Br HeartJ 1993;70:443-447

Importance of oxygen-haemoglobin binding tooxygen transport in congestive heart failure

RobertM Bersin, Michael Kwasman, Debra Lau, Cindy Klinski, Kevin Tanaka, PaymanKhorrami, Teresa DeMarco, Christopher Wolfe, Kanu Chatterjee

Cardiology Division,University ofCalifornia MedicalCenter, SanFrancisco, California,USAR M BersinM KwasmanD LauC KlinskiK TanakaP KhorramiT DeMarcoC WolfeK Chatter1eeCorrespondence to:Kanu Chatterjee, MB,FRCP, University ofCalifornia, San Francisco,Moffitt-Long Hospital,1186-Moffitt, SanFrancisco, CA 94143.Accepted for publication17 February 1993

AbstractObjective-To assess the importance of2,3-diphosphoglycerate (2,3-DPG) andoxygen-haemoglobin binding to oxygentransport in patients with congestiveheart failure.'Methods-In 30 patients with severe con-gestive heart failure, arterial, mixedvenous, and coronary sinus venous bloodconcentrations of 2,3-DPG were mea-sured and systemic output and coronarysinus blood flow were measured by athermodilution technique. Oxygen-haemoglobin affinity was expressed asthe oxygen tension in mm Hg at whichblood is 50% saturated with oxygen (P,O).Results-Compared with normal values,2,3-DPG was high in arterial blood(2.58 ,mollml, p = 0-01; 20*8 umollg hae-moblobin, p < 0-0001). Significant gradi-ents between arterial, mixed venous, andcoronary sinus blood 2,3-DPG concen-trations were also found (mixed venous =2'40 nmollml, p = 0.05 v arterial blood;coronary sinus venous blood = 2'23umorllml, p < 0-04 v arterial blood). P,0was correspondingly high compared withthe accepted normal value (mean 29-7mm Hg, normal 26*6 mm Hg, p < 0.001).Systemic oxygen transport (351 mI 021min/m2) varied directly with the forwardcardiac index (r = 0-89, p < 0-0001).There was no relation between systemicoxygen transport and arterial oxygencontent. Similarly, myocardial oxygentransport was found to vary directly withcoronary sinus blood flow. Calculationsof changes in cardiac index and coronarysinus blood flow at normal oxygen-haemoglobin binding indicate that a con-siderable increase in cardiac index andcoronary blood flow would be required tomaintain similar systemic and myocar-dial oxygen transport.Conclusions-In patients with severeheart failure increased 2,3-DPG andreduced oxygen-haemoglobin bindingmay be compensatory mechanisms thatmaintain adequate systemic and deliveryofoxygen to myocardial tissue.

(Br Heart _ 1993;70:443-447)

Systemic oxygen transport (the quantity ofoxygen delivered to tissues for aerobic respi-ration) may be impaired due to reductionsof either cardiac output, oxygen content of

arterial blood, or the haemoglobin content ofblood. In congestive heart failure, forwardcardiac output may be reduced, decreasingthe quantity of oxygenated blood carried tothe periphery. The oxygen content of arterialblood and the haemoglobin concentration arenearly normal and are not the main factorslimiting oxygen transport in patients withcongestive heart failure.

Despite reductions in the cardiac outputand systemic oxygen transport, systemic oxy-gen consumption is generally maintained bymeans of compensatory increases in tissueoxygen extraction.1 2 One of the mechanismsby which tissue oxygen extraction is enhancedis through a reduction in oxygen-haemoglo-bin affinity, favours the unloading of oxygento tissues.3 The reduction in oxygen-haemo-globin affinity is accomplished throughincreased synthesis of 2,3-diphosphoglycerate(2,3-DPG) in erythrocytes." The organicphosphate 2,3-DPG is high in patients withlung disease,9-11 cyanotic congenital heart dis-ease,12 13 and low output congestive heart fail-ure.'1'7 Little information is available,however, about the potential importance ofthe 2,3-DPG concentrations in erythrocytesto the enhancement of tissue oxygen extrac-tion in these clinical circumstances. In thisreport, we present data on 2,3-DPG concen-trations and oxygen-haemoglobin binding in30 patients with severe congestive heart fail-ure due to impaired left ventricular systolicfunction. We also evaluate the contribution ofa reduced oxygen-haemoglobin affinity to sys-temic and myocardial oxygen transport andits use in these patients.

