British Journal of Haematology, 1997, 96, 174–178

Evolution of blood coagulation and fibrinolysis parameters after abrupt versus gradual withdrawal of acenocoumarol in patients with venous thromboembolism: a double-blind randomized study B E R NA RD TA RDY, 1 B R I G I T TE TA R DY-P O N C E T, 2 S I LVY L A P O RT E -S IMIT S ID IS , 3 PAT R I C K M IS ME T T I , 3 H E RVE´ D E C O U S U S , 3 D E N I S G U YO TAT 2 AN D J EAN C. B E RTR AN D 1 Departments of 1 Intensive Care, 2 Haemostasis and 3 Clinical Pharmacology, C.H.R.U., Saint-Etienne, France Received 1 May 1996; accepted for publication 3 October 1996

Summary. A double-blind randomized trial was conducted to research a hypercoagulable state rebound after abrupt versus gradual withdrawal of acenocoumarol. 20 patients were included: 10 in the abrupt withdrawal group (AW) and 10 in the gradual withdrawal group (GW). Between days 1 and 15, F1 ‡ 2 was higher in group AW (P < 0:002). A significant increase of D-dimer with time was found (P < 0:001) without difference between the two groups. tPA and PAI-1 levels

remained stable throughout without difference between the two groups. No rebound phenomenon was observed. Four thrombotic recurrences were observed: group AW: 1, group GW: 3 (P ˆ 0:29). There is neither clinical nor biological support for a gradual anticoagulation withdrawal.

Thrombosis recurrence or ‘excessive’ increase in blood clotting factors after abrupt oral anticoagulant withdrawal (Poller & Thomson, 1964; Dinon & Van der Veer, 1969; Harenberg et al, 1983; Schofield et al, 1987) led some clinicians to taper anticoagulant treatment. Recent publications indicate that plasma levels of F1 ‡ 2 fragments and D-dimers enable evaluation of hypercoagulable states (Bauer & Rosenberg, 1987). Our objectives were, by comparing abrupt and gradual withdrawals of acenocoumarol in patients treated for venous thromboembolism, to (1) evaluate the evolution of plasma levels of these two marked, (2) research the existence of a hypercoagulation rebound state, defined as a significant increase of one of these markers during the period of treatment withdrawal compared with the value obtained at the steady state at the end of this period, and (3) research their influence on fibrinolysis.

versus gradual withdrawal of acenocoumarol. The study was approved by an ethics committee and all patients gave written informed consent. Patient selection. Inclusion criteria were: age > 50; acenocoumarol treatment instituted for at least 3 months for a venous thromboembolism documented by venography and/or angiography, acenocoumarol dosage of at least 3 mg/ d; a stable International Normalized Ratio (INR) between 2 and 3 for at least 1 month without change of dosage. Exclusion criteria were: malignancy, hemiplegia, cardiac insufficiency, nephrotic syndrome, coagulation-inhibitor deficiency, renal or hepatic insufficiency, severe venous thrombosis sequelae found on Doppler-US examination, antiplatelet treatment requirement. Treatment. Each patient was randomly assigned, in a double-blind fashion, to receive either placebo (group AW: Abrupt Withdrawal) or acenocoumarol (group GW: Gradual Withdrawal). After randomization, acenocoumarol or placebo was tapered as follows: days 0–6, two-thirds of initial dose of acenocoumarol; days 7–13, one-third of the initial dose and final withdrawal of treatment on day 14. Patients took the expected dose once a day at 8 p.m. Blood sampling and laboratory methods. Blood samples were collected between 8 and 10 a.m. at home for each patient on

METHODS Study design. The study was a double-blind randomized unicentre trial using two parallel groups to compare abrupt Correspondence: Dr Bernard Tardy, Service d’Urgences Me´dicales, CHRU Saint-Etienne, 42055, France.

174

Keywords: venous thrombosis, oral anticoagulant, hypercoagulable state, withdrawal.

