Control of Oral Anticoagulant Therapy*

ANNALS O F C L IN IC A L A N D LABORATORY S C IE N C E, Vol. 18, No. 6 Copyright 1988, Institute for Clinical Science, Inc. Control of Oral Anticoagulant Therapy* JAMES J. BIEMER, M.D. St. Joseph s Hospital and University of South Florida College of Medicine, Tampa, FL 33607 ABSTRACT In an effort better to standardize the control of oral anticoagulant therapy, it has recently been recommended that the prothrom bin tim e be reported in the form of an International Normalized Ratio (INR) based upon calibration of the locally employed thromboplastin with an International Reference Preparation. It has been demonstrated in our laboratory that the INR does minimize the differences in results which ensue from variations in the source of thromboplastin and instrumentation and should hopefully allow for better interlaboratory comparisons in the future. Recent studies have also suggested that many American patients tend to be over anticoagulated and at greater risk for hemorrhage. Based upon those findings, over 40 percent of our specimens w ere above the currently recommended levels. Introduction Oral anticoagulants have been em ployed for many years for the prevention of recurrent thrombosis, and control of dosage has largely been guided by the almost universally available one stage prothrombin time. These agents, divisible into the coumarin and indanedione type compounds, act similarily as inhibitors of vitamin K with major effects on some of the coagulation zyom ogens, namely factors II, VII, IX, and X, and proteins C and S. Normally before these proteins are released into the circulation and d ependent upon the presence of vitam in K, glutamic acid residues are * Send reprint requests to James J. Biemer, M.D., converted to gamma carboxyglutamic acid.20'21 These residues bind calcium, perm itting the coagulation proteins to bind phospholipids for activation in the clotting scheme. In the presence of vitamin K inhibitors, the gamma carboxyglutamic acid resid u es are not form ed, lead in g to decreased plasma levels of factors II, VII, IX and X, and proteins C and S, and a concommitant rise in proteins induced by vitamin K absence (PIVKA s). With induction of oral anticoagulant therapy, factor VII levels begin to fall within a day followed by factors IX, X, and II reaching a stable anticoagulated state in approxim ately one week. The rise in PIVKA s may also retard thrombin formation,7 while reduced levels of proteins C and S might possibly have an opposite 0091-7370/88/1100-0421 $01.20 Institute for Clinical Science, Inc.

42 2 BIEMER thrombogenic effect. Creation of a possible hyper-coagulable state has been suggested during the induction phase when protein C levels fall early, parallelling factor VII, but before the full reduction of the remaining pro-coagulant factors II, IX and X. D espite broader understan ding of their physiologic action and their long history of use, there continue to be many controversies regarding the specific clinical indications as well as the dosage and duration of therapy with oral anticoagulants. In clinical practice, the ideal therapeutic range has been assumed to be ju st below the hemorrhagic level26 and has been guided by the one stage prothrom bin time. Although not based upon what now would be regarded as solid objective evidence, W right et al28 in the early 1950s suggested that optimal therapy was w hen the p a tie n t s p ro thrombin time was 2.0 to 2.5 times the control value. By and large this recommendation was accepted and has been adhered to in N orth America for the past 30 years, as evidenced by similar guidelines in many cu rren t textbooks of m edicine.1,12,15 An example of these p rev ailing recom m endations is b est summarized in Goodman and Gillman6 which states, Patients on chronic therapy usually should be m aintained at a one stage prothrombin activity of 25 percent, which, expressed in seconds, is about twice the normal baseline of 12 seconds. However, the so called therapeutic range is based m ore upon the avoid an ce of b le e d in g th an on th e achievem ent of a proven therapeutic effect. Although Moschos et al17 suggested in 1964 that those recom m endations resulted in excessive anticoagu la tio n, o n ly m o re re c e n tly h av e randomized studies evaluating different anticoagulant intensities with clinically re la te d outcom es begun to address o b jectiv ely th e q u estio n s posed by H irsh.9,10 Clearly one portion of the confusion stems from lack of standardization of the prothrom bin tim e which is perform ed w ith a w ide variety of reagents and instrum ents. For many years after the prothrom bin tim e was introduced by Quick in 1935, many laboratories prepared their own throm boplastin reagents. These have gradually given way to more uniform, frequently