arteritis, with the typical symptomsof headache, jaw claudication, visual changes, andoculomotor-nerve palsies, clearly does not explainmost of this patient's s)'mptoms. Horvever. temporalarteritis has been reported to occur simultaneousl,vwith other vasculitic syndromes.22'23

Hypersensitivity vasculitis encompasses a heteroge-neous group of disorders that presumably result froma h,vpersensitivity reaction to endogenous or exoge-nous antigens.re The vasculitis mav a{fect any organsystem, but the skin is usually the most severely afÏect-ed; other organ systems are generally much less mark-edly affected than they'are in cases of necrotizing vas-culitis. Immune-complex deposition is believed to bethe inciting mechanism. The antigenic trigger may beexogenous, as in Henoch-Schônlein purpura, serumsickness, mixed cryoglobulinemia. and drug-inducedvasculitis, and the disease is occasionally postinfec-

Vol. 331 \o. 19

tious. None of these causes are relevant in this case.Hypersensitivity vasculitis can also be caused by anendogenous trigger,2a as with systemic lupus erythe-matosus, rheumatoid arthritis, and Sjôgren's syn-drome and more rarely with other collagen-vasculardiseases. The absence of the classic clinical findings ofsystemic Iupus erythematosus, rheumatoid arthritis,and Sjôgren's syndrome and the negative tests forantinuclear antibodies and rheumatoid lactor makethese diagnoses improbable. This form of vasculitiscan also be part of a paraneoplastic syndrome,rT butthere is nothing in the case record to suggest a malig-nant tumor.

Necrotizing vasculitis includes Wegener's granulo-matosis, lymphomatoid granulomatosis, polyarteritisnodosa, the Churg-Strauss syndrome, and the polyar-teritis overlap syndromes.

In lymphomatoid eranulomatosis various organsystems are invaded by angiocentric polymorphonu-clear inflammatory infiltrates that also contain lym-phocytes. histiocvtes, plasma cells, and atypical lym-phoreticular cells. Vascular occlusion and necrosis oftissue occur.2+'25 The lungs, kidneys, skin, and centraland peripheral nervous system may all be involved.The histologic and clinical features of the disease aresimilar to those of a malignant lvmphoreticular neo-plasm, and a fatal lymphoma develops in many pa-tients. However, the lung lesions are often cavitating,peripheral neuropathy occurs in only about 7 percentof the cases,26 and leukopenia and anergy to skin test-ing are usually seen. Neuropathy is a central lèature ofthis case, no cavitating lung lesions were seen, andthere was neither anergv nor leukopenia. For thesereasons the diagnosis of lymphomatoid granulomato-sis is unlikely.

The first subtype of necrotizing vasculitis to be rec-ognized as a distinct entity was Wegener's granuloma-tosis,27 an illness characterized by a granulomatousvasculitis. It can affect virtually any organ system andis often accompanied by fever. malaise, anemia, a highsedimentation rate, and mild eosinophilia.2'1'28 In-volvement of the peripheral or central nervous systemand the skin is common.2s'30 However, the disease usu-ally presents with severe upper respiratory tract mani-festations, including rhinorrhea and purulent nasaldischarge often complicated by mucosal ulceration orseptal perforation. In addition, the absence o[ eye andkidney involvement and the negative test for antineu-trophil cytoplasmic antibodies, although not sufficientto rule out the diagnosis, make it improbable.

The second subtype to emerge as separable frompolyarteritis nodosa is the Churg-Strauss syndrome,3rwhich is characterized by fever, a high sedimenta-tion rate, and peripheral eosinophilia. The peripheralnervous system is frequently involved. Skin involve-ment is more frequent in the Churg-Strauss syndromethan in polyarteritis nodosa, especially in the form oftender nodular lesions that may be found in the limbsand scalp, as in this case.32 Extravascular eosinophilic

CASE RECORDS OF ]'HE \IASSACHUSETTS GENER.{L HOSPIT.{L

infiltrates, as seen in this patient's temporal-artery-biopsy specimen, mav be found even in the absence ofperipheral eosinophilia.32 The Chure-Strauss syn-drome has also been called "allergic granulomatosisand angiitis" because of the histologic finding of ne-crotic eosinophilic exudates with a proliferation olepithelioid and giant cells, but these sranulomas arefrequentl,v not found, even at autopsy. A recent set ofcriteria proposed for the diagnosis of the Churg-Strauss syndrome33 emphasizes that the syndrome isdistinguished from polyarteritis nodosa on clinicalrather than histologic grounds, including the presenceof asthma, a very high peripheral eosinophil count,and a documented history of allergy other than asth-ma or drug sensitivitv. The findings in this case do notfulfill these criteria.

Polyarteritis nodosa is the diagnosis that I favor. Itinvoh,es small and medium-sized arteries in multipleorgan systems, including the kidneys, muscles, bow-el, liver, and skin.3{'35 The lungs are generally lessinvolved than in the Churg-Strauss syndrome, al-though some studies indicate that they are affected inup to 25 percent of cases. Because of the Iung involve-ment in the case, an argument might be made forclassiÿing the illness as the polyarteritis overlap syn-drome. This term. when used to describe cases ofoverlap between polyarteritis nodosa and the Churg-Strauss syndrome,36 has generally been resen'ed forcases with a combination of symptoms that are pecu-liar to each disease, such as a case in which the patienthas severe asthma and bowel infarction.

In over hall the cases of polyarteritis nodosa theperipheral nervous system is a site of vasculitis in theform of either mononeuritis multiplex or symmetricsensorimotor polyneuropathr,. The vasculitis involvesepineurial arterioles 50 to 300 ;r,m in diameter, caus-ing a characteristic central fascicular pattern of nerve-fiber damage.'ri The generalized symptoms of polyar-teritis nodosa include fever, malaise, and m,valgia, andthe laboratory findings include a high white-cell countand an elevated sedimentation rate. A positive hepati-tis B surface antigen or antibody titer is seen in 30 to50 percent of cases but is rarely, if ever, found inpatients with pulmonary involvement.sB The vasculitisresults in aneurvsmal dilatations and beading and oc-clusions of the medium-sized arteries, and in cases

with atypical symptoms or equivocal biopsy findingsthe diagnosis is aided bv arteriographic examination,which shows aneurysms of the renal, hepatic, and vis-ceral vasculature.:i5

The diagnosis of polyarteritis nodosa is made bybiopsy. In this case the lung was the most clearlv andmost prominently affected organ. The ground-glassappearance on the CT scan suggests an alveolitis orpossibly microgranulomas3s; in this case it suggests a

generalized alveolitis. A lung biopsy could have beenperformed, but I would have favored a sural-nervebiopsy or a nerve and muscle biopsy as a less morbidprocedure. The diagnostic yield of a nerve biopsv is

t 298 THE NE\\' ENGLÀND JOURNAL OF \{EDICINTI Nor'. 10. 199-i

lairlv high and is improved with rhe addition of amuscle biopsy'.lu A biopsy' of an electroph.vsiologicallyabnormal nervc is believed bv some obsen'ers to bemore likely to shorv evidence of vasculitis.+r I u'ouldhave expected the biopsv to show perivascular andtransmural inflammatory-cell infiltration and patchy.asymmetric nerve-fiber loss within the nerve fascicles.

Making the diagnosis of polyarteritis nodosa is veryimportant to the patient, since the disease may re-spond u,ell to immunosuppression. In one study thefive-year survival rate for patients receivins a resimenincludins prednisone and cvclophosphamide rvas 90percent, as compared n'ith a fir.e-vear survival rate ofl3 percent in untreated patients.si

Dr. E. Tessl Hpor.ry-lVnyrr: Dr. Cros. rviil r.outell us u'hat the clinical diagnosis rvas before the diag-nostic procedure?

Dn. Cnos: Our reasoning v!'as verv similar to thatof Dr. Haves. \\'e thought that the patient had a vas-culitic neuropathy, probably caused by polvarteritisnodosa.

CrrNrclr DrecNosrs

Polyarteritis nodosa, with sensorimotor axonal neu-ropathv.

Dn. Mrcrrerr, Heyrs's DracNosrs

Polyarteritis nodosa.

Parrror,ocrcar DrscussloN

Dn. HEor-sv-WnvrE,: The diaenostic procedureconsisted of a muscle biopsy and a sural-nerl'e biopsv.The muscle fibers varied greatly in size, rvith groupingof small fibers and many hypertrophied fibers - find-ings consistent with neurosenic atrophy (Fig. 2). Thefiber-type grouping indicates that reinnervation hadoccurred. In addition, necrosis of fibers and intersti-

Figure 2. Muscle-Biopsy Specimen (x120).Groups of small angulated Jibers and hypertrophied fibers withcentral nuclei are present. A tew deeply stained necrotic fibers

and interstitial fibrosis are also visible.

tial fibrosis w'ere present, susgestins ischemia super-imposed on a chronic neuropathr'. 1'he histoeram ofmuscle-fiber size u,as broader than normal, w'ith anexcess of both small and larse iibers (Fig. 3). In con-trast to the usual bell-shaped curve. this cur\re is prob-

8roU)

o€u-§tq)

310F

S20o0)-oIE

o*ioF

Type 1, 59%Mean size. 45.0;rmN=163

10 20 30 40 50 60 70 B0

Fiber Size (pm)

fype 2, 41o/o

Mean size, 38.6 pmN=112

90 100>100 10 20 30 40 50 60 70 B0 90 100>100

Fiber Size (ptm)

Figure 3. Histograms of Type 1 and Type 2 Fibers in the Muscle-Biopsy Specimen.Both histograms are broad and shifted to the left. The number oT type 1 fibers is increased. The mean size of the type 2 fibers is reduced.

\'bl. 331 No. 19

Figure 4. Sural-Nerve-Biopsy Specimen (x220).There is fibrinoid necrosis ol the vessel wall, with an adjacenl

inf lammatory inf iltrate.

.i - ., ,. -t',. ' , ,, .. i,'. , f ,.". ., I ,

. :l.s_ ..; '- - '! j -:' ,*'--,' i - r, -:1_.' _ .

r 299

in the medium-sized mvelinated fibers. In retrospect,the inflammatory cells in thc temporai-artery-biopsvspecimen may have bcen a manifestation of the polr'-arteritis nodosa.

Dr<. Ar-laN H. Ropprn: f)r. Havcs, do you believethat this t1.pe o[ generalized axonal ncuropathv is sim-plv the summation of multiple mononeuropathieslvhen there is no clinical and electrophysiologic evi-dence of a single nerve being preferentiaily or com-pletell' aflècted, or should we accept the existence of-arelated syndrome ol polyneuropathy that is due tomany fascicular infarcts rather than to cumulativemononeuropathies? It is the summation of these in-complete nen'e lesions that is likelv to resemble a gen-eralized sensorimotor poll'neuropatht'.

Dn. Heyns: The rvidespread small-vessel chansesseen with vasculitis may result in lolr,-grade ischemiao\:er a length of nerve or mav cause microinfarcts olthe nerve that ciinicaily resemble an axonal neuropa-thy rather than the larser nerve infarct that lve gener-aliv associate with mononeuritis multiplex.

Dn. Htnrry-Wunrt,: 'fhe changes in the suralnerve in this patient consisted ol multiple areas ofdegeneration of individual fibers, not a total loss offibers within a single area.

Dn. DaNrpL L. jvlr,xras: The patient rvas dis-charged eight days after the biopsv, uith a regimen ofprednisone (60 mg per dav) and cvclophosphamide(100 mg per day).

ANarolrrcar Dra,cNosrs

Pojarteri tis nodosa, with neuropathl.

A»orNpuuDn. Cnos: \\rithin three days after discharge the

fever and slvelling had almost completely resolved.Three months thereafter the dose of prednisone wastapered over a period of eight months. 'fhe dose ofcvclophosphamide was reduced one month later to 50mg per day. One vear alter discharge the patient rvasreceiving this dose of c1'clophosphamide as mainte-nance therapy. She is asymptomatic except for persist-ent acral dysesthesias of her exfemities, and she per-forms her household tasks '"vithout anv di{Êcultv.

RrrrnrNcrs1. Chang RW, Bell CL, Hallet M. Clinical characteristics and prognosis of

vasculitic mononeuropathy multiplex. Arch Neurol 1984;41:618-21.2. Anatomical classification of PNS disorders. ln: Schaumburg HH, Spencer

PS, Thomas PK. Disorders of peripheral nerues. Philadelphia: F.A. Davis.1983:7 -23.

3. Metabolic neuropathy: diabetes. In: Schaumbur-q HH, Spencer PS, ThomasPK. Disorders ofperipheral nerues. Philadelphia: F.A. Davis. 1983:41-55.

4. Ganda OP. Markedly increased erythrocyte sedimentation rate. hyperfibrin-ogenemia, and peripheral vasculæ disease in diabetic patients: associationwith clinical implications. Am J Med 1988;85:584-5.

5. Siltzbach LE. James DG, Neville E, et al. Course and prognosis ofsrcoido-sis around the world. Am J Med 1974.51:847-52.

6. Poole GW. The diagnosis of sarcoidosis. BMJ 1982:285:321-2.7. Stern BJ. Krumholz A. Johns C. Scott P. Nissim J. Sarcoidosis and its

neurological manifestations. Arch Neurol 1985:42:909-17.8. Fanburg BL. Sarcoidosis and other granulomatous diseases of the lung. New

York: Marcel Dekker. 1983.

9. Rohrbach MS. DeRemee RA. Pulmonary sarcoidosis and serum angioten-sin-convening enzyme. Mayo Clin Proc 1982',57:64-6.

CÀST] RECORDS O} HE \I,\SSÀCHUSI]]'TS GENER.\I, HOSPIl"\L

,.s-. " **= **.É, ..1*, 1 ;iliifiCâffi*" ; ,{r* } U.,

,§'*-;':.;*., * r*" - .'"t" * . t-'§

Ot{.}>î.''*. _ **{;:Ji. --ç

-L--. -'--

- é * tt

- ^ ',1 :* -rl--1.,;:',+t :-r-q -:. .-.,. -'.--r "ül; l,t t*i.*

:ji1l Ei.f,i*rr'i:;æ§r'i.i*r;H = ;;;rt

Figure 5. Sural-Nerve-Biopsy Specimen (x90).An epineurial vessel, with surrounding fibrosis and mild inflamma-

tory-cell inTiltration, is shown.

ably biphasic, particularly for the type 2 fibers. Themean fiber sizes were only slightly smalier thannormal. The fat surrounding the sural nerve containeda medium-sized vessei with fibrinoid necrosis in itswall (Fig. 4) and an adjacent inflammatorv-cell infii-trate. In the epineurial connective tissue there wasanother vessel with extensive adventitial fibrosis (Fig.5), which is consistent with polyarteritis nodosa.Teased nerve-fiber preparations showed multiple my-elin ovoids, and the 1-pr,m epon-embedded sectionscontained axonal sprouts indicating wallerian degen-eration of the nerve fibers, with most ol the fiber loss

l 300 'IHE )i'Etr\: ENGLAND JOURNAL OF' I.'IEDIC INE Nov. 10. 1994

27

28

l0 Chusid MJ, Dale DC, West BC, Wolff SM. The hypereosinophilic syn-drome: analysis of fourteen cases with review of the literature. Medicine(Baltimore) 197 5:54:l-27.Fauci AS, Harley JB, Roberts WC, Fenans VJ. Gralnick HR. Bjomson BH.The idiopathic hypereosinophilic syndrome: clinical. pathophysiologic, andtherapeutic consideralions. Ann Intem Mred 198297 :'7 8-92.Enright T, Chua S, Lim DT. Pulmonary eosinophilic syndromes. Ann Aller-gy 1989;'62:277-83.Wichman A, Buchthal F. Pezeshkpour GH, Fauci AS. Peripheral neuropa-thy in hypereosinophilic syndrome. Neurology 1985135: I 140-5.Monaco S, Lucci B, Laperchia N, et al. Polyneuropathy in hypereosinophil-ic syndrome. Neurology 1988;38:494-6.Said G. Lecroix-Ciaudo C, Fujimura H, BIas C, Faux N. The peripheralneuropaahy of necrotizing arteritis. Ann n-eurol 1988t23:461-5.Dyck PJ, Benstead TJ. Conn DL. Stevens JC, Windebank AJ, Low PA.Nonsystemic vasculitic neuropathy. Brain 1987;l l0:843-53.Hawke SH. Davies L, Pamphlett R. Guo YP, Pollard lD, Mctæod JG.Vasculitic neuropathy: a clinical and parhological study. Brain l99l;114:2175-n.Leavitt RY, Fauci AS. Pulmonary vasculitis. Am Rev Respir Dis 19861

134:149-66.Scott DG. Classification and treâtment of systemic vasculitis. Br J Rheuma-tol 1988:27:251-3.Fulmer JD. Kaltreider HB. The pulmonary vasculitides. Chest 1982;82:615-

Shelhamer JH, Volkman DJ, Parillo JE, Lawley TJ. Johnston MR, FauciAS. Takayasu's arteritis and its therapy. Ann lntem Med 1985:103:l2l-6.Amato MBP, Barbas CSV, Delmonte VC, Carvalho CRR. ConcunentChurg-Strauss syndrome and temporal arteritis in a young patient with pul-monary nodules. Am Rev Respir Dis 1989:139:1539-42.Frayha RA, Abu-Haider F. Polyârteritis nodosa masquerading as temporalarteritis. J Rheumatol 19791.6:76-9.Nadeau SE. Collagen vascular disease: vasculitis. systemic lupus erythema-tosus ând rheumatoid arthritis. Semin Neurol 1985:5:324-43.Fauci AS. Haynes BF, Costa J, Katz P, Wolff SM. Lymphomatoid granulo-matosis: prospective clinical ând therapeutic experience over l0 years.N Engl J Med 1982;306:68-74.Kâtzenstein AL, Carington CB, Liebow AA. Lymphomatoid granulo-matosis: a clinicopathologic study of 152 cases. Cancer 19191;43:36O-73.

Wegener F. Uber eine eigenartige rhinogene Granülomatose mit besondererBeteiligung des Arteriensystems und der Nieren. Beitr Pathol Anat 1939;102:36-68.Wanen J. Pitchenik AE, Saldana MJ. Granulomatous vasculitides of thelung: a clinicopathologic approach to diagnosis and treatment. South Med J

1989182:481-91.Finkelman R, Munsat T, Mandell H, Adelman L, Logigian E. Neuromuscu-lar manifestations of Wegener's granulomatosis: â câse report. Neurology1993143:6 t 7-8.Drachman DA. Neurological complications of Wegener's granulomatosis.Arch Neurol 1963:8:145-55.Churg J, Strauss L. Allergic granulomatosis, allergic angiitis, and periarteri-tis nodosa. Am J Pathol 195l'.27:277-301.Lanham JG, Elkon KB, Pusey CD, Hughes GR. Systemic vasculitis withasthma and eosinophilia: a clinical approach to the Churg-Strauss syndrome.Medicine (Baltimore) I984;63:65-81.Masi AT, Hunder GG, Lie JT, et al. The American College of Rheumatol-ogy I 990 criteria for the classification of Churg- Strauss syndrome (allergicgmnulomatosis and angiitis). Arthritis Rheum 1990;33:1094-100.Guillevin L, Du LTH, Godeau P. Jâis P, \Vechsler B. Clinical findings andprognosis of polyarteritis nodosa and Churg-Strauss angiitis: a study in I 65patients. Br J Rheumatol 198827:258-64.Lightfmt RW Jr, Michel BA, Bloch DA, et al. The American College ofRheumatology 1990 criteria for the classification of polyarteritis nodosâ.Anhritis Rheum I 990:33: 1088-93.tJavitt RY, Fauci AS. Polymgiitis overlap syndrome: classifrcation andprospective clinical experience. Am J Med 1986;81:79-85.Kissel JT, Mendell JR. Vasculitic neuropathy. Neurol Clin 1992l,10:761-81.Stokes LT. Turner-Warwick M. Lungs and connective tissue disorders. In:Munay JF. Nadel JA, eds. Textbook ofrespiratory medicine. Vol. 2. Phila-delphia: W.B. Saunders, 1988: 1479.

Chiles C. Putman CE. Techniques for interpreting pulmonary opacities inthe ICU. J Crit Illness 1994:9:198-206.Panegyres PK, Blumberg PC, Leong AS, Boume AJ. Vasculilis ofperipher-al nerue and skeletal muscle: clinicopathological conelation and immuno-pathic mechanisms. J Neurol Sci 1990;100:193-202.Wees SJ, Sunwoo IN, Oh SJ. Sural nerve biopsy in systemic necrotizingvasculitis. Am J Med l98ll'11:525-32.

il

t2.

lJ.

t4.

15.

16.

t7.

18.

r9.

20.

2t.

22.

30.

31.

32.

33.

36.

37.

38.

39.

40.

23.

24.

25.

J+

35

26

The Massachusetts General Hospital wishes toacknowledge the generous support of

Glaxo lnc.,

whose sponsorship makes possible the continuedpreparation of the Case Records.

35-N,IILLIMETER SLIDES FOR THE CASE RECORDS

Any reader of the Journal who uses the Case Records of the N{assachusetts General Hospital as a medical teaching exerciseor reference material is eligible to receive 35-mm slides, with identiÿing legends, of the pertinent x-ray films, electrocardio-grams, gross specimens, and photomicrographs of each case. The slides are 2 in. by 2 in., for use with a standard 35-mmprojector. These slides, which illustrate the current cases in the Journal, are mailed weekly from the Department of Pathologyand may be retained by the subscriber. Each year approximately 400 slides are sent to each subscriber. The cost of thesubscription is $550 per year. Application forms for the current subscription year, which began on March I, 1994, may beobtained from Lantern Slides Service, Department of Pathology, Massachusetts General Hospital, Boston, Massachusetts021 14 (Telephone [617] 726-4369).

Vol. 331 -\o. 19

Harrv L. Grecne II, ll.DEterutii e l/ite Presitfu nl

Pcrcl \4/. llhthran. \1.D., ChairStrrart Anfàng, \I.D.Scott FairficldBarbara.f. \IcNcil. lI.D.

Erlrvirrd W Campion, \I.f).Robcrt Stcinbrook. lI.D.

Alan H. DeCherner', \I.D.\Iarshall \1. Kaplan, \I.D.

Ronald A. Arkr; \,I.D.

John T. Harrinuton, \{.D.Leonard J. \Iorse. \I.D.

E»tt<x.s

.fcronrc P Kassircr, \7.D.. Edilonin-Chief\{arcia Ansell, \7.1)., Ërucutii,e Editor

Arnolcl S. Relman, \1.D., Etlin»-in-Chief Enttritus

Depult,Editors

Gregorv D. Curliran, NI.l).

Açsociale Edilors

F'ranklin H. lipstein, II.D.Robcrt J. \la1'cr, \LI).

Clharles S. Arnorosino,.JrË.rerulii,e Secrelan'

Williani G. Lavcllc, trI.D.. l,ice ChairHenrv H. Banks, M.D.

.fames F. \{cDonough, Nl.l).Samucl O.'fhien \I.D.

Robert S. Schrvartz, lt{.f).Robcrt D. fltigcr, \{.D.