Patients and methodsThirty patients with congestive heart failureand New York Heart Association (NYHA)class III or IV were studied. The underlyingcause of congestive heart failure was coronaryartery disease in 27 and idiopathic dilatedcardiomyopathy in three. All patients had leftventricular ejection fractions of 40% or less.They were all studied as inpatients at theMoffitt Hospital, University of California,San Francisco, for the management of con-gestive heart failure. This study was approvedby the Institutional Review Board and allpatients gave informed consent. Pulmonaryartery and coronary sinus catheters wereplaced by standard percutaneous techniquesand their positions were confirmed by fluo-roscopy. Cardiac indices (1/min/m2) weremeasured by a thermodilution technique'8

443

on July 7, 2021 by guest. Protected by copyright.

http://heart.bmj.com

/B

r Heart J: first published as 10.1136/hrt.70.5.443 on 1 N

ovember 1993. D

ownloaded from

http://heart.bmj.com/

Bersin, Kwasman, Lau, Minski, Tanaka, Khorrami, et al

with the patient supine and fasting. A mini-mum of three measurements were made foreach patient and these were -averaged.Coronary blood flow (ml/min) was estimatedby a continuous thermodilution technique'9with Wilton-Webster coronary sinus catheters(Altadena, California), and infusion of 5%dextrose in water at room temperature at arate of 46 ml/min. Oxygen saturations of arte-rial, pulmonary arterial (mixed venous), andcoronary sinus blood were measured directlywith a haemoximeter (Radiometer, modelOSM-2, Copenhagen, Denmark). Haemo-globin content was measured by the cyan-methaemoglobin method. Blood gases weremeasured in arterial, mixed venous, and coro-nary sinus blood with an automated bloodgas analyser (Corning Medical, model 168and 178, Medfield, Massachusetts). Oxygencontents were then calculated directly fromthe measured data with the formula: 02 con-tent (vol %) = 02 saturation/100 x haemo-globin content (g %) x 1-34 (mUg)+0*0031 Po2mm Hg.The other indices were calculated as

follows: systemic oxygen transport (mlo2/min/m2 = arterial 02 content x cardiacindex x 10; myocardial oxygen transport (ml02/min) = (arterial 02 content x coronaryblood flow)/100; systemic oxygen consump-tion (ml o2/min/m2 = tissue oxygen extractionx cardiac index x 10; myocardial oxygenconsumption = coronary blood flow/100 x(arterial 02 content - coronary sinus 02 con-tent).

Oxygen-haemoglobin affinity is expressedas the oxygen tension (mm Hg) at whichblood is 50% saturated with oxygen (P50).The P50 of blood was calculated for eachpatient from directly measured oxygen satu-rations and oxygen tensions of mixed venousblood measured by the method ofSeveringhaus.20 Measured oxygen tensionswere corrected to pH 7 40 before the calcula-tion of p50.21 Whole blood concentrations of2,3-DPG were measured in triplicate on arte-rial, mixed venous, and coronary sinus bloodsamples deproteinised in 0 5 M cold perchlo-ric acid with an NAD/NADH spectrophoto-metric assay (Sigma Chemicals, St. Louis,Missouri).22The contribution of reduced oxygen-

haemoglobin binding to oxygen transport wasassessed as follows. The directly measuredoxygen tensions and pH values for each bloodsample on each individual patient wereapplied to the normal oxygen-haemoglobinbinding curve to determine normalisedoxygen saturations (P50 = 26X6 mm Hg).Normalised oxygen contents were then deter-mined from normalised oxygen saturations.Normalised oxygen contents were used torecalculate values for coronary blood flowand cardiac index assuming constant oxygenconsumption (systemic and myocardial oxy-gen transport and oxygen consumption).Comparison of the calculated values for coro-nary blood flow and cardiac index with mea-sured values gives a quantitative estimate ofthe amount that coronary blood flow and

cardiac output would need to increase tomaintain constant concentrations of oxygendelivery and consumption if oxygen-haemo-globin binding were normal.

