INTRATHECALFIBRINOLYSIS WITH STREPTOKINASEINTUBERCULOUSMENINGITIS

BY A. P. FLETCHER1

(From the Wright-Fleming Institute of Microbiology, and St. Mary's Hospital,London, W. 2, England)

(Submitted for publication May 26, 1953; accepted July 1, 1953)

The introduction by Cathie (1) of intrathecalstreptokinase to prevent and treat blockage of thecerebrospinal pathways led to hopes that the treat-ment of this serious complication might be greatlyimproved; and that prevention and removal of theexudate in this disease might shorten the courseof treatment and render chemotherapy more ef-fective. These hopes were sustained by later re-ports (2, 3). On the other hand, Eadie (4, 5) andLorber (6), both with equal numbers of treatedand control cases, were unable to demonstratebenefit from the use of intrathecal streptokinase.The value of streptokinase in the treatment oftuberculous meningitis is therefore debatable.

The demonstration by Tillett and Sherry (7)that streptokinase produced an active fibrinolyticsystem in the pleural cavity, and was of greatclinical value in the treatment of clotted hemo-thorax (8) and empyema (9), suggested re-ex-amination of its value in tuberculous meningitis.The success of these authors appeared in part to bedue to the fact that they had shown active fibrino-lytic enzyme to be present in the pleural fluids oftheir treated cases, and had used measurements offibrinolytic activity to assess dosage and controltreatment.

A difficulty encountered in the work on tuber-culous meningitis has been the proper assessmentof dosage to produce free lytic enzyme; in fact noadequate measurements were performed by any ofthe quoted authors. In view of these uncertainties,clinical trial is not a satisfactory method of ap-proach in deciding the merits or demerits of thistreatment, since the immediate clinical assessmentof such cases is unreliable, and long term assess-ment takes at least two years.

A quantitative laboratory investigation of thesystem employed, therefore, seemed warranted inorder to eliminate unfruitful methods of therapy.It has been shown (10, 11 ) that streptokinase has

I At present: Merck International Research Fellow,Bellevue Medical Center, New York.

no fibrinolytic properties but acts on a serum fac-tor named plasminogen to form an enzyme withfibrinolytic properties named plasmin. However,specific antibodies to streptokinase exist in plasma,and may be increased by the injection of thissubstance, or may follow naturally occurring hemo-lytic streptococcal infection. The action of plasminmay be partially or completely inhibited by natu-rally occurring antiplasmins (Figure 1).

Measurements of the components of this systemmight be expected to supply the information whichappears to be essential for the rational control oftherapy.

Streptokinase (Lederle) as a consequence of itsmethod of preparation contains a substantial quan-tity of streptodornase (streptococcal desoxyribo-nuclease). This enzyme has been shown to exertimportant therapeutic effects upon purulent exu-dates containing desoxyribose nucleoprotein (12,13). Observations in this laboratory made uponmeningeal exudate taken from autopsied cases oftuberculous meningitis have shown that this exu-.date is composed chiefly of fibrin and fibrinousmaterial. Feulgen staining demonstrated com-paratively small amounts of nucleoproteins, andin vitro experiment confirmed that fibrin lysingsystems played a predominant role in the lysis ofthis material. For these reasons the effects ofstreptodornase have been ignored, and this com-munication deals solely with streptokinase action.

Similarly, since the action of the protease in-hibititor known to contaminate streptokinase prepa-rations appeared to be quantitatively unimportantin the concentrations of streptokinase used, thisfactor is not further considered.

Units of measurement

Units of streptokinase and plasminogen werethose of Christensen (14). One Christensen unitof plasmin is the quantity of enzyme required tolyse in 30 minutes at 35° C. a fibrin clot derivedfrom 1 ml. of 0.2 per cent bovine fibrinogen. This

69

A. P. FLETCHER

Enzyme Precursor (Plasminogen)+ Activator (Streptokinase)

Enzyme (Plasmirif

It Inhibitor (Antiplasmirn

Lyses Fibrin, etc.FIG. 1. SCHEMEOF SYSTEM

unit will lyse approximately 0.5 mg. of pure hu-man fibrin under the same conditions.

