Correlation of Plasma Coagulation Parameters With Thromboprophylaxis, Patient Characteristics, and Outcome in the MEDENOX Study

The incidence of venous thromboembolism (VTE) in surgical patients is well known. The clinical benefit and cost-effectiveness of thromboprophylaxis in this patient population is well established, as it is for certain categories of medical patients such as those who suffer from stroke or myocardial infarction.1 The situation is less clear in most medical populations because of the heterogeneity in the studied patient populations and in the design of the available trials, as well as the lack of uniformity in the evaluated endpoint.

The MEDENOX study2 was a controlled, multicenter, double-blind, randomized study designed to evaluate the outcome of patients older than 40 years who were newly admitted for an acute medical condition and thought to be at moderate risk of developing VTE. The study had 2 objectives: (1) to establish the actual incidence of VTE in this population using the most accurate evaluation technique possible, and (2) to determine the efficacy of 2 different regimens of the low-molecular-weight heparin (LMWH) enoxaparin for the prevention of deep vein thrombosis (DVT) and pulmonary embolism.

Similar to all LMWHs, enoxaparin has a high anti-Xa– anti-IIa ratio. Although both activities have been correlated with the administered dose of LMWH,3–5 anti-Xa has been the most widely used measure for assessing LMWH activity. In animal models, no strong correlation between ex vivo anti-Xa activity and antithrombotic effect has been demonstrated,6–10 but some clinical studies have found a significant correlation with thrombotic or hemorrhagic outcome,11,12 and others have given negative results or weak correlations.13,14 Few data are available for anti-IIa activity.

D-dimers and thrombin-antithrombin complexes are considered markers of ongoing procoagulant activity, and their normal plasma level is probably a result of normal physiologic clotting activity in vivo.15 D-dimer concentrations below a cut-off value of 0.5 μg/mL (0.5 mg/L) measured with a sensitive technique have shown a good negative predictive value for the diagnosis of thromboembolic disorders,16–18 and their use as a diagnostic tool for VTE has been extensively studied.19 The evaluation of the relationship of D-dimers with thrombotic events before and after prophylaxis could thus help in predicting the risk of developing VTE and the potential benefit of thromboprophylaxis in the patients of the MEDENOX study who were admitted for an acute medical illness.

Therefore, during the MEDENOX study, a plasma library was collected to perform such analyses.

This substudy of the MEDENOX study investigates whether there was a correlation between plasma coagulation parameters and patient characteristics, thromboprophylaxis or the incidence of VTE in the context of the study.

This study was a phase III, prospective, multicenter, parallel- group, randomized, double-blind, placebo-controlled study evaluating the efficacy and safety of 2 different doses of enoxaparin (20 or 40 mg administered subcutaneously once daily for 10 ± 4 days) (Aventis Pharma, Antony, France) compared with a placebo for the prevention of VTE.2 

A total of 1102 patients who were admitted to a hospital for an acute medical illness and considered at moderate risk of VTE were enrolled in the MEDENOX study. Patients older than 40 years, with a projected hospital stay of 6 days or more and immobilized for 3 days or less were considered for inclusion in the MEDENOX study if they had congestive heart failure (New York Heart Association class III or IV), acute respiratory failure not requiring ventilatory support, or one of the following conditions if associated with at least one additional VTE risk factor: acute infection without septic shock, acute rheumatic disorders, or an episode of inflammatory bowel disease. Additional VTE risk factors were as follows: older than 75 years, cancer, previous VTE, obesity, varicose veins, hormone therapy, or chronic heart or respiratory failure.2 Exclusion criteria included women of child- bearing age if they were pregnant, breast-feeding, or not using contraception; patients who had suffered a stroke or undergone major surgery within the previous 3 months; patients with thrombophilia; patients with a serum creatinine concentration greater than 1.7 mg/dL (150.3 μmol/L); patients with bacterial endocarditis or other conditions that could increase the risk of hemorrhage; and patients requiring anticoagulant therapy and those who had received anticoagulant therapy for more than 48 hours. A full, detailed list of inclusion and exclusion criteria is provided by Samama et al.2 

The primary efficacy outcome was VTE between days 1 and 14; venography was systematically performed at day 10 (±4).

