Extended prophylaxis with bemiparin for the prevention of venous thromboembolism after abdominal or pelvic surgery for cancer: the CANBESURE randomized study


Professor Vijay V. Kakkar, Thrombosis Research Institute, Emmanuel Kaye Building, Manresa Road – Chelsea, London SW3 6LR, UK.
Tel.: +44 207 351 8301; fax: +44 207 351 8317.
E-mail: president@tri-london.ac.uk


Summary. Background: There is not enough clinical evidence to make a strong recommendation on the optimal duration of thromboprophylaxis using low-molecular weight heparins (LMWH) in patients undergoing major cancer surgery. Patients and methods: CANBESURE is a randomized, double-blind study which enrolled patients admitted for abdominal or pelvic surgery for cancer. They received 3500 IU of bemiparin subcutaneously once daily for 8 days and were then randomized to receive either bemiparin or placebo for 20 additional days. Bilateral venography was performed after 20 days and evaluated blinded. The primary efficacy outcome was the composite of deep vein thrombosis (DVT), non-fatal pulmonary embolism (PE) and all-cause mortality at the end of double-blind period. Major venous thromboembolism (proximal deep-vein thrombosis, non-fatal pulmonary embolism and venous thromboembolism-related deaths) was also evaluated. The primary safety outcome was major bleeding. Results: Six hundred and twenty-five and 488 patients were included in the safety and main efficacy analyzes, respectively. The primary efficacy outcome occurred in 25 out of 248 patients (10.1%) in the bemiparin group and 32 out of 240 (13.3%) in the placebo group (relative risk reduction 24.4%; 95% CI: −23.7–53.8%; P = 0.26). At the end of double-blind period, major venous thromboembolism occurred in 2 (0.8%) and 11 (4.6%) patients, respectively (relative risk reduction 82.4%; 95% CI: 21.5–96.1%; P = 0.010). No significant difference was found in major bleedings. Conclusions: Four weeks compared with 1 week of prophylaxis with bemiparin after abdominal or pelvic cancer surgery did not significantly reduce the primary efficacy outcome, but decreased major venous thromboembolism (VTE) without increasing hemorrhagic complications.


Major abdominal and pelvic surgery for cancer puts patients at a high risk of post-operative venous thromboembolism (VTE) [1]. Such patients show a 2-fold increase in risk for post-operative deep-vein thrombosis (DVT) and a 3-fold increase in risk for fatal pulmonary embolism (PE), when compared with patients without cancer having similar surgery [2]. Based on the results of several randomized, double-blind clinical trials addressing the efficacy and safety of low-molecular weight heparins (LMWH) in preventing the above complications, LMWH are currently recommended for in-hospital prophylaxis [2–4]. However, in such patients the risk of VTE persists for several weeks after discharge from hospital [5–7]. A meta-analysis of three randomized clinical trials (one double-blind and two open) including 1104 surgery patients with and without cancer revealed that patients receiving extended LMWH prophylaxis for 1 month after major surgery showed a significant reduction in total and proximal DVT when compared with those receiving in-hospital prophylaxis alone [8]. However, these studies had a number of limitations, for example none of them included all-cause mortality in the primary efficacy outcome. Thus, extended prophylaxis as a therapeutic strategy, has not received a strong recommendation from international guidelines [2–4]. To provide additional clinical evidence, we performed the present double-blind, randomized study in patients undergoing abdominal or pelvic surgery for cancer and evaluated the efficacy and safety of prolonging thromboprophylaxis with the second-generation LMWH, bemiparin [9].

Patients and methods

Study population

Consecutive patients aged 40 years or older admitted to undergo elective, open, curative or palliative surgery for a malignant disease of the gastrointestinal (excluding esophagus) tract, genitourinary tract or female reproductive organs, were included in the study. The exclusion criteria were: active hemorrhage or high risk of bleeding; known hypersensitivity to unfractionated or fractionated heparins, radiological contrast media or anesthetic drugs; tumor of/ surgical intervention in the central nervous system within the previous 6 months; endocarditis; treatment with oral or parenteral anticoagulants within 5 days before surgery; history of heparin-induced thrombocytopenia or a baseline platelet count < 75 000 μL−1; severe renal or hepatic insufficiency; severe arterial hypertension; VTE within the previous 3 months; unable to comply with the study treatment and/or follow-up; cava filter; receiving prohibited medications; and pregnancy or lactation. Surgery for cancer in the liver, biliary tract or pancreas was also excluded.