One unit of antiplasmin is the amount of in-hibitor that will prevent the fibrinolytic action of1 unit of enzyme after 10 minutes incubation at350 C. Experiment showed that 95 per cent ofthe antiplasmin activity was fixed at the end ofthis time.

When measuring plasminogen it is assumed,under the favorable experimental conditions em-ployed, that complete conversion to plasmin occurs.The inaccuracies inherent in this assumption andin the method have been stated previously (14).

The system of measurement then becomes:One unit of plasminogen when activated becomes

one unit of plasmin which is inhibited by one unitof antiplasmin.

Assay methods for plasmin involving a sub-stantial degree of dilution will demonstrate enzymewhich is not available for fibrinolysis, since theeffect of non-specific inhibition is altered by dilu-tion (15). The method of assay for plasmin em-ployed in this investigation requires minimal dilu-tion, and measures only the "free" enzyme avail-able for fibrinolysis with an accuracy ± 10 percent (15). The unit of enzyme activity is a sub-stantial one, and it is probable that a level of ac-tivity equivalent to 1 plasmin unit per ml. C.S.F.is adequate for the requirements of therapy in thecerebrospinal spaces.

METHODS

Plasminogen was estimated by the original Christensen(14) technique, with the modification that, where a moreaccurate figure was required, a second titration with anarrower range of dilution about the point of 1 unit ac-tivity was performed. Plasmin, antiplasmin, and therequired dosage of streptokinase were estimated by themethods of Fletcher (15). Streptokinase antibodies were

not found to exert much influence in fluids of the cere-brospinal spaces and were not estimated as a routine;when required, a modification of the Christensen (14)technique was used.

RESULTS

Plasminogen content of the cerebrospinal fluid intuberculous meningitis

Since streptokinase is inactive in the absence ofplasminogen, 71 determinations were made onspecimens of C.S.F. from 26 cases of tuberculousmeningitis. The findings are illustrated in TableI, along with estimations on normal lumbar C.S.F.

It is seen that enzyme precursor is present inmuch greater amounts than occurs in normalC.S.F. Noteworthy differences in plasminogencontent between ventricular and lumbar fluids arepresent. Cases suffering from spinal block showedin lumbar fluids taken from below the block ex-tremely high plasminogen levels, in some casesas great as those found in the serum. High plas-minogen values were found in association withhigh levels of C.S.F. protein, although the cor-relation coefficient did not reach the level ofstatistical significance.

Antiplasmin content

The inhibitor content of 28 specimens of C.S.F.from 18 cases with tuberculous meningitis arerecorded together with normal findings in Table II.It is seen that inhibitor is present in every fluidexamined, and that significant differences areagain apparent between ventricular and lumbarfluids. The inhibitor content of these fluids ap-pears to be low when compared with their plas-minogen content. However the interactions of theother components of the system result in these in-hibitor contents exerting a significant effect uponthe difficulties of inducing fibrinolytic activity.

TABLE I

Plasminogen content of C.S.F.

Site of withdrawaland number of

fluids

Normal lumbar (10)Ventricular (29)Lumbar (34)Lumbar (8)

(spinal block)

Averageplasminogen

contentunits/mi.

62050

500

Rangeunits/ml.

2- 162- 64

16- 128250-1,000

70

INTRATHECAL FIBRINOLYSIS WITH STREPTOKINASE

TABLE II

Inhibitor content of C.S.F.

Site of withdrawaland number of

fluids

Normal lumbar (12)Ventricular (12)Lumbar (11)Lumbar (6)

(spinal block)

Averageantiplasmin

contentunis/ml.

0.51.55

13

Though the antiplasmin content of cerebrospinalfluid tends to rise with increasing protein levels,the correlation does not reach the level of statisticalsignificance.

