Direct oral anticoagulants versus vitamin K antagonists in patients with atrial fibrillation and cancer a meta-analysis

Atrial fibrillation (AF) is the most commonly diagnosed clinical arrhythmia and its prevalence increases with age, up to 18% at 85 years of age [1]. AF confers an increased risk of cardiovascular complications, including a fivefold risk of thromboembolic events, as such stroke and transient ischemic attack (TIA) [2], therefore anticoagulant therapy is recommended on the basis of individual thrombotic risk determined by CHA2DS2VASc risk score [3]. Given the high prevalence of malignances and AF in the elderly, the progressive aging of population will probably lead to an increased prevalence of cancer in AF patients. Currently, up to 25% of AF population has comorbid cancer [4]. On the other hand, AF is commonly diagnosed in cancer patients and may be related to shared comorbid states, inflammation, direct tumor effect, complications of cancer surgery or anti-cancer therapy [5‐8]. Anticoagulant management of AF population with cancer is challenging because of an the increased propensity for both thrombosis and bleeding of this population [7]. As a result, the search for an acceptable anticoagulation treatment is a major clinical issue, currently unsolved. Direct oral anticoagulants (DOACs) have been demonstrated non-inferior to vitamin K antagonists (VKAs) for stroke prevention in non-valvular AF patients [9‐12] with even better safety profile. Therefore, current European Heart Rhythm Association guidelines [13] recommend DOACs over VKAs as preferred anticoagulation strategy in patients with AF who are eligible for DOAC therapy. However, these recommendations cannot be extended to AF patients with malignances because in randomized clinical trials (RCTs) of DOACs for stroke and systemic embolism prevention in AF, cancer patients were underrepresented. Nevertheless, post-hoc analyses of RCTs of DOACs [14‐17] and retrospective population or cohort studies [18‐23] have shown promising results of DOACs compared to VKAs in non-valvular AF patients with cancer. Therefore, we aimed to systematically assess the available evidences in the literature regarding the safety and efficacy of DOACs in comparison to VKAs in patients affected by non-valvular AF and cancer.

To our knowledge, this is the largest meta-analysis comparing efficacy and safety of DOACs versus VKAs in patients with malignancies. The main findings of our study are as follows: (1) DOACs use resulted in lower rates of any stroke or systemic embolism, as compared to VKAs use; (2) DOACs were associated with safer profile risk than VKAs, as the use of DOACs resulted in a statistically significant reduction of major bleedings and intracranial or gastrointestinal bleedings; (3) in comparison to VKAs, DOACs were found to be non-inferior for the outcomes MI, cardiovascular death, all-cause death, major bleeding or non-major clinically relevant bleeding and any bleeding.

The best anticoagulation management in cancer patients is still debated in consideration of the unique clinical risk profile carried by malignancies. Indeed, cancer patients have higher rates of venous thromboembolism (VTE) and arterial thrombosis for inflammatory cytokines, tumor vascular invasion and vasculotoxic cancer therapies, whereas cancer-related thrombocytopenia and chemotherapy-related bone marrow suppression increase bleeding complications [6‐8]. As a result, in the past years concerns about bleeding complications and paucity of data have led to an underuse of DOACs in cancer patients with non-valvular AF. As reported by Ording et al. [20] from Danish population-based medical databases, only 15% of patients with cancer and AF are currently prescribed DOACs (vs. VKAs) in clinical daily practice. However, mounting evidences are demonstrating that DOACs could represent a valid choice in patients with cancer. Actually, Select-D Trial and Hokusai VTE Cancer trial respectively demonstrated that rivaroxaban and edoxaban were non-inferior to low-molecular-weight heparin (LMWH) in treatment of cancer-related VTE [28, 29], although at cost of increased bleeding complications. As a result, rivaroxaban and edoxaban are currently recommended for VTE treatment as alternative to LMWH in cancer patients with low gastrointestinal and genitourinary bleeding risk, low drug-drug interactions with DOACs and on the basis of patients’ preferences [30, 31]. The recently published ADAM VTE trial and Caravaggio Trial have shown the efficacy of apixaban as compared to dalteparin in the prevention of recurrence of cancer-related VTE, with similar bleeding rates among the study arms [32, 33]. Our meta-analysis is consistent with the effectiveness of DOACs in the management of cancer-related thrombosis. A previous meta-analysis by Deng et al. [34] found that DOACs were associated to statistically significant reduced rates of the composite outcome SSE, but no differences were found for the outcomes ischemic stroke. In line with previous report, in our analysis DOACs were related with lower rates of SEE, but in addition we found a significantly reduction of the outcomes ischemic stroke and hemorrhagic stroke in comparison to VKAs. Interestingly, better efficacy in preventing thromboembolic events was associated with a reduced risk of major bleeding or gastrointestinal and intracranial bleedings as compared to VKAs, while no differences where found for the outcomes major bleeding or clinically relevant non-major bleeding and any bleeding. An apparently better safety profile of DOACs in comparison to VKAs in cancer patients has never been clearly demonstrated so far, as in the meta-analysis by Deng et al. [34] DOACs showed borderline significant reduction of major bleeding and the reduction of gastrointestinal or intracranial bleeding was not consistent among different sensitivity analyses. Conversely, we found a statistically significant lower rates of major bleeding and gastrointestinal and intracranial bleeding that was consistent in all sensitivity analysis but one including only post-hoc analyses of RCTs. The better results obtained with DOACs vs VKAs on thromboembolic and bleeding events may be driven by different factors, as the pharmacodynamic and pharmacokinetic features of VKAs, whose anticoagulation activity relies on TTR. As previously reported by Kim et al. [35], the obtaining of an optimal range of international normalized ratio (INR) is difficult in patients with malignances receiving cancer therapy, so that the prevalence of patients with active cancer reaching a TTR > 60% during follow-up is as high as 10%. Although a suboptimal TTR during VKA therapy reduces anticoagulant activity and accounts for lower VKA efficacy, Kim et al. [19] showed that also in patients with optimal TTR DOACs still were more effective and safer than VKAs. Moreover, the benefits of DOACs over VKAs could be enhanced in patients with active cancer status, that has been defined in different ways across studies as a newly diagnosed cancer or a cancer receiving therapy during the study period or a cancer that received specific therapy within 6 months/1 year before starting anticoagulation (Supplemental material, Table S2). Patients with active cancer are more likely to undergo invasive anticancer treatment, such as surgery or biopsy, or pharmacologic anticancer therapy that may interact with anticoagulant drugs. Hence, active cancer is more likely associated to anticoagulant therapy interruption, reported as high as 69.2% by Fanola et al. [15] and 29% by Melloni et al. [16], for safety concerns about surgery or drug-drug interactions. In this setting, DOACs may offer advantages over VKAs in terms of both efficacy and safety outcomes, because of their short onset time, short half-life, low inter- and intra-individual variability and drug-drug interactions. Moreover, VKA interruption requires heparin bridging with increased risk of bleedings. The outcomes of DOACs in patients with active cancer have never been largely described so far, but as noted above, this is a high-risk population in which imbalance of thrombotic and bleeding risks may lead to serious outcomes. Hence, we performed a sensitivity analysis pooling data regarding efficacy and safety of DOACs vs VKAs only in patients with AF and active cancer and we showed for the first time in this peculiar population that DOACs use is related to stronger thromboembolic risk reduction and more favorable risk profile than VKAs, lowering the rates of the outcomes SSE, ischemic stroke, major bleeding and gastrointestinal or intracranial bleeding.