Patients underwent surgery with general or spinal anesthesia, with an estimated duration of over 30 min.

Study design and drugs

The CANBESURE study was a prospective, double-blind, placebo-controlled, randomized trial. Before randomization, all patients received once daily (o.d.) subcutaneous injections of bemiparin 3500 IU (0.2 mL) for 8 ± 2 days, the first dose starting 6 h after surgical wound closure. A computer-generated randomization list with random numbers in permuted blocks of four was used. The randomization was stratified per center and each number was assigned to patients in strict consecutive order of inclusion. The patients were then randomly allocated to receive o.d. subcutaneous injections of either bemiparin sodium 3500 IU (0.2 mL) or placebo (0.9% sodium chloride, 0.2 mL) for 20 ± 2 additional days. Bemiparin is a second-generation LMWH obtained by alkaline depolymerization of heparin sodium, with a mean molecular weight of 3600 Da, a long half-life (5.3 h) and an anti-Xa/anti-IIa activity ratio of 8:1 [9,10].

The day of the randomization was defined as day 1. Bilateral ascending venography was performed on day 18–22, 2 days before the last injection. A follow-up was performed 74–90 days after randomization.

Simultaneous administration of low doses of acetylsalicylic acid (< 125 mg day−1), or non-steroidal anti-inflammatory drugs with a half-life of < 20 h was permitted. Intermittent pneumatic compression, electrical stimulation or vena cava filters were prohibited, but gradual compression stockings were allowed.

The study was performed in accordance with local regulations and the Declaration of Helsinki. The protocol was reviewed and approved by the ethic committees of each participating center and written informed consent was obtained from all patients before entry.

The steering committee designed the study and had overall responsibility for the integrity of the data and, accuracy and interpretation of analyzes. This clinical trial is registered in ClinicalTrials.gov (reference number NTC00219973).

Outcome measures

The primary efficacy outcome was defined according to the former EMEA guidelines [11] as the combined incidence of total documented symptomatic and asymptomatic DVT, non-fatal PE and all-cause mortality at the end of the double-blind period. Secondary efficacy outcomes included combined incidence of any DVT, non-fatal PE and VTE-related mortality, as well as the isolated incidence of components of the aforementioned outcomes during both periods. When the revised EMEA guidelines came into effect [12], a protocol amendment was implemented before unblinding the data to prospectively evaluate the incidence of major VTE (composite of symptomatic and asymptomatic proximal DVT, non-fatal PE and VTE-related death) at the end of the double-blind and follow-up periods.

All deaths and symptomatic thromboembolic events were blinded, centrally evaluated and adjudicated by an independent committee composed of three experts. Non-fatal PE was verified by perfusion/ventilation lung scintigraphy, pulmonary arteriography or spiral computed tomography. VTE-related deaths included all deaths with a confirmatory autopsy, those anteceded by an objectively proven manifestation of VTE and sudden, otherwise unexplained deaths.

Elective bilateral venograms performed on day 20 ± 2 after randomization were blinded and centrally evaluated by an independent committee of five experts on vascular radiology, based on a previously described technique [13]. Venography was considered suitable by the central adjudication committee if both distal and proximal deep veins of both legs were viewed; patients were included in the efficacy analysis if a thrombus was detected in any of their deep veins even if the whole deep venous system was not viewed. All suspected symptomatic DVT cases before day 20 ± 2 were confirmed by means of unilateral venography or Doppler-ultrasound within 3 days from the onset of symptoms. When the result of these tests was negative, the patient remained in the study and the scheduled venography was performed according to the protocol.

The primary safety outcome was the incidence of major bleeding events during the double-blind period as adjudicated by an independent data safety monitoring board (DSMB) according to at least one of the following criteria: fatal bleeding, clinically overt bleeding associated with a fall in hemoglobin level of 20 g L−1 or more, clinically overt bleeding leading to transfusion of two or more units of packed cells or whole blood, retroperitoneal or intracranial bleeding, or clinically overt bleeding warranting treatment cessation. Secondary safety outcomes were: incidence of major bleedings from randomization to the end of the follow-up period and the incidence of minor bleedings (episodes that were clinically relevant but did not qualify as major bleedings) during both study periods. Small hematomas (not greater than 5 cm in diameter) at the injection site were not counted as bleeding events. The incidence of thrombocytopenia, serious adverse events and discontinuations as a result of adverse events were also recorded.