Lce Goldman, II.f).\lorton N. Srrartz, Il.I).

l3() I

Marl'Ellcn Avery; NI.D., BostonDar i<l Blunrenthal, \'I.D., lI.P.P., tsoston

Johu A. Collins, trI.f)., HamiltonL. f)avid Hillis. \{.D., Dallas

Àttilio \Iaseri. \I.D., Rome

Recd E. Pleritz, \I.D., Ph.D., PittsburghWilliam Silcn, NI.D., Boston

Bcrt Vogelstcin, II.l)., tsaltinroreRichard P. lVenzcl. \I.D., tr[.Sc., lota Citl

THE t(E!V DNGI-A.NDJOURNAL OF IIEDICTNE

The New England Journal of MedicineOrvned and Publishcd bv the Nlassachusetts }lcdical Societl'

Francis X. Van Houten, \1.D., Prridant

Tttr: Cotrrrrrrrr oN PuBr.tcÂ'r'ro\s oF THti \l,tss.tt.;truslrLs llrt»r:rt, SoctLl\

Hans-Olov Adarni, M.D., L-ppsala

John C. Bailar, III, l\I.D., Ph.l)., IlontrcalEugcne Braunu'ald, \I.D., Boston

Haner R. Coltcn. II.D.. St. LortisPaul H. Lange. \{.D., Scattle

Kenneth \{cIntosh, \{.D., BostonLovis P Rolland, It{.f)., Nor York

J.P. Soulillou, \1.D., \antesSterrn E. lVeinberyeç \I.D., Boston

John K. Iglchart, i\*alioml Corrupondenl

Nadia A. Roscnthal, Ph.D., Consultant in Moletular ,\Iedicine

Robcrt S. Schrvartz, \\.D. Bool; Ret'ieu Editor

Alexarrdcr tI. !\'alkc4 l\I.D, Dr.P.H.,James H. Ware, Ph.D., l\hlter lVillctt, \LD., Slatislirul Con.çullanls

Stepherr E. Cinto, trlanager ofEditorial ProducliotIlonnic Cooper,,lIanager o/' ConQuler S(n'ices

Patriciir \I. Larnv, )lonage r of Editorial Supporl ,ÿeniæs

Lrrrririne \\i Loviglio, ,llanager of ÀIanuscript Edi|ine

EDtr()RI^t, Bo,\Rt)

Karen H. Antman, \f.D., New YorkPetcrJ. Barnes, \1.D., L<»rdur

Christine K. Casscl, II.D.. ChicagoAnthonv S. Fauci, lI.D.. Bcthesda

Bob Lôrrenberg, \1.1)., Ph.D., RottcrdamRobcrt Nlichcls. II.D.. Noç Vrrk

Stcvcn A. Schroetlcr, NI.f)., PrincetonHarold C. Sox, \I.l)., Lebanon, N.H.

Iiulrontrl Ofil(:l.s

Victoria Alexandcr, \Ianuscript Editor; Cherrl J. Alkon, Editorial Assistant. Inrages in Clinical \Iedicine/Corrcspondence; Timothr S. Andcrson,

Editorial Assistantl Briana Dohertv, Editorial Assistant; Leigh S. I)uPu1, Nlanuscript Assistant; NIichr:llc Frayman, Netrvork Support Spccialist;Barbara Galc, Manuscript Editor: Dalc R. Golden, Assistant \Ianager ol [ditorial Support Scrviccs; Rachcl Hart, Research Assistant; Susan

L. Kaplan, Editorial Production Lavout Artist; Sandra S. \Iclean, \Ianuscript Editor; David 1\1 \IcNecl. CorrcsJrondcnce Rcpresentalivc; flcnn'

Lisa A. Scott, Editorial Procluction l)esktop Specialisti Natasha T. Scaman, Editorial Assistant; ilIarih,n Seaquist. Reccptionist; \Iarie P. Soullierc,Research Assistant; Deborah A. Stonc. Assistant \lanager of Editorial Produr:tion; Pamcla S. Strvjerrski, Editorial Production Proofieadcr:

Amv IVcst, \Ianuscril:t Assistant.

Robert l). Bovenschulte, Vicc Presidcnt fàr Publishing

PROSI'EC'II\:E authors should consult "lnfirrmation for Authors." rlhich appears incach issue and mav bc obtained from tlrc Journal Edirrrial OlÏcc (addrcss lrlou).-\RTICLÈS rrith original marerial arc acceptrd lirr considcration sith rhe under-stirndiog that, c\ccpt Ibr abstracts. rro prrt ol rhe data has bcel publishrd. or rrill lxsubmitted for publicatit»t rlserrherc, belbrr appearing hcre.\!A'I'ERIAL printc<l  the.lournal is corcrcd ll copvright. No part o[ this publica-tion mal bc reproduced rri(hout rlritten pcrnrission.'l'hc Jounal rlres not hold itscllrcsponsible for stattrncnts madc bt anl conlribùlor.STATII\IIIYIS or opinions cxpressed in tlrt'./orraal reflect thc vicrrs ol thc author(s)

and do uot rcprcscnt olficial policl of the llassachusctts \[edical Socittr unltss sostaled.r\L1'HOUGH all arltrutising matcrial accepted is expeclcd to conlirrrn ro tthicaltnedicirl stanrlards. ilc(cptilncc docs not implr cntlorsclrt:nt b,t thc Journ«1.

SUBSCRII'TIO\ PRICES: US$151i.00 p'r rtar (interns, residcnts, and sttrdents

US$99.(10 pcr tear). Within the L-.K. onlr: !78.00 pcr ,vear (interns, rcsidents, and

students !50.00 pcr vcar). Send parnrcuts and corrcspondence to: NEJII, /,, Europc-an -\{agazine Distributir»r Gnrbl{, lilescbcckstr: 96, 10623 Bcrlin. GERi\lAN\'.Please includt'curr('rrt nlailing label rvith renerval order InJapan, Y2(i,500 per lcar(intcrns, residcnts, and students *17,200 pcr lcar). Send ordcrs (o: \ankodo Co.,Ltd., +2-6. Hongo 3-chorne, Burkro-Ku, Tokrr I I3,Japan. I:or rr»rrpkte subscrip-tion inlornration, plcasr src thc IJusitrcss Inlôrnration page in this issuc.

IiDI'|ORIAL OFFICIIiS: l0 Shattuck St.. Boston. \lA 021 l5-6()91, USA.'ltlcphonc: (617) 73+-9t|00. FAX: (617) 13+-'l+5t-.

BUSl^\ESri A.\l) SI-TBSCRIP|IO\ OIII(]IIS: Il-10.\Iain St..'l\hltharn. IIA 02I5+-l6+9, USA. IAX: (617) 893-0+13.

t 30!

SUBCLINICAL HYPERTHYROIDISM _ JUSTA LOW SERUM THYROTROPIN

CONCENTRATION, OR SOMETHING MOREP

Str-cn the introduction 10 1'cars ago o1'assavs I'or sc-rum th)'rotropin that are sensitive cnough to distin-guish betn'een normal and lou' concentrations, thcmeasurement of serum th1'rotropin has becomc tht: sin-glc best tcst of thyroid function. The reason is that anincrease in the serum thyroxinc (T,) or triiodothyro-nine (T,,) concentration sufficient to cause clinically'ap-parent hyperthyroidism inhibits thvrotropin sccretionsubstantialll; just as a decrease in the serum thi'r'oidhormone concentration sufficient to cause hy'pothvroid-ism increases it.

Serum th1'rotropin concentrations are usuallt' undc-tectable in patients rt'ith overt h1'perthl'roidism' Thevma1'be detectable but lon'in patients r'r'ith thvrotropindcficicnc,v caused by hypothalamic or pituitary disease

and in patients with nonthyroidal illness. Patients inthcse tr,vo groups can bc identified b1' the clinical con-text, historli and physical examination and b1'thcir lou'or lorv-n<lrmal serum T, and 'f, c<lncentrations. Hon'evcr, lo\^' serum th1'rotropin concentrations are ntostcommonly encountercd in patients rvho havc [èu'or noclinical manifestations of hy'perthy'oidism and normalserum T, and T, concentrations - a disorder calleclsubclinical hyperthvroidism.

What are the frcquencl,, natural history; and risks o['subclinical h,vpcrthyroidism? Somc ans\vcrs to tht:sequcstions are provided by Sau'in et al. in this issue ofthe Journal.t Among 2007 clinically cuthy'roid persons60 vcars of age or older n'ho rlere examined inFramingham, \{assachusetts, Ii'on.r l97tl through l9t}0,including l15 u'ho \r'ere recciving thvroid h«rrmonetherapy; l2 percent had subclinical h1,perth1'roidism. Insever;rl other sun'evs o[' oldcr l]ersons, thc frequenc,vavcraged about .[ percent2-r (it is lorver among youngerpersons). The variation can be explainecl largely b1,

differcnces in what was considered a lou, serum tht-rotropin concentration, in the frequenc,v <ll' nontoxicgoiter in the rcgion «-rf'the studl; and in u'hether pa-tients rcceiving thyroid hormone therapli in u'hom theIrequency- of subclinical h1'perthl'roidism is substantial,u,erc included.

In an1, clinicallv cuthl'roid person found to havc alolv serum th1'rotropin concentration, serum free 'I'rshould be measurcd. A normal value, albeit usuallvabove average , conlirms the prescnce of subclinical h1-perthyroidism. Stricth's1>caking, the serum T,, crlncen-tration should also be normal in these persons, but thatmcasurenlcnt is not important ftrr decisions about man-agemcnt. What is important is asscssment ol'the possi-bilitl,that the person has pr<-rblems that mipçht be at-tributcd to cxcess th1'roid honnonc.

Orrc o[ these problems is atrial fibrillation, ir uell-knou,n mani{èstation of' hvperthvroidism, especially inthe t'lclcll1,. Arnor.rg patients rvith atrial fibrillation anclno cardiovascular clisease, a lèu'har.'e subclinical hlpcr-

THE \L\\' ENGI,A\D.JOUR\AI- O[ I{EDI(;I.\E -\or'. 10. I99-1

thyroidism.'Evrùence that the an\thmia is rt:vcrsirrlc

r.,,ith antith,vroicl therapv in these patients is sparse'

Among patients rvith ovcrt hvperthvroidism, ho\t'ever,atrial fibrillation clis;rppcars in 60 percent rçithin fourmonths after thc initiation o[ antithvroid therapl'.6 Oth-er carcliovascular effects of subclinical h1'perthl'roidisminclude increases in the pulse rate and lhe frcquenclof atrial prematurc contractions and incrcascs in leftventricular mass and contractilitl.; Anothcr potentialproblem is accelcrated bone loss, causccl bv increasedbonc resorption. The mineral densitv of' the sltine , fe-mur, and other bonv sitcs is decreased to a variable de-gree in postmenopausal lvomen n'ith subclinical hi'perth,vroidism, but not in men or \'()Llnger \\'omen.lrHorvever, in I ltlO patients receiving thrloicl hormone,the rate of hospitalizatior.r for lracture (and also forischemic heart diseasc) during follou'-up u'as similaramong those u'ith lou' zrnd those uith nortnal serumthyrotropin cr»rcentrations.r) Patients rvith subclinicalhyperthl'roidism can also have muscle rve:akness, neu-

ropsychological clyslunction, or other clinical manifès-tations of hvperthl'roidism in elderll' patients.ro In oth-er \\rords, the process might not alu'a1's bc subclinical.

lVhat is the risk «r[ atrial Êbrillation or ol' the subse-

quent development ol'overt hy'perthl'roiclism in patientsu,ith subclinical h1'perthvroidism?'fhe studl bv Sartinct al. addresses thesc risks ancl is uniquc for its largesizc and long f<lllon.up.l Among thc 2(X)7 subjccts, r'vho

rverc ftrllou'ed for up to l0 vears, atrial Iibrillatirln dc-vcloped in 192 (10 percent). As t:omparecl u'ith thegroup rvith normal serum thvrotropin ('onccntrations,the risk o{'atrial Iibrillation rvas 3.1 timcs higher in thegroup u,ith lou' serum thvrotropin conccntrations and1.6 times higher in thc group u'ith slightlv lorv concen-trations. Onl;'2 of these 192 persons also hacl clinicalll'evident hy'perth,vroidism, although more ma1'have hadelcvated senrm T, or'f , concentrations. Three personsin the trvo lorr'-serut-n-th1'rotropin groups and one inthe normal-senrm-thvrotropin group subsequently hadhy'perthr-roiclism but not atrial librillation.

Lou'serum thyrotropin concentrations tnav be tran-sient as n'ell as largch'innocent. h.r thc studv b1'Sau'inet al., 19 o1'36 persons in the lorv-serum-thvrotropingroup had norrnal scrum thvrotropin values at somelater time. Among 66 patients n'ith subclinical hvper-thy'roidism in a British general practice rcstudied one

year later, 26 (39 pcrcent) still had lou'serum th;'rotro-pin conccntratiotrs (one had rlr,crt hvperthvroidism)and ,t0 (61 perccnt) had norrnal concentratiotrs.r In an-other studv ol 53 euthl'roicl paticnts at a primarv carecenter in Su'edcn u'ho had lovn'scrum thvrotropin con-centrations, 27 (5 I pcrcent) had normal values t\1'o tothree l'ceks lzrter.r Iu short, thc onll clefincd risk of sub-clinical h1'perthvroidism is atrial fibrillation. Subclini-cal h1.perth1'roiclism olïen clisappears, ancl progressionlo ovcrl h1 llert h1'roidisnt is uttt otrrmott.

Thc natural history' ol' subclinical l.rvpcrthvroidism is

variable because it has several causcs -

rthich, in turn,alfect managemcrnt.'lhc rnost c()mmon causc is th1'roid

Vol. 331 \o. 19

hormone therapl'. Treatment is simplc: rr:ducc thedose, unlt'ss the patient h:rs thvroid carcinoma ancl therisk of recurrence is det:med greatel- than the risk ofsubr:linical h1'pcrthvroiclisn.r. Therc is probablv no nccdto rccluce the dose in a patient n'hose serum thlrotropinconcentration is onlr, slightlr, k»r', irccause there ma1' besomc fluctuation betrveen claih'doses, becausc the risksol slightlv lorv values in othcrrçisc healthl' paticnts arelon, and because frequetrt rnanipulations o{'the dosagecomplicate therap,v ancl increase its cost.

C)ther in-rportant causcs of pcrsistent subclinical h1-perth).roidisnr are n-rultinodular goiter, solitarl' thl.roidaclenoma, ancl subclinical Graves' discase, clefinr:cl as

persistent protluction ol'small arnounts of' thvroid-stim-ulating antibodics. Uttlbrtutratchl Sarvin ct zrl. do notpror.ide inlirrmati«rn about the frcqucncv of goiter intheir subjects. Autonomouslv lunctioning thvroid tissuc<lf any' tr,pe ma1' slou'I1' cnlarge, cause h1'perthr-roidism,or both. u'ith time. Each of thesc disordcrs could bemanaged b1' thc arlministration of radioiodine, butlarge dt-rses are olten neeclcd bt:cause thc fractional thy-roirl uptake is normal. 'fhis tl.rcrapv might be appropri-ate Ibr a patient nith othcr risk Iàctors fbr atrial fibril-lation or for lorv bone densitv anrl fi'ar:ture, but it is adrastic interr,ention lbr a siler.rt biochernical abnormali-n'. Thc same considerations applv to therapl'rvith anantith,vroid clrug, u'hicl-r cntails thc addition:rl risk ofside e{fects and thc probable need {br long-term, some-timcs cven lifelong, treatment. As noted ab«l'e, subclin-ical hvpcrthvroidism mav be transient, caused by pain-less (silent) thvroiclitis or a norlthvroidal illness, u,hichnccd not be sevt:re.

The increased risk ol' atrial hbrillation and possibl,vother problems means that subclinical hvperthvroid-isr-n amounts to more than.iust a lorv scrum th1'rotropinconcentration. And as mcasurcments o[- serum thy-rotropin become mrlre u'idelv used as the initial test of

SOUNDING BOARD

FOR EVERY DOLLAR SPENT _ THE COST.SAYINGS ARGUMENT FOR PRENATAL CARE

Puut Ic spending lbr prenatal care in the UnitedStates has becn justiiicd in rcr:t:nt 1'ears prir.narilv bvthe cost-savings argument. Prenatal c:rre, it is argr-red,can prevent tl'rc costs ancl mcdical complications asso-

ciated lr,ith lou'birth rreight; it is public hcalth spend-ing^ that pa1's lbr itsclf. 'fhis proposition seems in-tuitir,ch' rcasonable and stlpports a popular publicpolicr'. \Iorcovcr, the finclings of thc boclv ol rescarchue consider here havc givcn this asscrtion consiclerablc'rveight atrd remarkablc precision. !-erv other claims inmeclicinc or in public hcalth can bt: cncapsulated intothe statement th:rt Iôr evcn. $1.00 spenl, §tl.70r (or

thyroirl lLnction, more people with this condition n'illbe identified. If thc patient alreadv has atrial fibrilla-tion, other atrial arrhythmias, other cardiac disorders,or accelerated bone loss, to nhich small degrees of thy-roid hormonc exct: ss might contribute, antithyroidtherap,v should be scriously considered. For the remain-dcr ol patients, no intervention should be undertakenuuless subclinical h1-perthyroidism persists for severalmonths. Er''en iI' the patient's serum thyrotropin con-centration rcmains lolv, balancing the risks of the dis-ease against the problems of antithl'roid therapy leadsme to thc conclusion that careful follou'up rather tl.ranintencntion is thc most prudcnt policy..

RoerRr D. Urrcrn, \{.D.

RrrrnrNcrsl. Sawin CT, Geller A, Wolf PA. et al. Low serum thyrotropin concentrations

as a risk l'actor for atrial librillation in older persons. N Engl J Il1ed 1994:

331:1249-52.2. Eggertsen R, Petersen K. Lundberg P-A, Nystrom E. Lindstedt G. Screening

for thyroid disease in a primary care unit wrth a thyroid stimulating hormoneassay with a low detection limit. BMJ 1988;297:1586-92.

3. Sundbeck C. Jagenburg R, Johansson P-M, Eden S, Lindstedt G. Clinicalsignihcance of low serum thyrotropin concentration by chemiluminometricassay in 85-year-old women and men. Arch Intern Med l99t;l5l:549-56.

.1. Parle JV. Franklyn JA, Cross KW, Jones SC, Sheppard MC. Prevalence andfollorv-up of abnormal thyrotropin (TSH) concentrations in the elderly in the

United Kingdom. Clin Endooinol (Oxf) l99l;34:77-83.5. Woeber KA. Thyrotoxicosis and the heart. N Engl J Med 1992327:94-8.6. Nakazau'a HK, Sakurai K. Hamada N, Momotani N, Ito K. Management of

atrial fibrillation in thc post-thyrotoxic state. Am J Med 1982172:903-6.7. Biondi B, Fazio S. Carella C. et al. Cardiac ef'fects of long tem thyrotropin-

suppressive therapy wirh levothyroxine. J Clin Endocrinol Metab 1993;77:

33.1-8.8. Faber J, Galbe AM. Changes in bone mass during prolonged subclinical hy-

perlhyroidism due to L thyroxine treatment: a meta-analysis. Eur J Endo-crinol 1994:130:350-6.

9. Lecse GP. Jung RT. Guthrie C, Waugh N, Browning MCK. Morbidity in pa-

tients on L-thyroxine: a comparison of those u'ith a nomal TSH to thosewith a suppressed TSH. Clin Endocrinol (Oxf) 1992i37:500-3.

I0. Stdt DJ, Mcl-ellun AR, Finlayson J. Chu P, Alexander WD. Elderly patientswith suppressed serum TSH but normal liee thyroid hormone levels usuallyhave mild thyroid overactivity and are at increased risk of developing overth1'perthyroidism. Q J Med l99l:78:77-84.

$2.57'or $3.311'J) rvill be saved. Yet this is the commonf<,rrm o[ arguments usccl to justify- public spending onprenatal care.

That lr.omcn should seek regular medical care dur-ing normal, healthy pregnancies u'as Êrst rvidcly rec-ommended at the bcginning ol'this centur,l-.rr In theUnited States, social reformcrs and nurses introducedthe {lrst organized programs o{' prenatal care, whichled to impressive reductions in nconatal and maternallrortalit\'.r; The cost effcctiveness of prenatal care wasassumcd lrom thc start. A l9l3 article in the Boston

ùIedical and Smgical Journal stated, for example, "Thecost of $ I .16/paticnt suggests its economic feasibilit.v atthis nominal cost."7

In the latc l9tl0s, researchers attempted to measure

the rcturn in savings that could be expectcd from ex-

penditures for prenatal care' f'heir published conclu-

È]DlTORIAI | 303

I 30+ ]'HE NEIV ENGL\ND.JOURNAI, OF \IEDICINE \ov. 10, 199.1

Table I. Methodologic Shortcomings of 11 Studies of the CostSavings Due to Prenatal Care.+

Problems in estimating the effectiveness of prenatal care in reducingthe incidence of low birth weight

Noncomparable controlsr e-r5

Unsupported assumptionsrr r: r7

Problems in cstimating the cost of providing adequate prenatal care

Underestimation of the cost of comprehensive prenatal cuelj I2 r7

Underestimation of the cost of overcoming nonfinancial barriersto access to prenatal carer'ie17

Problems in estimating savings in the cost of postnatal careInaccurate detemination of costs of postnatal carer r'e ri !5

Oversimplification of the relation between changes in the frequencyof low birth weight ancl actual cost savingsr-r'e'r I ri r6

."ï:ï'r1,:ï,:Ti ;i::,§,i::: * I I studies Bdh Korcnbrot i and Lennie et

6 percent in the treatment group.r0 The estimates inthe studies based on calculations rÿere more conserva-tive. For example, in the Institute of N,Iedicine stud);the cost-savings estimate \\'as based on a reduction inthe percentage of babies u'ith low birth u'eight from11.5 to 9.0 perccnt. In estimating the elfectiveness ofprenatal care in reducing the frequency ol lorv birthu'eight, all l2 articles \r'ere allected by one of tw'o llaws.

Noncomparable Controls

Self-selection bias rvas a problem in all four cross-sectional r1r.li"r.t2-t5 Random assignment to studvgroups, rvhich rvould entail u'ithholding adequate pre-natal care from some prcgnant women, rvas not possi-ble . In realitli the women themselves chose their studygroups. This is problematic because, as the Institute oftrIedicine report asserts, women r.vho choose to obtainprenatal care, or choose to obtain earh'prenatal care,are likell' to be diflèrent from those rvho do not receiveadequate prenatal care in fundamental 1\'ays that mayinfluence the outcome of the pregnancy and that can-not be controlled statisticalll,.3 Il the tr,vo groups differlvith respect to factors other tl'ran their use of prenatalcare, factors that might be associated rvith adverse out-comes, such as income, living conditions, stress, health-related behavior, and maternal health status, then itrvould be inappropriate to attribute cliflerences in out-come solely to diflèrences in prenatal care. In the othercmpirical studies,r'!r-rr the studl' groups may not in facthave been comparable r,vith respect to income,l'e sub-stance abuse ,lr' maternal age,l0 citizenship,l0 and moti-vation and ability to seek timely care.l0'11 In trvo arti-cles,l'e historical controls r,l'erc used over a period whenrates of low birth rveight were declining. One studyllu'as flarved b1'a common bias in obsen,ational researchon prenatal care; an association betw-een low birthweight and fcwer prenatal care visits may be foundsimpl-v because preterm delivery itself curtails the num-ber of prenatal visits.

Unsupported Assumptions

Each of the four studies using h,vpothetical calcula-tions2'3'16'17 rvas based on the assumption that providingadequate prenatal care'"vould result in a decrease in

siotrs were quoted uncriticall,v in both the scientific lit-erature and the popular press and lvere instrumental inshaping government debate on the expansion ol N{ed-icaid coverage lor pregnant u,omen.sThe public's làsci-nation with numbcrs carried this process along, andthe cost-savings argument has become commonplacervhenever prcnatal care, preventive health care, or rc-form o{'health care financing is mentioned.

A critical look at the evidence lor the claim that thecost of pror,'iding prenatal care is offset by savings inthe cost of postnatal care reveals a far more complicat-ed picture. Serious flarvs in research methods undercutthe r.'aliditl'of the studies of'cost savings resulting fromprenatal care.

Tnr Booy oF RESEARcH

To identify studies on the economic elfects of prena-tal care, rve revielved articles in refereed.journals andgovernment documents from 1975 through 1993. Over100 studies on the efÏèctiveness and economics of pre-natal care rvere identified. Only 12 speciÊcally ad-dressed the issues of cost and cost savings.

Four o{' the articlesl'e-ll tvere based on three naturalexperiments (both Korenbrotr and Lennie et al.l)rverebased on the Obstetrical Access Project). In eachexperiment, the "treatment" groups received compre-hensive prenatal care; the controls received either noprenatal care or routine care. Four studiesl2-li rverecross-sectional analvses ol eroups ol patients. Each ofthese eight articles used an empirical studv clesign

-that is, each rncasured or estimatt:d thc rates of lon'birth u'eight and costs lor actual populations. The re-maining four article s2:r'r6 r7 used synthetic or hypotht't-ical calculations of cost savings. In three of these,2:l'r7

the authors began u'ith the assumption that providingimproved prenatal câre to specified groups ol'womenwould reduce the proportion o1'babies rvith lou, birthrveight. The costs saved as a result ol this assumed de-cre ase in the percentage of babie s u,ith lolv birth n'eightn'ere used to calculate net savings. The other studyr6askecl the question in reverse: Ho'rv much of a reductionin the proportion of babies rvith lorv birth u,eight rvouldbe needed to cover the cost of improved prenatal carefor lou-income u,omen?

