The Effect of Rivaroxaban Exposure and Clinical Risk Factors on Efficacy and Safety Outcomes in Patients with Nonvalvular Atrial Fibrillation

Introduction: Rivaroxaban, an oral factor Xa inhibitor, reduces the risk of stroke and systemic embolism in patients with nonvalvular atrial fibrillation (NVAF) when given in fixed doses without routine coagulation monitoring. An exposure-response analysis was conducted to evaluate the impact of rivaroxaban exposure and clinical risk factors on efficacy and safety outcomes in patients with NVAF to assess whether concentration-based dose adjustment would improve the benefit-risk profile of rivaroxaban in this patient population.

Methods: Efficacy/safety outcome data were taken from the rivaroxaban-treated group enrolled in the ROCKET AF trial (NCT00403767) in which 14 264 patients with NVAF were randomized to receive either rivaroxaban 20 mg once daily (15 mg in patients with moderate renal impairment) or dose-adjusted warfarin. Individual rivaroxaban exposure metrics (peak concentration, trough concentration [C_trough], and daily area under the curve at steady-state) were derived based on an integrated population pharmacokinetic (PK) model (which was previously developed using pooled data from 4918 patients in seven clinical trials across all approved rivaroxaban indications) that incorporated PK-related individual covariates and prothrombin time values measured in a central laboratory. Composite efficacy outcomes were 1) ischemic stroke and non-CNS systemic embolism (ISNSM), and 2) ischemic stroke and non-CNS systemic embolism and death from all causes (ISNSMD). Safety outcomes were 1) major plus non-major clinically relevant (NMCR) bleeding, and 2) major bleeding (MB). Kaplan-Meier analyses stratified by quartiles of exposures were performed with efficacy and safety outcomes. A multivariable Cox model was applied to link individual exposures and clinical risk factors related to efficacy/safety with the time-to-event outcomes.

Results: For efficacy outcomes, the event rates were 2.2% and 5.1% (154 and 357 out of 7061 patients) for ISNSM and ISNSMD. Kaplan-Meier analyses showed no apparent trend among the quartiles of exposure metrics and the efficacy event rates. Cox regression analyses revealed no association between exposures and the two efficacy outcomes (e.g., p=0.733 and p=0.82 for the association of C_trough with ISNSM and ISNSMD, respectively). For safety outcomes, the event rates were 20.7% and 5.6% (1475 and 375 out of 7111 patients) for major plus NMCR bleeding and MB, respectively. Kaplan-Meier analyses showed that cumulative major plus NMCR bleeding and MB event rates increased when rivaroxaban exposure levels increased. Cox regression revealed that the hazard ratio (HR) for these two safety outcomes increased with rivaroxaban exposure. The impact of increasing from the median C_trough to the 95^th percentile was a HR of 1.26 (p=0.00002) for major plus NMCR bleeding (Figure 1) and a HR of 1.25 (p=0.021) for MB (Figure 2). However, as shown in Figures 1 and 2, these HRs appeared to be of lower magnitude when compared with several clinical risk factors (e.g., HR 1.47, p=0.00089 and HR 2.55, p<0.0001 for history of gastrointestinal bleeding vs no history; HR 1.35, p<0.0001 and HR 1.41, p=0.0045 for age > 75 years vs age 65-75 years; HR 1.27, p=0.006 and HR 1.93 p<0.0001 for low baseline hemoglobin values [hemoglobin <13 g/dL for males and <12 g/dL for females] vs normal).

Conclusions: The relatively small impact of exposure on the HR for safety outcomes and lack of relationship for efficacy outcomes suggest that for the approved dosing regimen, rivaroxaban exposure is not an important risk factor for the occurrence of either bleeding or ischemic events. Several clinical risk factors are more impactful than rivaroxaban exposure on these efficacy and safety outcomes. Therefore, it is unlikely that concentration-based dose adjustment would further enhance the benefit-risk profile of rivaroxaban in patients with NVAF.

View largeDownload slide

View largeDownload slide

Close modal