Statistical analyses were performed on P50data with unpaired t tests. Normal values forP50 (26-6 mm Hg) were taken from the publi-cation of Severinghaus in which 10 healthynon-smoking volunteers were studied.20 Dataon 2,3-DPG concentrations were comparedwith the normal population by single group ttests and the population mean value was 2 1,umol/ml in whole blood.22 Statistical compar-isons were made of 2,3-DPG data betweenarterial, mixed venous and coronary sinusblood samples with paired t tests. Differencesbetween groups were considered significantwhen p < 0 05. All statistical analyses wereperformed with a Macintosh computer and aStatview statistical software package (version1 1, Calabassas, California), and the data arepresented as means (SEM) unless otherwiseindicated.

Results2,3-DIPHOSPHOGLYCERATEThe mean (SEM) arterial concentration of2,3-DPG in patients with heart failure was2-58 (0.18) jumoUml in whole blood, or 20-8(1 4) ,umol/g haemoglobin, which is higherthan reported normal values. A gradientbetween arterial, mixed venous, and coronarysinus 2,3-DPG concentrations was also found(fig 1). Mixed venous 2,3-DPG concentra-tions were significantly lower (2-40 (0 17)pmol/ml p = 0 05) than arterial concentra-tions and coronary sinus 2,3-DPG concentra-tions were the lowest (2.23 (0.22) ,umol/ml, p= 0 038). The arterial venous gradient for2,3-DPG might be explained by a larger thannormal gradient for hydrogen ions and car-bon dioxide. The blood pH was 0 07 unitslower in coronary sinus blood than in arterialblood, and the carbon dioxide tension was14-2 mm Hg greater (table 1). The mixedvenous pH was on average 0-05 units lowerand the carbon dioxide tension 8-7 higherthan arterial blood, indicating arterial tovenous hydrogen ion and carbon dioxide gra-dients about 50% greater than normal.

E

0E

0-

co)C;

1~~~~~~~~~~~~~~~~~~~~~~~0~~~~~~~~~~

0*5

n.n,- - -Arterial Mixed Coronary

venous sinus

Figure 1 Simultaneous measurements of2,3-DPG inarterial, mixed venous, and coronary sinus blood. Note theprogressive gradient between arterial, mixed venous, andcoronary sinus 2,3-DPG concentrations.

444

on July 7, 2021 by guest. Protected by copyright.

http://heart.bmj.com

/B

r Heart J: first published as 10.1136/hrt.70.5.443 on 1 N

ovember 1993. D

ownloaded from

http://heart.bmj.com/

Oxygen transport in congestive heartfailure

Table I Blood gases

Arterial Mixed venous Coronary sinus

pH 7-454 (0-001) 7 404 (0-001) 7-383 (0-001)Pco2 (mm Hg) 34-2 (1 1) 42-9 (0 9) 48-4 (1-2)Po2 (mm Hg) 98-7 (6 9) 31-7 (1-4) 18-8 (0.8)HCO3 (meq/l) 23-7 (0 7) 26-7 (0 6) 28-7 (1-0)

Values are means (SEM). Pco,2 carbon dioxide tension; Po2,oxygen tension; HCO3, bicarbonate anion.

OXYGEN-HAEMOGLOBIN BINDINGCorresponding measures of oxygen-haemo-globin binding (P50) were high in all but oneof the 30 patients studied (fig 2). The meanP50 was 29-7 (04) mm Hg v 26-6 mm Hgnormally (p = 0 001). There were no correla-tions between 2,3-DPG concentrations andthe measured P50.

SYSTEMIC OXYGEN TRANSPORT ANDCONSUMPTIONSystemic oxygen transport was 351 (16) ml02 min/m2, and was related to cardiac index(2-31 (0 15) 1/min/m2). Arterial haemoglobincontent was 12-5 (0-04) g/dl and oxygen con-tent was 16&0 (0-5) vol%. A significant linearrelation between systemic oxygen transportand cardiac index was found as expected (r =0-89, p < 0-001) (fig 3). There were no corre-lations between systemic oxygen transportand arterial oxygen content (fig 3).

MYOCARDIAL OXYGEN TRANSPORT ANDCONSUMPTIONMyocardial oxygen transport was 21P8 (2 9)ml oJ/min and consumption was 16&7 (2 2)ml o2/min. As expected, myocardial oxygentransport was dependent on coronary bloodflow (r = 0 97, p < 0 0001 (fig 4). Myocardialoxygen consumption was also related to coro-nary blood flow (r = 0-96, p < 0-0001), butnot myocardial oxygen extraction (fig 4).