Transformation of plasminogen to plasmin bystreptokinase

The addition of increasing concentrations ofstreptokinase to biological fluids results in an in-crease of plasmin activity until with high concen-trations a plateau of activity is reached; the lowestconcentration of streptokinase inducing the maxi-mal enzyme activity has been designated the "op-timal" streptokinase concentration. It has beenshown that the "optimal" dose of streptokinase

-.i

IVI-i

z

U.J& S

does not result in complete conversion of plasmino-gen to plasmin (15). These measurements weremade on a partly purified plasminogen preparationcontaining negligible amounts of inhibitor and withan activity corresponding to 5,000 Christensenunits per mg.N. Figure 2 shows that when the"optimal" concentration of streptokinase is em-ployed, a high conversion of precursor to enzymeoccurs at low precursor concentrations, but that athigh precursor concentrations the conversion ratebecomes low. The effect of this inefficient trans-formation of plasminogen to plasmin is to reduceconsiderably the potential enzyme activity of fluidsof high precursor content. This increases therelative importance of the apparently low anti-plasmin content of the C.S.F., so that fluids ofhigh plasminogen content have only a small lyticpotentiality.

Maximal plasmin activity

A number of factors unfavorable to the ap-pearance of "free" fibrinolytic enzyme have beendiscussed. It was thought that these factors, thepresence of streptokinase antibodies, the inefficiencyof streptokinase activation, and the content of anti-

CONVaESIONOF PLASMINOGENTO PLASMIN WIat

OPTIMAL STwPTOKINASt DOSAGE.

FIG. 2. AVAILABLE PLASMINOGENCHRISTENSENUNITS/ML.

71

A. P. FLETCHER

TABLE III

Plasmin levels with optimal streptokinase

Site of withdrawaland number of

fluids

Normal lumbar (10)Ventricular (10)Lumbar (19)Lumbar (7)

(spinal block)

Range ofenzymepotencyplasmin

units/mi.

0.25-1.51 -31.5 -51 -2.5

plasmin might result in a complete suppression offibrinolytic activity even with optimal amounts ofstreptokinase. Titrations of 56 cerebrospinalfluids from 23 patients were made with varyingamounts of streptokinase; in each case free fibrino-lytic enzyme was shown to be present providedthat a suitable amount of streptokinase was addedto the fluid. It was found that where insufficientstreptokinase was used, the unfavorable factorsresulted in a complete suppression of fibrinolyticactivity. This implies a minimum streptokinaseconcentration below which physiologically activeenzyme is not released, and below which clinicaleffects attributable to fibrinolysis cannot be ex-

pected. Table III shows the maximum enzyme

ranges recorded in 36 cerebrospinal fluids from 18patients where the "optimal" amount of strepto-kinase was employed. It is seen that despite thewide variations in plasminogen and inhibitor con-

tent demonstrated previously, the potential lyticpotency of fluids from different sites is very

similar.

The clinical requirement of streptokinase

The C.S.F. enzyme levels in Table III were ob-tained using streptokinase concentrations of 25 to100 units per ml. C.S.F. for the ventricular andlumbar fluids, and 100 to 500 units per ml. C.S.F.for the lumbar fluids taken below spinal blocks.These streptokinase concentrations would be un-

safe to use clinically, since assuming a C.S.F. vol-ume of 120 ml. the intrathecal dose would be2,500 to 15,000 units streptokinase.

It was decided to investigate a lower level of ac-

tivity, and Table IV shows the concentration ofstreptokinase required to induce I unit of plasminactivity per ml. C.S.F. Table IV shows that a

lower concentration of streptokinase is requiredto activate ventricular fluid than lumbar fluid; this

difference is in line with the other differences incomposition noted in Tables I to III. Ventricularfluids required 1.5 to 30 units streptokinase per ml.C.S.F. to produce an enzyme potency of 1 unit perml. C.S.F., lumbar fluids required 3 to 60 unitsstreptokinase per ml. C.S.F., while lumbar fluidsfrom below spinal block required 30 to 120 unitsstreptokinase per ml. C.S.F. to produce the samelytic potency. For 45 out of 58 fluids the ineffec-tive streptokinase concentration was one step be-low that required for 1 unit activity, i.e., if a fluidrequired 10 units streptokinase per ml. C.S.F. for1 unit of enzyme activity, 6 units streptokinase -perml. C.S.F. produced no free enzyme. For 17 fluidsthe ineffective dose was two steps below the doserequired for 1 unit of enzyme activity (see TableIV).