Statistical analysis

Sample size was calculated taking into account the occurrence of DVT and symptomatic PE in a previously published clinical trial on extended prophylaxis in a similar patient population with another LMWH [14]. This showed an incidence of 12% of late events with 1 week in comparison to 4.8% with 4 weeks of thromboprophylaxis, i.e. 60% relative risk reduction. We calculated that 446 assessable patients were necessary to demonstrate significant differences, with a two-sided type-I error of 5% and a power of 80%. Assuming that venography would not be assessable in 15% of patients, it was estimated that randomization of 526 patients was required.

The safety population included all randomized patients who had received at least one dose of the study medication.

Main efficacy analysis was performed on the ‘modified intention to treat’ population, which included all randomized patients who received at least one dose of randomized treatment and had an assessable venogram, or documented symptomatic DVT or PE, or died during the double-blind period.

Categorical data were compared using the χ2-test or Fisher’s exact test when requirements for the former were unfulfilled. The size of the effect of the active treatment was calculated by relative risk reduction (RRR) with the 95% confidence intervals (95% CI). P-values were two-sided, with a type I error rate of 5%.

All statistical analyses were made with sas version 9.1.3 (SAS Institute Inc, Cary, NC, USA).


Study population

Between July 2005 and February 2008, 703 patients were enrolled in 34 centers in three countries. Of these, 626 patients were randomized, 625 of which were included in the safety population and 488 patients in the modified intention-to-treat population. The reasons for exclusion were similar in both study groups (Fig. 1). Median duration of the double-blind treatment was 20 days in both the bemiparin and placebo groups. Both study groups had comparable baseline demographic and surgical characteristics, venous thromboembolic risk factors (Tables 1 and 2) and median duration of bemiparin treatment during the pre-randomization open period (8 days).

Figure 1.

 Patient disposition.

Table 1.   Baseline characteristics of patients in the safety population
 Bemiparin (n = 315)Placebo (n = 310)
  1. Data are mean (SD, range) or n (%), unless otherwise specified. *Data available for 309 and 298 patients, respectively. Data available for 315 and 309 patients, respectively. Data available for 314 and 308 patients, respectively. §Data available for 314 and 309 patients, respectively.

Age (years)64.1 (10.3, 37.1–89.7)64.6 (9.9, 40.0–90.7)
Gender (female)158 (50.2)137 (44.2)
Body mass index (kg m−2)*26.0 (4.3, 16.5–42.6)25.7 (4.0, 13.7–38.3)
Ethnic origin
 Caucasian313 (99.4)310 (100)
 Others2 (0.6)0 (0.0)
Venous thromboembolic risk factors
 History of venous thromboembolism2 (0.6)2 (0.6)
 Obesity58 (18.4)58 (18.8)
 Varicose veins37 (11.7)40 (12.9)
 Congestive heart failure11 (3.5)11 (3.6)
 Ischemic heart disease61 (19.4)63 (20.5)
 History of stroke and/or hemiplegia8 (2.5)3 (1.0)
 Thrombocytosis11 (3.5)6 (1.9)
 Chronic obstructive pulmonary disease15 (4.8)9 (2.9)
 Inflammatory bowel disease§3 (1.0)2 (0.6)
Table 2.   Surgical characteristics of patients in the safety population
 Bemiparin (n = 315)Placebo (n = 310)
  1. Data are n (%) or mean (SD, range). *Data available for 315 and 308 patients, respectively. Data available for 315 and 307 patients, respectively.

