Venous thromboembolism in cancer clinical trials: recommendation for standardized reporting and analysis


Agnes Y. Lee, Division of Hematology, Diamond Health Care Center, University of British Columbia, 2775 Laurel Street 10th floor, Vancouver, BC V5Z 1M9, Canada.
Tel.: 604 875 4952; fax: 604 875 4696.


Carrier M, Khorana AA, Zwicker JI, Lyman GH, Le Gal G and Lee AYY on behalf of the subcommittee on Haemostasis and Malignancy for the SSC of the ISTH. Venous thromboembolism in cancer clinical trials: recommendation for standardized reporting and analysis. J Thromb Haemost 2012; 10: 2599–601.

Venous thromboembolism (VTE), including deep vein thrombosis (DVT) and pulmonary embolism (PE), is a common morbidity and major cause of death in patients with cancer [1]. Thrombosis of other venous sites, including splanchnic circulation, renal veins and cerebral sinuses, as well as catheter-related thrombosis, also occurs. Although the risk of VTE is greatly dependent on the type and stage of cancer, it is also influenced by the type of cancer treatment (surgery, chemotherapy, hormonal therapy, immunomodulatory drugs and antiangiogenic agents), use of supportive measures (transfusions, erythropoietin stimulating agents and catheterization) and patient-related factors (age, previous VTE, immobility and co-morbidities) [2–6]. To accurately estimate the incidence and prevalence of VTE and determine the need for thromboprophylaxis, rigorous methodology using standardized definitions and analytical methods is needed.

The most appropriate research strategy to estimate the risk of VTE is a prospective cohort study using predefined definitions and meticulous and complete follow-up in a well-defined population. Such studies are expensive and require considerable time and resources. Consequently, the incidence of VTE in patients with cancer has been derived largely from retrospective studies from single institutions, post-hoc analyses of population-based or administrative type databases, adverse event reporting from randomized controlled trials that evaluate different cancer interventions, and systematic reviews of the published literature. In addition to the imprecision of using retrospective data, the accuracy of the incidence of VTE estimated using these types of studies is further limited by the wide variation of definitions for VTE used in individual studies. It is also important to consider if competitive risk analysis was performed to appropriately account for death as a competing risk with the thrombotic event of interest.

Conventionally, clinical trials in oncology use pre-established definitions from the National Cancer Institute’s Common Toxicity Criteria (version 2. 0 or 3.0) or National Cancer Institute Common Terminology Criteria for Adverse Events (version 4.0) (Table 1) to classify adverse events. Unfortunately, the definitions for various thrombotic events and the grading of the severity are not suitable for reporting VTE. Ideally, the classification of thrombotic events should discriminate between: (i) arterial vs. venous thrombosis; (ii) deep vs. superficial vein involvement; (iii) location of the involved deep venous site(s); (iv) catheter-related vs. not catheter-related DVT; and (v) symptomatic (expected) vs. incidental (unexpected) events. These distinctions are important because these events have different histories, require different treatments and vary in their outcomes. Without appropriate definitions and consistent reporting of VTE, it is impossible to accurately determine if there is a need for an intervention (i.e. thromboprophylaxis) and if systematic reviews can legitimately summarize the data across studies. A recent systematic review assessing the rate of VTE in patients with multiple myeloma undergoing immunomodulatory therapy highlights this problem and raises the concern that thromboprophylaxis is recommended for myeloma patients receiving a thalidomide- or lenalidomide-based regimen based on moderate to poor quality evidence (iv). Finally, these classifications for toxicity reporting were shown to underestimate the incidence of VTE and miss a majority of events when chart audit was conducted in patients with advanced colorectal cancer [7]. Therefore this could lead to a significant underestimation of the burden of VTE in certain cancer populations and potentially not detect an increased risk of VTE associated with certain types of chemotherapy regimens.

Table 1.   Toxicity criteria scales for reporting adverse events in oncology trials
Cancer Institute Common Terminology Criteria (version 3.0)
  1. CVA, cerebrovascular event; DVT, deep vein thrombosis; PE, pulmonary embolism. *A disorder characterized by occlusion of a vessel by a thrombus that has migrated from a distal site via the blood stream.

