Immunization by bovine thrombin used with fibrin glue during cardiovascular operations

Immunization by bovine thrombin used with fibrin glue during cardiovascular operations Development of thrombin and factor V inhibitors Brief case histories of three patients aged 58, 38, and 44 years are reported. AU underwent cardiovascular operations. Subsequently hemostasis test abnormalities developed between the seventh and eighth postoperative days after exposure to bovine thrombin used with fibrin glue. These were characterized by an increased activated partial thromboplastin time (64 to 147 seconds), prothrombin time (19 to 24 seconds), bovine thrombin time (> 120 seconds) and a markedly reduced factor V level «10 % in two patients and 16% in the third patient). A patient plasma dilution of 1 in 200 with a normal plasma pool was necessary to correct bovine thrombin time. No fast-acting or progressive inhibitor against factor V could be detected by coagulation tests, and fresh frozen plasma perfusion had no effect. Plasmapheresis was performed preventatively to avoid bleeding, and factor V levels stabilized at around 50 % after two to four exchanges. Immunologic studies showed that the inhibitors were directed not only against bovine factors but also against human ones. Therefore factor V decrease could have been the result of rapid clearance from the circulation of complexes formed with a nonneutralizing inhibitor that is not detected by clotting tests. These antibodies were purified by standard methods and immunoaffinity. Fast immunization could be explained by a prior sensitization to bovine thrombin exposure during previous operations. It is suggested that bovine thrombin used with fibrin glue contains small amounts of factor V and may be responsible for these abnormalities. This is in agreement with previous literature reports. However, these described neutralizing factor V inhibitors, which were easily detected. (J THORAC CARDIOVASC SURG 1993;105:892-7)

Micheline Berruyer, BS,a Jean Amiral, PhD,b Patrick Ffrench, MD,a Jean Belleville, MD,a Olivier Bastien, MD,c Jean Clerc, MD,c Alain Kassir, MD,d Susanne Estanove, MD,c and Marc Dechavanne, MD,a Lyon, France

Erin glue (fibrin sealant) is a new technologic advance used for reducing the risk of bleeding during operations.' It has been used in a variety of surgical fields: cardiovascular, thoracic, transplantation, head and neck, oral, gastrointestinal, orthopedic, neurosurgical, and plastic surFrom Laboratoire d'Hematologie et INSERM U 331," Departernent d'Anesthesie Reanimation," Centre Regional de Transfusion Sanguine.? Hopital Cardio-vasculaire et Pneumologique Louis Pradel, RP. Lyon Montchat 69394 Lyon Cedex 03 and SERBIO,b Genevilliers, France. Received for publication March 6, 1992. Accepted for publication June 11, 1992. Address for reprints: Micheline Berruyer, BS, Laboratoire d'Hematologie, H6pital Cardio-vasculaire et Pneumologique Louis Pradel, B.P. Lyon Montchat 69394 Lyon Cedex 03, France. Copyright © 1993 by Mosby-Year Book, Inc. 0022-5223/93 $1.00

892

+ .10

12/1/40169

gery. The glue is prepared by mixing bovine thrombin not only with human coagulant proteins such as fibrinogen, fibronectin, factor XIII, and plasminogen but also with bovine aprotinin and calcium chloride. The formed clot adheres within seconds and is slowly reabsorbed in approximately lO days by fibrinolysis. Major complications of fibrin glue products appear to be rare, but anaphylactic reaction has been described.s 3 Bovine thrombin has also been reported to be responsible for induction of thrombin inhibitors, especially when glue is used during cardiac operations.r" The biologic abnormalities observed are characterized by a very prolonged thrombin time (> 120 seconds), and mixing studies suggest the presence of thrombin inhibitors in patient plasma. When inhibitors are directed against bovine thrombin, they are not usually responsible for hemorrhagic complications unless they coexist with factor V inhibitors.v"

The Journal of Thoracic and Cardiovascular Surgery Volume 105, Number 5

In this study we report three patients in whom bovine thrombin inhibitors along with a severe decrease of factor V developed. The presence of an inhibitor against factor V could not be demonstrated by coagulation methods, and only enzyme-linked immunosorbent assay (ELISA) testing was able to show it. Plasmaphereses were successfully practiced and restored the factor V activity, but the thrombin inhibitor still remained.