# 1997 Blackwell Science Ltd

Venous Thromboembolism and Oral Anticoagulant Withdrawal days 0 (before the first reduction), 1, 4, 8, 11, 15, 19 and 22. Blood samples were collected at 48C on citrated vacutainer tubes (Becton Dickinson, France) and centrifuged within 1 h of collection (20 min, 48C, 2500 g). The platelet-poor plasma aliquots were frozen at ÿ808C. The following biological parameters were measured: prothrombin time with results expressed as INR values (Ne´oplastine CL20, Diagnostica Stago Laboratories, France); Prothrombin fragment F1 ‡ 2 (Enzygnost F1 ‡ 2, Behring Institute, France); D-dimer, tissue-type Plasminogen Activator antigen and Plasminogen Activator Inhibitor 1 activity (Asserachrom D-dimer, Asserachrom tPA, Stachrom PAI-1, Diagnostica Stago Laboratories). Measurements were performed blind in duplicate. Follow-up. Patients were examined on each day of blood sample collection, by a clinician unaware of their treatment. After day 22, patients were interviewed by phone once a month for 12 months. Only recurrent venous thromboembolism proved by radiological tests were considered for analysis. Statistical analysis. Wilcoxon-Mann-Whitney rank sum test and Fisher exact test were used for quantitative data and qualitative data respectively. To test significant differences in biological parameters (INR, F1 ‡ 2, D-dimer, tPA, PAI-1) during the study, data were analysed using a repeatedmeasures analysis of variance (ANOVA). All calculations were performed with SAS1 software. Biological parameter results were expressed as mean values and standard error of the mean (SEM). RESULTS Twenty patients (five men and 15 women, mean age 69 6 10 (6 SD)) were included without statistical difference between the two groups on inclusion (Table I). On inclusion,

INR values, obtained for each patient from outside laboratories, were in accordance with inclusion criteria (INR mean value of 2.4 in AW group and 2.6 in GW group). When these INR were estimated from plasma aliquots at our own laboratory, results were not significantly different (2.6 and 3.7 for AW and GW groups respectively) (Table I). All patients completed the 3 weeks of treatment withdrawal. One patient in group AW had taken one dose of treatment on day 15 instead of stopping on day 14. Biological results In both groups, INR values decreased significantly with time (from 3.15 on day 0 to 0.9 on day 22) (P < 0:002). Moreover, INR values differed significantly according to the treatment withdrawal scheme (P < 0:007); a faster decrease of INR values was observed after abrupt withdrawal (P < 0:02) (Fig 1). In both groups there was a significant increase of F1 ‡ 2 values with time (from 0.5 nmol/l on day 0 to 2.15 nmol/l on day 22) (P < 0:001). Moreover, F1 ‡ 2 levels were significantly higher in the group AW (P < 0:002) during the first 2 weeks but there was no difference after day 15. As all F1 ‡ 2 mean values were lower during the study than at the steady state (day 22), no rebound effect was found in either group (Fig 1). In both groups there was a significant increase of D-dimer values with time (from 369 ng/ml on day 0 to 654 ng/ml on day 22) (P < 0:001) without a significant difference between the two groups (P ˆ 0:14). As all D-dimer mean values observed were lower during the study than at the steady state (day 22), no rebound effect was found in either group (Fig 1). In both groups, PAI-1 and tPA values were almost stable with the time (P ˆ 0:51 and P ˆ 0:70 respectively) without

Table I. Clinical and biological characteristics of patients at time of inclusion (mean 6 SEM or frequency). Normal range Age (years) Sex (M/F) Level of DVT* Popliteal Femoro-iliac Pulmonary embolism Acenocoumarol Dosage (mg) Duration (months) INR F1 ‡ 2 (nmol/l) D-dimers (ng/ml) PAI-1 (UA/ml) tPA (ng/ml)

0.40–1.10 < 400 < 10 1–12

175

Acute withdrawal (AW) (n ˆ 10)

Gradual withdrawal (GW) (n ˆ 10)

P values

69:8 6 3:7 3/7

68:0 6 2:7 2/8

0.70y 0.85z

2 6 8

4 4 7

3:5 6 0:2 4:3 6 0:3

4:2 6 0:3 3:9 6 0:2

2:6 6 0:2 0:60 6 0:06 399 6 82 17:2 6 4:3 8:3 6 1:8

3:7 6 0:6 0:43 6 0:06 340 6 46 12:3 6 3:5 5:3 6 0:9

0.94z 0.50z 0.07y 0.31y 0.10y 0.07y 0.54y 0.38y 0.17y

* In four patients (two per group), pulmonary embolism was found without deep vein thrombosis. y Since the Wilcoxon test and the Student test led to the same statistical conclusions, only P value from the Student test is presented. z Fisher exact test.