In each of these articles, u,e identified methodologicproblems (Table l) that could have resulted in theoverestimation ol'the cost savings due to improved pre-natal care. These problems are discussed in more detailbelou'.

Pnost,eMs rN ESTTMATTNG THE ErrrcrrvnNBssOF PRENATAL CARE IN REDUCING THE INCI.

DENCE Or LOW Brnrn \{rrCUrThe apparent (or assumed) elfectiveness of prenatal

care in reducing the frequency o[ lou'birth rveight var-ied enormously among the studies. Itr general, the em-pirical studies claimed the greatest reductions; onestudy found the proportion of low-birth-u,eight babiesto be 21 percent in the control group, as compared rvith

\'('1. :i.i I \o. l!l sot'\l)l\(; IJ().\RI)

the 1r'e<lrrcnr:\'of low birth rrt'ight. ()nlv tl.rt' r'eports ltr'thc Instilute of \[crlic'incr an([ the ()llice ot"fcchnologv,\sscssrrrt ntl'' expli<'itlv t onsiclerecl tht' t'r'itlcncc fbr thcellt'ctiveness ol l)rcnlrtal cat'c. l'he ilillhols of' boththe sc r-cports couclurlccl that thc rçeieht ol'thr cviclencesupl;oltecl thc elli'ctiveucss o1' prenatlrl care. Recetrtstu<lics. lronevcr-. have n«rt bccn irble to rlcnxrnstrate thee f ft.ctivetrcss ot' prcnatal calc, t'spt'cialh' routine ltlt'-t.urtul t:lrrc. in I'cclrrcing the inciclcnr:c o[' lorç birthn t'ight.l'-ll

Asirle lionr tliis clLralitativc assurrptit)li. thcse Iàur'stuclies u'ere als«r bascrl on cluantitatilr' cstinrates ol'thenlagnilude ol' thc tcrltrclion in the li'erluencr' <-rf k»r'birth u't'i--.ht. Tht:sc cstir.natcs art: clitital to thc ck'ter-rninatiorl riI cost savillgs. 1-he Institrrtc ol' \Icclicint'sttrclr"i u as basccl ott tltc suLgcon eerrclal's objt'ctivt', cs-tablishecl in 1980. ol l "5 l)cr(cnt rat('ol klu,birth-rveieht ovetall ancl a nurxinrul)r r.rt(' ol !) pelccnt irrhigh-risk subpoltul:rli«rns" lrv 1990. l'his sturh l)rcselrtstto clirlencc as to rçlrelhcl thl' surg('on gcneral's goalcoul(1. in Iact. bc achicvctl ll.thc kincl ol-prt'rralal crrrt:nr':riIablc Ii»' the ('osts use([ in thc calctrlittion. Auotltt'rsttrclr'l stt'atilit'tl uortten accorrlitrg to crlur.ation zrnc[

cltosc the propol'tiorr of'lorr-l;ilth-\\'eits^ht babics boln to\\onlcn rçho harl lc('cive(l acleclr-rate llrcnittal care as th('olltcolne variabk'. l'his stuch uas baserl on thc assunrlt-tion that the use «rl'prt'natal cur'c \\'l1s the onh ftrctorlhal ex1>lainccl di11èr't'nct's in tlre inciclcnce o{- lou'-birth-rçcieht babies bctrçr'('n groul)s o['uon.rt'r.r uith sin-rilar'lcvcls ol- edttcalion. ISLrt it is likclr. that. cvt'n rçithingroulls delinec[ bv ctlttcation lcvel. nonrcu u'lro t'cct'ivcinarlecluate prenatal ( rll'c nl:rv hzrve a ]righel trnclcllr.ingrisk ol' ach'crse outconrL:s ol prcgnarrcv thun \\'()nr('nuho rcceive a(lequatc ('arc.

'fhe othel sturlv in tlris groupl; cliliclccl lorç-birth-rçeicht inlants into sis categol'ies. f'lrt' atrthor assunreclthat intensive plenatal sun'eillanct' lulc[ inter'\enti()lluorrlrl be efft'ctive t:noush to cause 20 pelcrr-rt o[' thcinlirr-rts in each clrtcgor\- to be shifit'cl to the ncxt hight:rbirth-neight czrtcB()r\'. -\o.justifir:ation lirr this assunrl)-tiorr. or fbr the choice ol'tlre 20 pcrcent lisule. uasgiu'n.

Pnorrtus rN ESTTMATING THE Cosr or PnovrotNcAorquarr Pnexaral C,lRe

l'hc estimatecl costs o['ltrenatal care varie(l cor-rsicler--

ablv auror.rs sttrrlit's. rauging il'orn $lllJ0r"r7 to $1.0-l2rI

I)er pregnarlcr: In arlcliti«rn to the inrplccision anrl po-tential inaccuracv ol' (l-resc cstinr:rtes. tu'o other prob-lcnrs allèctcd cstinrates «rl'thc cost ol provirling l)rena-tal carc to lon-incotric \\'omcn at high risk of havingkrrr'-llilt lr-rr ei ght lralrit's.

Underestimation ol the Cost ol ComprehensivePrenatal Care

Thcre is an enrcrgillg conscrlsr-ls that clli'ctivc l)rer)a-tal care lor rlomeu irt liigh risk ol haling lorç-birth-neigl'rt l;:rbir:s. thr.rsc rnost likelv to bc the targets o['

lttrltlic Proqr:tnrs. Irt'crls to be n.rr»'c intt,rrsivc (anrl rlorc

cxpt'nsive) than routirrr' prenirtal calc lil'nonrcn at lorr'1'lr[.]r':' Nevcltheless, threc o1' thc lirul strrrlics trsinslnpothctical calculatiotrsl i l" atrcl all thc closs-srr'tiottaIslurliesll Il \\'cl'c basccl on thc assunrpti()n tltirl lotrtint:pt'ettat;rl care \\1)ul(l bc cllèttivt'enough to pt'ocluct'lhcassume(l rlccl-casc in the fi't'cprcncr. «rl- lorr birth \\.cightin the talgt:t groul)s. -l'hus. these sttrclies nt:n h:u'c un-clercstin.raiccl il',c ti'rr" r'ost rrl plovi<[inq cllet iire l)rena-titl «'at'r'.

Underestimation of the Cost of Overcoming NonrinancialBarriers to Access to Prenatal Care

Ilnsuring that lou-inconrc \\onlcn lett'ivt' plenatalcart' recltrires lror'(- th:ut sinrplr paving- rlre bill lôr tl-risc'rlt' .l'l-l! Othcl l;allit'r's thal rnust bc o\'(f l(onr( irrcltrclethc scarcitr.ol healtlr carc llloviclt'rs nilling trr 1>r'ovidt:

l)r('nirtal cale to krrr'-iucr.>nrt- \\'olll(.ll! thc lat'k ol' trltus-l)(»'tiltion atrcl chikl i'alt' lirl sucli n-orrrcn. <[ilhcrüticsuith cnr-ollnrcnt rerluilcl'llents l'crr' ptrblic ploulalns. thec rrlt ural insensit ivil v ol' ltro9..rartrs. lack ot nrot ivatiorr onthc part of'rr-onren. au([ the l:elic[' that 1;r't.natul calc is

not inrpoltant. Tht'sc nonfinancial ballicls alc clill]cultt() overcolnc. ;rntl cflirlts to clo so rrrav ltt' erpensirc.\onc of thc sturlie s n c t'xnrnint:cl c,rnsiricr'. cl thc cost <-rl'

arklrt'ssing thcsc r.rou[]nant ial balricls.

PRosmNrs rN ESTTMATTNG SAvTNGS rN THE CosrOF POSTNATAL CARE

llstitnates o[' t«rst savitrgs alit'r- tleliven ut'rc alsohighll variirblc. ritluing- [r'onr $3]7rr to S13.6 l(itper'l)regnanc\'.

-fl'rc pr.actical clilficultics ol acculult.lv csti-

nrirting the cosl of' treatinc thc ''avclast"' lon-bilth-ueight inlànt ancl cxcessive sinrpli{itittion in c'alcul:rt-ine thc estimates nrnv hart lecl to an urcrcslinration ofcost savings.

lnaccurate Determination of Costs of Postnatal Care

The Institutc ol' IIctlicirrc ; and OlÏct' ol''li'chrroloer'Asse ssntetrt l'' sturlit's t't:liecl on t'laltolalt' r'alculationsbast'cl on stale ot' turtiotial cstitnates oI t'harccs lor'lrospitalizatic)n an(l lates ol' r.nortalit\ anrl nrolbic[in'alrlong lou-birth-u'eiqht inl'ants. This inIirrnralion istlillicult to cletenniur' pt't'cisclr'. 'l'he costs ol l)ostnatalcart' lor tht' coutlol gr()rU) \\-cre un:rr-ailal;lc lirr one olthe tl'rlee n:rtural ('\l)crirnentslr'; onlr- otrc o[- thc lhrt'cust:cl «lata ft'onr actual hospital bills.l" I-hc noncolnl)a-I'ablc stuclv groups in all the cmpilical strrclics castclorrlrt ou tlre clainr that rlifli'rences in c'osts lt'llect sar-ings attl-ilrtrtal)lc to l)r('natal cart'.

Oversimplification of the Relation between Changes in theFrequency of Low Birth Weight and Actual Cost Savings

l)ata fi'onr thc sirnulation lx' Schrvzrrtzr; suege st thatthc n-ragnitr-rcle ol' the s;u'ir.rss clcpcnc[s heavilr. onuhetlter pretratal care is ellcctive in plt'r,cnting birthrvciehts at the upper or thc k»r'er cnrl ol'the k»r-bir-th-rvcight clistribution. Recluctions in the pt'rccntage ol'in-[)rnts n ith mo(lcriltelv lorç birtl'r u'cigl'rt (1500 to 2191) u)hale ;r large inllucncc oll thc overall l)('l'('('llt:rg(' ol-lou.-

bir.th-rveight babies but a smaller elièct on costs. Re-clucing the proportion of infants u'ith r,en, lorv birthut'ight (undcl 1500 g) has a much grcater cllèct oncosts. In Iàct, in the projections ltv Schrvartz, alntosthalt (12 percent) of the estin-rated savings u'as attriltut-able to changes in birth-u'cight catcgorl' among thevclr-lon'-birth-u'eisht inlants, .iust 1.6 ltcrccrrt o[ allinlànts and I7 percent o[' infants rveighing less than2500 e. Bv con-rparison, inlants born u,eigl'ring 2000 to2{99 g accountecl lor ô0 pelcent of k»r'-birth-rreight in-far.rts but onll-24 percent ol'the projected cost savings.Without a more exact picture of the eflect of prenatalcare on the distribution ol birth u'eiehts, savings can-tlot be accuratell predictccl. In aclclition, problerrrs re-latcd to social clisadvallt:rge, substance abuse, anclurcdical clisorders in nen'borns ancl mothcLs ale clisplo-poltionatcll'associalec[ u ith lou birth n eight. T]rus, theincreased costs of caring Iirr lorr-birth-rreight inlàntsrna\- not clisirppear simplv as a rcsult of recitrcing thefi-r'cluencv of lorv birth rveight.

IupLrcarroNs FoR FUNDTNG oF PnnNnrar CenrPnocnaus

Saldell has arguerl that it u':rs in largc part thc cost-savinss argument lirr prenatal carc thiit cunvirrccclpolicl rrakers, lcgçisl:rtors. and taxpavers to Iirncl prc-natal care pr.oerams.s Dcspite thcir limitations, thcstudics rçe har.e discussed u'ele instnrmcutal in pro-n-rotinu this policl, sincc thcl'ploviclecl pre cise numeri-cal values for projectcd savitrgs, expressecl clorln to thcpenrn'1'ol evcrr.clollar sl)cnt. The evidcncc tl'rat plena-tal cale pavs {-or itsell is simph. not strong enoush tonrelit the virlr-ral certaintr. rçith rvl.rich this clzrint hasbecu cspr.rr-rsccl. 1'here is considcrable risk in basinghealtl'r irrogranrs zrnd public health polio'orr inlirrma-tion that is t-nore optir.nistic than scienti{it:, for scvcralf citsolts.

'L.hc stuclv rcports van in the clegrec to tvhich the au-thot-s acknorr,ledgc the uncertaintit:s inherent in theirresearch; Itou't:r'cr, enrphatic ancl overlv precise conclu-sions havc olten obscurcd the limitations of the rlata.Preciselv becausc these studies have been trsecl to sup-port lar'-reaching rncdical and public health polio,their shortcor.r'rir.r{s, and the completc disregarcl Ibrthose shortcomings, slrou]d bc of particular conc-e ln.

l.hesc alticlcs irnplicitlv alguc that there is a simplrnrerlical remeclr' filr problems that are probablv nrani-Ièstations of cleepll' rooteci sor:ial and cconomic lactors.In lar-gc part because of their assumed authoritr; it hasbeelt comnronltlace. but misleading. to hold up ltrenatal('arc

- especialh, tl-re relativelv inexpcnsive sen ice s

provickrd to u'omen at lou'risk -

15 116 solution to tht:

ltroblenrs o{'chilclrcn bom in povert\'. The currcnt pul)-lic perception of prenatal care oversirnpli{ies the dil'-Iictrlties ol dclivering prcnatal care to \\'onren u,ho cloltot lto\\' receive it, overestin.rates thc ber.refits ol prcna-tal care, and contributes to the r.r.rcclicalization ol'com-plex soc:ial problems. Stuclics that hzrve er,aluatecl the

TFIE \t-\\. E\Gr..\-\]) l ou R-\.\L oF -\ I Fl,D I C: I-\ H \oi'. 10. l9!)l

cl}ècts ol rccent expansior.rs ol public lirnclins for pre-nirtal carell lrr t:onlirm that pr-eventing adversc pclina-tal outcomcs rrill bc ncither sinrple nor cheap.

l{ieorous cvicleuce ancl a l;road nnah'sis alc inrltor-tant, not onlv tbr the selling of'this policl but also {Lrrthe acccptance ol' luturc public health initiatives. Be-cause tlte cost-savings argunrent has dor-rinatcd thc dis-cussiorr, publich' Iirnded preu:rtal care rlav be aban-cloned il it tr-rrns out not to pal lol itscll. I1' tht:prr4>«rltion ol babies u'ith lou' birth ueiglit ancl rlit:costs ol their care do not actuall)'decrease, those re-sponsible Ibr these programs, and for {uture programsthat protnise to serve n1o[e\; mar' fincl themselves u.ith-oLrt sru)l)ort.

lhc valuc :urd inrportancc ol'enstrring act:ess 1o acl-('(lLrate prenatal cu.c arc not thc issue. It is halcl t<-r

in'rasine a 1'ct1'('at ll'onr tlic enlichlened legislation tl.rathclps lou'-incolle rlolren reccir-c r.cconrmcndccl nredi-cal care dulinq thcir 1;reer-rancies. It is evt:n l,rossiblcthat rvith bt'ttcr clata. thc cost savings duc to bettcr pre-tratal care coulcl be convincinglv clenronstratecl. \er.cr-tltelcss, on the basis of t:xistins studics, it u'oulcl beinapprol;riate, il'not irnpossiltle, to atterlll)t such au es-tillate.

Four steps should bc taken in ltroruotine ancl stuclr-ing prenatal care in the luture. Iilst, uc sl-ror-rlcl berttort' circunrspect

- ancl less paltisan

- in interpret-

ing ancl citilrg the conclusior-rs o1' scientific stuclies. rc-garciless of theil provcuance . Sccorrcl, u'c shoulcl n-rakccven' ellirrt to cva[ratc tht' costs ancl benefits ol' r-necli-

cal carc bc(orc it bccorlcs established in the protès-sion's or the pul;lic's n-rincl as the stanclarcl of carc. Ihisis as true Ii»' potentiallv u-ictcspreacl ltrevcntivc ltlrr-granrs. srrch :rs prenatal cilre, as it is lbl a ncu- ch-ug orrureclical tcchnolosr. It mav bc tor-r late to ntcasure ade-cluatclv tl'rc oveliill cost savings associated rvith prena-tal czirc, but there arc oplloltLurities to sturlv specificcoml)on('nls of' ltrenatal cirrc pr'osl)('ctir-elr'. Fbr cxar-n-p[e. t'csearch shoultl cxamine rrhcther dillèrcnt n'roclclsanrl lcr.els o{'cale mav lte optirtral fol clilfcrcnt groul]sol'u'clmeli.

Third, u-e shoulrl lecoe'r-rizc that nreasuling the costsand bcnclits ol' anv trczrtment is neithcr siml;lc norstraiehtftrru'iircl. It is u'rr.possiblt' that prcnatal cart' isbcnelicial in less easih' r'ncasulecl \\'avs

- fol exan'rPle,

bv proclucing hcalthiel and happiel prcenancies anclprorrititine bettel relationships rvith health care provid-ct's. Thcse, in turn, lnav cncourag'e better ltalcnting,r-r-rore completc childhood imnrunization, ancl inrprovt-ments in other hcalth-related behavior. In this reeald,the comtncnt bv Enkin an<l Chalnrers on the value o1'

preniltal carc is apt: "trlanv thinss that lcallv countcannot be counted."i1

Finalh. rçe shoulcl corrsi<ler nl'rcthcr cost saritrqs istlie appropriatc criterion ltv uhich to .judgc prcnatalcarc programs. It is ter.r.rpting to assllme that in orclerlbr thcse pl'ograms to be valtrable, thev re:rll1' shoulclsave morc than the\ cost. \'ct $ herr $ e rccluirc prenatal

\k)1. 33 I \o. l9

care , and other prcvcntive health carc, to pai'f'or itself,\\'e lrav be inadvcrtentlv denving valuable bcncfits tosocietr'. lt mav be better to ask not "Horç much doesthis salc?" but, rathcr, "Hou' much is this u'orth?"

Group Health CooperativeSeattle. WA 981 l2

University of WashingtonSchool of Public Healthand Community Medicine

Scattle. WA 98195

J,rsl: Huvn\cro\, \[.]).

FRI:ol.Rrcx A. Cosxr,r.L,\{.D., \I.P.H.

Acldress reprint requcsts to [)r. f]onnell at thc f)epartnrcnt ol'Hcalth Services. SC-117. Unilersitr of \\hshington. Scattle. \VA9t r95.

Supportcd iu part bv a grarrt (\IC.J !)0{3) fionr the \Iaternal ChilclHcalth Bureau of the Health Rcsourccs ancl Scrvices Àdministration.

RnreRrNcrsL Korenbrot CC. Risk rtduction in pregnancies of low-income women: conl-

prehensive prenatal care through the OB Access Project. IMobius 198-1:-l(3):

34-.11.2. Corsky RD- Colby JP Jr. The cost el'fectiveness of prenatal care in reducing

low birth weight in New Hampshire. Health Serv Res 1989;24:583-98.3. Institute of Medicine. Preventing low binhweight. Washington, D.C.: Na-

tional Academy Press, 1985..1. Leavitt fW. Brought to bed: childbearing in America. 1750 to I950. New

York: Oxtord University Press. 1986.

5. Wertz RW. Wenz DC. Lying-in: a history of childbinh in America. Expand-ed ed. New Haven. Conn.: Yale Unirersitl Press. l9lJ9.

6. Thompson JE. Wulsh LV. Merkatz lR. The history of prenatal care: cultural.social and medical contexts. ln: Merkatz lR. Thompson JE. eds. Neu per-spectives on prenatal care. New York: Elsevier. 1990:9-30.

7. Huntington JL. Relation of the hospilal to the hygiene of pregnancy. BostonMed Surg J l9l3;169:763-5.

8. Sardell A. Child health polic.v in the U.S.: the paradox o[ consensus. J HealthPolit Policy Lau' 1990:15:27 I-30.1.

9. Lennie JA. Klun JR. Hausner T. Low-binh-weight rate reduced by the Ob-stetrical Access Project. Health Care Financ Rev 1987:tt(-l):83-6.

10. Moore TR, Origel W. Key TC. Resnik R. The perinatal and economic impacrof prenatal care in a lou-socioeconomic population. Am J Obstet GynecolI 9tl6: I 54:29-33.

I L Leppert PC. Namerow PB, Cost averted by providing comprehensive prena-

tal care to teenagers. J Nurse Midwilèry l9tl5:30:285-9.12. Lereno KJ, Cunningham FC. Roark ML. Nelson SD. Williarns ML. Prenatal

care and the low binh weight infant. Obstet Gynecol 1985:66:599-605.

l 307

Nlorales WJ. Vaughn BJ. Diebel ND. The cost of no prenatal care. J Fla \{edAssoc 1985:72:852-5-Schramm Yr'F. Weighing costs and benefits o[ adequate prenatal care tbr12.023 births in Missouri's Nledicaid program. 1988. Public Health RepI 99:: I 07:6-17-52.Wilson AL. Munson DP. Schubot DB. Leonardson G. Stevens DC. Does pre-natal care decrease the incidence and cost of neonatal intensive care admis-sions l Am J Perinatol 1992:9:28 1 -4.Oftice of Technology Assessment. Healthy children: investing in the tïture.Washington. D.C.: Govemment Printing Office. 1988. (OTA-H-31-5.)Schwartz RM. What price prematurityl Fum Plann Perspect l989l2l:170-1.Hall MH. Chng PK. MacCillivray I. ls routine antcnatal care uorthwhilelLancet l98O:2:78-80.Peoples MD. Grimson RC. Daughtry GL. Evaluation of the eflects of thc

North Carolina Improvecl Pregnancy Outcome Project: implications lbrstate-level decision-makin-c. Am J Public Health 1984:7.1:5.19-5.1.

Strobino DM. Chase GA, Kim YJ. Crawley BE. Salim JH. Baruffi G. Theimpuct of the Mississippi Improled Child Health Project on prenatal cueanrJ lou' birthueight. Am J Public Herlth l9E6:76(3):27+-8.Piper JM. Ray WA. Criflin MR. Ell'ects of Medicaid eligibiliti' expanrion on

prenatal care and pregnancy outcome in Tennessee. JAMA 1990:26.{:2219-23.Guyer B. Medicaid and prenatal care: nccessar)'but not sutÏcient. JAMAI 990:26-l:126.1 5.Buescher PA. Smith C. Holliday JL. Levine RH. Source of prenatal care andinfanl birth weight: the case of a Nonh Carolina county. Am J Obstet Gy-necol 1987:156:20-1- I0.Sokol RJ. \4bolf RB. Rosen MG. Weingarden K. Risk. anteparlum care. and

outcolne: jmpact of a rnatemity and intiint care project. Obstet Gynecol| 980:56: I 50-6.Oakley A. Rajan L. Grant A. Social support and pregnanc) outcome. Br J

Obstet Gynaecol 1990;97: 155-62.Currl' MA. ed. Access to prenatal care: key to preventing los binhweight.Kansas City: Arnerican Nurses' Association. 1987.Poland ML. Ager JW. Olson JM. Bariers to receiling udequate prenatalcare. Am J Obstet Gynecol 1987:157:297--303.Sr Clair PA. Smeriglio VL. Alexander CS. Connell FA. Niebyl JR. Situation-al and linancial bruriers to prenatal care in a sanrple of lou.income. inner-city women. Public Health Rep 1990:105126.{-7.Lazarus ES. Falling through the cracks: contradictions and barriers to carcin a prenatal clinic. Med Anthropol 199Ol.12:369-87.

Haas JS. Udvarhelyi IS. Monis CN. Epstein AM, The ettect of providinghealth coverage to poor uninsured pregnant women in Massachusetts. JAMAI 993:269:E7-9 I .

Etlèctiveness and satisfaction in antcnatal care. ln: Enkin M. Chalmers l.eds. Eilèctiveness and satisfàction in antenatal care. Clinics in dcvelopmen-tal medicine nos. 8l/82. London: Spastics lnternational Mcdical Publica-tions. 1982:266-90.

SOU\DINC BOARI)

ti

t5

20

2l

l+.

16.

t7.

18,

19.

13.

25.

26.

21.

28.

