Meta-Analysis of Antiplatelets in Patients with Atrial Fibrillation: A survival Model
Farkad Ezzet, Itzela Correa
Pharsight a Certara Company, St Louis, MO, USA
Objectives: Warfarin reduces the risk of stroke in patients with atrial fibrillation; however, due to multiple interactions, coagulation monitoring and dose adjustment are necessary. We report analysis and comparisons between treatments from 3 trials of new thrombin inhibitors using meta-analysis and exploiting properties of survival models.
Methods: Efficacy was measured by events of stroke or systemic embolism for apixaban vs. aspirin (N=5600) [1], dabigatran vs. warfarin (N=18000) [2] and ximelagatran vs. warfarin (N=3400) [3]. We digitized figures of cumulative hazard Rates (CHR) and obtained rate and time (month) variables. Combined data were modeled to determine linearity or nonlinearity of the hazard rate (HR). The same approach was used for events of major bleeding using data from [1]. The estimated rates were used to 1) compare between treatments, 2) simulate events and compare with observed, 3) design non-inferiority studies and 4) optimize clinical trial design. Under linear hazard, time to event was simulated using an exponential distribution. Virtual clinical trials were obtained and comparisons were made by fitting Cox regression models. The es timated HR was used to explore design of non-inferiority studies
Results: CHR for stroke and system embolism was estimated to be linear. Because of similar inc/exc criteria and baseline characteristics, study effect (fixed or random) was small, although improved the fit. Estimated rates were .30, .14, .11, .13, .09 and .13 (%event/month) for aspirin, warfarin, ximelagatran 36 mg bid, dabigatran 110 mg bid, dabigatran 150 mg bid and apixaban 5 mg bid, respectively. SE was < .003. Warfarin event rate was 1.68% or 151 events/year, in agreement with observed [2]. Using data up to 12 months [1], major bleeding was also linear. Estimated rates were .1 and .125 (SE = .002), for aspirin and apixaban. Using warfarin (rate=.14) as a reference and assuming similar efficacy of a comparator, simulations suggested 3000 patients/arm (no dropouts) would be sufficient to establish non-inferiority using a hazard ratio margin of 1.46. Using dabigatran 150 mg bid as a ref erence (rate=.09), sample size needs to increase to > 4000/arm.
Conclusions: 1) Cumulative hazard rates extracted from public literature provide adequate information to characterize drug attributes. 2) Survival analysis allows comparisons between treatments and provides a practical tool for designing and optimizing clinical trials
References:
[1] Connolly SJ, Eikelboom J, Joyner C, et al (2011) Apixaban in Patients with Atrial Fibrillation, N Engl J Med; 364:806-817
[2] Connolly SJ, Ezekowitz MD, Yusuf S, et al (2009) Dabigatran versus Warfarin in Patients with Atrial Fibrillation, N Engl J Med; 361:1140-51.
[3] Executive Steering Committee, (2003) Stroke Prevention with the oral direct thrombin inhibitor ximelagatran compared with warfarin in patients with non-valvular atrial fibrillation (SPORTIF III): randomized controlled trial, The Lancet; 362, 1691-98.