BritishJournul of Huematoloyy, IQ79,41, 563-572

The Determination of Antithrombin 111 by Radioimmunoassay and its Clinical Application

VIVIAN CHAN, T. K. CHAN, VIVIAN WONG, S. C. Tso AND D. TODD

University Department ofMedicine and University Department of Obstetrics and Gynaecology , Queen Mary Hospital, Hong Kong

(Received 15 May 1978; acceptedforpublication 16 August 1978)

SUMMARY. A radioimmunoassay (RIA) had been developed for the determination of antithrombin 111 (AT 111) in man. The detection limit was 25 pg/dl. AT 111-RIA level and biological activity (anti-Xa) was significantly correlated (r=0.737, P<o.ooI). Plasma levels-in 36 healthy males (mean 'I: SD, 19.9k2.5 mg/dl) and 21 healthy females (19.1 k2.4 mg/dl) were similar. Serial A T I11 measurements in normal menstruating females showed lower levels during midcycle and higher concentra- tions during menstruation. In carcinomas, the AT 111 levels were lower than normal, particularly in hepatocellular carcinoma. In cirrhosis of liver, the levels were mar- kedly decreased and in some patients were below that found in congenital AT 111 deficiency. Patients with deep vein thrombosis and patients with heart valve replace- ment had lower levels than normal, while patients with cerebral vascular occlusion had normal levels. The possible use of AT I11 as a diagnostic tool of post-operative deep vein thrombosis was demonstrated in one patient after hysterectomy. The increased sensitivity, specificity and precision of this type of assay offer distinct advantages over existing methods of AT I11 estimation.

Antithrombin 111 (AT III), an a2 globulin, is one of the natural inhibitors of coagulation in blood. Besides inhibiting thrombin (Abildgaard, 1967) and factor Xa (Yin et a l , 1971a), it is also capable of inactivating other proteolytic enzymes (XIIa, XIa and IXa) involved in the intrinsic pathway of blood coagulation (Harpel & Rosenberg, 1976). It is likely that AT 111 plays an important function in modulating these activated proteases produced during intravascular coagulation. Its anti-Xa activity is markedly accelerated by a low concentration of heparin (Yin et a l , 1971b) and this is thought to be responsible for the efficacy of small dose subcutaneous heparin regime in preventing venous thrombosis (Kakkar et at, 1971; Interna- tional Multicentre Trial, 1975). Although AT 111 only accounts for approximately 50% of the total antithrombin activity in plasma, and a1 antitrypsin, ct2 macroglobulin each contribute about 25% of the total (Lane et al, 1975), a decrease in AT 111 in congenital deficiency predisposes such patients to thrombosis (Egeberg, 1965). Furthermore, the finding of a low pre-operative AT 111 activity in patients who subsequently developed post-operative venous thrombosis (Sagar et al, 1976a, b) suggests the important regulatory role of this protein.

Current methods for the measurement of AT I11 depend on either its various biological Correspondence: Dr Vivian Chan, University Department of Medicine, Queen Mary Hospital, Hong Kong.

0007-1048/79/0400-0563$oz.oo 0 1 9 7 9 Blackwell Scientific Publications

563

564 Vivian Chan et a1

activities or the immunoreactive protein concentration. In view of the multiplicity of its biological activities and the overlapping effect of other protease inhibitors, it is not surprising that conflicting reports on the changes in AT 111 levels in various physiological and disease states have been reported (Lane & Biggs, 1977). Additionally, most biological assays measure the inhibitory function of the protein on thrombin (von Kaulla & von Kaulla, 1967; Yue et al , 1974) or factor Xa (Gitel & Wessler, 1975), which is time-dependent and the subsequent quantitation of the residual clotting activity of the substrate is prone to inherent errors. The recent introduction of synthetic chromogenic substrates to measure residual thrombin (Ode- gard et al, 1975) or Xa (Odegard et al , 1976) activity is a definite improvement, but these methods still lack specificity and are time-dependent, thus preclude the handling of large numbers of samples.