In a subset of 224 patients, blood samples were taken before thromboprophylaxis was administered, at randomization (day 0), and at the end of the prophylaxis period, 3 to 4 hours after the last dose of enoxaparin at day 10 (±4) (Figure 1).

The participation of recruiting centers in the plasma library was voluntary, but all patients recruited in those centers were included in this prospectively planned substudy. Given the relatively small study population size, care was taken when interpreting data from this substudy.

Tests were performed blindly in a central laboratory (Laboratoire de Thrombose Expérimentale, Université Paris VI, Paris, France). Anti-Xa and anti-IIa activities were measured using an amidolytic assay with a chromogenic substrate (Stago, Asnières, France), and D-dimer concentrations were determined with an enzyme-linked immunosorbent assay (Mérieux, Marcy-l'Etoile, France); thrombin-antithrombin concentrations also were determined with an enzyme-linked immunosorbent assay (Behring, Marburg, Germany).

The distributions of anti-Xa activities, D-dimers, and thrombin-antithrombin complexes at baseline and day 10 are presented graphically using box plots. Statistical comparisons of the distributions between groups (eg, groups receiving different prophylactic regimens, patients with or without DVT) were performed with the Wilcoxon rank sum test. Changes in anti-Xa activity from baseline to day 10 were compared between age groups and between patients with or without cancer at entry using analyses of variance. All P-values are 2-sided. P- values <.05 were considered indicative of statistical significance.

The complete results of the MEDENOX study have already been published.2 A total of 1102 patients were randomized in the study. Patient demographics and baseline characteristics were identical among the 3 groups, including the mean ages of patients (74.1 ± 10.6, 72.9 ± 10.1, and 73.1 ± 10.8 years in the placebo group, the enoxaparin 20-mg group, and the enoxaparin 40-mg group, respectively) (Table 1).2 

The overall incidence of VTE was 14.9% in the placebo group, and it decreased to 5.5% (a risk reduction of 63%) in the enoxaparin 40-mg group (P < .001). The incidence of proximal DVT decreased from 4.9% in the placebo group to 1.7% in the enoxaparin 40-mg group (a risk reduction of 65%) (P = .04). The group receiving 20 mg of enoxaparin had results similar to the placebo group.

Major hemorrhage occurred in only 1% of the patients during the period that thromboprophylaxis was given, and there were no significant differences between groups.

A subgroup of 224 patients was studied in the coagulation substudy. Figure 1 shows that the distribution of patients was similar for the 3 groups. Of the 189 patients on whom a venography was performed, 26 (14%) had a DVT, and 25 (11%) of the 224 patients showed some sign of bleeding.

The baseline characteristics of patients were similar in each group of this substudy as well as in the corresponding patient groups of the MEDENOX study (Table 1). The distribution of VTE risk factors and the reasons for the hospitalization of patients were also similar when comparing patients in each substudy group as well as in the corresponding MEDENOX study patient groups.

Table 2 summarizes the results of the laboratory endpoints. As expected, the median anti-Xa activity at day 10 in patients receiving 40 mg of enoxaparin was approximately twice that obtained with 20 mg (0.41 and 0.21 IU/ mL [410 and 210 IU/L], respectively), and there was only a mild increase in anti-IIa activity in both of the enoxaparin groups (0.03 and 0.02 IU/mL [30 and 20 IU/L], respectively). There was no significant difference in anti-Xa and anti-IIa activities in patients with or without VTE or in patients with or without bleeding (n = 25 and 198, respectively). Figure 2 illustrates the wide distribution of anti-Xa activities in each group receiving thromboprophylaxis. Anti-Xa activity did not correlate with the creatinine clearance rate in patients with a creatinine clearance rate greater than 30 mL/min (0.5 mL/s) (estimated with the Cockcroft and Gault equation) (Figure 3, A), with regard to age (Table 3; Figure 3, B) or the presence of cancer (Table 3). Anti-Xa levels were compared only in subsets of patients who had cancer or advanced age because these are well-known, readily identifiable, homogenous risk factors for VTE. Comparisons of Anti-Xa levels were not made in subsets of patients with other risk factors, as they were either not as uniformly identifiable and comparable or not as readily accepted as VTE risk factors.