These differences in composition between ven-tricular and lumbar fluids render it difficult tocalculate the clinical dose. Since cisternal fluidmore nearly resembles ventricular than lumbarfluid, the dose for each patient was calculated onthe assumption that the ventricular fluid assayrepresented a true sample of 80 ml. C.S.F. and thelumbar fluid represented a sample of 40 ml. Theuse of this simplification suggested that where thecerebrospinal circulation was free, 90 per cent ofpatients would require an intrathecal dose varyingbetween 300 and 3,000 units of streptokinase; inpractice the dose was calculated individually foreach patient. Though fluids with a high proteincontent required large doses of streptokinase fortheir activation, and fluids of low protein contentsmall doses, the correlation coefficient did notreach the level of statistical significance.

TABLE IV

Number of fluids showing at least 1 unit of plasmin activitywith various streptokinase concentrations

Streptokinaseunits/mi. 0 1.5 3.0 6.0 10 15 30 60 120

Number ofventricularfluids (24) 0 3 11- 15 22 23 24

Number ofcisternalfluids (10) 0 2 5 7 10

Number oflumbarfluids (18) 0 0 3 5 10 15 15 18

Number of lumbarfluids belowspinal block (8) 0 0 0 0 0 0 1 3 8

72

INTRATHECAL FIBRINOLYSIS WITH STREPTOKINASE

Correlation between in vivo and in vitro enzymeactivity

Where a number of unknown factors exist, cal-culations made from in vitro testing must be con-firmed by in vivo experiment. It was realized inmaking this comparison that inaccuracies con-nected with sampling and estimation of C.S.F. vol-umes had occurred, but it was believed that the re-sults were sufficiently informative to be valuable.

Table V shows the calculated enzyme activity,the calculated streptokinase concentration, and theenzyme activity obtained by sampling.

The calculated in vivo enzyme level was ascer-tained by in vitro titrations on samples of ventricu-lar and lumbar fluids from the patient, activatedwith varying quantities of streptokinase. Suit-able concentrations of streptokinase to give equalactivities in lumbar and ventricular samples hav-ing been found, the dose for injection was calcu-lated on the assumption that ventricular fluid rep-resented a volume of 80 ml. C.S.F. and lumbarfluid represented a sample of 40 ml. C.S.F. Thecalculated in vivo streptokinase concentration inunits per ml. was then obtained by dividing thetotal dose of streptokinase by 120.

When the ventricular or cisternal route wasused for injecting the streptokinase, ventricularand cisternal samples were withdrawn at 10 and30 minutes after injection from both sites. Thehighest level of enzyme activity obtained from

TABLE V

Enzyme assays of C.S.F. after intrathecal injection ofstreptokinase. Comparison with levels

suggested by in vitro testing

Calculated Calculated Actualin vivo in vivo in vivo

strepto- enzyme enzymekinase level level

concen- (from in vitro determinedtration testing) by sampling

units/ml. units/mi. units/mi.

S00 2.6 <0.1200 3.0 <0.1 spinal200 1.7 <0.1 blocks

50 0.5 <0.150 1.4 <0.150 2.8 <0.150 3.2 <0.134 1.9 <0.130 2.2 0.525 1.3 0.820 1.8 1.010 1.5 0.58 2.0 1.05 1.0 0.8

these samples was regarded as the in vivo enzymelevel. A few samples were obtained two hoursafter injection, but these samples invariably showedmuch less activity than earlier samples. In caseswhere streptokinase was injected below a spinalblock, the volume of the fluid below the block wasdetermined by the phenolsulphonephthalein dilu-tion method, and the dose of streptokinase wasfound using this volume as a basis for calculation.