# 1997 Blackwell Science Ltd, British Journal of Haematology 96: 174–178

176

Bernard Tardy et al

Fig 1. Evolution of INR values, F1 ‡ 2, D-dimer, PAI-1 and t-PA levels in acute withdrawal (AW) and gradual withdrawal (GW) groups. * Significant statistical difference.

significant difference between the two groups (P ˆ 0:81 and 0.11 respectively) (Fig 1). Follow-up No death occurred during the study. Four patients had a venous thrombosis recurrence (group AW: one; group GW: four; P ˆ 0:29), three within the 3 weeks after day 22. The early recurrence in group AW was associated with a marked F1 ‡ 2 and D-dimer levels increase after day 22 (3.3 nmol/l and 3204 ng/ml respectively) compared with the remaining AW patients (F1 ‡ 2 mean 2.45 nmol/l, 95% confidence interval 2.1–2.8; D-dimer mean 553 ng/ml, 95% confidence

interval 404–702). The two early recurrences in group GW were also associated with marked F1 ‡ 2 and D-dimer level increases after day 22 (3.3 and 1.9 nmol/l for F1 ‡ 2 and 854 and 745 ng/ml for D-dimer) compared with the remaining GW patients (F1 ‡ 2 mean 1.55 nmol/l, 95% confidence interval 1.15–1.95; D-dimer 414, 95% confidence interval 305–522). DISCUSSION Haemostatic markers levels after oral anticoagulant withdrawal were studied in two recent reports in which either a

# 1997 Blackwell Science Ltd, British Journal of Haematology 96: 174–178

Venous Thromboembolism and Oral Anticoagulant Withdrawal significant increase of F1 ‡ 2 levels or a transient increase of TAT levels was observed (Palareti et al, 1994; Genewein et al, 1996). However, these studies were not double-blind placebo-controlled trials. Secondly, diseases known to be associated with a hypercoagulable state (e.g. neoplasia, hemiplegia, renal or hepatic insufficiencies) were not clearly excluded in both studies. Thirdly, in Genewein et al (1996), there was not only a mixture of venous thrombosis and myocardial infarctions, but also a wide range of oral anticoagulant duration, and different kinds of oral anticoagulants. Patients studied by Palareti et al (1994) presented with arm venous thrombosis as well as with leg venous thrombosis and the duration of anticoagulation treatment ranged from 4 to 96 months. Finally, Palareti et al (1994) and Genewein et al (1996) included patients irrespective of age even though it has been demonstrated that the plasma levels of D-dimer and F1 ‡ 2 increased according to age (Bauer et al, 1987; Cadroy et al, 1992). Therefore their results a debatable and need to be confirmed. In our study, clinical and biological characteristics of patients were not statistically different on inclusion. Only INR values at inclusion, performed in our own laboratory and in different outside laboratories, are different (although not statistically significant). These differences between INR laboratories evaluation are well known and explained at least by the use of thromboplastin with different international sensitivity index (ISI), the use of different automated clot detectors or by the use of inappropriate control plasma (Hirsh et al, 1995). We found that the F1 ‡ 2 plasma levels increased significantly after acenocoumarol withdrawal (P ˆ 0:0006) and that this increase was earlier and more pronounced in case of abrupt withdrawal (P ˆ 0:005). Since only one early thrombosis recurrence was observed in the group AW, the clinical pertinence of an abrupt increase of F1 ‡ 2 remains to be demonstrated. We also observed that in the group GW the F1 ‡ 2 plasma levels increased slowly and remained at low levels (< 1:5 nmol/l up to day 15) even when the acenocoumarol dose was low (onethird of the initial dose) and INR value near 1. The fact that prothrombin activation can be significantly suppressed in vivo with the use of a low dose or oral anticoagulant was also observed in patients with inherited thrombotic disorders (Conway et al, 1987) and this whatever the F1 ‡ 2 assays used, radioimmunologic (Conway et al, 1987) or ELISA (Mannuci et al, 1991). We also found a significant increase of D-dimer values with time, whatever the mode of anticoagulant withdrawal and this had not previously been described. Palareti et al (1994) found that D-dimer mean levels were consistently normal throughout follow-up period with, however, individual variations from 35 to 448 ng/ml at baseline and from 49 to 7498 ng/ml after 3 weeks of withdrawal. As our study concerned only elderly patients, it is possible that, after treatment, D-dimer values return to baseline values with respect to age. During the 3 weeks following the abrupt treatment withdrawal we did not find any evidence of a hypercoagulable rebound although both F1 ‡ 2 and D-dimer levels increased during this time. We cannot exclude, however, a later occurrence of such phenomenon. Nevertheless, as, among a total of 52 patients