CONTRIBUTION OF REDUCED OXYGEN-HAEMOGLOBIN BINDING TO OXYGENTRANSPORTNormalisation of oxygen-haemoglobin bind-ing would result in a 6-2% rise in the mixedvenous oxygen saturation (p < 0-001) and a7-4 mm Hg rise in the coronary sinus oxygen

Figure 2 Plot ofmeasured mixed venousoxygen tensions andsaturations in the 30patients compared with thenormal adult oxygen-haemoglobin bindingcurve.

20 30

Mixed venou

EZ-.Et00.coccn

ai)

x0.2Eum

cn

8070601501403020101

10 010 o0890.

10 010 o 0o 010

0 05 1 1-5 2 2-5 3 3-5 4 4.-

800700600500400300200100

0

Cardiac index (I/min/m2)

0

0 oo0 000o00 000o0,Po% 00

0 0 0

2 4 6 81 0 12 14 16 18 20 22

Arterial oxygen content(ml 02/100 ml blood)

Figure 3 Relation betweenfonvard cardiac index andsystemic oxygen transport. Note that the arterial oxygencontent bears no relation to overall systemic oxygentransport-

saturation (p < 0-001) assuming no change inarterial oxygen saturation. Correspondingmixed venous oxygen content would rise 1-1vol % (p < 0-001) and coronary sinus oxygencontent 1-2 vol % (p < 0-001). Systemic andmyocardial oxygen extraction rates wouldtherefore fall about 1-2 vol% each (table 2).To compensate for this, the resting cardiacindex would need to be 31% higher to main-tain the same systemic oxygen delivery.Similarly, coronary blood flow would need tobe considerably higher to maintain the samemyocardial oxygen delivery and consumption(table 2).

-EC4

0

0

._

CL

Co

CUi)c

xco

L0

Co

0

0

70605040302010

70605040302010

0

r = 0-97

0

50 100 150 200 250 300 350 400 450

Coronary sinus blood flow (ml/min)

0

0

000

0

000 0 O 0

O O

0 2 4 6 8 10 12 14 16 18 20 22

Arterial oxygen content(ml 02/100 ml blood)

Figure 4 Relation between coronary bloodflow andI I myocardial oxygen transport There was no relation40 50 between arterial oxygen content and overall oxygen

transport to the myocardium; myocardial oxygen transportS P02 was related to coronary bloodflow, as expected.

C0Co4C

co

0OC4

0

a)~0X.)x

445

5

on July 7, 2021 by guest. Protected by copyright.

http://heart.bmj.com

/B

r Heart J: first published as 10.1136/hrt.70.5.443 on 1 N

ovember 1993. D

ownloaded from

http://heart.bmj.com/

Bersin, Kwasman, Lau, Klinski, Tanaka, Khorrami, et al

Table 2 Effect of normalisation ofoxygen-haemoglobin binding

Reduced °2 binding Normalised °2 binding p Value

P50 (mm Hg) 29-7 (0.4) 26-6 (0.3)

Oxygen transport in congestive heart failure

without heart failure, calculated P50 concen-trations were 26 1 (2 0) mm Hg.'6 Further-more, we found close agreement between theone point technique and the method of directmeasurement of P50.'6 With methods similarto ours, several studies have shown SDs ofabout 1 mm Hg in P50 measurements. In thepresent study, the SD of P50 values was simi-lar.'6 17 It is thus unlikely that this inadequacywould have influenced the results qualita-tively. We also did not measure systemic andcoronary haemodynamics concurrently in acontrol group. In a previous study, however,we determined system and coronary haemo-dynamics in patients without heart failure.39In these patients the cardiac index was 3-0(O 6) 1/min/m2. Coronary blood flow was 74 0(37) 1/min and myocardial oxygen consump-tion was 8-7 (4-2)ml/min. Thus in patientswith heart failure in this study, both coronaryblood flow and myocardial oxygen consump-tion were considerably higher, and cardiacindex lower than normal. The other limita-tion is that change in these adaptive mecha-nisms were not assessed during stress.Nevertheless, the results suggest that even inthe resting condition a number of compen-satory mechanisms are called upon to main-tain adequate delivery of oxygen to the tissuesof patients in severe heart failure, anddecreased oxygen-haemoglobin binding asso-ciated with increased 2,3-DPG is probablyone of them.