The measurement of AT 111 protein concentration by immunological methods is more specific. It is possible that the immunoreactive protein lacks biological activity, but in most studies with congenital AT I11 deficiency (Filip et al , 1976) and in various pathological states (Abildgaard et a l , 1970) these two parameters are closely related. A notable exception is the rare qualitative AT 111 deficiency (AT 111 'Budapest') reported in a Hungarian family (Sas et at, 1974). Immunodiffusion (Mancini et al , 1965) and Laurell rocket immunoelectrophoresis (Laurell, 1966) require high antibody concentration, lack sensitivity and the AT 111 levels measured are affected by complex formation with heparin (Andersson et al , 1977). A specific, sensitive and precise radio-immunoassay (RIA) should overcome these problems.

We report here the development of the first RIA for AT 111, preliminary results of A T 111 measurements in health and diseases and its possible application in the diagnosis of post-opera- tive deep vein thrombosis.

MATERIALS AND METHODS

Highly purified AT I11 (NIBSC 75/564) was kindly supplied by the National Institute of Biological Standards and Control, London, and used for standardization and iodination. One ampoule of standard (500 @g) was dissolved in assay diluent (1.0 mlo.06 M Barbital buffer pH 8.6 with 0.1% bovine serum albumin), divided into 100 p1 aliquots and stored at -2o'C. This stock solution was further diluted with assay diluent to give AT 111 standards of 0.8, 0.4, 0.2,

0.1,o.o~ and 0.02s mg/dl.

Preparation O J [ ~ ~ ~ ~ ] A T III Radioiodination was performed with Na1251 (Radiochemical Centre, Amersham) using a

modified solid-phase lactoperoxidase method (Karonen et al, 1975). 2.5 pg of AT 111 dissolved in iodination buffer (30 pl, 0.5 M phosphate buffer, pH 7.4) was reacted with diluted solid-phase lactoperoxidase (10 pl of I:IO dilution), NaiZ5I (I mCi) and hydrogen peroxide ( 5 p1 0f1:103

diluted solution) in a small autoanalyser cup. The mixture was shaken continuously for 30 min and the reaction stopped by the addition of buffer (200 p1 of 0.1 M phosphate buffer). The reaction mixture was centrifuged at 2500 rpm for 5 min and the supernatant removed for purification on a Sephadex G-IOO column (so x I cm). Phosphate buffer (0.01 M, pH 7.4) was used as eluant and 500 pl amounts of the eluate were collected into small polystyrene tubes at a flow rate of 0.5 ml/min.

Antithrombin ZZI 565

Preparation of Antiserum Specific-anti-AT 111-sera were raised in rabbits by multiple intradermal injections of small

amounts of AT 111 emulsified with Freund's complete adjuvant. 6 weeks later, a booster injection was administered in a similar manner. Antisera harvested 8 weeks after primary immunization were used in Laurel1 crossed immunoelectrophoresis of plasma and serum samples and showed a single peak in the a, globulin region. The avidity of the antiserum for AT 111 was further characterized by antibody dilution curves. Cross-reaction studies of the antiserum with a antitrypsin and clP macroglobulin were also performed. These were measured as the ratio of the amount of AT 111 required to displace 50% of the 1251-AT 111 from the antibody, to the amount of a1 antitrypsin of a2 macroglobulin which also gave 50% displacement.

Sample Collection A 4.5 ml blood sample was collected into 0.5 ml of 3.8% sodium citrate from 57 healthy

adult volunteers (36 males and 21 females), 17 patients with cirrhosis of liver, 13 patients with hepatocellular carcinoma, I I patients with other carcinomas, seven patients with cerebral vascular occlusion, 20 patients after heart valve replacement (1-5 years), 12 patients with deep vein thrombosis, and one patient with congenital AT 111 deficiency. Serial sampling was made on 12 patients undergoing hysterectomy (none of whom were given prophylactic heparin), for 2 d before and 5 d after surgery. In the patient with venous thrombosis, an extra sample was taken on day 10. In addition, sampling was made for 3 0 consecutive days in two healthy females. The samples were centrifuged 2500 rpm for 10 min and plasma stored at - 70°C until required for assay. All samples from the same subject were run in one assay to avoid inter-assay variation.

Equal amounts of fresh, citrated plasma from 36 normal male subjects (mean age 30 years) were mixed and stored in 500 p1 amounts at - 70°C. This comprised the control pool plasma.