Though the error of sampling by these proce-dures would be very large, earlier work (15) hadshown that the quantity of enzyme activity pro-duced in C.S.F. was not proportional to the con-centration of streptokinase, but approximatelyproportional to the log concentration of strepto-kinase, and this relationship would do much tominimize errors due to incomplete distribution.

It might be expected that adsorption of fibrino-lytic enzymes onto fibrin would occur in vivo andthus give rise to spuriously low levels of enzymeon sampling. The importance of this factor wasinvestigated by means of in vitro tests with models.

It can be calculated that in the spinal meninges,the maximum ratio of meningeal surface to C.S.F.volume is approximately 4 sq. cm. of surface to1 ml. of C.S.F. The general ratio of meningealsurface to C.S.F. volume is much less than this,and in addition only a part of this surface wouldbe covered with fibrin. Glass tubing of 4 mm.bore provides a surface to volume ratio of 1 sq.cm. per 1 ml. Thirty cm. lengths of 4 mm. glasstubing were coated with fibrin by introducing 0.4ml. of fibrinogen thrombin mixture (containing12 mg. bovine fibrinogen) along the bore of thetube with a fine pipette and rotating the tubesrapidly in the chuck of an electric motor untilclotting had occurred. Assays conducted on sam-ples of streptokinase activated C.S.F. respectivelyincubated in fibrin coated and similar glass tubesrevealed negligible differences over the range 3.0to 0.3 units per ml. plasmin. It was concludedthat over this range of enzyme activity adsorptiononto fibrin in vivo would not increase the samplingerror.

Table V shows that a great disparity exists be-tween the enzyme level calculated on the basis ofin vitro testing and the enzyme level actually foundin vivo. For instance, the first line of the tableshows that in a fluid taken below a spinal block

73

A. P. FLETCHERt

the calculated streptokinase level was 500 unitsper ml. which on in vitro testing should have pro-vided an in vivo level of 2.6 enzyme units per ml.The actual level as determined by in vivo samplingwas < 0.1 units per ml. In cases where largedoses of streptokinase were injected below spinalblocks no free fibrinolytic enzyme was produced,and in fact in cases where large doses of strepto-kinase were shown to be necessary by in vitrotesting the enzyme level was most disappointing.In cases where in vitro testing had indicated thata small dose of streptokinase could be used withsuccess, the disparity between the calculated andthe actual enzyme level found on sampling wasnot great; for instance the bottom line of Table Vshows that the calculated enzyme level with 5 unitsof streptokinase per ml. C.S.F. was 1 unit per ml.,and the actual level found on sampling was 0.8unit per ml. It is apparent that the use of highconcentrations of streptokinase in the spinal thecafrequently results in most disappointing enzymelevels, while the use of low concentrations of strep-tokinase is usually much more successful.

The reasons for this discrepancy are interestingand important. The evidence points strongly tothe fact that streptokinase in high concentrationsexerts an irritative effect upon the spinal thecaresulting in the in-flow of serous fluid with a re-sulting high concentration of inhibitor. Assays infour cases, where free fibrinolytic enzyme had notbeen demonstrated after high streptokinase dos-age, showed that the antiplasmin value had in-creased over 40 per cent in each case. Such anincrease in inhibitor would account for the fact thatfibrinolytic enzyme could not be demonstrated invivo. Though it was realized that fresh plasmino-gen would also enter the C.S.F. with the inhibitor,inspection of the data in Tables I and III showsthat this material would exert comparatively littleeffect. The practical value of this observation isclear; it indicates that the intrathecal injection ofstreptokinase may exert, in the absence of dem-onstrable clinical upset, an irritative effect uponthe spinal theca, which is sufficient to nullify itspotential therapeutic action.