Location of tumor
 Gastrointestinal tract257 (81.6)247 (79.7)
  Colorectal205 (65.1)207 (66.8)
  Gastric45 (14.3)37 (11.9)
  Other gastrointestinal7 (2.2)3 (1.0)
 Urologic23 (7.3)24 (7.7)
 Female reproductive organs34 (10.8)37 (11.9)
 Retroperitoneal1 (0.3)2 (0.6)
Type of surgery
 Curative*258 (81.9)243 (78.9)
 Palliative*57 (18.1)65 (21.1)
Tumor staging (TNM)
 Stage IA45 (14.3)25 (8.1)
 Stage IB45 (14.3)49 (15.8)
 Stage IIA72 (22.9)68 (21.9)
 Stage IIB22 (7.0)19 (6.1)
 Stage IIIA8 (2.5)14 (4.5)
 Stage IIIB41 (13.0)51 (16.5)
 Stage IIIC40 (12.7)30 (9.7)
 Stage IV42 (13.3)54 (17.4)
Duration of surgery (min)153 (61, 31–440)157 (65, 37–475)
Type of anesthesia*
 General only302 (95.9)290 (94.2)
 General and regional9 (2.9)15 (4.9)
 Regional only4 (1.3)3 (1.0)

Venography evaluation was performed in 528 patients and considered not assessable in 51 (23 in the bemiparin arm and 28 in the placebo arm).

Efficacy outcomes

Table 3 shows the efficacy outcomes up to the end of the double-blind and the follow-up period. The primary efficacy outcome occurred in 25 (10.1%) patients in the bemiparin group and 32 (13.3%) in the placebo group, leading to a RRR of 24.4%, which was not statistically significant (95% CI −23.7–53.8%; P = 0.26). When deaths induced by reasons other than VTE were excluded, the RRR rose to 39.5% (95% CI −2.7–64.4%; P = 0.06). Major venous thromboembolism occurred in 2 and 11 patients in the bemiparin and placebo groups, respectively, leading to a RRR of 82.4% (95% CI 21.5–96.1%; P = 0.010). Although the difference in the frequency of total DVT between the two groups was not statistically significant, only one (0.4%) patient in the bemiparin group, as compared with eight (3.3%) in the placebo group developed proximal DVT, leading to a RRR of 87.9% (95% CI 4.0–98.5%; P = 0.02).

Table 3.   Incidence of events in the modified intention-to-treat population (main efficacy analysis)
Outcome, n (%)Bemiparin (n = 248)Placebo (n = 240)RRR (95% CI) (%)P-value*
  1. RRR, relative risk reduction; DVT, deep vein thrombosis; PE, pulmonary embolism; VTE, venous thromboembolism. *P-values were calculated using the χ2-test or Fisher’s exact test, as appropriate. Primary efficacy outcome was defined as the combined incidence at the end of double-blind period of total documented symptomatic and asymptomatic DVT, non-fatal PE and all-cause mortality. Major venous thromboembolism was the composite of symptomatic and asymptomatic proximal DVT, non-fatal PE and VTE-related deaths.

Double-blind period
Primary efficacy outcome25 (10.1)32 (13.3)24.4 (−23.7; 53.8)0.26
 DVT19 (7.7)29 (12.1)36.6 (−10.0; 63.4)0.10
  Proximal DVT1 (0.4)8 (3.3)87.9 (4.0; 98.5)0.02
  Distal DVT only18 (7.3)21 (8.8)17.1 (−51.8; 54.7)0.54
 Non-fatal PE0 (0.0)0 (0.0)
 Death (all-causes)6 (2.4)3 (1.3)−93.6 (−665.1; 51.0)0.50
Any DVT, nonfatal PE and VTE-related death20 (8.1)32 (13.3)39.5 (−2.7; 64.4)0.06
Major venous thromboembolism2 (0.8)11 (4.6)82.4 (21.5; 96.1)0.010
Double-blind plus follow-up periods
Death (all-causes)8 (3.2)6 (2.5)−29.0 (−266.4; 54.6)0.63
Any DVT, nonfatal PE and VTE-related death21 (8.5)32 (13.3)36.5 (−6.9; 62.3)0.08
Major venous thromboembolism3 (1.2)11 (4.6)73.6 (6.6; 92.5)0.03

During the double-blind period two symptomatic DVT (one distal and one proximal DVT in the bemiparin and placebo groups, respectively) occurred before the scheduled venography. No symptomatic DVT was detected during the follow-up period.

There were nine deaths during the double-blind period without a confirmatory autopsy and/or a positive objective test on VTE prior to death. Four patients died because of sudden unexplained cardiac arrest (one and three in the bemiparin and placebo groups, respectively) whereas the remaining five deaths (one pneumonia, two systemic infections related to the surgical procedure and two gastric cancer progressions) occurred in the bemiparin group.