Assay Procedures In the RIA procedure, duplicate 50 pl aliquots of diluted plasma (I/XOO dilution with assay

diluent-0.06 M Barbital buffer, 0.1% bovine serum albumin, pH 8.6) or standard were incubated with [lZ51]AT 111 (10 ooo cpm) and antiserum (final dilution of 1/170 ooo), at a total incubation volume of 500 p1 in a small polystyrene tube (LP3, Luckham Ltd, England). Duplicate control tubes were set up with each standard curve, containing [lZ5I]AT 111 and diluent (diluent blank) or [ lP5I]AT 111, antibody and diluent (zero standard). After overnight incubation at 4"C, the antibody-bound and free AT I11 were separated using the second antibody technique. Sheep anti-rabbit IgG serum ( 5 0 pl of 1/10 dilution, Burroughs Well- come) and normal rabbit serum (so pl of 1/40 dilution) were added and incubated overnight a t 4'C, before centrifugation a t 2500 rpm for 40 min. The supernatant (containing free AT 111) was aspirated, and the antibody-bound AT I11 fraction in the precipitate was counted for 2 min in an auto-gamma counter.

The biological anti-Xa activity in 38 plasma samples was measured according to the method of Gitel & Wessler (1975) using commercially available reagents (Sigma Co., U.S.A.). The activity of each sample was expressed as a per cent of the activity of the control pool plasma.

566 Vivian Chan et a1

[1251]fibrinogen leg scan was performed on all patients undergoing hysterectomy, as described by Kakkar et al (1970).

RESULTS

Assessment of [ 12511AT III Chromatographic purification of the iodination mixture on Sephadex G-roo resulted in

two peaks of radioactivity (Fig I ) . The immunoreactive material in the first peak represented 1251-labelled AT 111.90% of this labelled peptide was bound in the presence of excess antibody even after correction for non-specific binding in the diluent blank tubes. The second peak represented the unreacted 1251.

Assessment of Antisera Antibody titre rose progressively between 8 and 16 weeks after primary immunization. On

two successive bleeds the highest titres which gave 50% binding of [1251]AT I11 were 1:170 ooo and 1:230 ooo respectively (Fig 2). The antiserum exhibited insignificant cross-reaction with a I antitrypsin and a2 macroglobulin (0.0001 YO and o.oooz% respectively).

Sensitivity, Specijicity and Precision Under the present conditions, the working range of the RIA was 12.5-400 ng AT 111 per

tube and the detection limit was 2 s pg/dl. Doubling dilutions of plasma produced inhibition curves parallel to that caused by serial dilutions of unlabelled standard AT 111 (Fig 3 ) . Precision was studied by performing 10 analyses on the control pool plasma on the same and on seven

.. FRACTION NUMBER

FIG I . Chromatographic purification of 1251-labelled AT I11 using Sephadex G-loo. The distribution of radioactivity (A) and the percentage binding in the presence of excess antibody (0) are shown. The first peak ofradioactivity was due to the [1251]AT 111 and the second peak represented unreacted 1251.

Antithrombin I I l 567

2 0 Ilj, I I I I 1

1:5 1:lO 1820 1i40 1880 1,160 1:320 1:640 1~1280 x 103

ANTIBODY DILUTION

FIG 2. Antibody dilution curve from sera harvested on two separate occasions. Gradual increase in antibody titre was noted from the later bleed (7-6).

1 I-

FIG 3 . Parallel inhibitions of doubling dilutions of three plasma samples with the AT I11 standard curve (0).

different days. The coefficients of variation for the ‘within’ and ‘between’ assay precision were 2.0% and 6.7% respectively.

Correlation of Immunological and Biological A T I I I Activities The relationship between radioimmunological AT 111 concentration and its anti-Xa activity

was shown in Fig 4. For ease of comparison, the RIA results were expressed also as per cent of

Vivian Chan et al

. /

I

6 100

z 0 E t /.

.. “ I -/: :

r - 0.737 p < 0.001

I I

0 50 100 150 ANTI - Xa ( X CONTROL POOL )

FIG 4. Correlation between radioimrnuriological AT 111 concentration and its anti-Xa (biological) activity.

the control pool plasma. The correlation between the two parameters was significant (rz0.737, P < 0.001).

