Tumor Site and Metastases as Predictors of Venous Thromboembolism Recurrence among Active Cancer Patients with Incident Deep Vein Thrombosis or Pulmonary Embolism: A Population-Based Cohort Study.

Background: Active cancer is an independent predictor of recurrent venous thromboembolism (VTE; hazard ratio=2.2–4.2). However, whether the risk of VTE recurrence among active cancer patients can be further stratified by tumor site or presence of metastases is uncertain.

Objectives: To estimate and compare the cumulative incidence of VTE recurrence by tumor site and presence of metastases among active cancer patients with incident VTE.

Study Design and Population: Retrospective cohort study of Olmsted County, MN residents with active cancer who lived >1 day after a first-lifetime deep vein thrombosis or pulmonary embolism over the 35-year period, 1966–2000 (n=418). 25 patients with oral (6), bone (1), skin (4), soft tissue (3), eye (1) and other/unknown (10) were excluded, leaving 393 remaining patients.

Measurements: Baseline demographic and clinical characteristics, tumor site, presence and location of tumor metastases, and date of first recurrent VTE. Tumor site and metastases were tested as potential predictors of time-to-recurrent VTE using the Cox proportional hazards model.

Results: 114 of 393 patients with active cancer developed recurrent VTE over 764 person-years of follow-up. The cumulative incidence of recurrent VTE at seven, 30, 90, and 180 days, 1, 5, and 10 years was 1.9%, 10.3%, 18.3%, 22.3%, 28.7%, 44.5%, and 53.5%, respectively. After adjusting for age and gender, tumor site was an independent predictor of first VTE recurrence (p<0.01). The hazard of VTE recurrence was over 2-fold higher for lung cancer patients compared to the lowest recurrence group (kidney, bladder and other genitourinary cancer; HR=2.04, 95% CI: 0.88, 4.70, p=0.096), and the hazard for brain and pancreatic cancer was nearly 2-fold higher (HR=1.9 for each, p=0.22, 0.24, respectively). The hazard of recurrence among hematological; breast and ovarian; gastric, colon, and liver; and prostate cancer patients did not differ significantly from the lowest recurrence group. Univariately, the presence of metastases at cancer diagnosis was not associated with recurrent VTE (HR=1.16; 95% CI: 0.79, 1.69, p=0.45).

Conclusions: Tumor site is an independent predictor of VTE recurrence after adjusting for age and gender. In particular, lung, brain and pancreatic cancer patients have the highest risk for recurrence and these patients should be considered for secondary prophylaxis. Disease severity, as reflected by metastatic disease at cancer diagnosis, is not a predictor of recurrence.