There were five additional deaths during the follow-up period; two deaths in the bemiparin group were because of a sudden cardiac arrest and a gastric cancer progression, and the remaining three deaths occurred in the placebo group as a result of ischemic stroke, pancreatitis with renal failure and complications related to re-operations.

DSMB considered all deaths in the bemiparin group during the double-blind and follow-up periods to be unrelated to the study drug.

Safety outcomes

There were three major bleeding events during the double-blind period (Table 4): an upper gastrointestinal hemorrhage and a rectal hemorrhage that occurred in the bemiparin group, and a gastric hemorrhage in the placebo group. Within the follow-up period there was another major bleeding (rectal hemorrhage) in the placebo group (Table 4). None of the bleeding events were fatal or affected a critical organ.

Table 4.   Incidence of bleeding events in the safety population
Bleeding event, n (%)Bemiparin (n = 315)Placebo (n = 310)
  1. *Major bleeding during the double-blind period was the primary safety outcome.

During the double-blind period
 Major bleeding*2 (0.6)1 (0.3)
 Minor bleeding1 (0.3)1 (0.3)
During the double-blind plus follow-up periods
 Major bleeding2 (0.6)2 (0.6)
 Minor bleeding1 (0.3)1 (0.3)

The incidence and type of adverse event were similar in both groups (data not shown). No case of severe thrombocytopenia (platelet count less than 75 000 μL−1), or increases three times above the normal range of liver enzymes (AST or ALT) was detected during the double-blind period. Thirty-five patients (17 and 18 in the bemiparin and placebo groups, respectively) had serum potassium levels greater than 6.0 mmol L−1.


The trial shows a RRR of 24% on the predefined efficacy outcome that was not statistically significant. However, this RRR increased to 40% (P = 0.06) when deaths for reasons other than VTE were excluded from the analysis, and achieved a statistically significant reduction of 82.4% (P = 0.010) when the analysis was confined to the composite outcome of major VTE (proximal DVT, symptomatic non-fatal PE and VTE-related deaths). This advantage was not offset by an increase in risk of bleeding, and persisted throughout the 2-month follow-up period.

The incidence of overall VTE was slightly lower in the short-term prophylaxis group (13.3%) and higher in the long-term prophylaxis group (8.1%) than that reported in a recently published meta-analysis (14.3% and 6.1%, respectively). However, the figures are within their 95% CI (11.2–17.8% and 4.0–8.7%, respectively) [15].

Interestingly, we also obtained a notable risk reduction on proximal DVT, as demonstrated in a previous study [16] and recently confirmed in two meta-analyses [8,15].

Our results reinforce data reported in previous randomized trials addressing the clinical benefits of extending thromboprophylaxis in major abdominal and pelvic surgery through the use of LMWH [14,16,17], although each study presented caveats. In the ENOXACAN II study, almost one-third of the patients did not undergo venography or had un-interpretable venograms, and the authors did not describe the method used for randomization [14]; the FAME study was open-label and included a substantial proportion of patients with benign disease [16] and, the Lausen study which was stopped before recruiting the planned number of patients was open-label including almost one-third of patients with benign disease and a quarter underwent thoracic surgery [17]. In addition, all-cause mortality was not included in the main efficacy outcome in the above studies. Despite these caveats, taken together with our data there is increasing evidence that prolonging LMWH prophylaxis up to 4 weeks in patients undergoing major abdominal and/or pelvic surgery is associated with a favorable benefit-to-risk ratio, similar to that reported in studies of extended thromboprophylaxis in major orthopedic surgery [2,3]. Furthermore, a recently published economic analysis has demonstrated that post-discharge prophylaxis with LMWH up to 28 days is cost-effective for cancer surgery patients [18].

The strengths of our study are: double-blind design, prospective inclusion of consecutive patients, a priori definition of the study outcomes assessed by independent adjudication and safety committees, inclusion of sudden deaths that occurred without apparent explanation as potential PE events [12,19], high rate of adequate phlebographies assessed by five independent radiologists from three different countries, and the low rate of patients lost to follow-up. The main limitation is the proportion of patients in whom venography was not performed.