AT I I I in Health and Diseases Fig 5 shows plasma AT I11 levels in the various groups of subjects. The levels in 36 healthy

males (mean +SD, 19.9+2.5 mg/dl) were similar to those in 21 healthy females (1y.1k2.4 mg/dl). Patients with cirrhosis of liver had low levels (9.6h2.4 mg/dl) and in nine of them the levels were below 10 mg/dl. Those with carcinoma had significantly reduced levels (15.8 k0.9 mg/dl, P<O.OOI) and three out of 13 patients with hepatocellular carcinoma had levels below 10 mg/dl. Patients with active deep vein thrombosis had low levels (15.8k1.9 mg/dl, P<O.OOI) which were significantly different from normal subjects. The level in one patient with congenital AT 111 deficiency was 9.6 mg/dl. Levels were normal in patients with cerebral vascular occlusion (19.1 k2.2 mg/dl) but significantly lower in patients with heart valve replacemant (16.5 ho.4 mg/dl, P<o.ooI), although all were within the normal range. Serial AT 111 concentrations in two healthy females over a period of 3 0 consecutive days were marked by a transient increase during menstruation (35.9% and 55.8% above the mean respectively) and a decrease during midcycle (24.0% and 18.5% below the mean respectively).

A T 111 in Post-operative Deep Vein Thrombosis The changes in AT 111 level of patients before and after hysterectomy are shown in Fig 6. The

results are represented as per cent deviation from the patient’s own mean basal value and for the group who did not thrombose after surgery their mean f SD AT 111 changes are shown, In these I I patients AT 111 concentrations fluctuated within 10% throughout the post-operative period. Results of [‘251]fibrinogen leg scan were negative in all I I cases. The remaining patient

Antithrombin I11

1

;. .. .. - ..

.*' ... a:::.

:

.- .. -. *.:

*.

..

..

2 h

A 1 :

0

.. 7

NS pq.001

N O R M A L N O R M A L CIRRHOSIS CARCINOMA D V T C V A H V R MALES FEMALES OF LIVER LIVER 6 ACTIVE A

OTHERS 0 AT ill DEF.0

FIG 5. AT 111 levels in various groups of subjects. The mean level in each group is represented by a horizontal bar and P values indicate significant difference from normal male subjects using Student's t test. o=patient with congenital AT 111 deficiency; I>VT=deep vein thrombosis; CVA =cerebral vascular occlusion; HVR= prosthetic heart valve replacement.

L " " ' 1 ' " ' I -1 0 1 2 3 4 5 6 7 8 9 10

D A Y

FIG 6. The mean changes in AT 111 levels of I I patients before and after hysterectomy (0) are shown and the shaded area represents +SD from the mean. Levels of the one patient who developed post-operative deep vein thrombosis (0) are shown separately.

5 70 Vivian Chan et a1

exhibited a continuous decrease in AT 111, reaching a minimum of - 17% on the fourth day after surgery. The level then gradually returned to basal value by the tenth day. This prolonged decrease in AT I11 concentration was associated with a positive [1251]fibrinogen leg scan.

DISCUSSION

We describe here, for the first time, a RIA for the measurement ofAT 111. Earlier methods were insensitive, time-dependent or lacked accuracy. The present system has various distinct advantages. It offers greater sensitivity, with a detection limit of 25 pg/dl as compared to the milligram range in immunodiffusion. Timing is not critical, which is different from the clotting or chromogenic methods. The methodology is simple and a high degree of ‘within’ and ‘between’ assay precision was easily achieved.

Enzymatic iodination of AT I11 was preferred because of the iodination damage induced by the chloramineT procedure. The *251-labelled AT 111 was stable for up to 6 weeks when stored at - zo°C and the sensitivity of the assay could be improved by delayed addition of radioactive tracer, after a 2 h pre-incubation period.

The preparation of antisera of high titre presented no problem since all the rabbits immunized produced antisera suitable for use in RIA after 10 weeks.

Actual quantitation of AT I11 was made in the RIA using NIBSC 75/564 as standard, where each ampoule contained 500 pg AT 111. In healthy male subjects, the mean AT I11 level was 19.9+2.5 mg/dl, which is lower than that previously reported (Heimburger et al, 1971). This probably reflects the purity of the AT 111 standard and in all RIA using unextracted samples (direct assay), the lower values simply indicate increased specificity of the antiserum (Chan, 1974).