Materials and methods Coagulation studies. Coagulation studies were performed according to the manufacturer's recommendations with the Amelung KC 10 coagulometer: activated partial thromboplastin time (aPTI) with Automated aPTI from Organon Teknika, Boxtel, The Netherlands; prothrombin time (PT) with Automated Neoplastin; reptilase time with reptilase; D-dimer by latex agglutination test with the D-Di-Test from Diagnostica Stago, Asnieres, France; and fibrinogen with Fibrinomat from Biomerieux, Marcy-I'Etoile, France. Strictly human deficient plasmas for clotting factor assays (Immuno AG, Vienna, Austria) were used for determining the activity levelsof the different factors. This point was important to avoid confusing the interpretation of coagulation factor assays based on bovine substrate plasmas." Thrombin time was routinely measured using bovine thrombin (3.3 IUjm!), from Hoffmann-La Roche & Co., Basel, Switzerland, or human thrombin (Fibrindex) (2.5 IU jm!) from Ortho Diagnostic Systems, Raritan, N.J. A rapid- or slow-acting inhibitor against factor V was looked for. Patient plasma was mixed with normal plasma (I: I) at room temperature or 37° C and tested immediately or after 30 minutes' incubation. A slow-acting inhibitor was also looked for after 2 hours' incubation with use of the same technique as described for factor VIII inhibitor testing.!" Thrombin inhibition by antithrombin III with or without heparin was performed as has already been described. I I Human thrombin (1.33 IU jml), from Diagnostica Stago, and chromogenic substrate S 2238 (2.5 IlmoljL), from KabiVitrum, Stockholm, Sweden, were used. Enzyme immunoassay for antiplasma protein antibodies. The presence of specific antibodies to coagulation proteins developedin patients was analyzed by ELISA. Capture antigens tested were bovine thrombin, human thrombin, prothrombin, factor V (Diagnostica Stago), and the bovine thrombin that was used in topical bovine thrombin (Tissuco!) Immuno AG, Vienna, Austria). These proteins (200 III at 10 Ilgjml in a phosphate buffer, 0.05 mmoljL; at pH 7.5) were coated overnight on type I micro-ELISA plates, flat bottom, purchased from Nunc, Roskilde, Denmark. After being washed, 200 III of test plasma, diluted I: I00 in dilution buffer (phosphate, 0.05 mmoljL sodium chloride, 0.15 moljL, 0.1% polysorbate 20 [Tween 20] at pH 7.5) was placed in the microwells and incubated; after being washed again, captured antibodies were detected by introducing 200 III of the second antibody, anti-IgGAM peroxidase conjugate at 2 Ilgjml in dilution buffer. Alternatively, specific determination of IgG or IgM antibody was done by using either anti-IgG or anti-IgM peroxidase from Biosys, Compiegne, France. Two hundred microliters of conjugate at 2.5llgjml was used. After incubation, followed by washing, color development was obtained with the substrate a-phenylenediamine (0.4 mgjml in phosphate citrate buffer, 0.05 moljL, at pH 5.0, and

Berruyer et al.

