𝗙𝗮𝗰𝘁𝗼𝗿𝘀 𝗙𝗮𝘃𝗼𝗿𝗶𝗻𝗴 𝗥𝗲𝗴𝘂𝗹𝗮𝘁𝗼𝗿𝘆 𝗔𝗰𝗰𝗲𝗽𝘁𝗮𝗻𝗰𝗲…

• Large, unambiguous treatment effects: In nearly all approved cases (Zolgensma, Kebilidi, Casgevy), the treatment effect was so dramatic that bias alone could not plausibly explain the results. FDA has explicitly stated that large effect sizes overcome external control limitations.
• Well-characterized natural history: Approvals were more likely when the untreated disease course was predictable and well-documented through established registries (DEM-CHILD for CLN2, PNCR for SMA). Prospective natural history data is preferred but retrospective data has been accepted.
• Ultra-rare populations with high unmet need: FDA showed greatest flexibility for diseases affecting very small populations (AADC deficiency: ~150 patients worldwide) where randomized trials are genuinely infeasible. Orphan Drug, Breakthrough Therapy, and Fast Track designations correlated with acceptance.
• Pre-specified analysis plans: Successful applications had pre-specified statistical analysis plans, including control arm selection, matching criteria, and sensitivity analyses. Post-hoc analyses were viewed skeptically.
• Objective, clinically meaningful endpoints: Motor milestone achievement (Zolgensma, Kebilidi), survival (Skysona), and clearly measurable biomarkers (Hemgenix Factor IX levels) were more readily accepted than subjective endpoints.
• Innovative within-patient designs: The intra-patient randomized design used by Vyjuvek and within-patient baseline comparisons (Hemgenix lead-in period) were viewed favorably as they control for patient-level confounding.

OR Regulatory Pushback

• Inadequate comparability between treatment and control populations: The RGX-121 rejection explicitly cited age and cognitive impairment differences between trial patients and natural history controls. Baseline covariate imbalance remains a primary FDA concern.
• Subjective endpoints combined with external controls: The eteplirsen controversy centered on whether endpoints were sufficiently objective for valid external comparison. FDA advisory committees consistently flag this combination.
• Insufficient pre-specification and unmeasured confounding: Troriluzole’s CRL highlighted design flaws and lack of pre-specification. FDA expects externally controlled trial protocols to be finalized before initiation, not selected post-hoc.
• Small effect sizes relative to potential bias: When treatment effects are modest, FDA cannot distinguish real effects from systematic differences between groups. The troriluzole review explicitly required a “large treatment effect” to overcome external control biases.
• Reliance on surrogate biomarkers without clinical validation: While biomarker endpoints have been accepted (Elevidys micro-dystrophin for accelerated approval), using unvalidated surrogates alongside external controls compounds uncertainty.