Model-based meta-analysis to evaluate optimal doses of direct oral factor Xa inhibitors in atrial fibrillation patients

Abstract

Key Points• Simulations suggested that the dose reduction of rivaroxaban would decrease fatal events associated with major bleeding.• Dose optimization of FXa inhibitors might further enhance their therapeutic benefit.The noninferiority of direct oral factor Xa (FXa) inhibitors (rivaroxaban, apixaban, and edoxaban) in treatment of atrial fibrillation were demonstrated compared with warfarin by several large clinical trials; however, subsequent meta-analyses reported a higher risk of major bleeding with rivaroxaban than with the other FXa inhibitors. In the present study, we first estimated the changes of prothrombin time (PT) in 5 randomized trials based on reported population pharmacokinetic and pharmacodynamic models and then carried out a model-based meta-analysis to obtain models describing the relationship between PT changes and the event rates of ischemic stroke/systemic embolism (SE) and of major bleeding. By using the models, we simulated the optimal therapeutic doses for each FXa For apixaban and edoxaban, no distinct change in the overall mortality was simulated by dose modification. This study suggested that the current dose of rivaroxaban might be excessive and would need to be reduced to decrease the excess risk of major bleeding.

Model‐Based Meta‐Analysis: Optimizing Research, Development, and Utilization of Therapeutics Using the Totality of Evidence

“…These dose-response simulations generated a hypothesis for further optimization of rivaroxaban therapy via selection of a lower dose than that currently approved, while concluding that the approved doses of apixaban and edoxaban are likely close to optimal. 57 This example illustrates the power of mechanism-informed MBMA leveraging relationships across the PK-PD-outcome continuum. The outputs of this type of analysis on approved agents also provide valuable insights to inform intelligent development of the next generation of drugs with comparable mechanisms of action, highlighting the power of MBMA as a quantitative reverse translation tool.…”

“…The drug‐specific PK/PD models for increase in PT coupled with the MBMA relating PT ratio to these event risks were subsequently used to compare simulated benefit–risk profiles of a range of doses of each drug flanking the current approved doses. These dose‐response simulations generated a hypothesis for further optimization of rivaroxaban therapy via selection of a lower dose than that currently approved, while concluding that the approved doses of apixaban and edoxaban are likely close to optimal . This example illustrates the power of mechanism‐informed MBMA leveraging relationships across the PK–PD–outcome continuum.…”

“…A recent example of application of MBMA to hypothesis generation for postapproval therapeutic optimization was for factor Xa inhibitors in atrial fibrillation. In this case, dose‐response relationships were developed along both the dimensions of benefit (reduction in incidence of ischemic stroke/systemic embolism) and risk (increase in major bleeding) to characterize the optimal dose . The analysis leveraged published population PK/PD models of the effects of rivaroxaban, apixaban, and edoxaban on prothrombin time (PT) as a drug effect biomarker to simulate the relative increase in PT at doses used in randomized trials of these drugs.…”

“…Exposure/dose-response MBMA models of efficacy and toxicity outcomes, leveraging mechanism-informed model structures where feasible, to simulate the benefit/risk profile of alternate doses and regimens and generate hypotheses for postapproval dose optimization. Illustrative Applications: References 56,57 MBMA, model-based meta-analysis; PK, pharmacokinetic(s); PK/PD, pharmacokinetic/pharmacodynamic. gov and their subsequent publication may contribute to reducing publication bias for clinical studies.…”

“…In this case, dose-response relationships were developed along both the dimensions of benefit (reduction in incidence of ischemic stroke/systemic embolism) and risk (increase in major bleeding) to characterize the optimal dose. 57 The analysis leveraged published population PK/PD models of the effects of rivaroxaban, apixaban, and edoxaban on prothrombin time (PT) as a drug effect biomarker to simulate the relative increase in PT at doses used in randomized trials of these drugs. Subsequently, the relationships between the modelestimated time-averaged PT ratio and risk reduction for ischemic stroke/systemic embolism or risk increase for major bleeding for the meta-dataset of trials across the three factor Xa inhibitors were characterized in the MBMA.…”