893

containing 0.03% hydrogen peroxide) for 3 minutes; the reaction was then stopped with sulfuric acid (3 moljL) before proceeding to the measurement of absorbance at 492 nm. All incubations, except for coating, were performed for 2 hours at room temperature. Sodium chloride (0.15 moljL) containing 0.1% polysorbate 20 was used as washing solution. At the same time, control experiments were run in a similar way on noncoated plates or plates coated with a neutral protein (such as protein C). Bidimensional electrophoresis offactor V. The first migration gel was prepared as follows: 1% agarose A 37, supplied by IBF, Villeneuve la Garenne, France, was melted by heating in a trisodium citrate buffer (10.3 gmjL), pH 8.8, supplemented with hirudin 0.1 ATU* j ml. Gel (12 m!) was poured on a 100 X 85 mm electrophoresis support and let to cool at room temperature. At the cathode side down corner (1.5 ern from edges) a weill 0 mm in diameter was punched out. The plate was placed in an electrophoresis chamber and connected to the migration buffer (sodium veronal 10.3 gmjL, trisodium citrate 1.45 gmjL, hirudin 0.1 ATUjml, pH 8.8) with Whatman No. 3 paper wicks, presoaked in the same buffer. Tested plasma (100 Ill) was introduced into the well, and the electrophoresis was run for 2 hours at 2.5 mAjcm, under refrigeration. During this time the second migration gel was prepared by heat melting 1% agarose in sodium veronal (2.57 gm/L), Na2 ethylenediaminetetraacetic acid (1.30 gmjL), and hirudin buffer (0.1 ATU jm!) at pH 8.8. The gel was let to cool in a water bath kept at 56° C. When the first migration was over, the deposition well was emptied, washed three times with buffer, and the plate was removed from the electrophoresis chamber. The part of the gel not containing the electrophoresed proteins was removed. Rabbit serum (2%) specific for human factor V (Assera V, Diagnostica Stago) was then added to the second migration gel maintained at 56° C and mixed. This gel was poured on the plate to replace the removed gel and let to cool. Then it was transferred to the electrophoresis chamber and connected to the migration buffer (sodium barbital, 2.57 gmjL, Na2 ethylenediaminetetraacetic acid, (1.30 gmjL) hirudin 0.1 ATU jml, at pH 8.8) with Whatman No.3 paper wicks soaked in this buffer. The electrophoresis was continued overnight at 2 mAjcm. After the migration, the plate was washed in physiologic saline, then in distilled water, covered with wet filter paper pads, and dried in an oven. It was stained with Coomassie Brillant Blue and destained in a mixture of 45% ethanol, 5% acetic acid, and 50% distilled water. Factor V and prothrombin antigens assays. Factor V and prothrombin were measured by electroimmunoassay as described by Laurell.I? Migration conditions were identical to those of the second migration of bidimensional electrophoresis of factor V. Antiprothrombin or antifactor V rabbit serum (Assera II or V, Diagnostica Stago) was used at a 1% concentration. For factor V antigen determination, 15111 oftest or control plasma was introduced into a 4 mm diameter well. For factor II antigen, the volume was 5 III with a 1:2dilution of plasma, diameter was 3 mm, and the migration gel was free of hirudin. Rocket height of a test plasma was converted to percent factor V or prothrombin antigen from a reference curve obtained with serial dilutions of a normal pooled plasma electrophoresed on the same plate.

*ATU

=

Antithrombin units.

The Journal of Thoracic and Cardiovascular Surgery May 1993

8 9 4 Berruyer et al.

Table I. Coagulation test parameters in three patients on the day of diagnosis (8 to 11 days after operation andtopkalbowneexposur~

Patient I

Platelet count (109/L) PT (sec) aPTT (sec) BIT (sec) BIT + protamine (I mg/rnl) BIT inhibitor titer HIT (sec) Reptilase time (sec) Fibrinogen (gm/L) D-Dimer (mg/L) Factor V level (%) Activity Antigen Neutalizing factor V inhibitor Prothrombin level (%) Activity Antigen

Patient 2 Patient 3

Normal range

120-400

215 24 64 >120 >120

280 26 147 >120 >120

160 19 70 >120 >120

1/200 26 18 5.3 8

1/200 64* 20 5.6 <0.5

1/200 54* 20 6.7 4

<10 <12.5 Neg

<10

16 12.5 Neg

60-120 60-120

58 65

60-120 60-120

89 130

Neg 65

II

30-36 21 18 25 16 1.8-4 <0.5

BIT inhibitortiter wasdeterminedby mixingpatient plasmawitha normalplasma poolin differentratios.The ratio that gives the firstnormalized BIT wasconsideredas the titer. PT, Prothrombintime; aPTT. activatedpartial thromboplastin time; BTT, bovinethrombin time; HTT, human thrombintime.-, not done. 'Patients 2 and 3 received heparin. HIT normalized as for patient I whenheparin was stopped.