commercially prepared reagents; but great differences continue to exist, namely, the thromboplastins of rabbit tissue origin commonly used in North America versus those prepared from hum an tissues such as the B ritish C om parative T hrom boplastin (BCT), and bovine brain origin favored in Scandinavia. The rabbit brain thromboplastins in commercial use in North America are generally known to be less sensitive to the reduction in the vitamin K-dependent clotting factors than standardized hum an brain throm boplastin commonly used in Great Britain.10 Secondly, and w hile citing surveys cond u cted by th e C ollege of A m erican Pathologists and the C enter for Disease Control, it has been shown that the type of instrum entation used in performing th e prothrom bin tim e has alm ost as much effect as the type of thromboplastin.4,24,25 In response to these issues, the World H ealth Organization (WHO) has prepared a primary International Reference Preparation of human brain throm boplastin to promote standardization of oral anticoagulant control.14,27 Subsequently, an in te rn a tio n a l c o m m ittee reco m m ended adopting a uniform calibration system based upon an International Normalized Ratio (INR) derived from the calibration of commercial thromboplastin reagents against the International Reference Preparation. The validity of this system has been confirmed in international collaborative studies.13,16,22,23 It has been proposed that manufacturers of thromboplastins used in oral

anticoagulant control should indicate the relationship of each batch of their material to the W HO International Reference Preparation by a num ber which describes the comparative slope (or sensitivity). This is presently referred to by the WHO as the International Sensitivity Index (ISI). M anufacturers should also provide a table or graph showing the relationship betw een the conventional term s of expression of results of the prothrombin time test and the INR. The INR is calculated by the equation, INR = Rc, where R is the prothrom bin ratio (patient prothrom bin time/mean normal prothrombin time) and c is the comparative slope (ISI) of the throm boplastin used.13 Manufacturers of commercial thromboplastins in the United States are now beginning to make available batches of reagents for w hich the ISI has been determ ined. It was, therefore, the purpose of this study to evaluate the feasibility of utilizing these reagents and adopting the reporting of the INR along with our currently em ployed m ethods and results. Furtherm ore, it was our wish to evaluate the degree of anticoagulation achieved in our community and compare those results with more recent recommendations and define w hether or not the utilization of the INR served to standardize better those results. Methods One hundred six blood samples were obtained from 42 patients who had been on oral anticoagulants for one month or longer. The blood was drawn into glass evacuated tubes with Siliconized stoppers* containing 3.8 p ercent sodium citrate anticoagulant. The blood was centrifuged and transferred into plastic * Venoject, Terumo Products. CONTROL OF ORAL ANTICOAGULANT THERAPY 423 tubes for testing w ithin two hours of phlebotom y. Prothrom bin times w ere performed on the Fibrometerf with rabbit brain thromboplastin^ (Thromboplastin C, batch # TPCD - 399, ISI not available). The rem ainder of plasma was frozen and transferred to another laboratory where prothrom bin time determ i nations were performed in batches utilizing Thromboplastin C (ISI values 2.28 mechanical and 2.53 photoptical), and S im plastin (ISI values 1.81 mechanical and 1.9 photoptical). Each plasma sample was tested in duplicate with both reagents utilizing the Fibrom eterf representing a mechanical system, and the MLA 800 1a photo-optical method. Normal values had been previously established as follows: Thromboplastin C /F ibrom eter - 11.7 seconds; Thromboplastin C/MLA 800-12.4 seconds; Simplastin/Fibrometer - 11.9 seconds; and Simplastin/MLA 800-12.3 seconds. Results were expressed in seconds, ratio of patient prothrombin time to normal time, and the INR, and compared w ith the recom m endations for antithrombotic agent usage by the American College of Chest Physicians and the National H eart Lung and Blood Institute National Conference on Antithrombotic Therapy.2 Statistical Methods A two-way, blocked analysis of variance was used to determ ine w hether or not the use of different data reporting methods, (i.e. prothrom bin time, ratio of prothrom bin tim e to normal, or INR), would result in significant differences in the stability of the reported results when t Baltimore Biological Laboratory, Cockeysville, MD. American Dade, Inc., Miami, FL. I General Diagnostics, Morris Plains, NJ. * Medical Laboratory Automation, Inc., M ount Vernon, NY.