177

with venous thrombosis, randomized to have a gradual versus an abrupt oral anticoagulant withdrawal (Palareti et al (1994) and this study), three early thrombosis recurrences occurred in each group, there is also no clinical evidence for rebound phenomenon. This result contrasts with two recent non-controlled studies in which either a transient increase of TAT levels either a high level of myocardial infarct recurrence was observed after oral anticoagulation withdrawal (Genewein et al, 1996; Grip et al, 1991). The different patient risk profiles could be one explanation as well as the indication of oral anticoagulation and length of treatment before withdrawal. We did not find any significant change of the fibrinolytic activity after treatment withdrawal. Although increased fibrinolytic activity under oral anticoagulation has been suggested (Van Werch & Van MourkAlderliesten, 1993) fibrinolytic activity does not seem to be strongly influenced by the mode of oral anticoagulation withdrawal. In conclusion, in the population studied with low thrombosis risk factors, there is neither clinical nor biological support for a gradual rather than an abrupt anticoagulation withdrawal.

ACKNOWLEDGMENT This work was supported in part by grants from the Caisse Primaire d’Assurance Maladie de Saint-Etienne. This study was presented in part at the XVth International Congress on Thrombosis and Haemostasis (Jerusalem, June 1995). An abstract was published in Thrombosis and Haemostasis, 1995, 73, 838. REFERENCES Bauer, K.A. & Rosenberg, R.D. (1987) The pathophysiology of the prethrombotic state in humans: insights from studies using markers of hemostatic system activation. Blood, 70, 343–350. Bauer, K.A., Weiss, L.M., Sparrow, D., Vokonas, P.S. & Rosenberg, R.D. (1987) Aging associated changes in indices of thrombin generation and protein C activation on humans. Journal of Clinical Investigation, 80, 1527–1534. Cadroy, Y., Pierrejean, D., Fontan, B., Sie´, P. & Boneu, B. (1992) Influence of aging on the activity of the hemostatic system: prothrombin fragments 1 ‡ 2, thrombin–antithrombin III complexes and D-dimers in 80 healthy subjects with age ranging from 20 to 94 years. Nouvelle Revue Franc¸aise d’He´matologie, 34, 43–46. Conway, E.M., Bauer, K.A., Barzegar, S. & Rosenberg, R.D. (1987) Suppression of hemostatic system activation by oral anticoagulants in the blood of patients with thrombotic diatheses. Journal of Clinical Investigation, 80, 1535–1544. Dinon, L.R. & Vander Veer, J.B. (1960) Recurrent myocardial infarction after cessation of anticoagulant therapy. American Heart Journal, 60, 6–22. Genewein, U., Haeberli, A., Straub, P.W. & Beer, J.H. (1996) Rebound after cessation of oral anticoagulant therapy: the biochemical evidence. British Journal of Haematology, 92, 479–485. Grip, L., Blomba¨ck, M. & Schulman, S. (1991) Hypercoagulable state and thromboembolism following warfarin withdrawal in postmyocardial infarction patients. European Heart Journal, 12, 1225– 1233.

# 1997 Blackwell Science Ltd, British Journal of Haematology 96: 174–178

178

Bernard Tardy et al

Harenberg, J., Haas, R. & Zimmermann, R. (1983) Plasma hypercoagulability after termination of oral anticoagulants. Thrombosis Research, 29, 627–633. Hirsh, J., Dalen, J.E., Deykin, D., Poller, L. & Busseu, H. (1995) Oral anticoagulants: mechanism of action, clinical effectiveness, and optimal therapeutic range. Chest, 108, 231S–246S. Mannuci, P.M., Botasso, B., Tripodi, A. & Bonomi, A.B. (1991) Prothombin fragment 1 ‡ 2 and intensity of treatment with oral anticoagulants. Thrombosis and Haemostasis, 66, 741. Palareti, G., Legnani, C., Guazzaloca, G., Frascaro, M., Grauso,

F., De Rosa, F., Fortunato, G. & Coccheri, S. (1994) Activation of blood coagulation after abrupt or stepwise withdrawal of oral anticoagulants: a prospective study. Thrombosis and Haemostasis, 72, 222–226. Poller, L. & Thomson, J.M. (1964) Evidence for rebound hypercoagulability after stopping anticoagulants. Lancet, ii, 62–64. Schofield, K.P., Thomson, J.M. & Poller, L. (1987) Protein C response to induction and withdrawal of oral anticoagulant treatment. Clinical and Laboratory Haematology, 9, 255–262. Van Werch, J.W.J. & Van Mourk-Alderliesten, C.H. (1993) Fibrinolysis and oral anticoagulation. Fibrinolysis, 7, 47–49.

# 1997 Blackwell Science Ltd, British Journal of Haematology 96: 174–178