Case reports CASE I. A 59-year-old white man was in good health until December 1989, when a type I aortic dissection developed. He underwent partial aortic replacement and received fibrin glue with 11,000 V of topical bovine thrombin (Tissucol) perioperatively. In November 1990 he underwent reoperation for extensive aneurysm of the aortic arch and of the ascending aorta, and 10,000 V of bovine thrombin (Tissucol) was used during the operation. Preoperative hemostasis parameters were normal, but on the seventh day clotting times suddenly became prolonged: PT, 18 seconds; aPIT, 66 seconds; and bovine thrombin time (BTT) greater than 120 seconds, associated with a decrease in factor V (12%). On day 9 the coagulation tests were still disturbed (Table 1), and a neutralizing inhibitor against bovine but not against human thrombin was found. Moreover, mixing studies with use of control and patient plasma (I: I) showed a normalization of PT (12 seconds) and offactor V levels (80%). The same results were obtained with and without 30 minutes' incubation at room temperature or at 37° C. No progressive inhibitor against factor V was detected. Factors II, VIII, IX, X, XI, and XII levels were normal, and factor VII level was slightly decreased (48%). Perfusion of I L of fresh frozen plasma had no effect on any of the coagulation tests. On day II the same disturbances persisted; there were no bleeding complications, but clinically there was still a hemorrhagic risk. Therefore the patient underwent plasmapheresis, exchanging 1.5 times the plasma volume per day for 4 days in an attempt to correct the coagulation disorders. On day 15, 24 hours after the

end of plasmapheresis, the BIT inhibitor titer was I:20 versus I :200 at the beginning; factor V level was 56% and remained stable the next 15 days. It then normalized, and no further decrease was observed, but BIT remained increased (> 120 seconds). On the ninetieth day factor V level was 92%, and BIT was still prolonged (> 120 seconds). No hemorrhagic or thrombotic complication had occurred. CASE 2. Our second patient was a 38-year-old white man with a history, since 1983, of many valve replacements secondary to infectious endocarditis. In April 1989 he underwent his fifth aortic valve replacement and received 2500 V bovine thrombin with fibrin glue (Tissucol). In November 1990, he received his sixth aortic valve replacement with an Omnicarbon valve (Medical Inc., Inver Grove Heights, Minn.) and a coronary artery bypass graft was also performed. He again received 10,000 V of bovine thrombin with fibrin glue (Tissucol) perioperatively and needed postoperative circulatory assistance until day 6. On the second day he was reoperated on for tamponade, and 12,500 V of the same bovine thrombin with fibrin glue was again used. Cardiac transplantation was performed on day 3, and again 10,000 V of bovine thrombin with fibrin glue was used. Azathioprine (100 mg/day) and cyclosporine (40 mg twice daily) were administered from day 4. On day 8 the PT lengthened to 15 seconds versus II seconds preoperatively, and there was also a slight decrease of factor V (49%). On day II the same hemostasis disturbances that were reported for patient I were observed (see Table I): a neutralizing bovine thrombin inhibitor with a titer of I :200 was found and was associated with a decrease in factor V level. Apart from factor VII (21%) all the other factor levels (II, VIII, IX, X, XI, XII) were normal. He received only two plasmaphereses on days 12 and 13 under the same conditions as for patient I. After the second exchange, factor V level rose to 54%, but BTT remained elevated, and the inhibitor titer was I :40. On days 15 and 16 factor V levels were, respectively, 46% and 27%. He died on day 16 of right ventricular failure with bilateral pneumonia and sepsis. No hemorrhagic or thrombotic complications occurred concomitantly with the coagulation disorders. CASE 3. This patient was a 44-year-old white woman. She first underwent a mitral commissurotomy in 1981 and then had a mitral valve replacement (No. 25, St. Jude Medical, Inc., St. Paul, Minn.) in August 1990. The tricuspid valve was replaced by a biologic one in February 1991. At that time she received fibrin glue containing 10,000 U of bovine thrombin (Tissucol). She became septic on day 6 and had melena on day 8. From day 6 to day 7 the PT had increased from 12.5 to 14.5 seconds, and the factor V level had decreased from 72% to 39%. On the eighth day the same coagulation abnormalities as reported for patients I and 2 were noticed (see Table 1), but factor V level was not as low (16%). All the other factor levels (II, VII, VIII, IX, X, XI, XII) were normal. Nevertheless, she underwent three plasmapheresis exchanges: 1.5 times the plasma volume per day for 3 days to avoid more severe hemorrhagic complications. Factor V level recovered quickly (73%) at the end of the third exchange, but BIT, as for patients I and 2, was increased (> 120 seconds), and the inhibitor titer was 1:10. Moreover, thrombosis developed on the biologic tricuspid valve despite treatment with unfractionated heparin administered by continuous infusion (daily dose, 22,500 V), maintaining an AXa at 0.20 IV /m!. On day 12 a second tricuspid valve was inserted, and thrombectomy of the right atrium was carried out. Unfortunately, the patient died of cardiogenic shock on day 14.