4 2 4 BIEME-R the four different reagent-instrum ent procedures were applied. Each sample was viewed as a separate block, and the four reagent-instrum ent procedures and three data reporting methods were the two tre a tm e n t effects. P ro th ro m b in times were used as the dependent variable in this model. A natural logarithm was used as a variance stabilizing transformation before the analysis was conducted. The presence of a statistically significant two-way interaction in this model would indicate that the use of diffe re n t d ata re p o rtin g m ethods have resulted in differential stability of the prothrom bin results when the different reag en t-in stru m en t procedures w ere applied.3 T hree one-way, blocked analysis of variances were used to determ ine which of the three data reporting methods displayed the least variability when the different reagent-instrum ent procedures were used. In each analysis, each sample n u m b er co n stitu ted a block, and the reagent-instrum ent procedure was the single tre a tm en t effect. P rothrom bin times w ere the dependent variable in each of the these models. A natural logarithm was used as a variance stabilizing transformation before the analyses were conducted. Analyses w ere done separately for the different data reporting m ethods. The partial F-statistic assoc iate d w ith th e single m ain effect, obtained from each of the three analyses, were compared to each other to determ in e w h ich of th e d a ta re p o rtin g methods displayed the greatest stability over the different laboratory methods. Lower F values indicate greater stability. p e r day. The m ean prothrom bin tim e obtained by im m ediate analysis of the 106 unfrozen plasma samples was 18.4 seconds. Analysis after freezing resulted in mean values that were approximately 2.2 seconds longer by comparable methodology. The mean prothrom bin times of the 106 sam ples analysed by the various reagent-instrum ent com binations and expressed by the three reporting methodologies, (i.e., prothrom bin tim e in seconds, prothrom bin tim e to normal ratio, and INR), are displayed on table I. The mean prothrombin time expressed in seconds ranged from a low of 20.6 seconds to a high of 27.2 seconds. For the prothrom bin tim e to normal ratio, the range was 1.66 to 2.28, while the INR resulted in a range of 4.02 to 4.73. The results of the two-way, blocked analysis of variance indicates that there is a significant difference (p < 0.001), in the stability of the three different data presentation m ethods. The results of the three one-way, blocked analyses of variances indicate that the INR reporting TABLE I Prothrombin Time Mean Values Obtained on 106 Plasma Samples Tested by Various Reagent-instrument Combinations Thromboplastin ISI Instrument P.T. (sec) PT/N INR D1 * Fib 18.4 1.56 * D2 2.28 Fib 20.6 1.76 4.15 D2 2.53 MLA 20.7 1.66 4.02 G 1.81 Fib 27.2 2.03 4.31 G 1.90 MLA 25.1 2.28 4.73 Results The patients had been receiving Coum adin for an average of 13 m onths, (range one m onth to 180 months), and the mean dose of all patients was 4.8 mg Dl-Dade-Thromboplastin c. Lot # TP CD-399 D2-Dade-Thromboplastin C, Lot # TPCD-391A G-General Diagnostics-Simplastin, Excel, Lot # OC201 Kit # OC222 Fib-Fibrometer MLA-MLA Electra 800 *-ISI not available PT/N-Patient prothrombin time in seconds/normal time in seconds INR-International Normalized Ratio ISI-International Sensitivity Index

