Design and feasibility considerations for a phase 3 efficacy trial of the M72/AS01

Abstract

BACKGROUND: M72/AS01

METHODS: We conducted event-driven simulations using lower bound (LB) of the two-sided 95% confidence interval (CI) for VE(D). For IGRA-positive participants, assumptions included 1:1 randomization, 9000 participants/arm, 0.4% TB incidence/year, 55% true VE(D), 5% dropout/year, and two-year enrollment. Enrollment irrespective of baseline IGRA status (mixed IGRA-status population) and IGRA-negative-only scenarios were explored to estimate sample sizes and trial duration.

RESULTS: Simulations demonstrated that 110 events rule out a VE(D) 95% CI LB ≤10%, and 185 events rule out ≤25%, assuming ≥90% power and a true VE(D) of 55%. With 18,000 IGRA-positive participants, simulations projected a 90% probability of accruing 110 events within 3.5 to 4 years and 185 within 5.5 to 6 years. In the mixed IGRA-status population, few endpoints occurred among IGRA-negative participants, yielding insufficient power. Standalone VE(D) evaluation in IGRA-negative participants required large sample sizes (approximately 134,800) and prolonged timelines, indicating infeasibility. Accordingly, the selected primary objective of the phase 3 trial was to confirm VE(D) in IGRA-positive HIV-negative participants using LB of 95% CI for VE(D) > 10% after 110 events; secondary objectives include safety and immunogenicity in HIV-negative IGRA-positive; HIV-negative IGRA-negative; and HIV-positive individuals irrespective of IGRA status.

CONCLUSIONS: An IGRA-positive-enriched, event-driven phase 3 trial is feasible to confirm VE(D) of M72/AS01