In recent years, direct oral anticoagulants (DOACs) have been adopted as a treatment option for cancer associated thrombosis (CAT). Randomized trials comparing anti-Xa drugs to low molecular weight heparin (LMWH) showed treatment with DOACs conferred less risk of recurrent venous thromboembolism (VTE), but found higher rates of clinically important bleeding, especially in patients with gastrointestinal (GI) malignancies. Given these findings, there is a need for additional data regarding the safety of DOAC use in GI malignancies. Here, we report bleeding events of GI cancer patients treated with anticoagulation in a large centralized CAT clinic.


We evaluated a prospective cohort of patients referred to our CAT clinic from 8/2014-10/2019. Patients with primary gastrointestinal malignancies treated with therapeutic anticoagulation with LMWH or a DOAC for acute VTE were included. Bleeding was defined using the ISTH criteria for major and clinically relevant non-major bleeding (CRNMB). Bleeding rates were compared between luminal [anus/anal, colon, esophagus, rectal, stomach] and extraluminal GI cancers [gallbladder, liver/bile duct, and pancreas]. Patient characteristics associated with bleeding were evaluated with Fisher's exact test and the association of age with bleeding was analyzed by Wilcoxan rank sum test.


Of 463 patients with acute VTE, 73 patients (15.8%) with primary GI tumors were included in the analysis. Males comprised 57.5% of the population, median age was 62 (range 36-86), and 61.6% had stage 4 disease. Figure 1 shows a breakdown of tumor types. Enoxaparin was the most commonly used anticoagulant (n=48, 65.8%), followed by DOAC (n=25, 34.2%). Overall, 16 (21.9%) patients had a bleeding event within 6 months of treatment (7 major bleeds and 9 CRNMB). There was no difference in 6-month bleeding rate between patients treated with LMWH (n= 9, 18.8%) vs. DOAC (n=7, 28.0%), p=0.39. None of the clinical factors analyzed were significantly associated with bleeding (Table 1). There was no difference in bleeding rate in patients with luminal GI cancers vs. extraluminal GI cancers and no difference was found in a three-way association between site, treatment, and bleeding, p=0.40 (Table 2).


In our centralized cancer thrombosis clinic, patients with GI malignancies had similar rates of major and CRNMB when treated with LMWH or DOAC. In both cohorts, bleeding rates were high within 6 months of starting anticoagulation. There were no statistically significant differences in bleeding rates based on several clinical characteristics evaluated in this study. Although limited by a small patient population, this study adds to the knowledge of treating GI malignancies with DOACs. There is a need for further prospective evaluations regarding the safety of DOAC use in GI cancer patients and there remains an unmet need for antithrombotic treatments that do not increase bleeding potential.


McCrae:Novartis: Honoraria; Momenta Pharmaceuticals: Consultancy; Rigel: Consultancy; Dova: Consultancy. Khorana:Bayer: Honoraria; Pharmacyte: Honoraria; Pharmacyclics: Honoraria; Array: Other: Research funding (to institution); Merck: Research Funding; Leap: Research Funding; BMS: Honoraria, Research Funding; Seattle Genetics: Honoraria; Pfizer: Honoraria; Sanofi: Honoraria; Medscape: Honoraria; Leo Pharma: Honoraria; Janssen: Honoraria.

Author notes


Asterisk with author names denotes non-ASH members.