The failure to show an advantage in the primary efficacy outcome with the study drug deserves some comment. Possible explanations may include inadequate sample size, insufficient dose of the experimental drug and the selected study outcomes. Sample size was calculated based on the results achieved in the ENOXACAN II study [14], however, the CANBESURE study may not have been powered enough because the primary efficacy outcome included all-cause mortality. We selected the dose of bemiparin found to be effective and safe in preventing VTE in high-risk patients, such as those undergoing major orthopedic surgery [9,10] and those with a favorable benefit-risk ratio in normal clinical practice for thromboprophylaxis in cancer surgery [20]. The remarkable hypercoagulability that is associated with a variety of cancer patients, especially those in advanced stage, can account for small, often asymptomatic thrombi despite the administration of anticoagulants [1,21]. If true, the ability of bemiparin to reduce the rate of late proximal DVT, symptomatic VTE and PE-related deaths is meaningful and clinically relevant. The main study outcome was crucially determined by the distribution of deaths in the two study groups caused by disease states other than PE, and thus by definition not preventable by antithrombotic drugs. Indeed, all five such deaths occurred in the group of patients randomized to 1 month of extended prophylaxis and DSMB considered them to be related to the background disease or complications of cancer progressions and systemic infection and, therefore, unrelated to the study drug. The decision to include all-cause mortality among the study outcomes was taken to meet the criteria recommended by EMEA guidelines [11,12].

The positive findings in some secondary efficacy outcomes are highly consistent with the results obtained from previous similar studies but the influence of an increase in type I error cannot be eliminated.

According to current guidelines, all patients undergoing major surgical intervention for cancer should be considered for extended thromboprophylaxis with LMWH after hospital discharge for up to 4 weeks [2–4], but so far there has been limited evidence from available studies [8,15,22]. Although the tested bemiparin regimen failed to achieve a reduction in the primary efficacy outcome (which includes all-cause mortality), a statistically significant and clinically relevant reduction in major VTE was observed without increasing the risk of major bleeding. Our findings provide further evidence to extend the duration of prophylaxis with LMWH in cancer patients undergoing extensive abdominal and pelvic surgery.


We thank the patients for their participation in this study. We also thank the members of the study committees, investigators at each site and the following individuals from Rovi: Rosaura Maeso, Begoña García and Mayte Monreal for their study management, and Ignacio Ayani for his expert medical opinion.

Disclosure of Conflict of Interests

This study was sponsored by Laboratorios Farmacéuticos Rovi. S.A. J. Martínez-González is an employee of Rovi. J.L. Balibrea, V.V. Kakkar and P. Prandoni have received consulting fees as members of the trial steering committee. V.V. Kakkar and P. Prandoni have received lecture fees from Rovi. No other potential conflict of interest relevant to this article is reported. All authors contributed to the study concept, design and implementation, and to the content and development of this manuscript.


Steering Committee: V.V. Kakkar (United Kingdom), J.L. Balibrea (Spain), P. Prandoni (Italy).

Committee for Adjudicating Clinical Events: G. Agnelli (Italy), F. Lozano (Spain), M. Monreal (Spain).

Data Safety Monitoring Board: F. Fontcuberta (Spain), A. Gil (Spain), E. Rocha (Spain).

Committee for evaluating phlebographies: A. Hernandez-Lezana (Spain), X. Montanya (Spain), J. Muchart (Spain), M. Vigo (Italy), H. Walters (United Kingdom).

Principal Investigators: Romania (415 patients, 18 sites): D.M. Anastasiu, S. Bancu, M.F. Coros, A. Duse, C. Duta, N. Grigore, I. Ioiart, A. Maghiar, B. Martian, T. Nicola, S. Olariu, D. Pascut, M. Pop, D. Sabau, I. Simon, A. Spatariu, D. Valceanu, D. Tuculanu; Russia (24 patients, five sites): A.V. Djachuk, A.I. Gorelov, S.I. Gorelov, V.A. Kachenko, A.S. Pryadko; Spain (264 patients, eleven sites): A. Alarcón, A. Antequera, E. Arévalo, E. Bernal, E. De la Puerta, P. De Llano, J. De Oca, G. Errazti, R. Lecumberri, S. Lledo, A. Marinelli, L.F. Pérdigo, I. Seco.