Anatomy of a Failure: Lessons from 89 FDA CRLs that Resulted in Failure
This briefing analyses 89 FDA Complete Response Letters issued to products that were never approved. It counts how often each category of deficiency appears, separates the failures a sponsor can remediate from the ones that end a programme, and sets out where in development each fatal flaw was already visible.
What this report answers
- How often does each category of deficiency actually appear in a terminal CRL?
- Which deficiencies can a sponsor fix and resubmit, and which end the programme?
- Why do some sponsors receive a second and a third CRL for the same facility problem?
- How does the failure pattern differ between biologics, small molecules and combination products?
- How much of the risk now sits in clinical data reliability rather than in the science?
- What should regulatory, clinical, CMC and quality leaders be testing before they file?
85%
the share of the 89 applications the analysis records as carrying at least one major manufacturing or product quality deficiency - the base rate against which every other failure mode is read.
54%
the share the analysis attributes to unresolved facility inspection findings, the most frequent single deficiency it codes within manufacturing and quality.
What the analysis establishes
How the deficiencies are counted
Every letter is coded into four domains - manufacturing and product quality, clinical and efficacy, non-clinical and bridging, and device and human factors - with an incidence rate for each named sub-deficiency inside them and a short line from a real CRL against each. Letters routinely carry deficiencies in more than one domain, so the analysis records how the categories compound rather than treating them as alternatives.
Fixable against fatal
The organising question of the report is not which deficiencies are most common but which ones a sponsor can still recover from. It establishes a division between execution failures that survive a resubmission cycle and asset-level or data-level failures that do not, and shows how a deficiency that is individually survivable behaves differently when it sits on top of one that is not.
Why sponsors receive the same CRL twice
Several sponsors in the dataset were cited for unresolved facility findings, resubmitted, and were cited again on the same grounds. The report treats that repeat pattern as evidence about quality oversight of external manufacturing rather than as bad luck, and works through what a second inspection-driven letter signals to the agency about a sponsor's control of its own supply chain.
Where the agency stops trusting the data
A distinct group of letters turns on the reliability of clinical data rather than on what the data showed - inspection findings at investigator sites, adverse event capture, and analysis conduct. The report sets out how often that appears across the cohort, which named applications it affected, and why this class of finding behaves differently from every other deficiency in the dataset.
Failure patterns by modality
The cohort is split into biologics and biosimilars, small molecules, and combination products and devices, each given an approximate share of the 89 letters. Each modality is assigned its dominant failure reason and a secondary challenge, so a reader can see whether the risk concentrates in manufacturing systems, in clinical proof, or in the regulatory pathway itself.
Role-level checklists
The final chapter converts the dataset into checklists for four roles: chief medical officer, VP regulatory affairs, head of CMC and technical operations, and head of quality. Each names the specific pre-filing evidence that role should be demanding, anchored to the applications in the dataset where that evidence was missing, and states the failure mode now attached to the role.
How it was built
A systematic review of 89 Complete Response Letters released by the FDA, all for products that were not ultimately approved. Each letter was coded across four deficiency domains, with incidence expressed as the percentage of applications in which a deficiency appears; because one letter can carry several, the categories overlap and are reported that way. Named applications are cited with their NDA or BLA numbers and quoted briefly. The modality split is given as ProGen Search's own approximation. The cohort is deliberately positioned against an earlier ProGen analysis of a different set of letters, so the reader can see what distinguishes a terminal CRL from a recoverable one.
Written for regulatory affairs, clinical development, CMC and quality leaders preparing a filing, and for investors and boards running diligence on an asset that has already received a Complete Response Letter.
Contents
- Executive Summary: The Insurmountable Hurdles
- The Quantitative Landscape of Submission Deficiencies
- Deep Dive Analysis: Primary Failure Modes
- Modality-Specific Failure Patterns
- ProGen Search Recommendations: A Framework for Identifying Fatal Flaws
Sponsors and agencies named in the analysis
FDA, Accord BioPharma, Aldeyra, Applied Therapeutics, Celltrion, Defender, Fresenius Kabi, Gan & Lee, Lexicon, Lykos, Mapi, MedRz, Minerva, Novo Nordisk, Olympus, Orexo, PTC Therapeutics, Regeneron, Stealth
The full 16-page briefing is free. Tell us where to send it and it downloads immediately.