Anticoagulation and Major Bleeding or Recurrent Thrombosis in Patients with Isolated Cancer-Associated Splanchnic Vein Thrombosis: A Multi-Center Cohort Study

Introduction: The optimal treatment strategies for cancer-associated splanchnic vein thrombosis (CA-SpVT) are unclear due to a lack of high-quality evidence. To date, there are no randomized studies, and existing retrospective studies are predominantly single-center. We aimed to assess the association between treatment with anticoagulation (AC) and the incidence of bleeding and thrombotic outcomes in patients with isolated and bland CA-SpVT. Methods: We performed a multi-center retrospective cohort study in Boston and Houston, USA. Inclusion criteria included new SpVT diagnosed radiographically 2011-2022 in adult patients with active solid cancer or lymphoma (diagnosed or treated within the last 6 months) and ≥ 3 months of follow-up. Exclusion criteria included myeloproliferative neoplasms and leukemias, baseline use of AC, concurrent deep vein thrombosis (DVT) or pulmonary embolism (PE) at the time of CA-SpVT (not isolated), and well-characterized tumor thrombus or mixed tumor/bland thrombus. Outcomes were measured for 6 months following diagnosis of CA-SpVT and included major bleeding (MB) by ISTH criteria, usual-site PE/DVT, recurrence/progression of CA-SpVT, and overall mortality. To estimate the average treatment effect associated with AC on each outcome, we used inverse probability weighting (IPW) to account for measured confounders including cancer type, stage, SpVT location, and whether it was symptomatic (goal weighted standardized difference <0.1). Given significant differences in AC rates and outcomes for patients with hepatobiliary (HCC) vs. all others, HCC was examined as separate cohort in the weighted analysis. Results: A total of 437 patients (277 and 160 from each site) with incident, isolated, bland CA-SpVT were included in the final analytic cohort. Most common tumor types were HCC (n=192, 43.9%) pancreatic (n=90, 20.6%), and colorectal (n=48, 11.0%). Metastatic disease was present in 42.6% of patients. The majority (n=266, 60.9%) of SpVT were asymptomatic. Most SpVT occurred in portal veins (n=355, 81.2%), followed by mesenteric veins (n=113, 25.9%), and splenic vein (n=71, 16.3%). A minority of cases (n=144, 33.0%) received therapeutic AC. In the overall cohort, the 6-month incidence of MB was 9.8% (n=43), usual-site PE/DVT was 4.8% (n=21), SpVT progression was 11.7% (n=51), and mortality was 36.8% (n=161). Patients with isolated CA-SpVT and HCC had significantly different baseline characteristics and treatment patterns than those with other cancers. They were more likely to involve portal veins (91.2% vs. 73.5%, p<0.01) and less likely to be symptomatic (33.3% vs. 43.7%, p=0.03) or have metastatic disease (25.0% vs. 56.3%, p<0.01). Despite less likely to receive AC (19.8% vs. 43.3%, p<0.01), patients with CA-SpVT and HCC had fewer usual-site PE/DVT (2.6% vs. 6.5%, p=0.07) and similar incidence of SpVT progression (11.5% vs. 11.8%, p=1.00) and MB (10.9% vs. 9.0%, p=0.52). In the IPW-weighted analysis for patients with HCC (n=192), those treated with AC (n=38) had numerically higher risk of MB vs. untreated (n=154) (24.2% vs 9.1%, p=0.07), as well as usual-site PE/DVT (8.5% vs. 1.3%, p=0.16). Notably, several thrombotic events occurred after AC cessation due to MB. There was no difference in SpVT progression (11.5% vs 11.9%, p=1.00) or mortality (39.9% vs 32.9%, p=0.372). In the IPW-weighted analysis for patients with all other cancer types (n=245), those treated with AC (n=106) also had numerically higher risk of MB vs. untreated (n=139) (11.9% vs 6.7%, p=0.20); however, they had numerically lower risk of SpVT progression (7.9% vs. 15.4%, p=0.08). There was no difference in usual-site PE/DVT (6.3% vs. 5.9%, p=0.90) or mortality (39.3% vs 39.7%, p=0.95). Conclusion: In this multi-center retrospective cohort study, treatment of CA-SpVT with AC was associated with high rates of major bleeding. Notably, patients with HCC and AC had doubled MB rates compared to those with non-HCC, likely related to the underlying cirrhosis and coagulopathy. AC was associated with numerically decreased risk of SpVT progression in patients with non-HCC cancers. The lack of difference in usual-site PE/DVT likely reflects short duration of AC and residual confounding by indication. These data support a tailored approach to AC decision-making in patients with isolated CA-SpVT including considerations such as underlying cancer type and bleeding risk.