29,

30,

2.1

3t

l 308

CORRESPONDENCE

THE NEUROLOGIC BASIS OF FEVER

'lb the liditor: Saper ancl Bredcr provide a comprchensiverclieu'ol'the neurologic basis of Ièrtr (June 30 issuc).r Thcr-mc»'egulation invo]r'cs a complt'x intcraction of autononric,cndocrinc. and behavioral responses sovcrnecl largelv bv theI'rvpothalamus. Darlage in thc preoptic region, uhich lunc-tir»rs as a thermostiit ancl contains mct'hanisms lbr hcat <lis-

sipation, nlal cause hcat intolclance or thermostatic clvslirnc-tiàn, ancl ,râ f".'.. rcsponsc uill occur.l ' Lesic,n.; of theposterior hvpothalamus or brain stenl rra\- clestt'ov the path-u'a1-s {«rr autonomic and behaviural thcrmoregulation. rçithconsequcnt hr'potherrnia or poikilothertnr'.r

Thc authors do not clearll'dillèr'entiate bcnteen lèr'er anclnon-pvrogen-induccd hr-pclthermia. F-cvcr is a regulatecl clt--valiun ir) thc preoptic lcmP( r'alur'( srl l){)iilt in l.esl)ollse l() ( ir-culating pr-rogenic cvtokines. In cor.rtlast, in hrpcrthernriathere is no rcsettirlg of thc set 1>oirrt or'1rrroqenic lesponse;instead, the thermoreglllatorv mechanisrns are ovenlhehrredbv excessive heat ploduction, heat storage, or impaired heatdissipation.'r' r CIinicallr, the dillerentiatiou bclrteen [èr'er anrlhrperthermia is oftcn cli{Tcult and depencls largeh on a me-ticulous histolv taking, but it is pivotal fbr the managemcntol elevated core temperatule.3 + Hvperthclmia can easilv bcmistakcn for (intermittent) lever and licc versa. particularh'in paticrrts rritlr poikil,,rhrlrn\ ol lrcatstt,rkr'.;

6500 HB \ijmt:gen.the Netherlands

-\Lrnn-s A. \I.\cKH\zrl, \I.D.Glnr..rcu l".F.II. Prc'r'r*s, \I.D.

Au R.\I.\I. Henrrr s. \I.D.L ttiri lsitr Hospital \ijnregt rr

THE \E\\'L\GLA\D.lOUlt\Ài. OI: ]IEI)ICII\D \ov. 10, l99l

l. Saper CB. Breder CD. The neurologic basis offever, N Engl J Med 199'113-30r

I ti80-6.2. Saper CB. Autonornic disorders and thcir nranagenlent. In: Wi'ngaarden JB.

Sûrith LH Jr. Bennett JC. cds. Cecil texttrttok of ntedicine. l9th ed. Philadel-phia: W.B. Saundcrs. 1991:2091-8.

3. Simon HB. Hlperthermiu. N Engl J Med !993:-329:"183-7..1. Coodman EL. Knochel JP Heat stroke and other fbrms of hypertherrnia. In:

Mackosiak PA. ed. Fever: basic ntechanisms and managemcnt. Neu'York:Raven Press. 1991:267-87.

5. MacKenzie MA. Hermus AR. \\irllersheinr HC. et al. Poikilothermir in mun:pathophlsiologl and clinical implications. \'ledicine (Baltimore) 1991:70:257-68.

To the Editor: The revit:rç bv Sapel ancl Bredel is cxcellent.but their l:rst tlo sentcr)ces nrake placticing pecliatric ncurol-ogists shuclcler. The in.rplication that ne\r rnedications to pre-vent rreuronal daDiage iD stlokc nrav bc especiallY inrportantin treating chilch'en u'ith lèblile seizures isnorcs one o1'thcbasic làcts about Ièblile seiztrres that rte have leartrccl lronrthe \ational Collabolative Pcrinatal Ploject.r I lt has beenrrcll established that sirnple l'cbrilt' seizurcs causc no lleur-o-logic morbiclitv in chilclrcn. Such seizurt's ltave no ellcct on in-tclligence later in lile and alc not associated l'ith r-nolbiditvor r.nortalitr'. Bv iniplving in their closing scnterc('s that ncu-ronal protection for hr'perthcn.nia rnar-lle important in tt'cat-ing cl-rilclrcn u'ith lèbrile seizures, the authors ignole tl-ris

impoltant fàct. ltrr rt'ars pediatric neut'okrgists have bectrtrving to clispel this r-nvth about the cllects of lèbrilc seizurcsin childrcn.

Houston. ]'X 77030

J,rrres \V. \Vrn:r-Ess, ]I.D.Universitr- oi'Texas

\lt:clical School. Houslorr

l. Nelson KB. Ellenberg JH. Prognosis in children u'ith febnle seizures. Pediat-rics 1978:61:710-7.

2. Ellcnberg JH. Nelson KB. Febrile seizures and Iater intellectual perfbrmance.Arch Neurol i 978:35: I 7-2 l.

3. Hirtz DG. Nelson KB. The natural history ol febrile seizures. Annu Rel Med198.1:3,1:451-7 I .

,1. Freeman JN{. The best medicine for tèbrile seizures. N Engl J Med 1992:327:I 161-3.

'lb the Etlitor: In their .*r, revierr, Drs. Saper anclBrecler stale that "r'cn high bodv tcrnperatures (abovc {{J"C)

Instructions for Letters to the Editor

Letters to the Editor are considered for publication (subject to editing and abridg-ment), provided they are submitted in duplicate, are typewritten and triple-spaced,and do not exceed 400 words of text, a maximum of five references, and one figureor table, Letters should have no more than three authors, and all should sign theletter, Please include a \,ÿord count, your telephone number, and your fax number (ifavailable).

Letters should not duplicate similar material being submitted or published else-rvhere, and they should not contain abbreviations. Financial associations or otherpossible conflicts of interest should always be disclosed. Submission of a letterconstitutes permission for the Massachusetts Medical Society, its licensees, and itsassignees to use it it the Journal's various editions (print, data base, and opticaldisk), in anthologies, revisions, and any other form or medium.

Letters referring to a recent../ournal article must be received within four weeks ofthe article's publication. To expedite receipt of such letters, we encourage authorsoutside the United States to communicate by fax (617.739.9864 or 617-734-4457),

We are unable to provide prepublication proofs, and unpublished material willnot be returned to authors unless a stamped, self-addressed envelope is enclosed.Receipt of letters is not acknowledged, but correspondents will be notified when adecision is made.

\,'ol. 33L\o. 19

can clearlv injure both the central nervous s.vstcm and otherbod1,s1stems." This has not been proved. About 20 percentol' children sccn in the emcrgencl' roorn have temperaturesover -l0oC,r but the,v usuallv havc a full recover\.. Il there ismorbiditv or mortalitl', it is due to the underlving disease. Theassociated fi'r,er mat, in lact be protcctive.

Fcr.er, commonlr.clue to viral infection, mav induce ièbrilescizures in 3 percent ol susceptible children. Temperaturesabove 40oC at the onsct are reported to be associated n'ith adecreased incidence ol' lebrile seizures.l In one studv leverabove 40oC in children rvith nrcningitis clid not indicate apoor prognosis, but all thc children presenting u'ith hypothcr-mia died.l

With lèver, unlike hyperthermia, bodv temperature is rvellregulated b1'a h1'pothalamic set point that balances heat pro-duction and heat loss eflectivelv so that the temperature doesnot excecd an upper limit ol'-12"C. Within this upper range,-[0 to 42oC, thcre is no evidence that thc lèr'er is injrrrious totissue.

Although it has been difficult to establish a critical thermalthresholcl (defined as the tempcraturc abovc rvhich tissuedarnage occurs) in humans, a temperature above -l2oC is like-11' to inducc such damage.+ Houever, temperatures that highare usuallv due to hvpcrthcrrnia, u'hich is not regulated cen-tralll'. Ternperatures ;lbove -12"C are uncommon,5 eind com-plications and mortalifi'lirom temperatures abor,e -12"C aremore closel'r related to the severitr,ol'the underlling diseasethan to the level of the temperatr-rre.

Sidcup, Kcnt DAl4 6LT,United Kingdom

A.S. Er-R.rosr, \LR.C.P.

Queen \Iar1"s Hospital

r 309

belorv). Establishing an arbitrar,v temperature threshold mar'be lcss important than understanding thc pathophvsiobgy olbrain injurt; rvhich mav occur at r.arious bodv tcmperaturesdepending on the physiologic state of'thc patient.

l'inallr; Whelcss and El-Radhi makc the important pointthat simple lebrile seizures (i.e., single, brief, nonlocal convul-sions) in children are nearly alu'ays associated rvith a gooclprognosis and thereforc do not rcquire treatment. The evi-dence supporting this vierv is quite strong, but unfortunatelynot all Ièbrile seizures are simple. 'fhc prognosis is muchn'orse for patients n'ith repeated or prolonged seizures associ-ated nith lever, since alebrile seizures later devclop in l0 to20 percent of these patients.r'b It is still unclear uhethcr thelever lorvers the threshold lirr a preexisting (but prer.'ious11,

clinicall,v silent) seizurc disorder or u,hether the prolonged orrcpeated seizures are themselves injurious to the brain. Holr.ever, recent advances in understanding the pathogenesis o[partial complex scizures suggcst that prolonged exposure olneurons to high levels of excitator)'amino acids during sei-zures may cause cellular damage.' The tendeno lor h,vper-thermia to depolarize excitable tissue mav cxacerbate thisprocess, rvhich is the basis lor our suggestion that neuropro-tective agents aimecl at blocking :V-methyl-o-aspartate rc-ceptors ma1' find a place in the treatment of lebrile seizurcs.This treatmeut lr'ould, of course, apply onl,v to complex lèbrileseizurcs. Simpie Ièbrile scizures are bv definition briel and notrepcated, and hencc are finishcd by the time the diagnosisis marle.

Boston, \IA 02215

Chicago, IL 60637

Cr.n'r-oRD B. S,\I,ER, \,I.D., PH.D.Beth Israel Hospital

CrtnisropunR D. Bnr:»ln, NI.D., Ps.D.Universin of Chicago

l. Simon HB. Hyperthermia. N Engl J Med 1993;329:.183-7.2. Yi PN. Cellulu ion content changes during and after hyperthermia. Biochem

Biophvs Res Commun 197 9 :9 | :17 7 -82.3. Yatvin MB, Cramp WA. Role of cellular membranes in hyperthemia: some

obseruations and theories reviewed. Int J Hyperthermia 1993:9:165-85.4. Laszlo A. Davidson T, Hu A, Landry J. Bedford J. Putalive deteminants of

the cellular response to hyperthermia. Int J Radiat Biol 1993:63:569-81.5. Verity CM. Ross EM. Golding J. Outcome of childhood status epilepticus and

lengthy febrile convulsions: findings of national cohort study. BMJ 1993;307:225-8.

6. Annegers JF. Hauser WA. Shirts SB, Kurland LL Factors prognostic of un-provoked seizures after febrile convulsions. N Engl J Med 1987;316:493-8.

7. Sloviter RS. The functional organization of the hippocampal dentate gyrusand its relevance to the pathogetresis of temporal lobe epilepsy. Ann Neurol1 994:35:640-54.

RISK OF RECURRENCE OF BIRTH DEFECTS

To tlte Editor: Lie et al. (July 7 issuc)r report that i{'a wom-an's first child had a birth delect, the risk ol'her second child'sbeing affected u'as increased, but that this excess risk in thesecond child l,as lou'er il'thc u'oman moved alter the birth olher first child than if she staycd in the same place. They at-tribute this differcnce to cnvironmrnt.

There are a number of other explanations lbr this result.First, u'omen u'ho move to another municipality are likely todillèr in a numbcr of s'avs from those nho do not move, w-ithregard to other factors related to the risk of a birth dclèct,such as socioeconomic status. Differences in this and othervariables ma1'explain at least some of the dilference attribut-cd bv Lie et al. to environment. Second, the ascertainment olsome birth delccts mav vary considerablyl Thc authors main-tain that the mother o{'an infant u'ith a birth delèct is notlikelv to overlook a similar defect in a subsequent inlànt. This

CORRESPO\DE\CE

1. El-Radhi AS. Carroll J. Fever in paediatric practice. London: Blackwell Sci-entific.1994:23.

2. El-Radhi AS. Banajeh S. Effccts of fevcr on recutrence rate of ttbrile convul-sions. Arch Dis Child 1989;64:869-10.

3. Wong VK. Hitchcock W. N{ason WH. Meningococcal intèctions in children:a review of 100 cases. Pediatr lntèct Dis J 1989:8:224-7.

.{. Bynum GD, Pandolf KB. Schuette WH. et al. Induced hypenhemia in sedat-

ed humans and the concept of critical thermal maximum. Am J Physiol I 978;235:R228-R236.

5. Sirnon HB. Hyperthermia. N Engl J Med 1993;329:.183-7.

The authols rcph':

To tlte Editor: NlacKenzie and colleagucs note the impor-tance ol'distinguishing lèr'er lrom h1'pcrthermia. trVe deliber-atell'did not focus on hyperthermia in our article, sincc it uascovered in detail b1'Sirnon in a recent revieu,in the Journal.lWe agree uith the points made b1'\IacKenzic et al. concern-ing the phvsiologic distinction bctrt,een lèr.er and hyperther-mia and the importance of- considering thc clinical sctting indilli:rentiatins bctrvecn the two at the bedside. Becausc leverand hvperthermia mav occur simultaneouslr. it is als'a1.s im-portant to look lbr infectious causes ol lever in patients rvithelevated bod)' tempcraturcs, even if there is an explanationlor l-r'n'perthermia.

Thc issue of'cstablishing a threshold Ibr thermal lissue in-jur1. in humans, raised br. El-Radhi, remains controversial. Anelcvated bodv temperature has graded t:llects on a varietr,ol'cellular processes ranging Iiom the s)'nthcsis o['nucleic acidsand protcins to changes in membrane lluiditv and intraccllu-lar ion conccntrations.2-r Long-term damage. at lcast to thenervolrs s)stcm, is likell'to bc associatecl \\'ith the release ol'excitator)'amino acids due to neuronal depolarization. Thisproccss mav be cxacerbated b1'h1,poxia (e .g., in pneumonia),ischemia (c.g., r'hen an elevated temperature precipitates acardiac al'rl1\lhmia), or prolonged or repeated seizures (see

is Iikclv t() bc tl'ue, horrcver. orrlr. li»'a major birth rlelèct. Rec-oqllition of the mole numcrous ntinor birth dcfccts lrr rnoth-ets oL mt'c'lical persor.rncl is likeh to bt- mort- raliablc. as isr'(-porlillg o1'thern, aud rninor clcli'cts ltlobabh contribute int-portantll to sorne o1- thc categories anallzed bl l-ic ct al.'fhus. ecogl althic dilli'rcnccs iu the recognitiorr anc[ rcpoltinqol'such clclècts coukl ac('ount lbl thc. authors' observatiorrs.fhe lorrel risk olclelècts in thc secontl chilcl o['a rromau s.hohacl chaneed ht'r citv of rcsiclenct sincc thc birth ol'her firstchilcl rçoulcl exemplil\'resrcssion to thc nrean. Also. it is notclear rçht:ther the anah'sis incluclccl delects detectecl bv post-nlortcnr t'rauritration. II'it rtid. thcn ccouraphic I'ar-irrtion ir.r

the fre<lrrencv of autolrsics coukl also t'ontriltute to tlie repot't-ed trend.

All tht:sc issues shoulcl be adch'essed btl'ore thc aurhors'staterrcnt tltitt "cnvilotrntctrt pl:rvs a strorlg J)ul't in lcpcatedrlelècts" can be accepted.

Lnrryl B. Iloor, \I.D.Unir clsin' ol' (lalilbrniir, Berkelcr,

School ol' Public Health

1. Lie RT. Wilcor AJ. Skjtenen R. A population-based stud]' of rhe risk of rc-currence of birth defècts. N Ensl J N,ted 199.1:33 I : 1,.1-

TLt lhr Editor: Lir: ancl his collcagues linc[ that a rnothcr'schaucc ol-havins a second babv uith a birth detèct is hah-edi{'sht' tnovcs to alx)thel' toln alitr thc bilth of her first babr'.I1. as thcl suggest. tl-ris is due to r1r1 cscrape li'onr tcratoeens,lathcr thal) to sonr(' ch:rractct'istit of ruoltilitr, sLrch as incomr,therc should be arr opposite clJèct rçht:n nrothcrs ol- normalcl'rilclrcn movc into atr altèctecl arca. ()1'such nrothels. 316 per100.000 (3 l0 ol' 89.:r8ll) hacl an abnolrnal second chilcl. ascotnparcd u'ith 235 pel 100,000 (15{ o1' 192.990) ol those uùostavecl put. It l ould be interesting to knorl u hether this cllcctis coulirrncil bv topoelaphic iinalr.sis.

fordingbrirlgc.

Berkelcr. (:-\ 9+720

l'Ht. \F.\\ t:\GLA\t)JOt R\.\L ()t \IEDI(:l\t \or'. 10. l!)9-l

Dr. Hughes-Davie s asks u,hether the patterns o[.recurrenccol birth clelrcts miqht be cluc to the concentration ol ter:rto-gens in spccific gcographic zrrcas. J.he seneral tcrrdencr-irrNonrav has becn to movL- fion lulal to ulban arcas, andsolrc' t\-pes ol dclêcts r.nav be dctcctecl bettel in urban :rlcas.'fhis mav have contributed to thc slightlr higher prcvalenceol dclècts in the seconcl inlànts of'liunilies rrhose Iirst inlantshad no delèct and rrho thcn movccl. Houever.. this nruld notcxplain thc decreased risk of'rccrrrrcnce aurong {irn-rilics rçhotnovr-cl alter having a hrst inlant rçith a dt'l't:ct. \\i: cloubt thatdillèrences in thc risk of reculrence ivith nioving are rclateclto dilli'rent background ratcs ol'clelècts, because thc back-grouncl tates clo not varl greatlv in clillèrent art:zrs o1'thecoulltr\:1 It seems more likeh'that làrnilics rrhose first inlirntshad a clefèct have sor.ne susccptibilitl to teratosens in theirimrnccliate envir.onmcnt (thc houst:holcl. thc neic-hbor.hoocl,thc u'orkplace. and so lbrth) and that this crrvilorrnrtrnr is

chaugccl bv r.noving.

Rolr- Trt1l, Lrr,. Prr.I).\lcclical Birtli Reqistn'

\-502 I Bcrgen, \oru'av ol \r»'tçar

Alr.r:r J. \\rrL{:clr, \1.D.. Prr.D.Rt'seat'ch Tlianele Park. National Institute of

NC 27709 Envilonmcntal Health Sciences

Ror.r' Sxl.rnvr\. Plr.I).\Icclical Birth Registlr'

-\-502 I Br:rgen, \oru'al ol'\or.l'av

l. Medical Birth Regisry of Norway. Annual report 1990. Bergen. Norua),:University of Bergen. I 99 I .

ACETAMINOPHEN POISONING AND LIVERFUNCTION

To lhe Editor:'Ilie suggcstion bv C)heung et al. (Junc 30 is-sue)l that an acetaminophcn overclosc itr ern alcoholic shotrlclbe treatcrl u'ith acetvlclstcine regardless o[' thc scrLlln acc-taminc4rl'ren concentration is unkrundecl an(l coutrar\- to thcavailablc cvidencc. The litelatulc includes lell ovt.r 10.000cascs ol'acetarninr4;hcn olerdosc.l r ancl the ar.ctuuulatcrl ex-pcrietice l'c»'lcluicle must involvc rrell ovcr 100,000 c:.rscs. De-spite this extt'nsirt crpcrienct', \\'e al-(, unu\\'are ol anv othetbona fide cxamplt:s ol tht: plrcrror.rrenon descrilled br' ()heungand colleagues. u'ho reported a cast'of fatal acetaminophcn-inclucecl hcpatic firilure irr an alcolrolic paticnt despite zr "non-I()\ic" set unr ar'ctantittolrltrn (()il( r.llltillioll.

One must first question 1hc acculacv ol'the histon- in thisctrse, l)articularh'$'ith |egarcl to tl'rc time ol'insestioll o{'acc-tanlillol)herl arrcl the possibilin'ol'repeatccl ingestion. It is in-conceivablc that the insrcrstion of 2j g of acctanrinophcrrloulcl rcsuh in a cr»rcentraticrn ol-onlv l2{ pg pt'r milliliterIirur hours alter ingestion, suggestine that either thc amourrtor thc time is inaccura(c. Such an clror is lirrthel suggestcdbv the occun'encc o{'severe hepatic lailule u'ith an undetecl-ablc serunr acctaminophen concelltlation içithin 50 hc>urs al'-tcr ingestion. In patients l ith such lirlminant toxic eliècts, tht:metabolism ol ace tarninophen quicklv becomcs neuliuiblc, re-sr-rlting in the pclsistenct-- o1' kxç le vels of circulatine ac( tamir)-ophen.

Remalkablt' caises clo occut', horrcver. an(l consider:rtionmust also be given to the implications ol'this case il the inlbr-nrati(»r providecl is corrcct. \\t: agree that both stuclit-s in an-iu-rals atrcl l-el)orts oI long-telm irrg^estion ol' a<etarninophensugsesl that alcoholics arc at ir.rcle:rsed risk firr heprtotrxitcflècts ol'the clrug. \\'hr. then. ..rre such ellècts not rcported

lJarnpshile SP6 2EJ,flnitccl Kingclrn-r

1..H. IIur.;r rrs-D.s'rns. F.R.C.P.Blt'ar.nore \Iitrsh

'fhe autlxrrs replr':

To lht Editor: f)r. I'Iook rzriscs tuo points irbout alternativ'intcrplctatior.rs o['our' <lala. I'hc first is that firrni]ies rçho nroverrar bc a sclcctcd erorrp. f'his is no doubt trtrc. \\t sought toclcal lith this 1>r'ol:lern iu tlre aralvsis l>r'making the cornpar-isorr n ithiu thc grltrp ol lantilies rvho hacl ntoved

- tliat is,

l;rnrilit:s rçlrose first inlants had no clclèct and lho then movedrltrc the lt-lèrencc groult lcrl thc làmilies rçltrse first iulântshacl a clefi'ct zrucl uho thcn nrovccl. His scconcl con<:cnr is pcr'-haps mort' sclious. He asks rlhethel clillèrenccs in the ascer-taiumclrt «r{'clcli.cts rnight account lor tlrc observcd trcnds. Inellèct. coukl thc lalc o1'ascertainr.ncnl ol'a birtlt clelèct in a

sccr»rcl inliurt sinrilal to rhe r»rc in thc fir'st ir-rlant be 50 pcr-ccnt lolcl irr u cli[Ièrent hospital? Althorrgh therc is consider-able unclct'uscertrlilunent r>f clclects ovcrall. rte clcl not lhinkthe di{-lt-rerrccs in the r';r1cs of asccrtaiumcnt coulcl bc tl'risl:rrge. cspecialh firr lamilies rrho al'cadl had an allèctecl in-lànt. \\t- crplorcd this prol;lcnr I;r- r"rnorins llrrnr the ar.rah sis

chrbftrot. the nrost comm(»r clclèct and ort: içith notol-ioush'poor asceltainrnent. If as(ertainnrerrt bias cr»rtributed to thcrcsults. the ellèct shoulcl be cspci:iallr.strong Ii».club{iiot anr[l'euku lbr tht'r'est. Lr Iact. ue liruud the opposite. Althouehutrknou n biascs nrar be afIècting our results. l e f ôuncl no er.irlerrcc that incornpletc ascertirinnreut is a conlriltrrtine firctor'.

Vol. l:i.:i I \o. I9

n.rot'e licrlrrentlv in alcoholics with iicetarninopherr ovcrckrscs.rl-flre

ust: ol'ir conscrtativc nor.noglam (25 pclcr:nt lorrel thanthe original clata-basecl vcrsion)l anrl liberal closes ol'acrtrl-cvsteitrt' (11330 nrg pcl kiloulirrn ol botlv rrcight over i-L thlce-clai- pelirlrl) are likcll r\plilnirtions and appeal to ovel'c()lncan\ disa(hantage associatc(l rrith alcoholism ol'other condi-tions.