Thromboembolism*Venous thrombosis (e.g. superficial thrombosis)Venous thrombosis (e.g. uncomplicated DVT); medical intervention indicatedThrombosis (e.g. uncomplicated PE, non-embolic cardiac mural thrombus); medical intervention indicatedLife-threatening (e.g. PE, CVA, arterial insufficiency); hemodynamic or neurologic instability; urgent intervention indicatedDeath
National Cancer Institute’s Common Toxicity Criteria (version 4.0)
Thrombosis/embolismNoneDVT, not requiring anticoagulantDVT, requiring anticoagulant therapyEmbolic event including PE

The primary objectives of this ISTH statement are to propose a standardization of the definitions for VTE and to propose that competitive risk analysis be used to report the incidence of VTE in oncology studies. These are necessary to ensure accuracy and consistency and facilitate summary of results across studies.

Definitions of thrombotic events in oncology trials

We recommend that venous thromboembolic events be described according to the following criteria and rationale (Box S1).

  • 1 Arterial vs. venous thrombosis. These events are different in terms of treatment, outcomes and impact on quality of life. Arterial thrombotic events may be further classified according to the arterial circulation involved, such as myocardial infarction, peripheral vascular event or stroke.
  • 2 Deep vs. superficial vein involvement. Thrombosis and/or inflammation of a superficial vein in the arms typically occur at the site of a venipuncture or an intravenous catheter and often do not require systemic anticoagulant therapy. This complication is commonly referred to as superficial thrombophlebitis. Thrombosis of the deep veins requires systemic anticoagulation and can have significant and serious sequelae.
  • 3 Regional location of the involved deep veins. It is useful to distinguish thrombosis involving the legs vs. upper extremities vs. lungs (pulmonary embolism) vs. other sites (e.g. portal or cerebral circulation). Naming of the specific veins involved (e.g. right common femoral vein), as well as a description of the extent of the thrombotic burden (e.g. subsegmental vs. segmental PE) is strongly encouraged.
  • 4 Catheter-related vs. no catheter involvement: catheter-related thrombosis commonly occurs in the veins of the upper extremities and central venous circulation. The natural history, outcomes and treatment response are different from thrombosis events not involving a catheter [8].
  • 5 Symptomatic vs. incidental: due to the increasing use of serial CT scans to follow disease response to cancer treatment, venous thrombosis has become a common finding in clinical practice. Such incidentally discovered events are usually unexpected, but they may or may not be associated with clinical symptoms. Because it remains uncertain whether such unexpected thrombotic events are clinically relevant or comparable in their natural history or response to therapy as symptomatic, expected thrombotic events, it is important to distinguish them in reporting and follow-up.

All VTEs should also satisfy standard diagnostic criteria for venous thrombosis in the general population [9–11] (Box S2 on-line appendix). Diagnosis must be confirmed using standard, objective imaging techniques (e.g. venous ultrasonography, venography, computed tomographic pulmonary angiography, etc.) and should not be based on clinical suspicion alone.

Analysis of thrombotic events in oncology populations

Competing risk analysis is the most appropriate statistical methodology to evaluate the probability of VTE in cancer studies [12]. In clinical trials that include patients with advanced malignancy, death is an important and frequent competing risk for thrombosis [13]. Consequently, using the Kaplan and Meier approach to assess the cumulative incidence of VTE, which ignores the effect of death, will overestimate the incidence of VTE [12]. Furthermore, when determining the effect of an intervention on the incidence of VTE in a randomized controlled trial, the Gray’s [14] test should be utilized to compare cumulative incidence rates of VTE in order to adjust for the presence of competing risks. Ideally, VTE should be included as a secondary or safety endpoint in phase III trials of cancer treatment studies if an increased risk of VTE is noted in early-phase studies to allow accurate accounting of this complication in a large sample of patients. Transparency in reporting both the relative risk and absolute risk reductions with thromboprophylaxis is also essential.

VTE is associated with a poor prognosis in cancer patients. Recently, successful anti-cancer agents (antiangiogenic, immunomodulatory agents, etc.) have seemed to increase the risk of VTE in cancer patients. However, inconsistent reporting of thrombotic events in cancer clinical trials undermines the accuracy of the true event rates, which can lead to inappropriate use of thromboprophylaxis and inadequately powered clinical trials to assess these important questions. We recommend standardizing the reporting and analysis of VTE to facilitate accurate and proper assessment of the burden of this complication and the impact of potential interventions.

Disclosure of conflict of interests

The authors state that they have no conflict of interest.