Ximelagatran and melagatran vs. low-molecular-weight heparin in major orthopedic surgery: relationship between efficacy and safety and timing of initial administration*


  • *

    Preliminary results of this analysis were presented at the International Congress of Thrombosis in Ljubljana in June 2004 as an oral communication (OC 065).

Paul Zufferey, EA3065 Groupe de Recherche sur la Thrombose, Unité de Pharmacologie Clinique, CHU Saint-Etienne Bellevue, 42055 Saint-Etienne cedex 2, France.
Tel.: +33 4 77 12 07 16; fax: +33 4 77 12 78 20; e-mail: paul.zufferey@chu-st-etienne.fr

Ximelagatran, a new orally administered antithrombotic, is a prodrug of the subcutaneous form melagatran acting through direct inhibition of thrombin. To date, six clinical studies have compared melagatran/ximelagatran with low-molecular-weight heparins (LMWH) in short-term prophylaxis (6–12 days) [1–6]. The results of these studies have shown conflicting data concerning the benefit–risk ratio of ximelagatran. However, various therapeutic schedules were used in these studies in terms of dose and timing of administration of melagatran/ximelagatran or LMWH. Thus, as has already been suggested for LMWH [7,8] or fondaparinux [9], the timing of the first administration may affect the efficacy or safety of melagatran/ximelagatran. With this in mind, we studied, using the techniques of meta-analysis and meta-regression, the benefit–risk ratio of the different regimen of melagatran/ximelagatran in preventing postoperative venous thromboembolic events in major orthopedic surgery according to the timing of the initial administration of this novel antithrombotic.

The selected studies for the meta-analysis were randomized studies comparing ximelagatran 24 mg b.i.d. whether or not preceded by subcutaneous melagatran with LMWH for preventing deep venous thrombosis (DVT) after total hip replacement (THR), or total knee replacement (TKR). The efficacy end points were total DVT (asymptomatic or symptomatic) and proximal DVT and were detected by mandatory venography at the end of treatment period or earlier if clinically suspected. Safety end points were severe bleeding and transfusion (allogeneic or autologous blood). All events were adjudicated by an independent adjudication committee and limited to the study period (maximum 12 days). The meta-analysis was performed using the logarithm of the relative risk (RR). The RR results are presented with 95% confidence intervals (CI). The effect of the timing of melagatran/ximelagatran initiation on its benefit–risk ratio was first evaluated by performing an analysis on the two groups of studies investigating, respectively, preoperatively administered melagatran/ximelagatran vs. LMWH, and postoperatively administered melagatran/ximelagatran vs. LMWH. The effect of timing was also evaluated by a fixed-effect meta-regression analysis of study-specific risk ratios by means of weighted linear regression [10]. The time used in the meta-regression was the mean time to first dose given to patients (observed in the study) or, if this was not mentioned in the article, the average time of the first administration according to the protocol.

Six studies with a total of 8450 patients were included in the meta-analysis. The summary of each study is presented in Table 1.

Table 1.  Melagatran/ximelagatran vs. LMWH: description of studies
ReferenceYearType of surgeryMelagatran/ ximelagatran and doseLMWH and dose (anti-Xa units)Study designTiming of first administration in relation to surgeryTreatment duration (days)Diagnosis of DVT*Number of patients randomized
Melagatran/ ximelagatranLMWH
  1. LMWH, low-molecular-weight heparin; THR, total hip replacement; TKR, total knee replacement; strat., randomization stratified by type of surgery; DVT, deep vein thrombosis; postop, after end of surgery; knife-to-skin, just before surgery.

  2. *At end of treatment.

  3. Average time to first dose.

US 203 [1]2001TKRXimelagatran 24 mg b.i.d.Enoxaparin 3000 × 2Open12–24 h postop12–24 h postop6–12Unilateral venography255
METHRO I [2]2002THR, TKR (strat.)Melagatran 4 mg b.i.d. for 2 days, then ximelagatran 24 mg b.i.d.Dalteparin 5000OpenKnife-to-skinpm day before surgey8–11Bilateral venography76
METHRO II, [3]2002THR, TKR (strat.)Melagatran 3 mg b.i.d. for 1–3 days, then ximelagatran 24 mg b.i.d.Dalteparin 5000BlindKnife-to-skinEvening before surgery7–10Bilateral venography760
METHRO III [4]2003THR, TKR (strat.)Melagatran 3 mg b.i.d. for 1–2 days, then ximelagatran 24 mg b.i.d.Enoxaparin 4000Blind4–12 h postop12 h preop7–11Bilateral venography2788
US 237 [5]2003THRXimelagatran 24 mg b.i.d.Enoxaparin 3000 × 2BlindMorning after surgery >12 h postop (20 h)Morning after surgery >12 h postop (20 h)7–12Unilateral venography1816
EXPRESS [6]2003THR, TKR (strat.)Melagatran 2 mg, then 3 mg b.i.d. for 1–2 days, then ximelagatran 24 mg b.i.d.Enoxaparin 4000BlindKnife-to-skin12 h preop8–11Bilateral venography2764