Median D-dimer (normal, ≤0.5 μg/mL [0.5 mg/L]) and thrombin-antithrombin concentrations (normal, <4 μg/L [40.8 pmol/L]) prior to randomization were high in all 3 groups, and there were no significant differences among groups (Table 2). At day 10 (±4), at the end of therapy, D- dimer concentrations were similarly elevated in the placebo and enoxaparin 20-mg groups, and they were unchanged compared with values before thromboprophylaxis. However, the median D-dimer concentration had significantly decreased in the enoxaparin 40-mg group compared with both the level before thromboprophylaxis (P = .05) and the placebo group (P = .04). These significant decreases in the D-dimer concentration correlated with a clinically beneficial effect of 40 mg of enoxaparin. After therapy, thrombin-antithrombin concentrations in both the 20- and 40-mg groups were significantly lower than in the placebo group (P = .007).

Figure 4, A and B, shows the distribution of D-dimer and thrombin-antithrombin concentrations, respectively, at days 0 and 10 (±4) and in each of the 3 groups. The percentage of patients with a decrease in D-dimer and thrombin-antithrombin concentrations from baseline to day 10 (±4) was significantly higher in the 40-mg group than in the placebo group (59.2% vs 50.1%; P = .05 for D- dimers; and 66.7% vs 44.4%; P = .01 for thrombin-antithrombin). There was a wide overlap of values among groups.

Table 2 also shows D-dimer and thrombin-antithrombin concentrations according to the incidence of VTE. In patients with VTE, median D-dimer concentrations were significantly higher both at presentation (P = .01) and at day 10 (±4) (P < .001) than in patients without VTE. No correlation of thrombin-antithrombin with VTE status was found. Figure 5, A and B, shows the distribution of patients with and without DVT according to D-dimer and thrombin-antithrombin concentrations, respectively, at days 0 and 10 (±4). All thromboembolic events occurred in patients with D-dimer concentrations higher than normal levels both at presentation and at day 10 (±4). The predictive value of individual D-dimers for VTE was poor because of the broad range and large overlap of results in each subgroup.

Because only 12 of the 60 recruiting centers volunteered to contribute to the plasma library, only about 20% of the patients recruited in the MEDENOX study were analyzed in this substudy. Thus, the interpretation of the results is limited by the relatively small study population. Nevertheless, significant differences could be observed between the 3 groups receiving different thromboprophylactic regimens, and the baseline characteristics of patients in the substudy and the corresponding MEDENOX study patient groups were similar. The results of this coagulation parameters substudy were acquired in parallel with a large- scale study of clinical thromboprophylaxis in which bilateral venography was used for detecting DVT, allowing a unique comparison of plasma coagulation parameters and the incidence of confirmed thromboembolic events.

As expected, anti-Xa activities correlated with the subcutaneous dose of enoxaparin, with the 40-mg dose giving a median activity nearly twice as high as the 20-mg dose. This correlation is consistent with the results of several studies.3–5,20 Anti-IIa activities were much lower than anti- Xa activities, and a dose effect was not detectable.

The relationship between the prevention of VTE and anti-Xa plasma activity is more controversial. In MEDENOX, in the group with 40 mg of enoxaparin and a higher median anti-Xa activity, there was an important reduction (63%) in the incidence of VTE. In contrast, in the group with the 20-mg dose and intermediate anti-Xa activities, no reduction in the incidence of VTE was observed. This is in accordance with the correlation between plasma anti-Xa activity and thrombosis observed in several surgical studies,11,21,22 whereas no dose-effect correlation was demonstrated in other such trials.12–14,23 Despite the clear difference in median anti-Xa activities, the large overlap of individual values among groups (Figure 2) precludes the use of anti-Xa activity as an indicator of clinical antithrombotic efficacy in medical patients. However, a correlation between anti-Xa activities and VTE cannot be excluded. It is not clear why there was a low, but measurable, anti-Xa activity in some patients in the placebo group at day 10. It is possible that this resulted from heparin in the intravenous lines used to collect blood samples.