The effect of a single streptokinase injection uponthe C.S.F. plasminogen level

Since the effect of streptokinase is to convertenzyme precursor to enzyme, there may be a fall

TABLE VI

Plasminogen level following streptokinase injection

Initialplasminogen

level

256128128

6432

Level6 hours

afterstreptoklinase

6464442

Level24 hours

afteractivation

128256

323216

in precursor level following an injection of strep-tokinase. Table VI shows the extent of the falland the rate of recovery to the former plasminogenlevel; these observations refer to patients with afree cerebrospinal circulation. It is seen thatthough an early fall in level is apparent, this fallis temporary, and the level has largely returnedto its former state in 24 hours. These findingssuggest that if a period of 24 hours is allowed toelapse between the first and the next streptokinaseinjection, the effect of the last injection will notbe hindered by lack of precursor. In patients withvarious types of cerebrospinal blockage the situ-ation is far from simple, and the results in suchcases have not been tabulated owing to their in-constant pattern and the multiplicity of factors atwork.

DISCUSSION

Streptokinase may be employed under threedistinct circumstances in the therapy of tubercu-lous meningitis, and these circumstances pro-foundly influence the nature of its employment andthe chances of success with its use. These uses are(a) prophylactically before obstruction of thecerebrospinal pathways has occurred, (b) for thetherapy of established or pending subtentorial andbasal cistern block and (c) for the therapy of spinalblock.

Streptokinase has been employed as a prophy-lactic measure in therapy by most workers. Thefindings reported here demonstrate that the rou-tine use of a dose of 100 Christensen units (600Cathie units) streptokinase will fail to producefree enzyme in a substantial proportion of cases.The optimal dose for intrathecal treatment appearsto lie in the range 300 to 3,000 units, with the ma-jority of patients requiring doses at the lower endof the range. Unfortunately the optimal dose canonly be gauged by laboratory assay. It is how-

74

INTRATHECAL FIBRINOLYSIS WITH STREPTOKINASE

ever certainly possible, and probably desirable, touse streptokinase in this manner provided its useis controlled by laboratory assay.

The investigations of Lorber (16) have re-vealed the high incidence and dire consequencesof subtentorial block, and it has also been ourpersonal experience that this complication is re-sponsible for much mortality and morbidity.

Despite the recognition of this complicationclinically, and the possibility of controlling the re-sultant hydrocephalus by ventricular drainage,the end results of treatment are poor. It wouldappear that streptokinase may prove effective inlysing the fibrin producing the block, since the en-zyme can be placed in close proximity to the le-sion by cisternal puncture, and assays on cisternalfluid have shown that lytic activity can be inducedby small doses of streptokinase. The wisdom ofleaving treatment with lytic factors to this stageis doubtful.

The easily diagnosed complication of spinal blockexerts a considerable fascination for those wishingto assess the effect of fibrinolytic therapy owing tothe ease with which the progress of treatment canbe ascertained. Assay of lumbar fluids taken frombelow spinal blocks has demonstrated that theproblem presents peculiar difficulties. The anti-plasmin content of the lumbar fluids ranged from11 to 16 units; the dose of streptokinase which failsto produce free enzyme may be as high as 34 unitsstreptokinase per ml. C.S.F.; the concentration re-quired to produce 1 unit of plasmin per ml. maybe double this figure, while the range of dose foroptimal activation is higher still. In vivo assayhas previously shown that unsatisfactory resultsoccur when dosage of this order is suggested byin vitro testing. The clinical experience may bebriefly summarized by the statement that 25 at-tempts to lyse nine spinal blocks have been madeand only one temporary success resulted. It wasconcluded that streptokinase alone on the basisof in vitro testing and clinical trial is an unsatis-factory agent for the treatment of spinal block.

SUMMARY

Examination of the components of the plas-minogen plasmin system present in the cerebro-spinal fluids of patients suffering from tubercu-lous meningitis suggests that the injection of

streptokinase is capable of inducing useful concen-trations of fibrinolytic enzyme. The concept of alower limit of effective dosage below which, owingto the presence of inhibitors, no free fibrinolyticenzyme is produced, has been advanced. It hasalso been shown that an upper limit of intrathecaldosage is also present, since not only will highconcentrations of streptokinase produce systemicupsets, but the local irritation of the spinal thecawill result in the failure of such therapy. Labora-tory examination of the cerebrospinal fluid is es-sential to the estimation of the correct intrathecaldose.