The previous finding of low AT I11 levels in patients with cirrhosis ofliver (Abildgaxd et al , 1970), which in many instances were lower than that found in congenital AT 111 deficiency, was confirmed in this study. In carcinoma, AT 111 levels were lower than normal and this would be compatible with the known increased risk of thrombosis in this condition. The finding of very low levels in some patients with hepatocellular carcinoma is of interest and probably indicates the degree of hepatic dysfunction. Patients with deep vein thrombosis had a lower mean AT 111 level than the normal subjects, and in two patients with active thrombosis the AT 111 levels were below the normal limits. The importance of AT I11 in arterial thromboembolisrn is not established. In this study, patients with cerebral vascular occlusion had normal levels while patients with prosthetic heart valve replacement had decreased levels.

The changes in AT 111 throughout the menstrual cycle are reported for the first time. These changes may be associated with alterations in circulating oestrogen concentration, since decreased AT 111 levels were observed during midcycle, when oestrogen level was high, and a transient increase occurred at menstruation, when oestrogen was low. This variation in A T 111 level at different phases of the menstrual cycle should be considered when changes in AT 111 in women before and after oral contraceptives are compared.

There are conflicting reports on the usefulness of serial AT 111 measurements in the diagnosis of post-operative deep vein thrombosis (Aberg et at, 1973; Yin et al, 1977). The lack of precision of previous methods of AT 111 estimation may account for this confusion. With the range of sensitivity and precision of the RIA, more meaningful results could be obtained in serial

Antithrombin 1Il 571

determinations. We have shown, in a small number of post-operative patients, that a sustained fall in AT 111 was associated with venous thrombosis. Further work in a larger series of patients is required to confirm this finding. This application of serial AT I11 measurements would be preferred to the [1251]fibrinogen leg scan and phlebography because no in uivo administration of radioisotope is necessary and, also avoids the discomfort of the latter procedure.

ACKNOWLEDGMENTS

We thank the National Institute of Biological Standards and Control, London, for the supply of NIBSC 751564, Dr R. Gordon, St Bartholomew's Hospital, London, for advice on iodination techniques, Professor A. S. Douglas, University of Aberdeen, for providing the plasma sample from a congenital AT 111 deficient patient and the International Development Research Council, Canada for financial support.

REFERENCES

ABERG, M., NILSSON, I.M. & HEDNER, U. (1973) Antithrombin 111 after operation. (Letter). Lancet, ii, 1337.

ABILDGARRD, U . (1967) Purification of two progres- sive antithrombins of human plasma. Scandinavian

journal of Clinical and Laboratory Investigation, 19, 190-195.

ABILDGAARD, U., FAGERHOL, M.K. 81 EGEBERG, 0. (1970) Comparison of progressive antithrombin activity and the concentrations of three thrombin inhibitors in human plasma. Scandinavian journal of Clinical and Laboratory Investigation, 26, 349-3 54.

ANDERSON, L.-O., ENGMAN, L. & HENNINGSSON, E. ( I 977) Crossed immunoelectrophoresis as applied to studies on complex formation. The binding of heparin to antithrombin 111 and the antithrombin I11 thrombin complex. journal of Immunological Methods, 14, 271-281.

CHAN, V. (1974) The assay of urinary thyroid hor- mones for assessing thyroid function. Annals o j Clinical Biochemistry, 11, 120-129.

EGEBERG, 0. (1965) Inherited antithrombin deficiency causing thrombophilia. Thrombosis et Diathesis Hae- morrhagica, 13, 516-530.

FILIP, D.J., ECKSTEIN, J.D. & VELTKAMP, JJ. (1976) Hereditary antithrombin I11 deficiency and throm- boembolic disease. American journal of Haematology , 2,343-349.

GITEL, S.N. & WESSLER, S. (1975) Plasma antithrom- bin Ill: a quantitative assay of biological activity. Thrombosis Research, 7, 5-16.

HARPEL, P.C. & ROSENBERG, R.D. (1976) a,-Macro- globulin and antithrombin-heparin cofactor: modulators of haemostatic and inflammatory reac- tions. Progress in Haemostatis and Thrombosis (ed. by T . H. Spaet), Vol. 3. pp. 145-189. Grune & Strat- ton, New York.

HEIMBURGER, N., HAUPT, H. 81 SCHWICK, H.G. (1971) Protease inhibitors of human plasma. Proceedings of the lnternational Research Conference on Proteinase Inhibitors, Munich, I970 jed. by H. Fritz and H. Tschesche), pp. 1-22. De Gruyte, Berlin.