m ethod is the most stable when the four reagent-instrument procedures are compared as follows: Data Reporting F-statistic M ethod PT in seconds 203.60 PT/N 213.86 INR 26.99 Each method of reporting prothrombin times was evaluated to determine if a given sample value was within the therapeutic range of anticoagulation, above that range, or below it, (for the prothrom bin time/normal ratio, the therapeutic range was defined as between 1.2 to 1.5, and for the INR 2.0 to 3.0). First, individual samples were compared to see how frequendy that sample fell into the same therapeutic range when tested by the various reagent-instrument combinations and reported by either of the two methods. As can be seen in table II, the INR led to total agreem ent of all four reagent-instrum ent combinations more frequently than the prothrombin time to normal ratio, although the differences were not great. Lastly, the 424 challenges, (i.e., 106 samples tested by four different reagentinstrum ent combinations), were evaluated to compare each reporting method in term s of how many samples would be regarded as at, above, or below the recom m ended therap eu tic range. Since patient diagnoses were not obtained as part of this study, it was not known how many of these patients may have had clinical indications for a higher therapeutic range. Therefore, all specimens were simply evaluated according to those recom m ended criteria which would apply most frequently, i.e., INR = 2.0 to 3.0 or PT/Normal ratio of 1.2 to 1.5. As can be seen in table III, all m ethods of reporting showed many specimens to be above the newly recom m ended therapeutic range. Specifically, those tests run CONTROL OF ORAL ANTICOAGULANT THERAPY 425 TABLE I I Prothrombin Time Ability of Each Reporting Methodology to Yield Same Therapeutic Conclusion PT/N (1.2-1.5) INR (2.0-3.0) All 4 Values Agree 76 (72%) 83 (78%) 3 Agree/1 Variance 20 (19%) 18 (17%) 2/2 Split 10 f 9%) 5 f 5%) Total Tested 106 106 PT/N-Prothrombin time in seconds/normal time in seconds INR-International Normalized Ratio on fresh plasma indicated 42 percent of the samples were above the therapeutic range; however, w hen analysed after freezing, 69 percent w ere above the th e r a p e u tic ra n g e if r e p o r te d by prothrom bin time/normal ratios and 56 percent when reported by the INR (table III). Discussion It has been demonstrated that importa n t v a ria b le s can be e x p e c te d in prothrombin testing owing to use of different commercial reagents and instrum ents as reported by others.4 10 19 24,25 T hese differences w ere seen despite many factors which tended to minimize the differences in our study, - namely the use of thromboplastins of a single anim al tissue source and batching of analyses which were performed in a single laboratory by the same technologist. Despite these stabilizing factors, a great variation was observed by us in the various reagent-instrum ent com binations when the results were expressed in seconds or prothrom bin to normal ratios. W hen expressed by the use of the INR, however, statistically significant stabilization of results betw een the various r e a g e n t- in s tr u m e n t c o m b in a tio n s occurred. Clearly in view of the frequent long term use of oral anticoagulants and

4 2 6 BIEMER TABLE III Prothrom bin Time Comparison o f I n te r p r e ta tiv e R e su lts Between R eporting M ethodologies I n i t i a l * P T /N IN R Above Therapeutic Range 45 (42%) 292 (69%) 239 (56?) Within Therapeutic Range 52 (49%) 114 (27%) 110 (26%) Below Therapeutic Range 9 ( 8%) 18 ( 4%) 75 (18%) Total Specimens Tested 106 424 424 * PT/N run on fresh unfrozen plasma PT/N-Prothrombin time in seconds/normal in seconds (frozen plasma) INR-International Normalized Ratio (frozen plasma) our increasingly mobile society, there is an urgent need for standardized interlaboratory comparisons. The use of the IN R re p o rtin g m e th o d a p p e a rs to address many of these needs, although when its ability to classify patients uniformly as over, under, or adequately anticoagulated according to recent recommendations,2 it did only slightly bette r th an th e tra d itio n a l m eth o d of reporting. Also of interest was the fact that all rep