We first performed an analysis on the two groups of studies investigating, respectively, preoperatively administered melagatran/ximelagatran and postoperatively administered melagatran/ximelagatran. Preoperative administration seemed to be more effective than postoperative administration as regards the effect on total DVT (preoperative melagatran/ximelagatran RR = 0.68 (95% CI, 0.51–0.89) vs. postoperative melagatran/ximelagatran RR = 1.14 (95% CI, 0.77–1.69); heterogeneity test between subgroups, P = 0.03) as well as proximal DVT [preoperative melagatran/ximelagatran RR = 0.37 (0.25–0.55), vs. postoperative melagatran/ximelagatran RR = 1.28 (0.59–2.77); heterogeneity test between subgroups, P < 0.01]. Conversely, preoperative administration seemed to increase the proportion of patients requiring a transfusion [preoperative melagatran/ximelagatran RR = 1.08 (1.04–1.14) vs. postoperative melagatran/ximelagatran RR = 0.95 (0.90–0.99); heterogeneity test between subgroups, P < 0.01], and also of severe bleeding [preoperative melagatran/ximelagatran RR =2.51 (1.60–3.92) vs. postoperative melagatran/ximelagatran RR = 0.85 (0.52–1.40); heterogeneity test between subgroups, P < 0.01].

Using a regression model, a significant correlation was found between the timing of first administration of melagatran/ximelagatran and the RR of efficacy and safety criteria evaluated in each of the selected studies. Such a timing–effect relationship was observed not only for total DVT (R2 = 0.60; test of slope, P = 0.04), but also for proximal DVT (R2 =0.90; P < 0.01) and for severe bleeding (R2 = 0.72; P = 0.03) (Fig. 1). The relationship for transfusion did not reach statistical significance although there was a trend in the same direction as severe bleeding.

Figure 1.

Meta-regression analysis of the timing of melagatran/ximelagatran 24 mg b.i.d. initiation and effect on proximal deep venous thrombosis (DVT) and on severe bleeding. All studies included in the meta-analysis are represented. PR: relative risk; open squares denote observed relative risks of proximal DVT; solid circles denote observed relative risks of severe bleeding. The curves are calculated by weighted linear regression by the reciprocal of the variance of the logarithm of PR; time of first administration-response curves are solid lines, and 95% confidence limit intervals are dotted lines. R2 is the coefficient of linear correlation; P is the probability that the slope of regression differs significantly from zero.

The results of our meta-analysis suggest that the timing of first administration of ximelagatran 24 mg b.i.d. whether or not preceded by subcutaneous melagatran may have affected the benefit–risk ratio of this novel antithrombotic vs. LMWH and that the magnitude of efficacy is negatively correlated with safety (higher the efficacy, lower the safety, Fig. 1). Therefore, the optimal timing of first administration of melagatran/ximelagatran 24 mg b.i.d. will depend on whether one favors the antithrombotic efficacy or safety.

Our analysis combined studies with different types and regimen of LMWH. Thus, the time–effect relationship is tentative and should be interpreted with caution. However, LMWH were started either 12 h preoperatively or more than 12 h postoperatively (none were started close to surgery) and recent guidelines suggest that for LMWH there are only small differences between these two schedules [11]. Furthermore dose of LMWH did not influence the melagatran/ximelagatran vs. LMWH comparison (data not shown).

If physicians are satisfied with the efficacy of LMWH and reducing the risk of bleeding is a major concern then the appropriate timing of melagatran/ximelagatran 24 mg b.i.d. could be an administration starting postoperatively (4–12 h), which though not necessarily more effective than LMWH, is perhaps safer (significant 5% risk reduction for transfusion), is easier to administer, and does not present any risk of heparin-induced thrombocytopenia [12].

However, it is uncertain whether 24 mg ximelagatran b.i.d. is the optimal dose as a recent phase 3 study showed that ximelagatran 36 mg b.i.d. was more efficacious and as safe as 24 mg b.i.d. when initiated the morning after surgery (median time 20 h postoperatively) [13]. These results raise an important issue for the development of antithrombotics for the prevention of postoperative thromboembolic events in orthopedic surgery. In the current context, it is indeed crucial that studies be planned to investigate the combined effect of dose and timing, as has been done for another antithrombotic, NAPc2 [14].

In conclusion, our results support the hypothesis of a relationship between venous thromboembolic or hemorrhagic risk and timing of first administration of melagatran/ximelagatran and suggest in terms of benefit–risk ratio a postoperative administration of melagatran/ximelagatran for thromboprophylaxis in major orthopedic surgery.


The authors thank Andreas Theodorou (MediBridge Medical Affairs, Vélizy, France) for English translation.