Aims: In a previous study, Singaporean Asians were found to have lower rivaroxaban plasma concentrations than Caucasians. This study attempts to identify predictors that may be associated with bleeding and stroke and systemic embolism (SSE) in Singaporean Asians taking rivaroxaban and apixaban. Methods: A total of 134 Singaporean patients on either rivaroxaban or apixaban for non-valvular atrial fibrillation were included for this study. Baseline characteristics were recorded at recruitment while bleeding and SSE events were recorded during a 1-year follow-up. Characteristics of patients with or without bleeds were compared using relevant statistical tests. Multivariable regression that included covariates with p < 0.1 from an initial univariable regression was performed to analyze predictors that resulted in higher risk of bleeding in patients. Results: Median creatinine clearance (CrCl) was significantly lower in patients on rivaroxaban who experienced bleeds as compared to patients who did not experience bleeds (61.5 vs 70.8 mL/min, p = 0.047), while concomitant simvastatin use was found to be independently associated with a six-fold increased risk of bleeding [Adjusted OR = 6.14 (95% CI: 1.18 – 31.97), p = 0.031] for rivaroxaban after controlling for body mass index, CrCl and having experienced a previous SSE. Conclusion: Our findings suggest that concomitant use of simvastatin with rivaroxaban may be associated with bleeding events in an Asian cohort. Further studies using physiologically-based pharmacokinetic modeling are required to investigate the drug-drug interactions between these drugs. Keywords: Atrial Fibrillation, Bleeding, Rivaroxaban, Simvastatin

Background and Purpose Rivaroxaban is emerging as a viable anticoagulant for the pharmacological management of cancer associated venous thromboembolism (CA-VTE). Being eliminated via CYP3A4/2J2-mediated metabolism and organic anion transporter 3 (OAT3)/P-glycoprotein-mediated renal secretion, rivaroxaban is susceptible to drug-drug interactions (DDIs) with protein kinase inhibitors (PKIs), erlotinib and nilotinib. Physiologically based pharmacokinetic (PBPK) modelling was applied to interrogate the DDIs for dose adjustment of rivaroxaban in CA-VTE. Experimental Approach The inhibitory potencies of erlotinib and nilotinib on CYP3A4/2J2-mediated metabolism of rivaroxaban were characterized. Using prototypical OAT3 inhibitor ketoconazole, in vitro OAT3 inhibition assays were optimized to ascertain the in vivo relevance of derived inhibitory constants (Ki). DDIs between rivaroxaban and erlotinib or nilotinib were investigated using iteratively verified PBPK model. Key Results Mechanism-based inactivation (MBI) of CYP3A4-mediated rivaroxaban metabolism by both PKIs and MBI of CYP2J2 by erlotinib were established. The importance of substrate specificity and nonspecific binding to derive OAT3-inhibitory Ki values of ketoconazole and nilotinib for the accurate prediction of DDIs was illustrated. When simulated rivaroxaban exposure variations with concomitant erlotinib and nilotinib therapy were evaluated using published dose-exposure equivalence metrics and bleeding risk analyses, dose reductions from 20 mg to 15 mg and 10 mg in normal and mild renal dysfunction, respectively, were warranted. Conclusion and Implications We established the PBPK-DDI platform to prospectively interrogate and manage clinically relevant interactions between rivaroxaban and PKIs in patients with underlying renal impairment. Rational dose adjustments were proposed, attesting to the capacity of PBPK modelling in facilitating precision medicine.