o rtin g m ethods suggested a large number of patients were over anticoagulated when compared to the recommendations of the recent National Conference on A n tith ro m b o tic T h e ra p y.2 Briefly, that report recommended maintaining patients on oral anticoagulants at an INR of 2.0 to 3.0, (corresponding rabbit brain thromboplastin ratio of 1.2 to 1.5), for prophylaxis of venous thromboembolism, treatm ent of venous thrombosis, and prevention of systemic embolism in patients with atrial fibrillation, valvular h eart disease, bioprosthetic h e a rt valves, and acu te m yocardial in fa rc tio n. O n ly in p a tie n ts w ith mechanical heart valves and in patients with recurrent systemic embolism was a higher effect recommended, namely, an INR of 3.0 to 4.5, equivalent to a rabbit brain thromboplastin ratio of 1.5 to 2.0. While the specific diagnostic indications for anticoagulation were unknown in our group of patients, it seems likely that the therapeutic goal for the majority of these patients would have been an INR 2.0 to 3.0. Of interest, however, and serving to illustrate the still prevailing huge divergence of opinion and the lack of uniformity in therapeutic goals for these agents is the synopsis for recommended ranges cited by Leoliger et al16 in 1985, extending from an INR of 2.0 to 20.0! In an international survey of dosage and control of oral anticoagulants, Poller and Taberner19 compared the mean dose prescribed in hospitals in 23 countries. While there was a constancy of overall mean doses, there were wide inter-laboratory differences betw een the mean doses of Warfarin. The lowest mean dose was 2.45 mg for one Hong Kong center, whereas one United States center prescribed a mean dose almost four times that amount of drug. In general, North A m erican cen ters w hich used rabbit reagent thromboplastins tended to prescribe h ig h er doses of W arfarin and advocate m ore intensive therapeutic ranges. These views were further emphasized in 1982 by Hull et al.11 They compared two groups of patients w ith proximal vein thrombosis which were treated with Warfarin. In the first group w here the Warfarin administration was regulated w ith th e M a n c h e ste r C o m p arativ e Reagent, (human brain thromboplastin), a lower dose resulted (average 4.9 mg per day) than in the group where Simplastin (rabbit brain thromboplastin) was used for the regulation of the Warfarin dose, (5.8 mg p e r day). W hile both groups of patients had sim ilarly low rates of recu rren t throm boem bolism, two percent, the group with more intensive Warfarin had a much higher rate of hem orrhagic complications, that is, 22 percent versus four percent in the less intensely treated group. W hereas the original therapeutic ranges set at the beginning of their study were 1.5 to two

CONTROL OF ORAL ANTICOAGULANT THERAPY 4 2 7 times the control for Simplastin and two tim es the control for the M anchester Comparative Reagent, Hull et al11 concluded that those recommendations lead to higher doses when the guideline for rabbit brain thromboplastin was used. F u rth e rm o re, th e low er dosage was associated with equal clinical effectiveness and a much lower incidence of hemorrhage. F urther support for the effectiveness of lower Warfarin doses was furnished by Francis et al5 in 1983. They compared th e safety and efficacy of W arfarin sodium with Dextran 40 in the prevention of venous thrombosis in patients at high risk for deep vein thrombosis after elective total hip or knee replacement. Warfarin was given in a two step regimen designed to avoid operative bleeding com plications w hile still p rev en tin g venous throm bosis. This was accomplished by a 14 day pre-operative Warfarin sodium dose sufficent to prolong the prothrombin times 1.5 to three seconds more than the control, and then sufficent