Lorg-tc|rr use ol alcoh()l appears to be relevant to the clis-cussion of' the rnana{enlcrlt ol' long-tt:rrrr overingestion of :rce-turninoplror. Erpcricnce sugge sts that ulcoholir:s, patients rc-cciving thcr':rp1' llrat causcs tlrc irrrluctiorr ol cr tochlor-nr:P--150 cnzvmes.l anc[ inlànts rr'it]r Icbrilc ilhrcsscsr are at sub-stantiallv higher risk lor licpatotoxic eflècts. Such ellects arecssentiallv ullhear'(l o['in patients rçithout these risk Iàct«rrs.

Givcn tht' high plcvalenct' ol both alcohol abrrse ancl acrrtea( etanliilol)llen ovcr'(lo:jc. ltlr\ ( hangc irr tlcllttl]e nt lecon'r-rlrcn(laliolls rçotrlrl havr a sul)stalltiul c[li'ct. Palticulalh rriththe cur'r'eIit rmphnsis otr cost c11i:ctivcness.:r chiurg,: in trt;.rt-rncrlt Ie(on)r1cn(lations on the basis ol this ol alrv olhcl |e-nlal'kal)l(' tuse l'rltrlrl be u l;ig stcp blckrr:rrcl.

^\I.\Rl'r \ .f . S-rrrr.xs'r'r-r-r, r\I. l).Drrtrt- R. [)ot'r;r-.rs. \1.D.\Iou.*rro R. D.»'.r, \I.D.

Poltlancl. OR 9720 1 Olcgon Hcalth Scienccs Univelsitr-

l. Cheung L. Pons RG. Nlever KC. Acetanrinophen trcatnlent trorlrogralr.;r.- Engl J N'led l99l:-1-3():1907 8.

f. Rurnack BH. Peterson RC- Koch GG. Amrra lA. Acetrminophen orerdose:662 cases rr ith evaluation of oral rcetvlc\ steine treatncnl. Arch lntern Medl9Ul:l-11:380 5.

-î. Snrilkstein MJ. Knapp GL. Kulig KW. Run)ack llH. Eflicacl' of oral N-ace-tvlcysteine in the treatmcnt of ucetunrinophen orerdose: anulysis of the Na-tional Multiccnter Studi ( 1976 lo I985). N Engl J Nleti lt)EE:319:1557-62.

-1. Bray CP Harrison P\'1. O Gradl'J(i. Tredger JNl. Willianrs R. Long-terrrl an-ticonrulsant therapy \\orsels outc(nre in paracctamol induced tulrninant he-patic fuilurc. Hunr Erp Toricol 1992:l l:265-70.

5. Henretig FN{. Selbst S\,I. Fonest C. ct al. Repeutc-d acetanrinophen orerdos-ing: causing hepa«r«rxicitS in childrcn. Cliu Pediatr 1989:18:525-8.

Tit tlrc Editor: I lln suLc I rr'ill rrot bt' thc onlv lJlitish lracleluho fèels nrovt'cl to lcplv to tlrc lcttct'lrr Cheulrg ct al. Thear:ctanrinr4rlren tr('atnrent ]lomogranr is inclcccl rçicleh' ust:cl ;rs

a guiclt-linc 1ôr' 1r't'atmenl irr paticnts u ith acctanrinoplrerroleldost's.r hert- as irr tlrc L'nitr-cl States. Horrever, in thiscountr\, il not in tlrt Unitcrl States, it is alreatly *ide11'r'etog-nizetl that cthanol, as \rell as nranr otlrcl hepatic cr tochlonralenzvnre-inclucing rlruqs, n)aY Cor)foun(l t[ris aPproach. lb| thisl'e:tsorl, it is consiclerecl loutirre l)rrlctice to hah'e thc level atrlhich tleatnrt'nt is corrsiclelr'(1. as prol)osccl in thc curr'('nt is-rjrre of tlr(' IJriti.çh -\'ational hi»uulatt'.)

If Cihcung et al. h:rcl takt'n this into account, the ac(:tanlin-ophett lcvt:l o1' I2{ pg per r)lilliliter irt lirut'hout s (rrhich is lcssthan 2(X) p.s per ulilliliter-. thc cutolT pr-rint in thc nomogram,l)ut nlolc than 100 g.g per nrilliliter'. hall thc cLrtolf point)\rould lraYc clictatccl tl'catnient §ith acctvlcrst(:ine, n(]t non-t|eatnlcnt. A dcclinc that bcciLrne irlcr,ocablt'lnight []a!ebeen plevetrtcd. As Cihcung ct al. pt-rint out. tlicre shoulcl be aIorç threshold Iôr tleatnrcnt rr ith acetvlcr stcinr. sincc tlrc haz-ar([s ol'su(h treatnicnt ale ncgligible.r

Fiilkirk FKI 5QE. A\I)RL\\' D.s rt,. lI.B.. Clu.ts.Unite<l Kins(lonr Falkilk an<l Distlict Roval Intrnrarv

L Runack BH. Peterson RC. Koch (iG. Anrara lA. Acetarljnophen oveLdosc:662 cases Iith eralurtion of oral ucetYlcysteine treatment. Arch Intcrn Medl98l:l1l:180-5.

l. British National Formulary. Nu.27. N,larch 199-1. London: British MedicalAssociatinn. 199.1:20- l.

.1. Miller LF Rumack BH. Clinical satèty of high oral doses of acetylcysteine.Semin Oncol I98.1: l0:Suppl I :76-tt5.

ISlt

7'o tlte llditor: Chr:ung et al. dcscribe their crperiencc litha 25-r'car<rlcl \\'oInall rçho hacl consllm(]d 25 g ol aeetanritr-ophen il a suicide attempt but u'as not treatecl l ith acetvlcls-teirrc bccatrse she rvas not coltsidered to ltc a risk orr thc l>asisol' the acctaminophen treatulcnt nolrogram. fulmin:rnt he-patic lailure subsequentlv developed, alrd she clied. 'I-hc lca-s()n, the authors conclude, is that shc had consumed Iargeanlounts ol'cthanol lbr a long period. Long-terrn usc of etha-nol \r'as thought tr> have altered the m(:tabolism ol'acetamin-ophcn sullicientlv to invalidale the plt:clictile valtrc ol' tl.re

acctatnitiophcn treatmcnt n()nlogranr fi)r' deterrnitling hep:rto-toxicitl,. 1'hus, thc authors conclude that a carelul historv ofalcohol usc should be taken lbr er-erv paticnt \ÿith an aceta-nlinophen overdosc.

There arc several serious problems u'ith the arrlhors' ltle-sr.lpl)ositions irnd conclusions in this cast'. Filst. 25 g ol lcc-tallrinopllcn (u'hich is rnoLe than six tinres tht: ntasiural lt:c-(»nmt:ndccl rlailv closc) is clcarlv in tht: toxit: r'alrgc.l Sc(olr(l!although thc acetarriinophell tl'c,rtmellt nonrogranr is ust:lirllot'populatiorr studirs. it shotrld ne\cr llt: uscd t() n'ithholcl :rbenign ancl potcntiallv lil-csaving trcalmcnt in thc c:rse ol'aknon'n ovt:r'close. Thircl, the rnagnitude bv u,hich alcohol cr.r-

hances the toxic ellècts o1'acetarninophen has nriver bccn es-

tal;lished iu humans. 'fhc indtrction ol- o lochrotnc P-+50 2lrlcllzymc acti\,it)'by alcohol is onlv trvolbld «r thleelirld ancl Le-

tLlrns to normal *'ithin fir'e da1,s,J suggestins that lecent alco-hol r-rse is rnore inlportallt than long-tclm use. Irurthemor(-.tlre rc<iuction in serum glutatl-rionc levt'ls in alcolxrlics is orrlr'nroclcst,:j and thc rcduction in hepatic glutathionc lcr-els <ltrr'-

ine ethanol consurnption. u,hcn controllt:d Ibl lirsting. has rrotbeen dernonstratecl ilr humans. l-inalh, ot]rer fàctors that nravincleasc the toric cfk'cts ol acctamillol)hen. such as a gL-nctic

1>t'cdisposition or làsting,+ rnav be evcn nlore inrl:ortarrt thana liistorr' «rf alcohol consumpti«rn. Thus, the casc b)- C)hcrrnget al. pr-ovides an argument fi.lr the treatntcnt ol'l)atie llts u ithacetatninophen ovcrcloses rcgardless <>l thc sclunr lcetanrin-ophen le\'('l l-lut lails to plovc that the toric ellècts of a(:('ta-ntitiophett are increased bv alcohol collsumpti(»r.

I uoulcl suggcst the lollouing appr<>ach. 'I'r't:at cven acr'-tanrinophetr ovcrdose (>4 g 1>cl dcciliter) rvith acetllcvstinc.Use the acct:rminophcn trcrrtn)rlt llonlogram to inlirrrl thep:rtient ol' fàrnih that in largc populations an acetaurinr4rhcrrlertl similar to thztt in the patient usuallr- results in he1;atic:

tosic ellccts or miuimal toxic efIècts, dcpending on the l(\'cl.

Pittsburgh, PA 15213

D,sl» Cl. \'\'rrlLc:orrr, Il.D.. PH.D.Universitl ol l'ittsbulgl'r

School ol lleclit:ine

l. Prescon LF. Roscoe P. Wright N. Brou'n SS. Plasma-paracetamol half-litèand hepatic necrosis in patients with paracetamol orerdosage. Lancet 1971:l:5 I 9-22.

2. Perrot N. Nalpas B. Yang CS. Bcaune PII. Motlulation of cytochronre P-150

isozl'rnes in hunran liver. by ethanol and drug intake. Eur J Clin Inrest 1989:

I 9:519-55.3. Lauterburg BH. Velez ME. Glutathione deficiency in alcoholics: risk tac«rr

firr paracetamol hepatotoxicity. Gut I 988129: I I 53-7.-1. Whitcomb DC. Block GD. Enhancecl susceptibilitl'to acetanlinophcn h!'pa-

totoxicit)': the ellèct\ of tasting and ethanol use. JAMA (in press).

The authors replr':

'lb tfu Editor: \\ihitcornb, r:iting Plcscott et al.,r states thatittgcstion ol'25 g ol acetarninophen clearh làlls in the rarrqeassociatccl rrith l'rcpatotoxic ef]ècts. Horlever, i.rn :rbs«rlutt: rt:-lation bc§r.ecn all ingestcd clrrse ol' acctaminopht:n ancl thc

I)lasma level or degree o1'hepatoto.xicitl has uot bcttt clt:at'lrt'stablished in hunrans. Accoldins to the acetanrinophen tlcut-

(]ORRESP()\DE,\C]L

t3r:

nlellt l)onrogriirll.r olrr l)aticnt's plasna acctaninophen le\el(in nricroelanrs pcr nrillilitcr') tras not iu thc loxic range; the

l)lasma levcl, ttot the total itrgcstt--d anrount. is comrnorrlv uscrlzrs a guiclt' to ltrt'clict toxicitr'. \\thitconrb also st:rtcs that thcinclr.rction of'cr-tochromc P-{50 enzr-nrr: activitv is onlv niildh-ittctcascd bv alcohol and rcturns to norrnal rvithin {ivc clavs,suggcsting that reccnt alcohol use is moLc irnport:rnt thanlong-term use. Houeveq the induction of crtochrome P-450cnzvnlc activitv br,alcohol can ranse {iom t$r- or tlireelcrlclto six- r.rl sevcr-rlbld ltccausr: ol lzugc variations in gt-netic anrlcnr,irournt:ntal firctorsl it t'etulns to nrtrnlil lithin firr: clavsonlr afiel abstilcnce.j C.lontinuctl rllrilv usr: rrr:rv not allorça l'ctul'n to normal c|tochrome P-{50 enzvme activit\,. Al-though the reduction o{'glutathione levcls in alcoholics rnavbe onlv modest, a modest dccline mav be sulficienl to increasethc toxic t:lïct:ts, if'toxic at:cumulation of at'etarninophen rne-tabolitcs frrrthcr cleplctcs glutathionc stores.

Srrrilkstein et al. suggest that the conrbination cll hcpatotor-ic efiècts uith a rrontoric zrcctarninopheu plasnra lcr.el that ur:clescliberi irl a long-tcrnr :rlcohol trser is "re rnark:rl:le" in thata silnilal case has not bccu rt:1:ortcd in stuclics invoh'ing overIt)0,(X)0 cases. Ho\r'cver, nrost studies har,e nnt becrn specifi-callv designccl to erplorr thc outcome in such paticnts. Firr cx-:u-r'rple. arnolg thc studies citcc[ ]tv Smilkstein et a[., onh' 5 ol'thc 662 cases in the stuclv lepr»trrd bv Rumack ct al. invohedtrontoxic acetalnirxll)hen levcls altd a histor'\' of' long-tcrnr al-cohol use l'ithout short-tcrm alcohol ingcstiou (rrhich alorrc:has becn suguestecl to bc protectir.e against thc hcpatuto\i('ellccts ol acetaminophen).: Iurthclmore. in the \ational\Iulticentt'r Studv lcpolted br Smilkstcin ct al..r patir:nts rr it[.r

a historv of alcohol trse l\'cre unilirrrrilr,clistribr"rtecl arriong thcsturlv groups itr otdcr to dccrcase conliruncling variables. r\l-cohol rrse \\'as not cvaluatcrl :rs al inrlcpcnclcnt risk Tact<lr. Anaccuratc historv inrlicatine thc :rmount and time ol- irrgcstionis critical I'or plcdictinu lrt--patotoric elïècts. Lr our prticnl,the rmsc't of'srlrptoms and the elcvation in thc livcr-enzvrnelevcls colrcl:rtccl rvith thc exl:ected clinical stirgcs of' acr:ta-rninophen toxicitr', suB-gcsting that thc time of insestiort rçasacc ur'zttc.'

Darie points out tirat hahing the acetamiuoplten level atrçhich treatnrcnt is consiclered is routine practicc ir paticutsu'ith an alteration in cvtochromirl enzvrnc incluction. Wc art:not a\\'are of'such a routine practicc in the Unitccl States, butthis approach shoulcl be consiclcred. \Ve agrce rlith \\'hitcombthal a 1;otentiallv lifèsaviug aucl bcnisn therapv should not l:errithhckl solclv on the basis of thc :rcet:rnrinol;hen levr:1. Thecase ol our paticnt demorlstl'ates that rcliancc on thc accta-ntinrDlten trcatùtent rrontogram as a guidc to thctapv canhavc firtal collse(luenc(:s. \\t do not aeree uith Snrilkstein etal. that a leconsidct'ation of cun'ent trcatment rcconrnrenda-tiorts uoulcl bc "a big stcp backl'arcl."

\Iaclison. \\iI 53792

Lrs.r Clrrr:urr;. \[.]).Krrru C. Xlr:r'rx. \'1.D.Univcrsitv of' \\IisconsinClinical Science (ientcr

L Prescott LF, Roscoe P. Wright N. Brown SS. Plasma-paracetamol halt:litèand hepatic necrosis in patient\ §ith paracettunol oterdosage. Lancet l97l:1:5 r 9-22.

2. Rurlack BH. Petcrson RC. Koch GG. Amara IA. Acetaminophen overdose:662 cases with evaluation of oral acetylcvsteine treatment. Arch Inrern Medl98l:l.ll:38(I5.

3. Perrot N, Nalpas B, Yang CS, Beaune PH. Modulation of cytochromc P-150

isozy'nes in hunran lirer. b1' ethanol antl drug inttrke. Eur J Clin Invest 1989:I 9:5.19-55.

-1. Srniikstein MJ, Knapp GL. Kulig KW. Rumack BH. Eflicacy of oral N-ace-tylcysteine in thc treatment ol acetaminophen overdose: analysis of the Na-tional Multicenter Study (1976 to I985). N Engl J Med 1988:319:1-557-62.

5. Lintlcn CH. Rurnack BH. Acctaminophen overdose. Enrcrg Med Clin NùrthAm 198-1:2:103- l9

]'Hti \L\\- E \GLA\D.JOUR\.\L OF .\lLl)lC: l-\li \or'. 10. l!X) l

MINERAL BALANCE IN POSTMENOPAUSALWOMEN TREATED \{ITH POTASSIUM

BICARBONATE

7b tht, Editor: St'bastian aucl collcagues (Jture 2,i issur)r le-p(»'t that the aclr-r'rinistralion ol' 1tc-rtassiuur lticalltonatc le-rluced uet acicl ancl calciurr crcretir»r arrcl hacl bencficizrlellccts on measures ol bont- tutnover in pr,rstnterropausal\\'onlen. l'lrcl sugecst that the "krng-tentr aclnrirristlatiotr of

1:otassium bir:arbonatc r.nar thelclble be t:llèctile in ltrelcnt-ing anrl tr('iltir)g l)ostnteltol)arlls:rl ostcc4tolosis." It is r;trcstion-ablc uhcthcr thc lesults ol this stuclr ar(. I'clrvanl to thc tastmajoritv ol uomcn.

The proti:in cont(:nt ol' the u,omen's clailv cliet (96 g per'60 kg of bodl'u eight) u as cousidcrablv higher than that cou-sunred l)\'nrost \\'onrcn. Accorcling- to (lilta collecterl b\ thc \u-tiolal Clcltt'r' fi»'IIealth St;rtistics iu the fir'st -\ational Htalthancl \utliti<»r E-runriuatiorr Suln:v (l!)76 thtr»tglt l9i3t)).r thcnrcan ploteirr intiikc o1' \\1)lner) .lit to (i I r-eals o1- agt' in th,:[,niterl States uas 5i s pcl clav an(l t]lc !)0th ancl 95th Pclt--crt-tile valtrcs nelt'90 ancl I()7 g per'<lar'. r't's1rt'ctirt'h'.'flrus. litltlcss than l0 pcrtcnt of tlrc Postnrcnr4rilrsal rçolt](]tt itt thcUnitetl Statcs consunring thc lalgc uluount of plote iu qir t'rr irrthe stuch'l>i-St'bastian ct al.. thc gcnelalizabilitv ol theil fincl-ings is opcn to qucstion.

Rais.ing- thc (lictar\ pr.tltt'iu intake is known to irrcrt'lsr uIi-tt:rt v cnlcirrm rxt Lction lrnrl rçolscn (alciunr l;:rLurt e.' Fbl cr-anrple, in atr analvsis ol'l6 publishccl stutlies invohinq- l5laclults. Kcrstr'ttel attcl Allctt+ culculatccl tlrat irr pels,,rrs rlrtirrgless than 200 g ol'protcin dailr, thc lclation benle,'rr l)rotrinintake arlrl urinan t:alt--iutn cxctctiou uas littr':u', srrch thatIbr cach 50-s increment in tlailr prot('in intake an ('\.tril 60 nls-

ol'raltiurn las ercrctecl in tht'utine. Thus. iI tltc usualclietan protein inllke ol the ur»nen stucliccl ltv Scbastian ctal. u'as sinrilal to the U.S.,rvelage lirr this age glorrp; tht'rnean (+Sl)) r'ecluction in ulinarv calciulrl crcretion rel)ortr-([(-6++ l9 Ins per clar'pcr (r0 kg) dtrrirre supplt'rrrentation rrithpolassium bicarlxrnatt upplorinr:rtes tlrc irrclcasc in <:al<itrurexcLction intluccrl lrv tlre consunll)lior] ol- llre crcrss l)r'()tcirrclulitre the str.rclr'.

1'hercfirrr:, potassiurrr ltic:irbonatc strpplenrentati()lr nrrl\have k:ss clfit acl in thc nrajoritr o1- postnrcnopausal \\'onren.nho ]rave lorrel plotein irrtakcs an<[ errclogenous zrcid kraclsthzrn thc \\onrell in the strrclr'. Until a stu(l\-ol'uonrt.n sirrnsrraller ar)rounts of ploteirr is conclucterl. the srr*=aestit)n lhtLtalkiLli suppler-rlentatioll lte trserl 1o prt\1'nt ()r tlcut olitcol)or'()-si\ ill l)r,\llltIll(,1).llrs.ll \\(,nr( n is l)rcilrirtllr'('.

Boston. -\IA 021 I I

Rtr:u.r«u J. \\'rxru, Pu.D.Tirfis l,'rrivclsitr

L Sebastian A. Harris ST. Ottawal, JH. Todcl KM, Nlorris RC Jr. I:rprored min-eral balance rnd skeletal nretabolisnr in poslnrcnopausal uomen treatc'd qithpotassiunt bicarbonate. N Engl J Med 199,1:-l-i0:1776-81.

2. National Center tbr Hcalth Statistics. Canoll -VD. Abraham S. l)resser CM.Dietarf intake source dara: United States. 1976-80. Vital und health statistics.Serics I l. No.23l. Wrshington. D.C.: Governnrent Printing Omce. I98-l:l+.(DHHS publication no. (PHSl E3-l6El.)

-3. Allen LH. \[bod RJ. Calciurn and phosphorus. In: Shi]s \'lË. Olson JA. ShikeM. eds. Modern nutrilion in hcalth and disease. Sth ed. Vol. l. Philadelphiu:Lea & Fcbiger. 199..1: I-17.

-1. Kerstctter JE. Allen LH. Dietarl, prolein increascs urinarv calciuu. J NutrI 989: | 20: I 3.1-6.

Thc :ruthors rtplr':

7b the Editor: l)r'. \\irocl is skeptital that yrotassiutn bicur-bonate rçill halt' a long-tcllr l;errelicial ellèct orr bone rnussbeclrtrsc tltu rlir:talv intirke ol g;rotcitt in lhe lrrtnen rlc sttrcliecl\\'as at thc high crrrl r-rl'tht nonrral lansc lirr licc-living sr,rl;-

\i,1. ll.l I ( t( )Rl{ t.tsPo\t)ll\(:E

jects.'fl-rc clailr intake of l)lot('ir] \ras !)6 g l)el'60 kg ot'boch-rvt ighr (62 l)er'(.cllt allimul an(l 138 pt:rct'trt rr'gctablt ltrotcitr).irith ll llercenl o1 llir totrrl Plotcin intlrkt cor.nine liorn ltlo-tein-errrichecl blcur[ lirrnrrrlatt'rl rrccotrlinq to thc r'('cil](' of'Lrrtz.i Clle alh. lol er irrtzrkts of ploteitr rrotrlcl genclatt' lou er'

anlour)1s ol aticl. 'l'hc cak irrnr-rrlrsting- lrnrl lront-rllrstinq cf-1ècts o1' 1r'oteitr-clt'rirtcl ut icl. lrorr t'r'rt. clo not abltrl;tlv corn-mencr'alt the Iriqlr cntl ol thc Irorrrt;rl Innq'r'ol proteitt int:rke.R:rtht'r'. tlrc rttagttiturL' ol ll pct'cult irrt ia itrchrt e<l bl rlietalr-plotcin irrtake. likt thc inriclertce o1'hip liactrrlc rrncl tlrc rle-t Lr';rst' irr lront' rnincral rlensitr. is :r < ontinrrorrs lunctit.rrr o1'the.ulount ol Plotcirt ingcstccl olct':r br',,iLtl lalrgr ol intake l»llcc-livirrq subjccls.r-r-\s Il':trttr' pLrts it. "'l'lrt' [calciulic] el'-li'r't crtcrr<l-r.r(l'oss th( lirll larrge ot'plotein inlakes. li'orn rlc-llcient to sulli'i1.1ulcl is rrot firtutrl olll\ \\'itlI'erct'ss'l)lotcinintakt."r l,r'err lith intakt's ol pt'ott'iu in thc rni<lcllc ol thcnolnlrl rlrrig(. \\'()nler) rlr ut lisk lirt :r ncgatirt tlrlcirnn l;al-arrcel thc rleqlec tlr'pcrrcls on srrclr furlors as talcirrm iutakr'.('stlogen stiLlrrs. :rrrtl llrt sorrrte ol clictarr ptott'itt (atritnal orrr.geIrrblt ).