It has been shown that the customary dose ofstreptokinase used for intrathecal treatment (100Christensen Units) will usually be insufficient toproduce free fibrinolytic enzyme. The variationbetween individual patients is considerable, but theoptimal dose for intrathecal treatment usually liesbetween 300 to 3,000 units of streptokinase; astandard dose cannot be stated. It is believed thatthese experiments indicate the reason for the vari-able clinical results obtained by various workers.

These investigations indicate that streptokinasemay be used with reasonable chances of success,either prophylactically, or for the treatment ofsubtentorial or basal cistern block; the chancesof successful treatment of spinal block by thismeans are small.

ACKNOWLEDGMENTS

I am indebted to Professor Sir Alexander Fleming andProfessor Robert Cruickshank for much encouragementand helpful criticism. I also thank Dr. W. D. W. Brooksfor advice in the clinical management of the cases; Mr.J. Tobar for excellent technical assistance in the labora-tory; and the Director of Lederle Laboratories Divisionfor a generous gift of "Varidase" streptokinase-strepto-dornase, and supplies of "Clotting Globulin."

REFERENCES

1. Cathie, I. A. B., Bacterial fibrinolysin, its possibletherapeutic application in tuberculous meningitis.J. Clin. Path., 1949, 2, 73.

2. Cathie, I. A. B., Streptomycin-streptokinase treatmentof tuberculous meningitis. Lancet, 1949, 1, 441.

3. Cathie, I. A. B., and MacFarlane, J. C. W., Adju-vants to streptomycin in treating tuberculous menin-gitis in children. Lancet, 1950, 2, 784.

4. Eadie, M. B., cited in Scottish fever group. Brit. M.J., 1950, 2, 1050.

75

A. P. FLETCHER

5. Eadie, M. B., Personal communication, 1952.6. Lorber, J., Streptokinase as an adjunct in the treat-

ment of tuberculous meningitis. Lancet, 1951, 1,1334.

7. Tillett, W. S., and Sherry, S., The effect in patientsof streptococcal fibrinolysin (streptokinase) andstreptococcal desoxyribonuclease on fibrinous,purulent, and sanguinous pleural exudations. J.

Clin. Invest., 1949, 28, 173.8. Sherry, S., Tillett, W. S., and Read, C. T., The use

of streptokinase-streptodornase in the treatment ofhemothorax. J. Thoracic Surg., 1950, 20, 393.

9. Tillett, W. S., Sherry, S., and Read, C. T., The use

of streptokinase and streptodornase in the treatmentof chronic empyema. J. Thoracic Surg., 1951, 21,325.

10. Milstone, H., A factor in normal human blood whichparticipates in streptococcal fibrinolysis. J. Im-munol., 1941, 42, 109.

11. Christensen, L. R., and MacLeod, C. M., A proteolyticenzyme of serum: Characterization, activation, and

reaction with inhibitors. J. Gen. Physiol., 1945, 28,559.

12. Sherry, S., Tillett, W. S., and Christensen, L. R.,Presence and significance of desoxyribose nucleo-protein in the purulent pleural exudates of patients.Proc. Soc. Exper. Biol. & Med., 1948, 68, 179.

13. Sherry, S., Johnson, A., and Tillett, W. S., The actionof streptococcal desoxyribose nuclease (streptodor-nase) in vitro and on purulent pleural exudationsof patients. J. Clin. Invest., 1949, 28, 1094.

14. Christensen, L. R., Methods for measuring the ac-

tivity of components of the streptococcal fibrino-lytic system, and streptococcal desoxyribonuclease.J. Clin. Invest., 1949, 28, 163.

15. Fletcher, A. P., Measurement of the components of theplasminogen plasmin system in biological fluids.Biochem. J., 1953.

16. Lorber, J., Studies of the cerebrospinal fluid circula-tion in tuberculous meningitis in children. Part II.

A review of 100 pneumocephalograms. Arch. Dis.Child., 1951, 26, 28.

76