The Journal of Thoracic and Cardiovascular Surgery Volume 105, Number 5

Berruyer et al.

895

Table II. Detection of IgG antibodies by using ELISA with coated bovine and human coagulation proteins on microplates Postoperative days Patient 3

Control

Day 9

Day 90

Day

Days

8

8-15

1.82 1.88

1.83 1.84

2.07 2.03

<0.07 <0.16

1.57 1.49 1.43 0.28

0.96 0.80 0.38 0.14

1.28 1.35 1.55 0.45

<0.13 <0.16 <0.13 <0.12

Patient I

Coated bovine proteins Thrombin from Tissucol Thrombin from Diagnostica Stago Coated human proteins Prothrombin Thrombin Native factor V Activated factor V

Norm.l pll.ml , ••1

Results are expressed in optical density at 492 nm. Proteins were all coated at the same concentrations (10 I'g/ml). Plasma dilutions were 1/100. Four control patients who received Tissucol but in whom coagulation disorders did not develop were tested at the same time.

Discussion The study of these three cases elicits the following remarks. Coagulation inhibitors developed in all after at least a second cardiac operation. It has been reported previously that this type of inhibitor can be generated after operations.i" and in particular after cardiac operations. The short interval between exposure to the thrombin preparation and development of the inhibitors is consistent with a prior sensitization," We are certain that patients 1and 2 had received bovinethrombin in the past, and it appears almost certain that patient 3 had. This immunization induced by the bovine thrombin that is used with fibringlue is, aboveall, characterized by a veryincreasedBTT but a normal human thrombin time and prothrombin level(see Table I). It is not corrected by protamine addition, so that no interference with heparin can exist. This inhibitor shows a high titer only against bovine thrombin. The observed decrease in factor V level,whichmay explainthe increasedaPTT and PT, ismore surprising; nofast-actingor progressive inhibitor has been shown against factor V activity. In addition, perfusionof fresh frozen plasma had no effect on the plasma levelof factor V in patient 1. Consequently, this raises the question of the presence of nonneutralizing inhibitors, which are veryrare but have been reported to explainprothrombin decrease associated with lupus anticoagulant.P: 14 Sincecoagulationtests were inadequate to analyze this theory, we developed an immunologic assay. Different hemostasis proteins were coated on ELISA plates and incubated with patients' plasma dilutions. The captured antibodies were of IgG type (Table II) and had been

Plttent 1 pt••ml Fig. 1. Bidimensional electrophoresis offactor V. The migration gelcontained hirudin to avoid factor V activation. Results show that intheplasma pool (top) therewas only thenormal arc of factor V.In the patient's plasma (bottom) therewas no normalfactor Varc,anda small amount ofactivated factor V was present (arrow).