In the MEDENOX study, any patient with a creatinine clearance rate lower than 30 mL/min (0.5 mL/s) was excluded from the study. Our data suggest that in patients with mild or moderate renal impairment who receive prophylactic doses of enoxaparin, the peak anti-Xa activities at day 10 (±4) are not correlated with creatinine clearance rate, which has been observed previously in the study by Mahe at al.24 Although our data confirm no correlation between anti-Xa and creatinine clearance, the possibility of a correlation if the area under the curve is examined is not excluded, as in the study by Sanderink et al.25 

A significant correlation between anti-Xa activity and bleeding has been observed in surgical prophylaxis trials,12,26 but the data from the MEDENOX study, which included nonsurgical patients, did not allow such a correlation to be observed because of the very low rate of major bleeding during the prophylaxis phase (1.1%, 0.3%, and 1.7% in the placebo, the enoxaparin 20-mg group, and the enoxaparin 40-mg group, respectively) and the small proportion of samples available for analysis.

Median D-dimer and thrombin-antithrombin concentrations at baseline were higher than normal and were similar in all 3 groups receiving a placebo, 20 mg of enoxaparin, or 40 mg of enoxaparin, thus supporting the hypothesis that this patient population already had an increased clotting activity on admission, even before immobilization in a hospital bed. This situation is probably a result of the underlying presence, in most of the patients, of one of the numerous possible conditions likely to increase plasma D-dimers, including the advanced mean age of the population.2,19,27 This baseline elevation of clotting markers has been observed in other studies of surgical14,28 and medical29 populations.

Median D-dimer concentrations were decreased at day 10 only in the population of patients treated with 40 mg of enoxaparin and were associated with a significant decrease in the incidence of VTE. In contrast, the 20-mg dose, despite producing a measurable anti-Xa activity, showed no biologic effect on the procoagulant activity as measured by D-dimer concentrations and no measurable reduction in the incidence of VTE. Thus, in the MEDENOX study, we could demonstrate a link between the biologic activity of enoxaparin and clinical efficacy. Nevertheless, as in other studies, we have failed to demonstrate an association between D-dimer concentrations and anti-Xa activity produced by enoxaparin.17 

In all groups of patients, D-dimer concentrations (but not thrombin-antithrombin concentrations), both at baseline and day 10, were significantly higher in those who developed VTE than in those who did not. Furthermore, no venographically confirmed DVT was found in patients with a normal D-dimer concentration (<0.5 μg/L [0.5 mg/L]), confirming the good negative predictive value of the D-dimer test.

These findings confirm the high sensitivity of D-dimer plasma concentrations to thromboembolic disorders.14,16,30,31 It would appear that, in the MEDENOX study, thrombin-antithrombin measurements were sensitive to the presence of heparin but had less discriminatory value with respect to the presence of thrombosis.

The important overlap of population distributions did not allow us to discriminate for the effect of prophylaxis with enoxaparin (Figure 4) or the presence of VTE (Figure 5) on the parameters of clotting activity. D-dimers are very specific for fibrin, but the specificity of fibrin for VTE is poor because fibrin is produced in a variety of conditions (eg, cancer, inflammation, infection, tissue necrosis), and normal D-dimer concentrations are higher in the elderly.19,31 One or several of those conditions were present in a large proportion of the MEDENOX population; therefore, the clinical usefulness of D-dimer and thrombin-antithrombin measurements for identifying the presence of VTE in individual patients was very limited, the poor specificity of the test resulting in a poor positive predictive value. In contrast, our data show that all patients with VTE had D-dimer concentrations higher than normal, an observation in accordance with the good negative predictive value of a normal D-dimer test value for the diagnosis of thromboembolic disorders.16–18 

In conclusion, anti-Xa activity correlated with the dose of enoxaparin but with no other laboratory or clinical parameter, including creatinine clearance greater than 30 mL/min (0.5 mL/s). The correlation of the higher anti-Xa activity reached with the 40-mg dose of enoxaparin and the incidence of VTE confirmed a dose effect of the anticoagulant in preventing VTE. Median D-dimer and thrombin-antithrombin concentrations were increased at presentation in this medical population. D-dimer concentrations were significantly decreased at day 10 (±4) compared with day 0 only in the group receiving 40 mg of enoxaparin that showed a 63% reduction in the incidence of VTE. D-dimer concentrations were higher in the subgroup with VTE, but no predictive value of individual results for VTE could be demonstrated.

The recently deceased A. Eldor, MD, was a member of the MEDENOX steering committee. The MEDENOX study was supported by a grant from Aventis Pharma.