Pol:rssirrrn bicrrllroturtr is ir ll()t.nl lrr.poclLlcirrric .rg.nt.,'rcrr uhen plottiu intake is not high. Spccilic:Llh. \\e strr(li('(lsir noltn:ll utt'tr t'atinq:r rrlrolt'-tirocl rliet cont:rinirtu 113 g olplott'in pt't' clitr plr' (i0 ku (unpublishr'rl cluta). 'l'lrt v tcceivccl90 to 100 nrlrx)l ()l potassiunr bicar'lron:lte l).r'([e\'li»'r'ightdars itltcl a bast'-lint'perio(l ol-t'iqht dais (ltie. 1). \\'ithinl8 horrr-s afit'r' tlrl initralion ol potassirrttr l;ic:u'lrorr:rt,. Lrli-

BeforeKHCO3

narv calciunr ercretion clccrr:asecl lrr about one thir(1. 'Ihe

nrean l;ast'-litrc ert'rctiou of net urci<l (37 rltrcl ptl clur) anclc:Llcium (122 mg [:].1 nrnlrl] pcl clar) rtas about lrirlf th:rt o1'

tlre l)ostnrenoparrserl rrrrntcn in otrr stuclr. .\tcolclinglr. .rrb-stantiallv lorr'cl rlietan intakcs ol' l)l'otcin than tlxrsr ol tht'\\'om('n irr oru strrclv rroulcl not prt,r'lucle lr ltcrrefit ial tl'fer:t o1'

potassiutn bitarl;onutt on borrc. il' thc hr ltocirlciLrlic t.[]èr t ot-potassiurl bit'arltorr:r{c 1;crsists cltrlirrg kruq-telnr lLrlnrirris-t rati()r1.

Àr'r'rrorY St:s.rs'r r.rr. \l.I).R. flt n t r: .\Ionrrs. .]R.. \1.D.

L triri r.itr ,l (.;rlilolrri:r.S;rrr ft.urtirt,r

l. Lutz J. Calciun balance and acid-base \tatus ol romen as rlÈelcd by incrcased protein intake and hr sotlium bicarbtmate ingestit». An J Clin Nut:I 98.1:i9:28 I -8.

2. Helnel' RP Protein intakc anrl thc calciunt ec(ntonl\. J Am Diet Arsoc I99l:9l: I 259-60.

-1. .{belou BJ. lloltbrd TR. [nsogna KI.. Cross-cultural associatiorr hr'tuc'cn di-etarv animal protein and hip trrcture: t hl pnthcsis. Calcil Tissue Int 1992:50:I -l-t.

-1. Hu J-F. Zhao X H. Paryir B. Cantpbell TC. Dir'tary intakes rrrd urinarr er-crctiorl of cûleiur)l and acids: I cross-scctionul stud\ ot \\omcn in (lhinr. ÀrnJ Clin Nutr l99l:58:398-.106.

5. llu I F. Zhao X-H. Jia J-8. Parpia [3. Carnpbc-ll TC. Dietarl calciunr antl honetlcnsitr ar»ong nrirlclle accd ancl elderll uomcn in Chinr. Art J Clin Nutrl99l:-58:l I 9-27.

PULMONARY-FUNCTION TESTING

7b tlte l-ditor: ltr (iliqro's lclierv ol ltulrnonlrlr-lirncti(,n t('sr-ine (.Julr'7 issue).1I l'as sutprist'cl l>r'tlrc stiitcnltnt. "Sonr('strbetorrps ('.o.. cirtirlr:tte sntokt-r's ancl pcoplc erltoserl toknotçrr agettts ol lutrg injulr, such as asbrslos or cliisocra-llatos) arc at higlrcl risk of lunc clisrascl scrcerling an<1 nrc»r-itoring alc rrlprolrriatc lirr'lhem." Thc reli:rcrrccs citt cl rlo notzlppr'al t() sul)port scrcerring ancl <lo not ollèr'or rcièr'lo <latathal rçoLrlcl .jLrstil-v spiromctric st lecttins ol'all snrokcls.2 ;

Tlie stanctalcl reii'renccs lirr thc eliLlrration anrl rer'orrrnlr:l-cl:rtiott of'sctt'ertitrg 1;roccclut'cs Iirl the gerrer:rl ltoprrlation or'

subpoptrlatiotts are the positi,,n l)ill)( r\ ()n \(rcrninq issue(ll» thc Clanaclian Tltsk l:orcc on the Perioclic Ilcalth E-xitrni-lualion. the L.S. Prt'r'entivt Sclviccs f usk ltrrcc. ancl thtAtlt't'ican Clollee(- of Phvsiciirns. \one rcconlmencl tlrc rrsc ol'sl)iromctr\ lirl screcnine ant po1;trlation or strbpopulation. Lracltlition. .r rt-c('llt leyiol ol-screctrirtg ptocr'<lulcs bv Sox cli<l

not rrrcnlion sprironrctrr rrs a possible scrt'cnirrg proceclule for'sruokcrs.l

Thert'are at lcast tl\'o qurstions tlrat lrart'to be lnsut'r't-rlbelirrc spirorn{rtr'\' can bt' cottsiclcrecl a rotrtirre sclc( linlt' t( st

Ior sr.nokers. l-irst. uill the rlcnronstration ol an ircccltlatetlcleclirre in lrrng lirncliou in the l0 to l5 pclccut ol'srnokrrsrlestinecl lo Iravt irir'florç obstrrrt:tion t'esult in a hisl]o'ratc ol'stnokirrg t:cssation thirn thzrt associrrtcd irith thc strorle rec-omnlcn(lation to quit sn.urking that cvcl'\' l)hYSician shoultlnrake to evt'rv srnokcr.'Scconcl. rçill the "nornral" ratc ol'(l('-clint'in lung lunctiorr in tht'r't-rnaining [35 to !)0 pt't'cerrt ol'stnokcrs provicle a llrcasr.lle of rcasstrrarrcc irncl rlett act li'onrthe plnsician's abiliti to irrlltrcr.rcc srnokcrs to cluit? Sinccthesc clut-stir-rns havc lrt to lte alsr«'r'ecl l;r'ttpploll'iiit(' stu(l-ies, it is l)l'cnrature to l'r'('ournlen(l sc reening ol srnokers l'itli.ipit ornt't Lr'.

Austin. 'l'X 7lt70lWtrst'or \\i.rtt'tr. ll.D.. \l.P.H.

SoLrtlr .\ustirr I l,'rrltlr I )ist r it t

L Crapo RO. Pulnonarl -funclion testing. N Engl J NIed 199+:l-l I ::-5--iU.

l. Amcrican Thomcic .Socien. TasL Group on .Screcning lbr Adull Rr'spiratory'Disease. Screening ibr adult respiratory discase. Am Ro'Rcspir Dis 198-l:I l8:76|t-7.1.

San I:t:urcisc o. (1.\ 9 l l+ll

DuringKHC03

AfterKHCO3

Coô^a

xtUc

=o),'Y @vi>(§Y\LO)ô^

t5

;P -40(u

_cO

ililï' l1

-60

1 3 5 7 I11131517192123Day

Figure 1. Effect of Oral Administration of Potassium Bicarbonate(KHCO3) on Urinary Calcium Excretion in Normal Men Eating aWhole-Food Diet Containing an Amount of Protein in the Middle

of the Normal Range (48 g of Protein/Day/60 kg).

The bars indicate the daily mean change in excretion relativeto that before the administration of KHCO., and the thin ver-tical lines above the bars indicate the standard errors. To con-

vert values for calcium to millimoles per day per 60 kg,divide by 40.

r3 t+ TItE \t-\\' E\GL-{\D.JOUR\.\L OF }IIrl)I(;I\E \or. lt). Ml-l

Salt Lzrkt- (lit\. Ll-f 8+l+3RouriR.'r' O. Cn.rpo, \1.D.

LI)S Hospital

3. HirnkinsonJL.Pulmonarvfunctiontestinginthescreeningofuorkers:guide-lines for instrumentation. performance. and inlerpretation. J Occup MedI 986:28: I 08 I -92.

,1. Sox HC Jr. Preventive health services in adults. N Ensl J Med 199,1:330:1589-95.

Dr. Crapo replies:

To lhe Editor: I apprcciate I)r. \\alr's comrnents and basi-calh'agrc('rçith thcnr. It uas rrot rnr intcr.rt to imlth'that allsrnokers sh«ruld bc scleenecl l'ith spilou-rctn'. I{, honever, I)r.\\iarr's re{t't'ctx:es to tlre position papcrs on scrccninq bv thc(lanaclian Task lbrcc and thc American Collegc ol Phvsiciansar(' triken to mean that there arc no indications Ibr usincspir'ometrv to screen ar)\'gl'oul) of-smokcrs, I disacree. \h'po-sition on screenins coir-rciclcs rrith that rif'the Amcrican Tlro-r':rcit: Socirn.r rlhich states, irr part. "Apparcntlv healthr, highlisk patients slxrulcl bc considcred for spilomctt'ic tpstinu ir\l)alt ol llicir lt'gul:rr eramination." As examples, thcv inclrr<lcht'm'r' smokers arrd people rvith occupuriiotral cxposurc to tlreinhalation ol- hazalcürus subst:urces.

as corrlparccl $ith voungcl paticnts. Accordilg to l9!]l clata

lrorn tl'rc -\alional Clentcl Iôr' Hc:rlth Statistics, thc asthrnu-lclated dt:zrth rate pcr l(X).000 p;rtients is [30 pcrcerrt liighelamorrg peoplc s'ho aLc 0o to 7l lears olrl than irnrotrg thoscrçho alc 5ll to 6{ r-r'ars olcl an<l three tinrt:s the ratc alllone-thosc J5 to 5-l vcars olcl.

With t.he adclition ol- salmctcrol to a thcrirpeutic lugintcrr.patients shorrl<l be instructe(l to usc thtil shc»'t-acling. inltaleclB,-adrorrrqic-l'eccpt()r :rgonist firr tht' rclicl'ol aculc bron-chospzrsm. Salnretcrol is inclicatt'<l lil' Iong-te rm rnainttnllce1r'eatnrcnt aud prcv'ntion ol l>ronchospasrn in 1)atielrts \\'itl')zrsthnla rtho arc l2 r'eals ol'ase or oklel ancl lccpile legular'trcat.ment rçith inhalecl. short-actineB,-agr».rists. It is ntrt in-clicated lor the I eliel-ol thc acrrte svrnl)tor)rs ol'astlinra. as cs-plicitlv statcrl in both the plo<lutt inli»'rnatic»r ancl thc in-slfr.lctions Iirr use ltv I)atier)ts. \\'hcn usccl :rs rt'conrtrtentlr'<1.saln.reterol ollèrs patit'nts sevt-r'al unique bencfits, ancl onso-ing u'orlclrvi<le monitolins ol'its s;r[ètr cLrcs rrot in<licatc an irr-ct'eltsctl t isk .rttr,,ttrg .lrl,'t lr plrtit trlr.

J.urrs ts.D. P,rr,ru.r. I[.]).K.rrrrlrr:r A. Rrr:r-rnu. \I.I).

J. Rrr:tr.rno'l'r rorrps«rr. Pn.rnrr.l).Rcscarch Tlianglc Palk, \Cl 277()9 (ilaxo

l. Jenkins MM. Price K. Pounsfbrd JC. ct al. Satet)'anrl efficacy of salmeterttlin elderll'patients uith asthnra. Am Rer Respil Dis 1992:l-15:Suppl:A65. ab-stract.

Daue CN. Cheesman MG. Pounclstirrtl JC. Salntctcrol is an ellective bron-chodilator in elderly pllients. Eur Respir J 1992:5:Suppl l5:20-lS-205S. rb-stract.

Stark ID. Luce P Inhaled salmeterol in c'lderlr patients s'irh rerersible air-u'ays obstruction. Eur Respir J I 99 I :-l:Suppl i -l:-j-31S. abstract.

Castle W. Fuller R, Hall .[. Palmer J. Screrent natir»wide surreillance study:comparison of salmeterol uith salbutamol in asthnratic patienls \\,hù requlreregular bronchodilator treatrrent. B\,lJ I 991:106: I 0-l-l 7.

National Arthma Education Progran. Expert Panrl reporr - guidclines ti»the diagnosis and management ol asthnra. Bethescla. Mtl.: National Institutesof Health. August 1991. lNIH publication no. 9l-30-12.)

CANCER THERAPY MEETS p53

To the Edilor: Kinzlel ancl \'irgelsteirr (.fuh 7 issuc)r elo-quenth' cliscuss thc clinical inrplicatir»rs ol' basit r(,searclt olrthc p53 tumor'-supplcssor scne. In kev st;rlements. thev srrB-gcst that p53 tnutatiotrs mav provide a gclr('tic bzrsis lbl cL'ugrcsisl:rnce anc[ thirt the lelation betleerr pil3 nrut:ttions anrlthelapeutic responsc shoulcl bc rtrifierl. hr lrt t. thc fi<.lcl

rnoves lapiclh-aheacl, and a pl3-rlcpencle nt cross-r'('sistare e t(,ionizing racliation ancl chcnlotll('r'il[)('utic agcnts can norç llecousidcrerl a rçell-cstablished ltht'rromcnon. It uccurs not orrlvas lt result o[' p53-rlependcnt apoptosis n]o(lulaling t r trrtorir'-itr'.1 but also becarrsc oI p53-clt'perrdent g('ne an)])lificalionirp53-clepcnclenl e-rplc'ssiorr ol- thc nrrrltirlnre-r'esistance gent'

l.lIDR I ).1 ltt<l pj3<lt prrrrk rrl t rll prrrliIi'r'rrtiorr.-''-f'lrerc is also stronq evirlt'uce that p5l3-clcpellclent cross-

rr-sistallce is indeed clinicalh lt'lt-r'ant. irr lh:rt :rl;rron'nal p513

ncgativeh'afli'cts thc I)r()gnosis ol u-raur tllnrol's. Since the ap-p('nrancc in tl'rc .Journtlof thc levierç article on pirl3 in l9!t3.r'"aclclitior-ral dirtu lr:ue bccn publishecl orr thc pleclictir-e rolc ol'p5l3 mrrtatior.rs an(l lh(' abnolrrral rspr-ession ol pjll protriu irr

a varictv of tur.r.rors. 'l'hus, p5li-clepenrlcnt closs-r't'sistuncr' is

a t:onrplex l)hcnomrrt()n u'ith it clcar clinical iml;ot'tance.

H. Pr.rr:n Rr tz. \I.D.CrH-+03 I Blsel, Su'itzerlanc[ Bast'l Lniversiti Hospital

I . Kinzler KW. Vreelstein B. Cancer therap)' nreet\ p53. r'. Engl .l NIed 199-1:

3.11 :19-50.l. Lo$e SW. Ruley.HE. Jacks T. Housnran DE. ps.3-Dependent apoplosi\ nrod-

ulates the c)totoxicit), ol lnticancer agents. CelI |993.71.t)51-67.

l. Anrerican Thoracic Society. Task Group on Screening lbr Adult Respira«rryDisease. Screcning fin adult respirak)ry disease. Am Ro, Respir Dis 1983:I 28:761,l-7-1.

RISKS OF SALMETEROL?

[o the ]rditor: I have a large gerizrtric-intemal mcdicinepracticc. I havt- cnthusiasticalll'adopted tht: use ol thr: salutt'-It'rol inhurlcr (trrice ir clar) fur-ml patients as a \\'av to irnptoveconrpli:rnct-. In the past thlec rnonths, trço eklerh pxtientsu ith rnocleratcll' sevele aslhma have hacl f iital respiratorr ar'-

r('sts at honre. Botli ucre lourrcl holding their inlralcrs. Bothh;trl bccn tolcl that ther-shoulcl expect a delav oI a half-lxrtrr'to an hour in the onsct ol- thc action of salmetclol and thattht'1'could use thoir l)revious inhalcr (albutcrol) lirr emcrgetr-cies, but thcv did not do so.

I bclieve that alth<>ugh tht' occurrenct: ol thesc t\\1) c.lsesnrav be er t'oincidencc, the use of salmetcrol mav pose an in-cteasccl risk amolrg the eiclerlll I suggest cauti(»t in thc usc oftliis tlrug until the l:ood ancl l)r'ug Administrzrtion irrvcsti-gates this matter.

Ph'mouth. \tA 02360F'r-,rrr N. lir\K[r.s'rEr\*. \I.l).

I l0 Long Pond Rd.

Glaxo, the nranulactureL of salmeterol, rcplies:

To the Editor: It is dillcult to comnrent on f)r. Finkclstein'scases as pr'csented, l:ecause hc provides fi'rr details. Since sal-lleterol ûrst bccat'ne availablc, in 19!10, hol'ever. ovcr I I ntil-liorr palit-nt-months ol' e\pcricnce havr bcen accumrrlatcclrrorklrtidc. Salmt'tt:rol l'as shonl to be rçell tok'r'ated ancl t'f'-Iective in 177 r:klt'rlv patients l'ith asthrna (mean age, )70x'at's) unrler conlr.)llcd conclitions in zr total ol ninc Glaxo-sponsored clinical trials.r-:l In addition, â lalge, doublc-blinct.randomizo(l surveillance studv conducte<l in thc United King-rlotn inrolverl over' 25,0[X) patients. alrproxirnatelr' -1000 oftr'ltotn u'ere over 65 r'cars ol age.r No cviclence ol'a c[r'ug-related incrcasc in the risk ol' molbirlitr' «rr mortalitv iu lrrvage group, includiug the elderlr'. r,r.as uotecl.

'l-he irrrl>ort:urcc ol'eclucation Iôr p:rtir.nts rçilli asthm:t can-tuot bc ovcrcmphasized.' llducation is particulallr irlportantlirr t:lderll'patients,:r group uith a creatcr potclttial for pr'«rl>-

lenrs in cornpliance and a highct-risk ol'clcatli li.onr aslhma.

\-ol. :l3l -\o. l9

3. Yin Y Tainskl N{4. Bischoll FZ. Strong LC. Wahl GM. Wild-type p53 re-stores cell c;cle control and inhibits genc amplification in cells rvith mutantp53 alleles. Cell 1992:70:937--18.

.1. Chin K-V. Ueda K. Pastan I. Cottesman MM. Modulation of activit)' of thepromoter of the humun MDR] gene h1' Rirs and p53. Science 1992:2-55:-159-62.

5. Hanis CC. Hollstein M. Clinical implications of the p53 tumor-suppressorgene. N Engl J Med 1993:329;1318-27.

6. Clinical implications of thc p53 tlrmor-supprcssor gene. N Engl J Mcd 199'1:330:86.1--5.

7. Bourhis J. Bosq J. Wilson GD. et Jl. Corelation bet$'een p53 gen( c\pre\sionand tumor-cell proliferation in oropharyngeal cancer. Int J Cancer 199,1:57:

{58-62.

POSTMORTEM VIABILITY OF HUMANIMMUNODEFICIENCY VIRUS

- IMPLICATIONS

FOR THE TEACHING OF ANATOMY

7b the Editor: I rçant to drarç r'our attention to thc viabilitr'ol the htrman immunodellcicnc\'\'irus (HIV) in cadavers andthc implicirtiorrs o[-the presence o{'live III\r during cmbalm-ing proccclures in arlatorn)' clepartments ancl during the dis-section ol' cnrbalmed cadavers b1' students.

It is thc policr- of'anatom\'(lcpartments to kr:cp thc periodbctrrcen the death ol'a patieut and thc cmbahning of the bodvas short as possiblc to minimizc degeneration. Rcccnt find-ings inclicatc that HIV irr zi paticnt lho dit-d of'AIDS is stillinlcctious at the timc of the bodr's arri\.al at the anatom) de-partmcnt ((-18 hours post mortcm). Inlectiorrs HIV has beenrcportccl in the pleural lluicl, pericardial fluid, irnd blood ofsuch patitrnts aiter storage at 2oC for up to 16.5 da1,s postmortem.r Viable Hr!' rças also rccovered lrom bone liag-ments, brain, bone marros, splccn, and lvniph nodes fi'om apatient rvith AIDS at autops). six clavs alter deathj

Recentll; an occupatiorlal HIV in{èction in a nursc l"ho\\:as pricked bt,a rreedle that had becn used on a drug addict\r'as Ieportcd:l; at the time of thc injurr, the sourcc patient \rasnegativc lor zrnti-HN'antiboclies blrt positi\.c t'ol p24 HfV an-tigen. This (:ase ol lransmission cluring tl're "lindorr period"ol HIV infèction sLrggcsts tl-rat the bodics of p;rtients *'ho havcdied ol'AII)S are potcntial carriers of inlectious virus, as areother bodies arri\-ing in dcl:artments ol'anatomr'. Therefore,univcrsal precautions should alt'avs be lollorvccl in embalm-ing rooms o{ allalomv clcpitrtnlents.

I3l:r

Despite a thorough search ol thc litcrature. \rc coulcl notfintl clata on thc (lisilllèctant propcl'ties ol' fluicls cor'r'unonlvused to cnrbalm c(r'pscs. The ernbiilnriug- tluid use(l in anilto-nrr' dcpartmcnts c(»ltains firativcs. clisinlèctants. g*h ccrol.and salts dissohecl in latcr. Thc most fi'cquentlv usecl Iira-tives and clisinlictiLnts are ethanol) fi»'nr.rlin, anrl Jrherurl. Lrsuspension tests, 251 pcl'cent ethanol ;rncl 0.5 pcrccnt f<rrnnl-clehr-de rvere sholtt to be ef-fective against HI\r.r IIoucvcr. itis not clear lhetht'r tltcsc concentriltions are also r:IÏèctive incadavers, firr several rcasons. First. in suspension tcsts. cell-{i-ee vinrs is tcstccl, l'hereas in hurnans. HfV- is also bczrlizcclrrithin cells. Seconrl. the concentr'.ltir.ln o1'thc c()llll)onellts ofenrbalming fluicl rlt:creascs as the\-dilluse thloughout tht: hu-ntan bod): 1hild, scveral classes ol' pl'ochrcts inclutling phe-nolit: agents al'e partialh' ol complctelr. inactivatcd bi' the

I)resencc of prott:inr':: this sensitivitv t<> organic load sirggr.'ststhat the elficicncv o['the clisinlectants will be much louel inc;rcl;n'ers tharr in in vitro tests.

On tl're othcl hauct, the combinati<.rn ol firatives an(l (lisin-lectants in en'rbalmirrg Iluicl coulcl bt: nrore eflèctile in inacti-vating HIV than cach agenl alone. Ncvcrthclcss, the 1>otentialhazard poscd bv enrbalmed col'pscs to mcmbers ol- alrat()nr\'clepartments necrls itlestigation. u'ith l)articular enpliasis onthe lensth of tirnc it takes to inactitatt: IIIV in all parts ol cm-balned hurnan bodics, including the bones (since HI\,'mar bespread bl acrosolsr).

ts-1090 Brussels. BclgiurnDttx Dn Cr.rl-rrun, Pu.D.Vri jc- Univcrsiteit Brussel

l. Douceron H. Deforgcs L. Gherardi R. Sobel A. Chariot P Long-lasting post-mortem viability of human immunodeficieno' r irus: a potential risk in lbren-sic medicine practice. Forensic Sci Int 1993:60:61-6.

2. Nyberg M. Suni J. Haltia M. lsolation of human immunode ticicncl r irus(HIV) at autopsy one to six days postmortem. Am J Clin Pathol 1990:9.1:'122-

5.3. Ippolito G. Puro Y De Carli G. Itâlian Study Group on Occupationll Risk of

HIV lnfection- The risk of occupational hunran immunodeficiener rirus in-fection in hcalth care s orkers. Arch Intern Med I 993: l -53: l,l5 I -8.

J. Sattar SA- Springthorpe VS. Surÿiral and disinfectant inactilation of the hu-man irnmunodeticicncy rirus: a critical revies. Rer Infèct Dis l99t:l-1:-130-7.

5. Scully C. Samaranal'akc L. Martin M. HIV; answers to comnron quc\rrons ontransmission. disinfection and antisepsis in clinical dentistry. Br Dent J 1993:175:1'7 5-9.

CORI{ESPO.\DT].\CIE

r3l6 THL \L\\' E,\GL-{\DJOUR\AL OF \IEDI(lI\E \ov. 10. 199{

that of the anuouncement, \'ears ago at a concert inLas Vegas: "EIvis has entcred the building." Disorient-ed, a young internist in lront ol' me leaps to his feet,spilling his lukerçarm criflèe. He shoots me a panickedlook, but I cannot reassure him about this train. I finclmvself r'r,ondcring, Has Elvis been sighted on this train?Am I likelt,to bun-rp into a brokenhcarted Ann-llargreton the platform?

There must be othcr trains 1r'e could board. \\'l'ratabout the Peace'frain, the Love Train, the Soul'l-rain,the \Iidnight Train to Georgia. or the one that lell offthe bridge over the River Kr.r,ai? We'r.e got to be carefulrçhich train rve hop aboard.