This meta-analysis is the largest comparing efficacy and safety of DOACs vs VKAs in patients with non-valvular AF and cancer so far, and clearly shows that DOACs may be considered a suitable anticoagulant agent in this challenging subgroup of patients. DOACs use was related to a more effective and safer profile as compared to VKAs and significantly lowered the rates of SSE, ischemic and hemorrhagic stroke, VTE, major bleeding and gastrointestinal and intracranial bleeding. In addition, DOACs represent a handy therapeutic strategy, not requiring frequent monitoring of INR and with less expected drug-drug interactions, providing a less burdensome alternative to a highly frail population. Moreover, our analysis confirms the favorable efficacy and safety profile of DOACs in cancer patients recently outlined by Giustozzi et al. in a meta-analysis of RCTs comparing safety and efficacy of DOACs vs LMWH in the treatment of cancer-related VTE [36, 37]. Although the choice of anticoagulant therapy should be tailored on patients’ preferences and bleeding risk profile, taken together these evidences are hypothesis-generating, suggesting that the use of DOACs may represent a reasonable choice in cancer patients with AF. Prospective randomized trials evaluating the efficacy and safety profiles of DOACs vs VKAs in cancer patients with AF are eagerly awaited, in order to give more confidence to physicians that are involved in clinical daily management of this troublesome population.

Our study presents some limitations. First, the meta-analysis only includes post-hoc analysis of RCT or retrospective population-based cohorts. The observational nature of reported data may affect the generalizability of our findings that should be considered as exploratory. However, prospective RCTs on this field are currently missing and we included in the analysis the best evidence produced so far. Second, all but three studies reported the adjusted RRs or propensity-score matched RRs. Including in the analysis unadjusted data for confounders may reduce the validity of the study. However, we performed a sensitivity analysis excluding the studies reporting only unadjusted RRs that confirmed the main findings of our study. Third, active cancer has been associated with worse outcomes [16, 27]. We performed a subgroup analysis on safety and efficacy of different anticoagulation strategies in patients with active cancer but the lack of information about type of cancer treatment and the heterogeneity among the studies regarding the definition of active cancer may have affected our results that need to be confirmed. Fourth, we did not perform an analysis on the basis of different cancer stages due to the lack of this information in the majority of the included studies. Fifth, TTR for VKA was reported only in few studies. The lack of TTR did not allow concluding that DOACs are superior to VKAs because a suboptimal TTR may have affected VKA safety and efficacy as shown by Kim et al. [19]. Finally, cancer population included in this meta-analysis is heterogeneous and some studies, as such the post-hoc analysis of ROCKET-AF trial [14] and the study by Sawant et al. [21], excluded patients with life-expectancy < 2 years and < 1 year, respectively, possibly with advanced cancer. However, the inclusion of large population-based, real-life studies in the analysis might have possibly overcome this pitfall.

In patients with cancer and non-valvular AF, the use of DOACs is associated with a significant reduction of thromboembolic and bleeding events and this result is consistent among patients with active cancer. Prospective randomized studies are needed to confirm our findings and address gaps in evidence.