The study was supported by grants from the National Heart,Lung, and Blood Institute (No HL-01791) and the AmericanHeart Association (No 88-1185).

RMB was a recipient of the Physician-Scientist Award ofthe National Institute of Health, and is a WinthropPharmaceuticals Grant-in-Aid Awardee of the AmericanHeart Association. CW was a recipient of the ClinicalInvestigator Award of the National Institutes of Health. KC isthe Lucie Stern Professor of Cardiology of the University ofCalifornia, San Francisco.

1 Cain SM. Oxygen delivery and uptake in dogs during ane-mic and hypoxic hypoxia. J Appl Physiol 1977;42:228-34.

2 Baim DS, Rothman MT, Harrison DC. Simultaneousmeasurement of coronary venous blood flow and oxygensaturation during transient alterations in myocardialoxygen supply and demand. Am Jf Cardiol 1982;49:743-52.

3 Bohr CHR, Hassenbalch K, Krog A. Ueber einen in bio-logicher wichtigen Einfluss, den die Kohlensaures-pannung des Blutes auf dessen Sauerstoffbinding ubt.Scandinavian Archives of Physiology 1904;16:402-12.

4 Benesh R, Benesh RE, Yu CI. Reciprocal binding ofoxygen and diphosphoglycerate by human hemoglobin.Proc NatlAcad Sci 1968;59:526-32.

5 Eaton JW, Brewer GJ: The relationship between red cell2,3-diphosphoglycerate and levels of hemoglobin in thehuman. Proc Natl Acad Sci 1968;61:756-60.

6 Bunn HF, Jandl JH. Control of hemoglobin functionwithin the red cell. NEnglJMed 1970;282:1417-21.

7 Torrance J, Jacobs P, Restrepo A, Eschbach J, Lenant C,Finch CA. Intraerythrocytic adaptation to anemia.NEnglJMed 1970;283:165-9.

8 Kilmartin JV. Interaction of haemoglobin with protons,CO2, and 2,3-diphosphoglycerate. Br Med Bull 1976;32:209-12.

9 Lenfant C, Ways P, Aucutt C, Cruz J. Effect of chronichypoxic hypoxia on the 02-Hb dissociation curve andrespiratory gas transport in man. Respir Physiol 1969;7:7-29.

10 Lenfant C, Torrance J, English E, Fiknch CA, Reynafarje

C, Ramos J, Faura J. Effect of altitude on oxygen bind-ing by hemoglobin and on organic phosphate levels.J Clin Invest 1968;47:2652-6.

11 Osli FA, Gottheb AJ, Delivoria-Papadopous M, MillerWW. Red-cell 2,3-diphosphogylcerate levels in subjectswith chronic hypoxemia. N Engl J Med 1969;280:1165-6.

12 Morse M, Cassels DE, Holder M. The position of theoxygen dissociation curve of the blood in cyanotic con-genital heart disease. JClin Invest 1950;29:1078-103.

13 Berman W, Wood SC, Yabek SM, Dillon T, Fripp RR,Burstein R. Systemic oxygen transport in patients withcongenital heart disease. Circulation 1987;75:360.

14 Mulhausen R, Astrup P, Kjeldsen K. Oxygen affinity ofhemoglobin in patients with cardiovascular diseases,anemia, and cirrhosis of the liver. Scand J Clin LabInvest 1967;19:291-7.

15 Woodson RD, Torrance JD, Shappell SD, Lenfant C.The effect of cardiac disease on hemoglobin-oxygenbinding. J Clin Invest 1970;49:1349-56.

16 Wilkinson PL, Graham BH, Moyers JR, Hamilton WK,Ports TA, Ullyot DJ, Chatterjee K. Changes in mixedvenous and coronary sinus P50 secondary to anesthesiaand cardiac disease. Anesth Analg 1980;59:751-8.

17 Metcalfe J, Dhindsa DS, Edwards MJ, Mourdjinis A.Decreased affinity of blood for oxygen in patients withlow-output heart failure. Circ Res 1969;25:47-51.

18 Ganz W, Donose R, Marcus HS, Forrester JS, Swan HJC.A new technique for measurement of cardiac output bythermodilution in man. AmjICardiol 1971;27:392-6.