Recall lbr a minute The Little Engine That Could. Itchugged along say'ing, "I think I can, I think I can."The problem Ior n're is that at this point in m1'career Ino longer think I can. In fact, I knou.I can't - not un-less I fill out endless rclerral forn'rs and obtain precer-tification {rom a clerk u.ith an eighth-grade education.NIoreor.er, thc little engine pulled a circus train. At my'age, do I reallr'\\'ant to share a compartmcnt rvith a gi-raife?

Rcmember that train bearing a\\.ay a mournfulHumphrev Bogart, the platlorm at the Gare de L1'onl,acant of an1'sign ol Ingrid Bergr.nan? It could be a sadtrip if I boarded the urong train.

"Think like a businessr.nan." The consultant is urg-ing us to jump on the PHO train. Nlissins it rvill meanexposing rnv patients (he refers to them as "marketshare") to risk. Ilarauding bands of Hl{O clerks n.illsu,eep dou'n at night from the parched hills and carrl'oll great numbers ol' mv patients. I u'ill be lelt alone,destitute.

The situation seems desperate, the landscapecharged u'ith danger. Can this PHO protect me? Wouldit be better simpll-to l.rire armed guards (equipped n'ithautomatic \\'eapons protected bv the National Riile As-sociation) to l'atch over mv patients? And u-hcre isl,Iarshall Dillon nou'that I need him? He's dou'n at thedepot, *'aving goodbl'e to \Iiss Kittr', n'ho has had thegoocl sense to board the train that I nor,r'fèel certain Ishould be on.

Run and jump, I tell m,vsclf, run and.jump! "The trainis leaving the station. Think like a businessman." OK,I r.r,ill. Hand me one of those golden parachutes 1'oubusinessmcn get. I'm readv to jump. Give me a push.With m,v luck I'll float dou,n right onto thc lVhite Houscgrounds, be taken prisoner immediatelr, b1' Socks andHillary; and paraded in ignominl.belôre the runningdogs of the press corps. "The horrorl The l.rorrorl"

Slr,eat soaks my bron'. Am I au'ake or asleep? I findm1'sell'running down the platform at the Gare de L1'on.A train is pulling au'ar'. \I1' girllriend from mcdicalschool (she u,as alu'a)'s smartcr than I u,as) is alreadvon board, leaning out a rçindou' atrd observitrg mr'progrcss rvith clinical detachmcnt. "Life is change."I must get on that train. I race to\l'ard her, gasping forbreath, arthritic knees about to gile out, nrv vision blur'-ring. Chest pains course up mv neck and dortn I.ut'leltarm as I near tl.re caboose. Surnmoning mv last nlotornellror-r, I fling myscll' f'orrçard.

OCCASIONAL NOTES

THE TRAIN IS LEAVING THE STATION

Ir is 7 a.m. I am sitting in a heroically proportionecllecture hall try,ing to lbcus mv e\:es. On a table in frontol me, slightll' blurred, sits rvhat n'as advertised as acontinental breakfast: lukervarm collèe in a styrofoarr-rcup and a Danish, n'hich is "continental" onlf in thcsense that it may be a leftover from D-day'.

I am recovering from a bad night on call. Several as-saults bv telephone designed to make me "aware " of aproblem at a nursing home reduced the architecture of'my sleep to rubble. RE\{Jess, m), mind \\'anders. Icatch m1'self thinking about a girl I dated in medicalschool.

" I he train is leaving the station." This phrase shootspast me like a cannonball across mv bou,. Althoughgrogg),, I am an,are that I have heard it before. Aboutfive minutes before. A speaker in the lront ol the roomhas been repeating it at regular inten'als, possibll' as asort of mantra.

A consultant from California is haranguing (in a

soft, Ne'rv Age rvay) the hospital's medical staff. Assem-bled in all our giorv at this earlv hour, becpers honkingnonstop, rve resemble nothing so much as a lost flockof rvild geese.

But u'e're not lost. \\'e'r,e attr:nded meetings like thisin the past. Battle-scarred veterans of the \{anagedCare Wars, many of us have alread,v been captured andcapitated, branded and corralled. lVe'r'e been H\'IO'd,IPA'd, PPO'd. We have seen our goocl names printed inslick brochures and touted among the populace. Wemake a grim-looking audience.

"Life is change." Another phrasc zings past me.lVhere does this guv get his material? The Californianis urging us to take advantage ol a ne rv concept itr man-aged care: the physiciar.r hospital organization, or PHO.His tone is urgent. There's not much time left. Wema1' be able to ensure our financial viabilitv in the Iu-ture if \'r'e are not averse to selling our practices to thisnerv PHO entit1,, lvhose dissipated representative sitsblandlv on sta€Je looking somer'vhat like MickJagger onIithium.

Am I read,v lor this? The idea that the NlanagedCare Express mav leavc me behind is beginning to\{,orry me. Can w'e find out a iittle more about thistrainl Where is it eoing? lVho is the enginecr? Aresleeping accommodations available or rvill $,e ride incattle cars? Is there a club car, a snack bar

- an1'place

n{rere I can get a hot cup of coffee?IIy mind is drifting. I imagine mvself seated on a

tiny train. A little girl in pigtails rvalks dorvn the aisleshaking a box of candr.. But instead of thc larniliar cho-rus "Good 'n'Plentl; Good 'n'Plent1" I hcar "singlepaver, single pa),er." I crane mv head out thc u'indou' toget a glimpse of the engineer. Sure enough, it's Cho«.1-

Choo Charlie. Is this the right train lor n.re?

"The train is leaving the station." Each time l're sa1's

this it sounds morc like a thrcat. The phrase clearlvhas a galvanizing eflect on the audience, much like

\:ol. 3lJl .\o. 1!) ts(X)K R}]\IIE\VS

Remembcr the scene in Itdiana.fones ond thc LastCrusade in u'hich tl-rat beautilul bloncle \zrzi temptressclives lor the Hoh' Grail but it just clLrcles her grasp?That's uhat hi,rppens to me. I lall hcarih'to thc tear-soakcd platform.

Crun-iplcd and bruist:cl, I iook up and u.ave goodbl'e{br the last time to rrtv departing girlliiencl. I see herturn slorvlr. li-om the liuclon'ancl melt into the embracc

of anothcr man. Evcn lrom this clistance, I recoqr-rizehis silhoucttc: it's Eh is.

I arvaken alonc, mv hcad resting Llncomfortabl)- onthe rock-hard Danish. Thc lecture hall is emptr,. Thespeaker eviclenth' has ridtlen o1Ï' ir.rto the sunsct. 'Ihetrain, thc tr.Ianagecl Care Express. has left.

PO. Box 30tiConcordville. PA 19-l3l BuRro-r J. Wrrlr.rrrs, II.D.

BOOK REVIEWS

Psycnra.rnrc Eprorurorocy: ASSESSMENT

CONCEPTS AND METHODS

(ThcJohns Hopkins Series in Psrchiatn' atrcl Neuroscit'trce.)È,clited br-Juan E. Ilczzich. \Iigucl R.Jorgt:. and Ihsan \[.Salkrun'r. 6I2 pp. Baltirnolc,Johns Hopkins Univcrsin' Press,1991. $6r. ISBN 0-U01u-+615-3.

Arncricans t:njov big nou:ls, biogt'aphics. and textbooks.\orç thev havc a big book on psichiatric cpirlemiologr. Thetitkr rnav reprcscnt a limitation. since rçith the prescnt tenden-cv lor psvchiatrv to cuclclle up to the nculosciences it louldbc nrore au ûnanl to spt:irk of neur-orpiclctrrioklgr'. Pclhapsthis is a selling point. 'l'his is zr bie book. but is it reallv a bookat all? It is a collection r>l publishecl picces, u ith conlnrntar-ics ar-rd shol't slrmm:rries ol'other articlcs. It is not the pro-ceeclir-rgs of a confèrcncc. r.rr a Frstschrili. or thc- invited vit'usof cxperts on a fixecl topic. The book is unusual, bccause allthc nratcrial hzrs bcc-r'r successlulh pcer-revicu'ccl ancl pr-rb-

lishccl elserçhcrc. So lhelc docs that leavt- tlte revicler? Gir-cn that I arn cited. along l'itli liiends ancl collcagues, lhatcan I.justiliablv sar'?

-fhe eclitors state that thc dcveloptnenl ol the book "start-ed içith the plt'paration of a course on assessmcnt itt psr-chiatric epidcmiologl at the Graduate School ol PublicHealth ol' the Univcrsitv of'Pittsl:rrlel-r." Fint-, but al u'hatIevcl. and Ibl teachcrs attcl studctrts alike.'Arc these articlcsprcparation for lectures aucl scminars. or aIe theY back-grotrnd reading? Thev arc lull ol detail anrl trnlikelv to holdthc l>cginner's attcntion. 'l.hosc u lto clo psvchiatric epicle rni-ologr', such as the peoplc \rho atterld thc biennial r:onlèr-ence ol' thc \\irlld Psvchiatric Associatiou, Ior esample. u'illrecognize namcs ancl substance. But thev are an in-gt'oup.not tht: rçidcr rçorld. f'hev r-nav cvr:n cluestion thc choice ofmatclial, sincc tl'rc al'ticlcs arc làrniliar brrt llot neccssirrilvc lassic.

So, apart fr'oni the stuclents in Pittsburgh, rçho rrill leaclancl benelit fi'om this book? It's hald to sa1, especiallr-since thc book conccntrates or1 assessmcnt, concepts. andnreth<lcls r.atheI than finclings. It carlllot conrparc. loI ex-anrple, t;ith Ps-ythiatric Di.çorders in .lncrica: 7'he EpidemiologicCatchnrcnt Area ,\tudl' (Let: N. Robins arrd Darrcl A. Rcgicr,cds. \erç York: Frcc Pless, 199l). u'hich recountcd theEpiclcmioloeic Catchmt'nt Area studies instigatcd bv Ros-alvnn C,'arter. For martr', that book, sholittg that melrtalillness was uiclesprcacl ancl mort' conrmon than manv oth-er lcll-knoun discases. rras a rcvelartion. It rças creitilgstul1. leading to grcater public alareness and improvecltrcatnlenl. Ol:r'iouslr'. though, l ithout thc in'rprovec[ reli-abilitv and vaIiditv me:rsurL'nrents of' pslchiatric phenom-en.r ()\'er thc p:ist lèrr clt-cadcs, the Epiclcnriologic Catch-ment Arca stu(lies l'oulrl not have comc about. \Icasuringthcsr phenomcna has bcen a success stor1, 1;r'obablv nr,rt

rçcll unde rstood eve n bv the re st ol mcclicine, iincl certain-lv not br societv in general.

Epiderniologv has bcen a popular scientilic pursuit ol'psvchiatrists because it irrvolves lhat thcv do clinically'r.ritlarge. The rcsult has bccn a bettcr description ancl classifr-cation of svmpton.rs, svncL'omes, and clisolders. This in it-self' shoultl lead to tnore eflective ancl efllcient trcatment.This book, therelbt'c, dcserves to be rrcll receir.ed as a trib-utc t<; all thc l'ork clcvotecl to rnt-thodologr'. The obr ious ca-vcat, hou.cver'. is that the unarçare rçill lind the detail tc-clious ancl the expcrt rçi1l knon it alrcadr-. Finallr; n'hoer,crattends that course in Pittsburgh lould gain lrom the in-clusion ol' rnorc on the historv o[' thc topic and lcss discus-siorr of irtclividrrlrl qucstiotrnaitcs.

Tolonto, ()N\I5T 1RB, Carracla

\I.R. E.r.s'r'n'o()D. N{.D.Clarkc Institutc o[ Pslchiatrr'

BrNcn Eerruc: NAtunn, ASSESSMENT, AND

TREATMENT

E,dited bl Clrristophr:r G. Failburn ancl G. Tercncc \Yilson.{19 pp. Neu York, Guilfbrd Pless, 1993. $+0. ISB\ 0-89862-99ri-0.

Research r)e\r-s on nutrition. cating- bchavior, and bodl'u'cight is t:urrcntlv câpluring headlincs in the nrcclical litera-tnrc ancl the popr,rlar press. 'I-his multiauthored book prt-rvidcs

a timelr; conprehensivc olervierç ol currcnt rescarch onbulimia n('rvosa! I'or lrhich binut: rating is a sine qua non, andexplorcs the rolc of' binge catirtg in anorcria n,'n'osa andobc sitr'.

In thcir'prefàce. the eclitors state, "Some books scem con-trived: ft us this book seemed lilmost :r necessilr'." Thc pro-spectivc reacler mal incleecl rçoncler wht:tlre r a book rçhose fo-cus is bingc eating can accurateh' portra)' thc complexbehaviolal :rnd mcclical svmptoms associated l'ith bulimiancrvosa ancl anoresia ncrvosa. In Iàct. this hcuristic themeu'r-rrks notal:lr, l'cll, lith a fi'n t-xceptions.'Ihe book is me-thoclicallv organizecl itrto majol sections covering clinicalcharactcristics, cpiclcrniologv and etiokrel; and cvaluationand treatment

- a stlucllrr(' that contributes to its readabil-

itv and acccssibilitv as a relcreuce 'rvork.The introcluction iittempts to clefine bingc cating

prisinglr- cor.nplex task -

and includes a làscinating histori-cal overviol ol- the mcdical litcrature on binge ealing. Forn«-rnspecialists hoping {-or an update on psvchiatric nosokrgvlbr thc eating disorders, this book proviclcs a cogent euidc tothe lcrurth t-clition of' the Diagnostic ond Slati.çtical -llanual oJ'

.\[ental Disorderr (DSNI-IV), rect'ntlv publishcd bl the Amcri-can Psvchiatric Association (Washington, D.C., 199-l). It alsoincludcs an oveLlicrr' of nerçll clelelopccl recommcndationsIbr the subclassification of paticnts uith l:ulima ncrvosa ac-colcling to rlhether thct-engegc in recurrcnt purqing;, ancl of

l).lticnts \\-itll .rnorcxia ncr\'osa accordins to \\'hethel' thcv en-g.r1', itt r'( r u|l ( nt I>irrqc eatirrq ,rr ptrrqirrg.

:\s nolecl in this book. thcre has lrectt less e\tcl]silc rr'-search on thc prclalent:e and role ol binge eating in ol;csitr.Rellccting thc neecl ['rrr aclclitional inlcrrmation in this ;rlca.thc DS\I-IV inclrrcles rcsearch critt:ria hrr' "bingc-eatins clis-olclcr" irr an appcnclir. In this contt-xt. thc psvchosocial Iac-tols. neclicul collscqLlences. and treatlnent of obesitv itself arr'l:rrgch lrovonc[ tht' scope of'Bingt Eating.

\\'ith a lirlrnat rcllectirrg scholalli: cletailed rcvicrrs of tht'Lesclrrch litt'ratlrlc, tlrt: book rcflccts points ol'r'icrv that alceclcctic ancl inclusive. A consiclelatiou ol'thc causc of thc bc-halior', lirl t'ranrple. cliscusses dcvelr4rmcntal, psicholoeiczrl,an<.1 psvchobiolrgic nie chanisnrs. Inclilidual cltapters o11 trcal-mcnt cliscuss plurrmacologit' interventions. short-telrrl psr-cho-Iosical :ipploachcs, ant[ pslcho<lrnanric l)svchotherapr: Clin-ical approaches ftl' the treâtmcnt ol' patients rr-ith rell'actolr'illncss, such as hospitalization, iLLc less lcll covelccl, rr:flect-ing a lcss estensivt- r'esczit'ch litt'ratttre itr thcsc are:rs.

Thc chaptels t'cflect the cxpcrtis(' and individtrirl points o{'

vicrr'o1'thc contlibutors. lho are active inlcstigators iu thefit'ld. Thc bibliog^raphv in cach chaptcr is conrprchetrsive ancl

up to datc. and the index is notablr-cor.upletc. Rcaders iu-vohr:d in clinical teaching nrav llote thc relative paucit\'of il-lusllative casc prcsclrtations. Thc conchrclinpç sectiou, h<xler.cr. contilins an extcnsirr:lv lt:searchecl intcrvicrr qucstionnairtIbl thc stanclarrlized asscssn.rent o['paticnts l ith cating clisor'-dets aurl a tlctailecl outline clesct ibing a cognitive-bchavioraltrcatmer)t ap1;roar:h lirr bulimia rler\'osa.

IIr sutturarr'. this is a user'-f iendh. colrpre]lcnsivc surnma-rv o1' thc current r"escirrch litcraturc on l;ulimia ntn't.rsa ancllclatcd eating disorclers. Gcneralists u ill fincl this lxrok a valu-able rcli'rence. Fol clinicians alcl resc:rlchers lho l'ork regu-larh' lith l)aticnts lho have eating clisorders, this slioulrl

l)r'()\'e a içelcomc hanrlbook.

Boston. lIA 0221iD.rrr» C. Jtur-tsor. \I.I).

Beth Israel Hospital

PRnvExsTRuAL DYSPHORIAS: MYTHS AND REALITIES

Eclitcrl br. .Judith H. Gold :intl Salli K. Sevcritlr. 262 pp.\\lLshington, D.C., Arnerican Psvchiatric Prcss, 1991. 532.rsB\ 0-rJB0+B-666-X.

'l.his rcsorrrce book l;rings us up to datc ou the Ihcts aboutprcrnenstrual dvsphoria lhilc sholiug hou thesc lat'ts areenrl;eclcled in a sociot'ultural contcst that is Ii'aught u-ith psv-chological r-ncanir.rg and r.nvthokrgv and tnuch itr trt'ccl o1'cotr-tinuecl studv ancl clalifi catiott. Pret.t-tcttstrual clvsphoric clisor'-

der. or plcme nstrual dr-sphoria, is the curLenth acccptecl tet'mIi» a controvelsial ct»rclition related to premelrstrllal svtt-drornc and late-lutcal-pl'rase cllsphoric clisordcr that has beenaclcled to the Iôur-th eclition ol' the Diagno.ttic and 'Stati.çtit«l.l I u rr ttal of .l Icttttr I Di,ordtr,.

'Ib nrcct thc tliagnostic criteria lirr prer.nenstrual d1'sphoricclisolclrr'. a \\'onrarl necds to dcmonstrate! l:r' kccping daililecorcls ol- trro nrenstrual c)clc':j, that dLu'ing thc last rçcck oltht' luteal phase in most mcrlstl'ual cr-cles, slr has at lcast fi\'ephr sical or t'rnotic»r:rl s\nl)toms that m:rrkedlv iutcrlelc l'ithhtr lork anc.l social life. Ol thc fir'r:, onc n-rust bc associatccliçit]r a moocl disorcler'. such as severe allective labilitt, in-ctcasr(l :urger or irritabi]itr. ol clcplession. Thcse srlrptomsnrust be present prcmenstruallr' (in the lutcal phirsc), l;r.rt ab-scnt in the u'cek altel thc onsct o1- mLuscs (the follicular'phasc): thcv cannot bc ahçals prescnt, r)or can thcv rcflectexaccrbations ol other conclitions. Prcrnenstmal dvsphoricclisorclct'lack' a l;iologic rnarker'. its cause is utrknorln. and itsnzrtur:rl histon- irnd cpidemiologr- have lrt to l;e clesct'ibecl. It

TIIE, ,\E\\' L-\Gr.A\l)JOL R\-\L OF ,\IEDI(:I\E \r». 10. 199.1

is a sct o['svnrptor.ns clcrivcd fi'otn sociocultural al)d persol]alpsvchokrgical cxpericnces [i-rr rrhich n-rcdical advice is soughtanrl a diagnosis is rnaclc.

\\ rmcn givcn this ncl tliagttosis nra\- l)e t'asicr to treat(h:ru the maloritr'ol'uomcn l'ho have svmptolns c()usistcntlith prenrenstrual sr,nclrornc but rrho do not meet the strictcliagnostic critt:ria. \Iost lorncn being secn lor prer-nenstrualsvuclrome have a complicatecl and inconsistcnt patteln ofs\-mptonrs. Plemenstrual svndrome rud premenstlual d1's-pholic clisolclt't catr be clistinguishccl bv esarnining the clailvrecorcls a \\onliur has kcpt of'her r.nenstmal crclcs. On thc bur-

sis o[' l'ristolr' :rlonr. manr' \\'{)nlen associatt lters, rnel, lanrilial,antl l ork prol:lenrs rçitlt the onset ol their r.nenscs. 'l'hc svnrp-totns neecl to I)c takel) scrioush ancl mav rcfleci a prolorrncllilè rlistLrrbante. 'fhev rnav also t'equit'c extcncled evaluatiot.t.\\bnrerr uho repot't havitrg pt'ctnt:nstrual svndrome but clo notn'rcct tht' strit t clitcria lor prencnstrual clvsphoric disoldcrirle particularh- likcli to have another psvchiatric disorcler.such as dvsthvmic clisolcler. grhobia, obsessive-cornpulsive clis-orclct, or alcohol- or substitttcc-abuse disr.rrder. Thcr- can bcl'ell scn'ecl bv lelèrral to :r psr-chiatlist lol' trcatmcnt of thcirprir.narv psvchiatric problcr-ns. A trcncl torrarcl pt'oqrcssirerrc.rlscning of' problems ovct scveLal urcnsttual cvclts in pto-spc( tivc lecoLcls can be a plcclrrsor of a nrajol mental clisor-cler and anothel rcason lbr a psvchiatlic refcrrarl.

'fhelc havc bct'u llr.rmcl'ons tl'catlllcnts lcrr premcnstt'ualsvnclrome , an(l the bettcr stuclies itre t-ell sLlnlniu'izrd in thcclrirlrtcr it Prtntanslruol D-ysphorias bv Rilcla-'li»ar zrnd col-leagues. 'l.he clTèctiveness ol tr-eatmt'nts li.rl thosc rrho mectthe criteri:r lor prunenstrual clisphoric rlisorclcr raiscs irlter'-r:sting qLrestions about tl'ris cliagnosis. A placebo resp(,lsc is

frecprcnt anrl substantial. 1'hough a horr.nonallr' corrclatcclclisorcler. pl-cmenstlual svnrlLorne is not consistenth'hclpcd bvtrea(ment rlith prorestcronc. Ovulation supprcssors such as

clanazol mavlork llut havc exccssivc sidc ef-lccts: the cxtrcmeof oopliorectomv has been usecl. Psvchotherapics have notbecn lcll stucliecl but ckr r.rot apl)eal' espccialh plomising as

tht: onlr- Itrnn ol' trcatment. \Ianr- sor.r.ratic treatlnents (such as

thosc invohir.rs alprazolan, bromocriptine, buspin»rc, lightthefap), melolazonc. naltre\one, ancl spilonolactonc) al'c su-pcrior to placcbo rçhen aclministcrecl cvclicalh. Ol 1>articular'interest is u'lx- alprazolar.n rrorks. aucl rçln' clcpenclencc andckrst escitlation do not seem to occur lhcn it is givcn in a clailr'dost' ol'0.25 to {.0 mg <»rlv in the slrnptomati( phase.

This book is particularlr'apptopriatc l'or avnccologists, pri-mar'\- cafc phr-sicians rr ho scc man\' \\'omtn rlith pt'cmenstru-al problcrns. ancl psvcliiatlists. It is au oxcelleut resoulcc hrllo1' carchrllr- clocumcnted and current str-rclies. \\'ith its bal-anct'cl prcsentzrtion ol- contrortrsic's, tl'ris book goes lal toconrbat the stign'r:r that has bcen associatcrl l'ith plcnrcrtstt-u-al svnclLor.r'rc.

Boston. -\IA 02115R«rennrlJ. Arrcr.. \1.D., \I.P.H.

Han'ar-d Ilcdical School

A Sruprr, Tnnonv oF THE SELF

Bv Davicl \V. \Iann. 176 pp. Ncrr \irrk. \\i\\i. Norton, 199-1.

$25. rSB\ {t-39'.3-70t72-7.