W arfarin post-operatively to prolong the prothrom bin time 1.5 times the control. Compared to Dextran, the Warfarin patients had a much lower incidence of venographically demonstrated thrombosis, (21 percent versus 51 percent) and fem oral or p o p liteal vein thrombosis (two percent versus 16 percent). F urtherm ore, the incidence of hemorrhage was infrequent and similar in the groups, and thus Francis et al4 concluded that the two step W arfarin provided highly effective prophylaxis of post-operative venous thrombosis without excessive risk of p eri-o p erativ e bleeding. There is a growing bulk of evidence which acknowledges that the generally accepted prothrom bin time ratios of 1.5 to 2.5 with rabbit brain thromboplastin result in a greater intensity of therapy than the British hum an brain ratios of 2.0 to 4.0, and that rabbit brain prothrom bin time ratios over 2.0 can explain the apparently greater number of bleeding complications reported in the United States.26 In 1986, the National Conference on Antithrombotic Therapy2 combining the talents of an international group of experts exam ined extensive objective and subjective evidence on this topic and reached similar conclusions recommending that where oral anticoagulants are used, a rabbit thromboplastin prothrom bin time ratio of 1.2 to 1.5 (INR 2.0 to 3.0) is appropriate. Only in the case of a m echanical prosthetic heart valve or recurrent systemic embolism was a higher ratio of 1.5 to 2.0 (INR 3.0 to 4.5) recommended. The question of w hether or not the conversion of all patient results to the INR as recom m ended by the International Committees for Standardization in H aem ato lo g y and T h ro m b o sis and Hemostasis13 warrants attention. Poller18 after three years experience asks, Is the scheme working satisfactorily? He indicates that some manufacturers are not yet calibrating their reagents correctly and that some coagulometers cause serious deviations from the correct INR. Furtherm ore, the INR cannot be relied upon during the induction phase of Warfarin treatm ent because of the varying responses of reagents to depression of individual coum arin clotting factors. Despite these limitations, however, the INR will tend to prevent gross differences in dosage betw een centers and should allow better comparison of published results. Lastly, with our increasingly mobile society, the availability of uniformly comparable prothrom bin time studies at different laboratories should greatly assist individual patients to systematically regulate their oral anticoagulant needs. Acknowledgments The author gratefully acknowledges the technical expertise of Ms. Caroline Glenn, B.S, M.B.A., with

4 2 8 BIEMER out whose assistance this work could not have been completed. Additionally, the statistical analysis of results was generously contributed by Mr. Eric Ross, Sc.M., Research Assistant, School of Public Health, University of South Florida, Tampa, Florida. References 1. B R A U N W A L D, E.: Harrisons s Principles of Internal Medicine, 11th ed., 1987. 2. D a l e n, J. E. and H ir s h, J.: American College of Chest Physicians and National Heart, Lung and Blood In stitu te National C onference on Antithrombotic Therapy. Arch. Intern. Med. 46:462 471, 1986. 3. D raper, N. R. and S m ith, H.: Applied Regression Analysis. New York, John Wiley and Sons, Inc., 1966. 4. E vatt,b. L., B r o g a n,d. T riplett,d. A. and WATER, G.: Effect of thromboplastin and instrumentation on the prothrombin time test. Clin. Lab. Haematol 3:331-342, 1981. 5. F rancis, C. W., Marder, V J., M cc ollister, E., and YaUKOOLBODI, S.: Two-step Warfarin therapy. Prevention of postoperative venous throm bosis w ithout excessive bleeding. 249:374-378, 1983. J. Am. Med. Assoc. 