19 Ganz W, Tamura K, Marcus HS, Donose R, Yoshida S,Swan HJC. Measurement of coronary sinus blood flowby continous thermodilution in man. Circulation 1971;44:181-95.

20 Severinghaus JW. Blood gas calculator. J Appl Physiol1966;21:1108-16.

21 Severinghaus JW. Simple, accurate equations for humanblood 0° dissociation computations. J Appl Physiol1979;46:599-602.

22 Rose AB, Leibowitz J. Direct determination of 2,3-diphos-phoglycerate. Anal Biochem 1970;35: 177-80.

23 Kennedy AC, Valtis OJ. The oxygen dissociation curve inanemia of various types. J Clin Invest 1954;33:1372-81.

24 Bellingham AJ, Detter JC, Lenfant C. Regulatory mecha-nisms of hemoglobin-oxygen affinity in acidosis andalkalosis. J Clin Invest 197 1;50:700-6.

25 Bersin RM, Chatterjee K, Arieff AI. Metabolic and hemo-dynamic consequences of sodium bicarbonate adminis-tration in patients with heart disease. Am J Med 1986;86:7.

26 Rose AB. Enzymes controlling 2,3-diphosphoglycerate inhuman erythrocytes. Federation Proceedings 1970;29:1105-11.

27 Von Planta M, Weil MH, Gazmuri RJ, Bisera J, RackowEC. Myocardial acidosis associated with CO2 produc-tion during cardiac arrest and resuscitation. Circulation1989;80:684-92.

28 Adrogue HJ, Rashad N, Gorin AB, Yacoub J, Madias NE.Assessing acid-base status in circulatory failure: differ-ences between arterial and central venous blood. N EnglJMed 1989;320:1313-6.

29 Weil MH, Rackow EC, Trevino R, et al. Difference inacid-base state between venous and arterial blood dur-ing cardiopulmonary resuscitation. N Engl J Med 1986;315:153-6.

30 Alella A, Williams FL, Williams CB, Katz LN.Interrelation between cardiac oxygen consumption andcoronary blood flow. AmJIPhysiol 1955;183:570-82.

31 Lombardo TA, Rose SL, Taeschler M, Tul.uy S, Bing,RJ. The effect of exercise on coronary efficiency in man.Circulation 1953;7:71-8.

32 Marcus ML. The coronary circulation in health and disease.New York: McGraw Hill, 1983:65.

33 Mehmul HC, Duvelleroy MA, Laver MB. Response ofcoronary blood flow to pH induced changes in hemo-globin-O2 affinity. YAppl Physiol 1973;35:485-9.

34 Rude RE, Tumas J, Gunst M, Kloner RA, DeBoer LWV,Maroko PR. Effects of ortho-iodo sodium benzoate onacute myocardial ischemia, hemodynamic function, andinfarct size after coronary artery occlusion in dogs. Am JCardiol 1983;51:1422.

35 Pantely GA, Oytama AA, Metcalfe J, Lawson MS, WelchJE. Improvement in the relationship between flow toischemic myocardium and the extent of necrosis withglycolytic intermediates that decrease blood oxygenaffinity in dogs. Circ Res 1981;49:395-404.

36 Bristow JD, Metcalfe J, Krall MA, Welch JE, Black JA,Dhindsa DS. Reduction of blood oxygen affinity in dogsby infusion of glycolytic intermediates. J Appl Physiol1977;43:102-6.

37 Graf H, Mayer G, Cada E, Muller NM, Bersin RM. Iserythropoietin a direct stimulator of 2,3-DPG? AnnualMeeting, American Society for Clinical Investigation,April, 1988 [abstract]. Clin Res 1988;36:482A.

38 Flenley DC, Fairweather U, Cooke NJ, Kirby BJ.Changes in haemoglobin binding curve and oxygentransport in chronic hypoxic lung disease. BMJ 1975;1:602-4.

39 De Marco T, Chatterjee K, Rouleau J-L, Parmley WW.Abnormal coronary hemodynamics and myocardialenergetics in patients with chronic heart failure causedby ischemic heart disease and dilated cardiomyopathy.Am Heartj 1988;115:809-15.

447

on July 7, 2021 by guest. Protected by copyright.

http://heart.bmj.com

/B

r Heart J: first published as 10.1136/hrt.70.5.443 on 1 N

ovember 1993. D

ownloaded from

http://heart.bmj.com/