-\Ieclicinc is lal rnore than arr cmpirical :lctivit\', ancl pl'x'si-cians are not merc technocrats. 'I-he underlring philosophiesof horç lre rcgarcl natur-e ancl oursclves cletcrmine the vervchallictrr ol' meclicint:. 'flrus, tlre vast cli{Ièrcnces bctucettn-reclicincs il China ancl the \\'est mav b(i traced to thcir dil-Ièr'ing vicrr-s of the l-orld. Ouc ol'tlic quandat'icrs itr \Vcstet'nthotrght and \\'estcrn meclicinc is the Cliirtcsian ntiud-boclvdualitr-. \iarious plrikrsol>hical lcsponses to this problem havcgiven lise to clillèrcnt psvchologies «rl' tl-re self . In me<licine,

Yol. ll:ll \o. l!)

our understarrcling ol' selllxrocl -

1ç11;11 it mearrs to be an in-dividual

- has implicatiotrs fol evervthing liom pslcl-riatt'ic

pra('tice to doctor-paticnt relationships. Thcolirs ol' thc selftherclbrc allèct nrattcrs rrrll bcrr»rrl t'sotclit' discrrssiorrsanrong [)hilosoPhers.

TIrc thesis o|-'1 ,Sinple Thcort'rtl tht,5'r'f is that (lrr self'hastlrrt'c clir.nensions: tirrrt-, bodr' (spacr'. loosch'r'it'rrtd). :rn<[ rt'-lation. 'l-hc iirst trl'o cutt'gories arr' st'lllerplallitt()t\

- \\-e

cztt lt Ir;lt ir pcrsonal histc»'r. ancl aul at\\.trcness ol'ottt' embocl-irncnt. But rçhat is "rclation"? \Iatttr has iticor'prrr:rtctl a donr-inant 20lh-ct-ntrlr\ l)hilosoi)lrical torrct'nr t'ith thc so-callrclsel[-othcr lclatiou thitt incluclcs not onlv our rt-l:ttionship rrithoth('r l)elsons iul(l our envilor-rr.nent (both social ancl natural)brrt also our rclationship uith orrrsclu:s. llann r.rses tho tctnl"reflcsivitr" to itrclir.atc sell'-alutrttt'ss, a psrrh«rkrgical <li-

rnt'usiou in rçhich \r'c rccognizc thul lhcrc is a scll. Bv lôctrs-ing otr sclf-cortsciousrtess, l'hich is the ittnt--rmost ltart ol'ottriclcntin. llann builds :r psrchologr tliat rrses sell-iclerrti{icationas i.r nlcilns ol'opetring- our psvcltt'. Bv t'splicith isolating thereflr'sive compol)ent ol out cor.tst iottsttcss. IIanrr hopes to de-velop nt'u categories of psrt'hiatrit th'sliurctintt atrtl it.nprovt'olu trnrlcrstanrling^ ol nolrnirl bchavior.

'l'hcrc is thc gerni oI a porrcrfirl lrslchologv in thc rel]crivcscll. l;ut I clo not belicle \lann lurs rlevelopcd the potcntial ofits intrigtring nlrrativt. polet'. I regalcl his thesis as:r preltrclt'ol perhaps zur allusion to a rich intt'llct'tual historv thzrt uillrevt'al u highh' nuanccrl aucl cornplt'x thcorr'. Illlher than ol--

lèr a rlt'tailcd (riti(lur. I lill onh indicate thc origins ofll:rnn's ploposirl ancl tlie g^t'nclal rlirt'ction in rr.hich he anclolhcrs ol'sirrrilat' thotrght utav be hcacle<I. That tlris purticularborrk sccms to lllc iIIr ina<lt'cluatc liruttclittion lor a ncrl psr.chologv basc<1 on rcflerivitv harrlh' torrtlernns tht' gcneritlctlirlt.

-'l ,\inllr Theo» of tht,5r'll originatt:s iu a complex histon'of'the philosr4rhr-o{- personzrl identitr: Hegt'l ancl his Mth-centtr-n'lirllorvcrs chunqerl tlx'srll'fl'orn att cntitv lo ir l)r-ocess.Thcir vicu las that tlrt's<'l{'tteith,'r is givcn Itol ettrt'rscs insor.nc firral lorrn lrtrt cun be c«rnstilutcd orth in t'clalion to itsobjcct

- llùcthrl th:rt is the cxtt'r'iol rvorlcl ol sonrc irrrrcr'

sensc ol'i<lcntitr'. Kiclkesaalcl \\'r-ot(' thart the sell'is "u reiationrçhich rclates itscll'to its oun sell. ;Ln<[ in rclating itsclf'to itsorlrr sell'rt'latcs itsell to anothcr."'l'ltis tk'linitiort nas expancl-ed ll'20th-ccntrrn phcnonrt'nologists. rr ho lcrcrrsccl ort hou rc-lations clcfinecl. ol' iruthenticiitecl. thc sell. Ancl latt'r thcorists.oltcn lelèr'rcd to ils l)ostnro(lcrnists. clt'constructcd the self as

conrplctt'lv clepcndent on a particulut' crtltnral anrl histolicaldc sign.

lLutn's thesis is a «:onlirst'cl rnixttrt't. of these tso viels. Otrtht' one hand seenringll ultau.arc ol tlte rich histon' ol' hiscentral iclc:r in philosopln: hc alfirnrs that "the st'll'i' cluintt's-scntiall\ rt'flcxivr." \'et. on the othe r hall(t, he fàils to clrau'thclirnrlanrcntal lesson li'«n.r.r that chalactr:r'ization: thc scll- as ancntit\ clissoh'es. \Iann corrt.cth' pcl ceilcs th:rt therc is a plob-lem. nùit h hc charactcrizcs as tht' st'lf 's indctcrn-rinacr'. Thishc fincls to be a nirtlrral conscqucncc ol Heisenbcrg s unccr-taintl principle. in rçhich. in quar.rturn mt:chanics. the sirnrrl-tan('ous clctcrmination ol'both th(' rnonrr:ntum anc[ thc posi-tiou of'a l).rrticlr is ir.upossible. Horr'. or Iirr that r.r.lilltcr', \\'h\'.rclations lbunrl in elcrncntarl parti('lc pllsics rescnrblc hu-rnan rclationships is uever cxplainc<I. I tkr not bclicvt'tliat thc:

prol-llem ol- defining the sclf'can bt' explzrinctl l;r'tht' meta-phvsics ol'quarks.

\lost Conler-nporar\ Phikrsophe|s regarrl thc sclf Problcnt:rs onc ol'scvcr:rl lrprs ot'imbroglios itrtpost:cl ltv thc linrita-tiorrs ol'our lzrneu:rge . ()nc viel . irr kt't'pirrg u ith llarrn's ou norit:nt;rlion, is tl-rat tl-rc self is nritlrt'r'srrbjcct nor obirct but rc-flexivitr' (in philosophital tt'rnrs. a dialt'ctic). Onlv vt'rbs cap-trrtt'such action. antl uc rkr uot lr'arlilv collal;sc rtlbs Iiotn

l3l!l

thcir diliderl subjt'ct-prcrlicatc status to a uniou, r'et this isuhat is rcquirecl. Ilann's sell'is lost in a grnmnrat'\'et to l)ear-ticulated. Hr still relèrs to the psr-chiatric patient as a sul)-

.jcct. :r multitlinrt'nsion:rl cntitv that he c:rnnot aclequatelv rlc-scribe. His rlt'scription làltcrs becatrse in nl' opinion hc issinrplv and ironicallv still lockecl into a pre-Ht-gcliarr psvchol-ogr. \Iann, uho persists in regarding the st'll'as alr entit\. onh'vaguelr- pcrccives tl'rc phikrsophical conunclnrnr.

It'rve seek a ucl psv'hologr. it is unlikelv to ernerst' li'or.t.t

the version clescribed in this book. A psvcliologv basecl on a

reflr:sivc lirn<liorr trrust dcvck4r .l gl'.tnrr.nilr- that pl;rces ortr-sclvt:s itr out'lorlcl, ncither as strbjects nol olr.jects, but ils ilr-teqratcd "sul).i(:ct-lcss verl)s" o['e\periencc. This is the proicctbt'gun l»' \\'illianr.Jamt-s, an(l onc that lenrains itrcornplete.-lir so situatt' the sclf is l rnost ambitious philosophical task.perhayrs inrpossil>le. ][anv are hailinq the cnd ol 1;hilosopln,not bt'causr the t'l:rssic yrroblcn.rs hzlt' been soh'ecl. bul bc-cause thrrt-- is grruirrg atrxit-tv that phikrsophical analvsiscannot rcsohe strch isstrt's as the scll. \cvertht'lcss. ert'ttllarçecl. dclivative clïorts such as \Iarrn's lemincl us that thcploblt'rn renrains ancl sce ks l'csponses.

Boston. \lA 022l5Ar.rRe» L 'Lrunr:x. ll.D.

Boston Univ'rsitr

Coron Arlas oF CLTNTCAL Evrnvorocvtsr'Keith L. \Ioore. T.V.\. Pcrsaud. iinrl Kohei Shiota. 2J2p1;., illustrutc<1. Phil:rcielphia, \\iB. Saunders, 199-1. SB9. ISts\0-72 I 6-+663-8.

This is an easilv rcad book that makes clnbr\1)log\-cas\ torrnclerst:rncl. IIorrever, I think it is geared n)ol'c t() llre sencrirlpublic than to tht rneclical profèssion. fhc book cleals uiththe cmbnrrlogic processt's lcading to congenital abnormali-ties.'I'ht:re :lle ûrore illustrations than te\t. and the illustra-tions arc beuutifirlll prest'ntcd and st'll annotate(I, :r misturt'o[' diagrarns, lètaI trltrasonogrlms. anrl photouraplrs.

\Ianl arcas arc onh skctchilr' <liscusscc[. Frlr exarnple,thcrc is no nrcntion ol' hcpiltic and pant:reatic :rbnormalitics.rvhcrcas alrnormalitics ol- thc car and erc are describc<l ingreat dctail. The chapter on the cardiovas('ulal'slstem in lirctcovcrs onlr thc hcart, lith verv little discussion of'the vascu-lar svstenr. ltrr instancc, therc is no clescription of'aortic c[e-

veloprnent apart li'om a bricl'rliscussiot-r of thc aortic al'ch('s.'l'here is also no nrention o['tltc rvork bvJapartese rcscarcherson snrall-bouel atrcsia in [ètal lambs. an abnormalitv causecll»' tlrc làilrrrc ol' crrrraliz:ttiott.

Hacl this l;ook gonc into nrolc cletail, it loulcl have been ausrfirl :rdclition lo the librarics o['pcdiatricians, petliatlic sut'-geons, racliologists, and ol;stelrit:al ultrasono{raphcrs.

Blackler. \Iancht:stcr Cl-rnot.lrl \I. D«ltc. C:tt.\I.\l9 2AA, Unitecl Kingdonr Ultiversitv ol \Ianchcster

Drsrasns oF THE BnoNcutorrsEditt'cl lrr- Garr R. F,plcr. { } } pp., illustrirtccl. \el York, RirvenPrt'ss. 199 1. S105. IsiB-\ 0-7{Jl7-0123-6.

Iixcept rvhcn caused bv noxious gascs or lirrnes, bronchiolar'clist'ases havc k»rg bccn viese<l as diseascs ol- childhoocl thutarc caused rnainlv br-microbes. Rccentlr. bronchiolar diseuseshave bcen recognized as clinicallv inrport:rnl clist'ases allict-ing all agc gr'(]r.rps. Onh' in tht' past clccatle lrarr' the clinicaland pathological lèatures ol rnanv ol- thcse diseascs bccncharactcrizcd su[ficientlr'[rn'trs to bcgin to undcrstand tlrt:n.riu ati orclcrlr' Iaslrion. Ttr mv kuol'letlge, this tcxtbook olJersthe orrlv tholough cliscussion of'diseases ol thc bronchiolcs. In

BOOK RE\/IE\\-:i

1 llil

the plcf:rcc. f)r'. I)1;lel states thal he clcsigncd tlrc book to bca cornpreltcnsivc rcfcrt'nce lirl the clinit:al ancl pathological[èatttrcs ol lrtrnrirtr bronchiollrr discrLses as \\'e cul'fenth un(lel:startcl therrr. I brlielt' hc actornplislrec[ this task quitc uell.\,'atiorrs chal)tels hatc great sti'cngths arncl sot.rrc l'eakncsses,but I r-ieiç thc book as it l-eltome urldition to ntv libralr:

Divaçe.c o[ tfu l]rutnthiolaç is a multiaulll»'erl \\1)l'k and ist]roclolt'ulitlcn in nrrrltiplc stvles. Somt'chaptt.rs ale crccl-lcnt. Othcrs are l)oring. al)(l therc is corrsidt'rabk. ovet'lap an(lrel)ctition. IIou,evcr', nrost cliaptcrs irre lt.ll tllittcn, utrcl thcnrarrr illrrstrations ancl tabk's ale trsualh clt:ar. Dr'. Ilpler'st-hoict' of contril;utors is velv goocl.

l'ht'organiz:rliotr ol-the book is unusual. It is divirlecl intolour scctiorrs. rçith thc filst clt'rrrterl to:u)iltollt\: l)atltolog\; l'('-s('at'(.h. anrl a clinical classiïication ol- thc llont'hiolal cliseas-es. 'l'hc nr'\t t\\() sr-t:tious consiclcr lrloncliiolar rliseascs Iilst.rs "rrir\rav t[iscllrlers" irr)cl th('n as "inlerstitial cliscases." Fronra lxrthogettctic slancllloint. I rçoulcl cr»rsiclcl all ftirms to brbrotrchiollrl anc[ rrrost {irlnrs to bc tapablt' of carrsinr.l intersti-tial rliseasc. rlepcnding on tlic severitv ol the injurr: thc chro-nicitr ol'tlrc injulr. antl thc abilitr-«rl- 1hc hrng t«r lcpail itsr'H.'fhis clivision al)l)ciu's artifir'ial but is pcllraps usecl to sepa-rate bt'ortchiolitis obliter':rns Ironr lrronchiolitis oblitelans or.ganizine Pneunrouia.

'|he sr:<oncl scr:tion inclrrckrs rlescrip-tions ol- srnoker"s lrronchioIitis. nlineral-dust llonchiolitis.clillirsc prrrrlrronclriolitis, :rncl the sl)c('tnrrlr of icliopathic anclknourr ciurscs ol bronchiolitis oblitcrans, 'l'hc thircl sectioncliscrrsses lt-spir:r(orv bronchiolitis (a lblrr-r o1' srrrokcr's bron-

BOOKS RECEIVED'Ihr rutipt ll lhtt lruils it atliuottlttl:lrtl. ttrl tlti' li:ling nust fu tgartltd a' ruffi-

.itùl ù,1iltù fot lht nurlt.fi oJ llu tndn: llaoJ;t lltat oppttt ta b( of pattitul«t ittilt'tt;,ill h ntitt,rrl n' ÿrut lttrntil:. I/a'Jotrrrrrl ùt' nul ltttllith utttliLilt.d ni ir;,r.

llto-rttDtr rr. 5t.lt:rr r:

Alzheimer Disease, Down Syndrome, and Their Relationship. Edited by J.[i\. Berg,H. Karlinsky, and A.J. Holland. 297 pp. New York, Oxford University Press. 1 993. $85.lsBN 0-1 9-262382-6.

Bacterial Pathogenesis: A molecular approach. By Abigail A. Salyers and Dixie D.Whitt. 418 pp.. illuslrated. Washjngton. D.C., ASN, Press, 1994. $44.95. ISBN1 -5558'1 -070-5.

Blood, Blood Products and HlV. Second edition. Edited by R. Madhok, C.D. Forbes. andB.L. Evatt. 276 pp. New York, Chapman & Hall. '1994. $74.95. ISBN 0-412-40400-

Bloom and Fawcett: A Textbook of Histology. Twelfth edition. By Don W. Fawcett.964 pp.. illustrated. New York. Chapman & Hall, 1 994. $89. ISBN 0-412-04691-1 .

Cells and Cylokines in Lung lntlammation. (Annals of the New York Academy ofSciences. Vol.725.) Edited by Michel Chignard, [,4arina Pretolani. Patricia Renesto,ând B. Boris Vargaftig. 380 pp., illuskated. New York, New York Academy of Sciences,1994. S110. tSBN 0-89766-855-3.

Cellular, Biochemical, ând Molecular Aspects oI Reperrusion lniury. (Annals of theNew York Academy of Sciences. Vol. 723.) Edited by Dipak K. Das. 506 pp., illustrât-ed. New York. New York Academy ot Sciences, '1994. $135. ISBN 0-89766-882-0.

A Color Atlas oI Urine Microscopy. (Châpmân & Hall N.4edical Atlas Series. No. 13.)By D.F. Birch, K.F. Fairley, G.J. Becker, and P Kincaid-Smith. 1 48 pp., illustrated. NewYork. Chapman & Hall, 1994. 5125. ISBN 0-412-47950-8.

Developmental Biology. Founh edition. By Scott F. Gilbert.894 pp., illustrated. Sun-derland, I\.4ass., Sinauer. 1994. $57.95. ISBN 0-87893-249-6.

Devèlopmental Biology: A guide for experimental study. By Mâry S. Tylet. 172 pp.Sunderlând, Mâss.. Sinâuer, 1994. S18.95. ISBN 0-87893-834-6.

DNA Topology. (ln Focus.) By Andrew D. Bates and Anthony Maxwell. 114 pp., illus-trated. New York. Oxford University Press. 1994. S15.95, ISBN 0-19-963349-5.

The Endocrine Pancreas, lnsulin Action, and Diabetes. (Endocrine Reviews Mono-graphs I2l.) Edited by Louis E. UndeNood.216 pp., illustrated. Bethesda, Md.. En-docrine Society Press, 1994. $55. ISBN 1-A79225-14-X.

The Functional Gastrointeslinal Disorders: Diagnosis, pathophysiology, andtreatment - a multinationel consensus. Edited by Douglas A. Drossman. with JoelE. Richter, Nicholas J. Talley, W. Grant Thompson. Enrico Corazziari, and William E.Whitehead.370 pp.. illuskated. Boston, Little. Brown. 1994. S87.50. ISBN 0-316-19342.9.

Gas Chromatography: A practical approach. (The Practical Approach Series.) Edit-ed by PJ. Bâugh.426 pp.. illustrated. NewYork, Oxford University Press, 1994. $47.tsBN 0-1 9-963271 -5.

\or'. 10. l!)11 [

chiolitis) bLrt is larg-rlr rlcvotccl tr.r a cliscussit.rtt ol'the icliopath-ic zrncl knorlr) caus('s ol' bronchiolitis obliterans ot'gellizing

1)ucutrtonizr. 'l'hc lbru'th and final scction clcals l'ith blonchi-

olar clisease in chilrh'or.Sevcral chirpters alr: srrpcrb. (ihapte r [i. bv Finklcstein and

(losio. is e-xccllent. It bcautilLrlll inteslates rk'scriPtions ol 1-la-thokrsr. l)athosellesis, arrr[ clinicirl cliscase. Sirnilarlr. Izinti'ssucc:irr< t chal)tcr olr a slol)a1 r'icu' ol' icliopatliic blonchiolitisol)litcrilns olq:urizing pneunroniu is esccllt:nt. II I had to selectotrc r:huptel ls outstanclitrg. it rrould be thc otre l;r Ciorclier,l'evrol. ancl Loilt:. lhich Iircuscs on blonchiolitis oblitelansoLgzurizing plrcrrrnonia as a nro(lel o(' inflantrnatolr' (airuarJItrng rlist'asc. This chaptcl is a svrrth.sis.l rctent cllitu on lrrnginjun ancl lellair that portrivs the eu'rrts n'sulting in clirricalcl i scase.

-l'he ncakest chapters are on ilnaging. ()ticn thelc appealsto be conlusi«.rtr about rvhethcr 1he autlx.rl is cliscussing ltron-t'hiolitis oblitt'r'ans orts-ilnizing pneunronia or l;r'onchiolitisoblitelans. \lani illustlatir»rs art' soo(l an(l rlill prov' usclirl,Itol'evt't

I rccomnrcntl tliis book to all clinicians aucl clinical scien-tists rrith an iutelcst in ltrns cliscascs. lt is not likr-:lv to behelplirl to st-ttr-ralists or busic rt:scalcht'r-s. It is a book that bc-gins to ill-rrninirte \ÿhal littlr: le knou aborrt cliscases o1'thel:t onchiolcs.

.f.rr:r D. IrrL.\rnR, \Lf).LLrivclsitr- ol' Alabanra

Birmirrgham, AL 3529I at Birrriing-ham

Glycobiology: A practical approach. (The Practical Approach Series.) Edited byM. Fukuda and A. Kobâta. 401 pp. New York, Oxtord Universrty Press, 1994. $68(cloth): S47 (paper). ISBN 0-19-963372-X (cloth): 0-19-963371-1 (paper).

The Liver: Biology and pathobiology. Edited by lruin M. Arias, with five others. '1628

pp., illustrated. New York, Raven Press, 1994. $265. ISBN 0-78'17-0133-3.Medical Virology. Fourth edition. By David O. White and Frank J. Fenner.603 pp.,

illustrated. San Diego, Calif.. Academic Press. '1994. §95. ISBN 0-12-746642-8.Molecular Genetics oT Haemostasis and lts lnherited Disorders. (Oxford Mono-

graph on l\redical Genetics. No. 25.) By Edward G.D. Tuddenham and David N.Cooper 585 pp. NewYork, Oxtord University Press. 1994. $145. ISBN 0-19-261661-7.

Molecular lmaging in Neuroscience: A practical approach. (The Practical ApproachSeries.) Edited by N.A. Sharif. 245 pp.. illuslrated. New York. Oxlord University Press.1994. S38. tSBN 0-19-963380-0.

Molecular Methods lor Microbial ldentitication and Typing. By K.J. Towner ândA. Cockâyne. 202 pp. New York. Chapman & Hall, 1 994. $44.50. ISBN 0-41 2-49390-x

New Findings in HSV-lnduced Retinitis in the Von Szily Modê1. By Manfred Zierhut.95 pp., illustrated. Boston, Butlemodh-Heinemann. 1994. S49.95. ISBN 9-070-43006-r.

Pathogenic and Clinical Microbiology: A laboratory manual. By Sharon S. Bow-land, Sherril Ross Walsh, Louise D. Teel, and Amy M. Carnahan. 389 pp., illustrated.Boston, Little. Brown. 1994. S25.95. ISBN 0-316-76049-8.

Physiochemical Hydrodynamics; An introduclion. Second edition. By Fionald F.

Probstein.400 pp., illustrated. New York, John Wiley. 1994. $69.95, ISBN 0-47'1-01011-1.

The Pituitary Gland. (Comprehensive Endocrinology.) Second edition. Edited by Hiroolmura. 520 pp., illustrated. New York, Fiaven Press, 1 994. $1 40. ISBN 0-781 7-0207-0.

Plasmids: A practical approâch. (The Practical Approach Series.) Second edition.Edited by K.G. Ha'dy. 252 pp.. illustrated. New York. Oxford University Press, 1 994.s33. tsBN 0-'1 9-963444-0.

Principles oI Benal Physiology. Third edition. By Christopher J. Lole. 200 pp. NewYork. Chapman & Hall. 1994. S24.50. ISBN 0-412-55520-4.

Receptor Autoradiography: Principles and practice. Edited by John Wharton andJulia Nil. Polak.335 pp., illustrated. New York, Oxford University Press, 1994. S59.95.lsBN 0-'1 9-262209-9.

Sherris Medical Microbiology: An introduction to infectious diseases. Third edi-tion. Edited by Kenneth J. Byan. 890 pp., illustrated. Noruâlk, Conn., Appleton &Lânge. 1994. S49.95. ISBN 0-8385-8541-8.

Stroke in Children and Young Adults. Edited by Jose Biller. with Katherine D.Mathews and Betsy B. Love. 259 pp.. illustrated. Boston, Butterworth-Heinemann,1 994. S69.95. tSBN 0-7506-9203-0.

Lt)I ( \il()\. Hrrr0u\. lJro(;R\pH\'. \\t) Pr IJLI( PRI\\

Allergies A-2. By Myron A. Lipkowitz and Tova Navarra. 352 pp. New York, Facts OnFile. 1994. S40. ISBN 0-8160-2824-9.

The Complete Guide to Prevenling Cancer: How you can reduce your risks. ByElizâbeth Whelân. 385 pp. Amherst. N.Y, Prometheus Books,'1994, $26.95. ISBN0-87975-890-2.

|H [ \]r\\' L,\(;l.A\l).JoL R\AL OF -\ II;DICI \ Il