purified from patients' plasma by immunoaffinity. In addition, they had been confirmed by immunoblot analysis in which antibody binding to bovine thrombin was tested. Indeed, these antibodiesrecognizedbovinethrombin but also, to a lesserextent, human thrombin and prothrombin. They also reacted with factor V, although among coagulation tests there was only an inhibitory effect on BTT. These data suggest that these IgGs are able to recognize both bovineand human thrombin, but there are distinct differencesin the relative binding affinities of the antibodies to the two different species of thrombin. Moreover, the inhibitory effect against factor V was essentiallyfor the native form, whileactivated factor V presented only a slight reactivity. Thus there were two major effectsagainst human pro-

896

..... Q 8

The Journal of Thoracic and Cardiovascular Surgery May 1993

Berruyer et ai.

2.,--------------.,

!

E :Il

oJ

100

50

';

::

<" >

...

=

oJ

O..l..--.-----.-----,-----r--.---.---' 8pm

9pm

toern

10pm llpm

.""..

12pm

Postoperoti ve doy

3.,--------------,

....t

oJ

1/200 j:

c

:cE

E

1/100 ;: I

::c

.

< c

8pm

9pm

teem

10pm llpm

Postoperotive doy

12pm

';

=

ll:I

Fig. 2. Efficiency of plasmapheresis on elimination of IgG antibodies (case 3). Bindingof antibodieswas tested by ELISA in the same conditions as for Table II. OD, Optical density; arrow, plasmapheresis. teins, the first on thrombin and prothrombin and the second on factor V. Prothrombin and factor V are immunologically very different, so we can presume that there are at least two antibody subgroups in each patient. Because it has been shown that bovine preparations contain factor V,6 we can conclude from these studies that the bovine thrombin with fibrin glue (Tissucol) is responsible for the induction of these antibodies. Other cases of acquired factor V inhibitor have been described, 15 but they are not associated with thrombin inhibitors. What are the biologic consequences of these antibodies? First, we can postulate, as has been reported for antiprothrombin antibodies and the prothrombin level decrease associated with lupus anticoagulant, 13,14 that factor V deficiency stems from rapid clearance from the circulation by antigen-antibody complex formation. It is noteworthy that antibodies preferentially recognize factor V rather than factor Va. This might explain why bidimensional electrophoresis (Fig, I) showed only the presence of trace amounts of factor Va that would not be cleared from the blood, confirming the immunologic levels observed (see Table I), Second, the antithrombin

antibody is directed against human thrombin and prothrombin, but it is difficult to state how it works. No inhibitory effect was present on neutralization of human thrombin by antithrombin III in the patients' plasma, Therefore this antibody is different from that of Lawson and colleagues," who found no binding on human prothrombin and an interference of thrombin neutralization by antithrombin III. We could speculate an interference of our patients' antibodies on the activation of protein C. These antibodies may also interfere with binding of thrombin on platelets or endothelial vascular cells, The true induced risk (thrombotic or hemorrhagic, or both) remains to be determined, It should be noted that patient 3 first had melena and then a valve thrombosis developed despite treatment with heparin, Was this immunization responsible for it? We cannot answer this question at this moment. No major bleeding occurred, as has been reported elsewhere for factor V inhibitor associated with a thrombin inhibitor': 6; the three patients were preventatively treated with plasmapheresis, however. Surprisingly, the efficiency of plasmapheresis was more evident on factor V activity (Fig, 2, top) and thrombin inhibitor titer (Fig, 2, bottom) than on their antigenic levels. Therefore we could think that these antibodies are of restricted heterogeneity, as has already been described for factor V inhibitor I 6 and for lupus anticoagulant associated with a decreased prothrombin concentration.l ' Plasmapheresis would be more efficient for certain antibodies that have the highest inhibitory effect on coagulation tests. As in other patients with IgG-type inhibitors at initial presentation, however, the effect of plasmapheresis was transient, probably because of postpheresis reequilibration of serum IgG with the extravascular compartment. Only patient I was alive 3 months later, and the bovine thrombin inhibitor (see Table II) remained at a high level. Factor V inhibitor had decreased but was still present, as demonstrated by ELISA (see Table 11), although factor V activity level was normal. As a corollary, this observation confirms that normalization of factor V in plasma does not necessarily imply complete remission of the disease and the absence of the inhibitor.l'' Recently it has also been shown that bovine thrombin could cause fatal hypotension immediately after injection.? All these findings suggest that human thrombin should be used in place of bovine thrombin to avoid these complications. In summary, this work confirms findings in other reports showing thrombin and factor V inhibitors after bovine thrombin exposure. In our patients, however, no neutralizing effect against human factor V could be proved, and ELISA was useful in detecting these IgG

