External controls are not an experiment. They are a twenty-five-year regulatory record — in rare disease, oncology, and gene therapy — with a guidance, a program, and dozens of approvals behind them. This page is that record, what the agency accepted and probed in each case, and where our statisticians have stood inside it.
Selected precedents by therapeutic area. “What the agency probed” is the part that matters: it is the list of questions your comparator will face.
| Product | Indication | External control | What the agency accepted · probed |
|---|---|---|---|
| Zolgensma 2019 | Infantile-onset SMA | Retrospective natural-history cohort, 23 patients | Accepted where untreated natural history is predictable and fatal (death or permanent ventilation by 24 months); the landmark case for gene therapy. |
| Skysona 2022 | Cerebral adrenoleukodystrophy | Matched natural-history controls | 72% major-functional-disability-free survival vs 43% in matched natural-history controls; rapidly progressive fatal disease, placebo deemed unethical, well-characterized trajectory. |
| Lenmeldy 2024 | Metachromatic leukodystrophy | External-control natural-history study | FDA’s basis of approval: two single-arm trials, an EU expanded-access program, and an external-control natural-history study together formed the adequate and well-controlled investigation. |
| Kebilidi 2024 | AADC deficiency | Natural-history comparison | Motor milestones never achieved in untreated disease; comparator credibility rested on the disease’s known course. |
| Roctavian 2023 | Severe hemophilia A | Prospective non-interventional study of factor VIII prophylaxis | A prospectively collected external comparator for bleeding rates; index-date and era alignment were central. |
| Fayuvi 2026 | Sanfilippo syndrome type A | Natural-history external control | Natural-history evidence carried a registrational decision; reliability assessed on systematic data collection, meaningful endpoints, and patient matching. |
| Product | Indication | External control | What the agency accepted · probed |
|---|---|---|---|
| Brineura 2017 | CLN2 Batten disease | Historical natural-history cohort | Proportion without a 2-point decline on the CLN2 rating scale vs natural history, with sensitivity and supportive analyses; the agency probed comparability of the rating instrument across studies. |
| Strensiq 2015 | Perinatal/infantile hypophosphatasia | Historical control cohort | Survival vs historical controls in a lethal pediatric disease. |
| Kanuma 2015 | LAL deficiency (infants) | Historical cohort | Survival beyond 12 months vs an untreated historical infant cohort. |
| Zokinvy 2020 | Progeria | Natural-history cohort | Mortality vs matched untreated patients from natural-history data; first approval for the disease. |
| Rethymic 2021 | Congenital athymia | Natural-history survival | Survival compared to the expected course of an untreated, fatal condition. |
| Product | Indication | External control | What the agency accepted · probed |
|---|---|---|---|
| Blincyto 2018 expansion | Ph-positive R/R B-ALL | External comparator from medical-chart review | Weighted analysis of patient-level standard-of-care data supported the accelerated approval; the original 2014 approval had used historical response context. |
| Bavencio 2017 | Merkel cell carcinoma; platinum-refractory urothelial | Historical controls | Accelerated FDA and conditional EMA approval from single-arm data compared to historical controls. |
| Ibrance 2019 expansion | Male breast cancer | Real-world data (EHR/claims) | Label expansion supported by real-world evidence; a model for RWE-based comparators. |
| Rozlytrek 2019 | ROS1-positive NSCLC | Real-world external control arm | RWD external control submitted alongside single-arm data. |
| Xpovio 2019 | R/R multiple myeloma | Real-world evidence | RWE supported the NDA; the agency’s decisions remained situation-specific. |
| Kisqali 2023 expansion | Male breast cancer | Real-world evidence | RWE-supported label expansion following the Ibrance precedent. |
| Product | Indication | External control | What the agency accepted · probed |
|---|---|---|---|
| Omegaven 2018 | Parenteral-nutrition-associated cholestasis (pediatric) | Historical control from hospital records | Two single-arm trials matched to a historical control arm; a pediatric, non-rare-disease precedent. |
| CAR-T class 2017– | Hematologic malignancies | Single-arm trials; external comparators in supportive and HTA analyses | Response-rate-based approvals, with external comparator analyses increasingly used to contextualize single-arm results. |
Sources. Jahanshahi et al., Therapeutic Innovation & Regulatory Science (2021) — FDA decisions 2000–2019; Mishra-Kalyani et al. (FDA Oncology Center of Excellence), Annals of Oncology (2022); FDA Summary Basis for Regulatory Action, Lenmeldy (2024); Molecular Therapy Advances review of gene and cell therapy approvals 2019–2025 (2026); FDA guidance, Considerations for the Design and Conduct of Externally Controlled Trials (2023). Product details are summarized from public regulatory documents and reviews; consult the primary documents for any regulatory use.
Brineura’s review probed whether the rating scale used in the natural-history cohort matched the trial’s. Endpoint definitions, scoring, and assessor training must be aligned before matching begins.
Zolgensma and Skysona were accepted because the natural history was rapid, well-characterized, and fatal. A comparator is only as credible as the disease trajectory it represents.
Eligibility alignment, era, prior therapy, baseline severity, and index-date rules. Blincyto’s comparator used weighted patient-level analysis, not aggregate historical rates.
Tipping-point analyses, E-values, and alternative comparators are what turn a single estimate into a defensible one. Pre-specified, not post hoc.
Every accepted case answers this first: too few patients, unethical placebo, a fatal predictable course, an accelerated pathway with a confirmatory plan. The argument for the design precedes the design.
Every statistician on our engagements brings a minimum of 15 years in regulated industries. Between them, they have built and defended the comparators that carried programs through review.
Send the protocol synopsis. Within 48 hours you get a written read on comparator sources, matching strategy, the precedent that applies, and a fixed-fee scope.