6. G o o d m a n, A. G. G., G o o d m a n, L. S., a n d G il m a n, A.: G o o d m an a n d G ilm an s T h e P h a r m acological B asis o f T h e ra p e u tic s, 7 th ed. N ew York, M acm illan Pub. C o., 1985. 7. H emker, H. C. and M uller, A. D.: Kinetic aspects of the interaction of blood-clotting enzymes. VI. Localization of the site of bloodcoagulation inhibition by the protein induced by vitamin K absence. (PIVKA). Thromb. Diath. Haemorrh 20:7 8-8 7, 1968. 8. H e r m a n s, J. van d e n Be sse l a a r, A.M.P.H., L o e u g e r, E. A., and van d e r V e l d e, E. A.: A collaborative calibration study of reference materials for Thromboplastins. Thromb. Haemost. 50:712-717, 1983. 9. HlRSH, J.: Is the dose of Warfarin prescribed by American physicians unnecessarily high? Arch. Intern. Med. 147:7 6 9-7 7 1, 1987. 10. H ir s h, J., D ey k n i, D., and P o l l e r, L.: Therapeutic range for oral anticoagulant therapy. Chest 89:11 5-1 55, 1986 (Supplement). 11. H u l l, R, H i r s h, J, J a y, R. et al.: Different intensities of oral anticoagulant therapy in the treatment of proximal-vein thrombosis. New Engl. J. Med. 307:1676-1681, 1982. 12. Hurst, J. W.: The Heart, Arteries and Veins, 6th ed. New York, McGraw-Hill Book Co. 1986, p. 1669. 13. ICSH/ICTH Recommendations for reporting prothrombin time in oral anticoagulant control. Thromb. Haemost. 53:155-156, 1985. 14. K i r k w o o d, T. B. L., and L e w i s, S. M.: Requirem ents for thromboplastins and plasma used to control oral anticoagulant therapy. W HO Tech. Rep. Ser. 687:81-99, 1983. 15. Krupp, M. A, Schroeder, S. A., and Tierney, L. M., Jr.: C u rrent M edical Diagnosis and Treatment 1987. New York, Appleton & Lange, 1988, p. 285. 16. Leoliger, E. A., van den Besselaar, A.M.H.P., and Lewis, S. M.: Reliability and clinical impact of the normalization of prothrombin times in oral anticoagulant control. Throm Heamost 53:148 154, 1985. 17. Moschos, C. F., Wong, P. C., and Sise, H. S.: Controlled study of the effective level of longterm an ticoagulation. J. Am. M ed. Assoc. 190:799-805, 1964. 18. POLLER, L.: Regulating the dosage of Warfarin for anticoagulants. New Engl. J. Med. 14:316, 20:1277-1278, 1987. 19. Poller, L. and Taberner, D. A.: Dosage and control of oral anticoagulants: An international collaborative survey. Brit. J. Haematol. 51:479-485, 1982. 20. STENFLO, J. and Suttie, J. W.: V itam in K- dependent formation of carboxyglutamic acid. Ann; Rev. Biochem. 47:157-172, 1977. 21. Suttie, J. W. and Jackson, C. M.: Prothrombin structure, activation and biosynthesis. Physiol. Rev. 57:1-70, 1977. 22. Thomson, J. M., Darby, K. V., and Poller, I.: Calibration of BCT/441, the ICSH reference preparation for thromboplastin. Thromb. Haemost 55:379-82, 1986. 23. Thomson, J. M., Tomenson, J., and Poller, L.: The calibration of the second primary international reference preparation for thromboplastin (thromboplastin, human, plain, coded BCT/ 253). Thromb. Haemost 52:336-342, 1984. 24. Triplett, D. A., Evatt, B. L., and van den Besselaar, A.M.H.P.: Proficiency testing and standardization of prothrom bin time: Potential use of throm boplastin calibration in th e U.S. C hapter 12. Thromboplastin Calibration and Oral Anticoagulation Control, van den Besselar, A.M.H.P., Gralnick, H. R., and Lewis, S. M., eds. Hingham, MA, Boston/Nijhoff, 1984. 25. van den Besselaar, A.M.H.P., Evatt, B. L., BroGAN, D. R., et al.: Proficiency testing and standardization of prothrom bin time: Effect of thromboplastin, instrumentation, and plasma. Am. J. Clin. Pathol. 82:688-699, 1984. 26. WESSLER, S. and Gitel, S. N.: Warfarin from bedside to bench. New Engl. J. Med. 311:645-652, 1984. 27. W HO Expert Committee on Biological Standardization: Thirty-fourth report. W HO Tech Rep. Ser. 1984; 700:7-25. 28. Wright, I. S., Beck D. F., and Mabple, C. D.-. M yocardial infarction and its treatm ent with anticoagulant. Lancet J:9 2-97, 1954.