The Journal of Thoracic and Cardiovascular Surgery Volume 105, Number 5

antibodies against human prothrombin, thrombin, and, especially, factor V. Finally, plasmapheresis used preventatively proved to be efficient to restore factor V level. We thank Dr. Nia Griffith for checking the English in this article before its submission. REFERENCES 1. Gibble JW, Ness PM. Fibrin glue: the perfect operative sealant? Transfusion 1990;30:741-7. 2. Berguer R, Staerkel RL, Moore EE, Moore FA, Galloway WB, Mockus MB. Warning: fatal reaction to the use of fibrin glue in deep hepatic wounds. Case reports. J Trauma 1991 ;31 :408-1l. 3. Milde LN. An anaphylactic reaction to fibrin glue. Anesth Analg 1989;69:684-6. 4. Stricker RB, Lane PK, Leffert JD, Rodgers GM, Shuman MA, Corash L. Development of antithrombin antibodies following surgery in patients with prosthetic cardiac valves. Blood 1988;72: 1375-80. 5. Flaherty MJ, Henderson R, Wener MH. Iatrogenic immunization with bovine thrombin: a mechanism for prolonged thrombin times after surgery. Ann Intern Med 1989; 111:631-4. 6. Zehnder JL, Leung LLK. Development of antibodies to thrombin and factor V with recurrent bleeding in a patient exposedto topical bovine thrombin. Blood 1990;76:2011-6. 7. Lawson J H, Pennell BJ, Olson JD, Mann KG. Isolation and characterization of an acquired antithrombin antibody. Blood 1990;74:369-70.

Berruyer et af.

897

8. Diez-Martin J, Sikking RA, Gilchrist GS, Bowie EJW, Fass DN. Development of anti-bovine thrombin antibodies following neurosurgical procedures. Br J Haematol 1990; 74:363-70. 9. Clyne LP, Komp DM. Circulating anticoagulant to animal plasmas in a cardiac patient. Am J Clin Pathol 1986; 61:640-4. 10. Kasper CK, Aledort LM, Counts RB, et al. A more uniform measurement of factor VIII inhibitors. Thromb Haemost 1975;34:869-72. II. Tran TH, Bondely C, Marbert GA, Duckert F. Reactivity of a hereditary abnormal antithrombin III fraction in the inhibition of thrombin and factor Xa. Thromb Haemost 1980;44:92-5. 12. Laurell CB. Quantitative estimation of proteins by electrophoresis in agarose gel containing antibodies. Anal Biochern 1966;15:45-52. 13. Bajaj SP, Rapaport SI, Fierer DS, Herbst KD, Shwartz DB. A mechanism for the hypoprothrombinemia of the acquired hypoprothrombinemia-lupus anticoagulant syndrome. Blood 1983;61 :684-92. 14. Fleck RA, Rapaport SI, Rao LVM. Anti-prothrombin antibodies and the lupus anticoagulant. Blood 1988;72:5129. 15. Nesheim ME, Nichols WL, Cole TL, et al. Isolation and study of an acquired inhibitor of human coagulation factor V. J Clin Invest 1986;77:405-15. 16. Chiu HC, Rao AK, Beckett C, Colman RW. Immune complexes containing factor V in a patient with an acquired neutralizing antibody. Blood 1985;65:810-8.