Comparative effectiveness and safety of apixaban vs. rivaroxaban in patients with cirrhosis and portal vein thrombosis: An analysis of real-world data Free

Background: Portal vein thrombosis (PVT) affects approximately 14% of cirrhosis patients and is associated with increased morbidity and mortality. Anticoagulation is recommended to treat PVT despite low-certainty evidence, largely derived from studies of heparin and warfarin. Recent evidence suggests direct oral anticoagulants (DOACs) may offer comparable effectiveness and less bleeding than warfarin. Apixaban and rivaroxaban are the most commonly used DOACs and differ in hepatic metabolism. Yet, there is no head-to-head comparison between the two drugs in a real-world cohort of cirrhotic patients with PVT.

Methods: We conducted a retrospective new-user active comparator cohort study using Optum's de-identified Clinformatics® Data Mart Database from January 1, 2013 to June 30, 2022. We included individuals who were ≥18 years; initiated treatment with apixaban or rivaroxaban (cohort entry date); had a diagnosis of cirrhosis identified by ≥2 outpatient or ≥1 inpatient claims using ICD-9/10 codes; had a diagnosis of PVT within 33 days prior to DOAC initiation; and had ≥183 days of continuous healthcare enrollment prior to cohort entry. We excluded individuals who used other oral anticoagulants at baseline and those with hepatobiliary malignancy.

The primary effectiveness outcome was a composite of hospitalization for hepatic decompensation (ascites, encephalopathy, hepatorenal syndrome, or spontaneous bacterial peritonitis) or recurrent PVT. Venous thromboembolism (VTE) was a secondary effectiveness outcome. The primary safety outcome was a composite of hospitalization for gastrointestinal (GI) or intracranial bleeding. We stratified the primary outcomes by individual components.

Follow-up extended from cohort entry until the earliest primary safety or effectiveness outcome, liver transplant, anticoagulant switch, disenrollment defined as a gap ≥ 30 days, or study end. We used an intention-to-treat approach.

We used propensity score (PS) matching to balance baseline differences. We matched apixaban to rivaroxaban users with up to 3:1 matching and a caliper of 0.2 standard deviations of the logit of PS. Covariate balance was assessed using standardized differences (<0.1 indicating acceptable balance). Cox proportional hazards models estimated marginal hazard ratios (HRs) with 95% confidence intervals (CIs). Sensitivity analyses included alternative caliper widths, exclusion of patients with baseline VTE or atrial fibrillation, shortening the baseline to 90 days, and expanding outcome definitions to outpatient claims.

Results: The unmatched cohort included 494 patients (378 apixaban, 116 rivaroxaban). Apixaban users were older and had higher rates of hepatic decompensation, chronic kidney disease, and diuretic use. After matching (206 apixaban, 106 rivaroxaban), baseline covariates were well balanced (all standardized differences were <0.1). Median follow-up was 155 days for apixaban and 317 days for rivaroxaban.

The primary effectiveness outcome occurred in 19 apixaban users (9.9 events per 100 person-years) and 12 rivaroxaban users (8.1 events per 100 person-years) (HR 1.01, 95% CI 0.50–2.06). Recurrent PVT occurred in 6 apixaban users and 1 rivaroxaban user; hepatic decompensation occurred in 13 apixaban and 11 rivaroxaban users. Fewer than five apixaban users experienced VTE, with none in the rivaroxaban group.

The primary safety outcome occurred in 6 apixaban users (3.0 events per 100 person-years) and 7 rivaroxaban users (4.7 events per 100 person-years) (HR 0.62, 95% CI 0.19–2.12). GI bleeding occurred in 6 apixaban and 7 rivaroxaban users; esophageal variceal bleeding occurred in fewer than five users in both groups. No intracranial hemorrhages were observed. Sensitivity analyses produced consistent results.

Conclusions: In this real-world cohort of patients with cirrhosis and PVT, apixaban and rivaroxaban were associated with similar effectiveness. Use of apixaban was associated with a lower incidence of bleeding than rivaroxaban, although the CI included the null value of one. These findings are consistent with its favorable pharmacokinetic profile in hepatic impairment and with observations in other populations. The study's small sample size and wide confidence intervals limited our ability to make definitive conclusions about the relative effectiveness and safety of the two anticoagulants. Larger prospective studies with granular clinical data are needed to guide anticoagulant selection in this high-risk population.