INTERNATIONAL MULTICENTRE TRIAL (1975) Preven- tion of fatal postoperative pulmonary embolism by low doses of heparin. Lancet, ii, 45-51,

KAKKAR, V.V., FIELD, E.S., NICOLAIDES, A.N., FLUTE, P.T., WESSLER, S. & YIN, E.T. (1971) Low doses of heparin in prevention of deep-vein thrombosis. Lancet, ii, 669-671.

KAKKAR, V.V., NICOLAIDES, A.N., RENNEY, J.T.G., FRIEND, J.R. & CLARKE, M.B. (1970) '251-labelled frbrinogen test adapted for routine screening for deep-vein thrombosis. Lancet, i, 540-542.

KARONEN, S.L., M O R S K Y , ~ . , SIREN,M. & SENDERLING, U. (1975) An enzymatic solid-phase method for trace iodination of proteins and peptides with 125iodine. Analytical Biochemistry, 67, 1-10.

LANE, J.L. & BIGGS, R. (1977) The natural inhibitors of coagulation: antithrombin 111, heparin cofactor and antifactor Xa. Recent Advances in Blood Coagulation (ed. by L. Poller), Vol. 2 , pp. 123-139. Churchill Livingstone, Edinburgh.

LANE, J.L., BIRD, P. & RIZZA, C.R. (1975) A new assay for the measurement of total progressive anti- thrombin. British Journal of Haematology, 30,

LAURELL, C.B. (1966) Quantitative estimation of pro- teins by electrophoresis in agarose gel containing antibodies. Analytical Biochemistry, 15,45-52.

MANCINI, G., CARBONARA, A.O. & HEREMANS, J.F. (1965) Immunochemical quantitation of antigens by single radial immunodiffusion. Internationaljour- nal oflmmunochemistry, 2, 235-254.

103-1 1 5 .

O D E G A R D , O.R., LIE, M. & ABILDGAARD, u. (1975)

5 72 Vivian Chan et al

Heparin cofactor activity measured with an amido- lytic method. Thrombosis Research, 6, 287-294.

ODEGARD, O.R., LIE, M. & ABILDGAARD, U. (1976) Antifactor Xa activity measured with amidolytic methods. Haemostasis, 5 , 265-275.

SAGAR, S. , NAIRN, D., STAMATAKIS,J.D., MAFFEI,F.H., HIGGINS, A.F., THOMAS, D.F. & KAKKAR, V.V. (1976a) Efficacy of low-dose heparin in prevention of extensive deep-vein thrombosis in patients un- dergoing total-hip replacement. Lancet, i,

SAGAR, S., STAMATAKIS, J.D., THOMAS, D.P. & KAK- KAR, V .V. (1976b) Oral contraceptives, antithrom- bin-I11 activity, and postoperative deep-vein thrombosis. Lancet, i, 509-5 I I .

SAS, D., B L A S K ~ , G. , B ~ N G H E G Y I , D., J A K ~ ) , J. & PALOS, L.A. (1974) Abnormal antithrombin 111 (antithrombin ‘Budapest’) as a cause of a familial thrombophilia. Thrombosis ei Diathesis Huemorrha- gica, 32, 105-115.

1151-1 154.

VON KAULLA, E. & VON KAULLA, K.N. (1967) Anti- thrombin 111 and diseases. Ameriranjoumal of Clini-

YIN, E.T., BYGDEMAN, S., WERLIN-BERGER, T . & TANGEN, 0. (1977) A new differential assay for plasma XaI (Antithrombin 111). Biological activity; predictor of post-operative DVT? First Florence Conference on Haemostatis and Thrombosis.

YIN, E.T., WESSLER, S. & STOLL, P.J. (1971a) Identity of plasma-activated factor X inhibitor with anti- thrombin 111 and heparin cofactor.journal ofBiologi- cal Chemistry, 246, 3712-3719.

Y l N , E.T., WESSLER, S. & STOLL, P.J. (197Ib) Biologi- cal properties of the naturally occurring plasma inhibitor to activated factor X. Journal of Biological Chemistry, 246,3703-3711,

YUE, R.H., STARR, T . & GERTLER, M.M. (1974) The rivanol method for the quantitative determination of antithrombin I11 in plasma. Thrombosis et Diatlr- esis Haemorrhagica, 31. 439-451.

cal Pathology, 48, 69-80.