Outcomes by Class of Anticoagulant Use for Nonvalvular Atrial Fibrillation in Patients With Active Cancer

Background The choice of anticoagulant agent for patients with nonvalvular atrial fibrillation (NVAF) in the setting of active cancer has not been well studied. Objectives The aim of this study was to compare the rates of cerebrovascular accident (CVA), gastrointestinal bleeding (GIB), and intracranial hemorrhage (ICH) in patients treated with direct oral anticoagulant agents (DOACs) compared with warfarin for NVAF in patients with active cancer. Methods This was a retrospective electronic medical record review of eligible patients treated at a cancer hospital. The outcome events were CVA; GIB; ICH; the composite of GIB, CVA, or ICH; and overall mortality. Propensity score matching (1:1) was conducted to select comparable patients receiving warfarin vs DOACs. Fine-Gray models were fitted for each outcome event. Results The study cohort included 1,133 patients (mean age 72 ± 8.8 years, 42% women), of whom 74% received DOACs (57% received apixaban). After propensity score matching, 195 patients were included in each anticoagulant agent group. When comparing warfarin with DOACs, there were similar risks for CVA (subdistribution HR: 0.738; 95% CI: 0.334-1.629); ICH (subdistribution HR: 0.295; 95% CI: 0.032-2.709); GIB (subdistribution HR: 1.819; 95% CI: 0.774-4.277); and the composite of GIB, CVA, or ICH (subdistribution HR: 1.151; 95% CI: 0.645-2.054). Conclusions Patients with active cancer had similar risks for CVA, ICH, and GIB when treated with DOACs compared with warfarin for NVAF.

Cancer is associated with both prothrombotic and increased bleeding risks, and the risks and benefits of anticoagulation must be weighed. Although data are limited on the preferred oral anticoagulant agent in patients with NVAF and active cancer, there have been previous randomized controlled trials and a recent metaanalysis that compared DOACs and low-molecular weight heparin (LMWH) in treating venous thromboembolism (VTE) in patients with active cancer. [7][8][9][10][11] These studies showed that DOACs were either noninferior or superior to LMWH in preventing recurrent VTE, with generally no increased incidence of major bleeding, although in certain studies clinically relevant nonmajor bleeding was increased compared with LMWH, primarily in patients with gastrointestinal malignancies. These data overall suggest that DOACs are safe in patients with cancer for the prevention of recurrent VTE but do not directly address their use in patients with NVAF. Current guidelines do not indicate which anticoagulant agent to choose for patients with vs without malignancy 12 ; furthermore, embolic and bleeding risk scores do not include the presence of active cancer or history of cancer. In this study, we aimed to assess the embolic and bleeding risk of patients with NVAF and active cancer and to compare ischemic and hemorrhagic events between patients receiving warfarin vs DOACs using adjudicated individual medical record review.    Table 1). The caliper width used was 0.2. The propensity score was obtained for each patient using this model, and 1:1 propensity scorematched cohorts were selected using the OnetoMa-nyMTCH macro, using a greedy algorithm. 16
In the whole cohort, the baseline risk scores for both thromboembolism and bleeding were higher for those on warfarin compared with DOACs (CHA 2 DS 2 -VASc Table 1). Approximately 42% were women, and the most common malignancies were hematologic (21%), genitourinary (20%), breast (15%), and gastrointestinal (12%). Aggregate numbers of types of cancer for patients listed in the "other" cancer category are listed in Supplemental Table 2. Differences in baseline characteristics on the basis of sex are provided in Supplemental Table 3.
When we artificially censored at 5 years, and before propensity score matching, 145 (20.5%) died, 59  Table 2). After matching, baseline characteristics were balanced between the DOAC and warfarin groups; Tables 1 and 2 show mean standardized differences before and after matching. In addition, a detailed description of differences between unmatched and matched cohorts is presented in Supplemental Table 4.

DISCUSSION
To our knowledge, we report one of the first realworld studies using clinical data in patients with active cancer and NVAF, comparing adjudicated outcomes between patients on DOACs and those on warfarin. We observed that there were no significant differences in CVA, ICH, and GIB in patients on DOACs compared with those on warfarin (Central Illustration).
Cancer is known to cause both a prothrombotic state and an increased risk for bleeding, which makes its particularly challenging to treat this patient population with optimal anticoagulation in the setting of NVAF. 17  Additionally, we did not observe any significant difference in composite event-free (CVA, GIB, or ICH) survival or overall survival in patients with active The majority of patients received apixaban and rivaroxaban in our propensity-matched cohort (47% and 46%, respectively) as displayed in Table 4.
Therefore, it is unknown whether the findings of our study are generalizable to all DOACs compared with warfarin vs predominantly apixaban and rivaroxaban.   Table 1.

DOAC vs Warfarin in Patients With Cancer
Although current guidelines support the use of DOACs as first-line agents for anticoagulation in patients with NVAF, they do not offer specific guidance for anticoagulation in patients with NVAF and cancer. 12, 26 As a result of the paucity of data, and a lack of clear guidance in a population with heightened concern for bleeding, DOACs have been shown to be underused in patients with cancer with NVAF. 27