The Consequence Library · How records are made and graded
On 27 March 2024, two days after FDA pre-announced a site inspection, a third-party vendor deleted electronic clinical outcome data in the Q-global system, including the associated audit trails, for all 47 subjects in a pivotal galactosemia trial. FDA's warning letter states it was unable to access, copy and verify records relating to the study, and that without the audit trails it could not verify the accuracy, consistency and completeness of data collected for the critical assessments used to measure primary and secondary efficacy endpoints.
Evidence caveat. No AI system was involved. The record is included because it is the clearest documented case of regulated evidence being destroyed by the party holding it, at the moment the evidence was required, which is the failure independent verification exists to address.
What happened
Applied Therapeutics sponsored a pivotal trial of govorestat in classic galactosemia, using electronic clinical outcome assessments operated by a third-party vendor in the Q-global system. FDA pre-announced a site inspection. Two days later, on 27 March 2024, the vendor deleted the electronic data for all 47 subjects, together with the associated audit trails. FDA inspected from 29 April to 3 May 2024 and issued a Form 483, followed by a warning letter dated 27 November 2024 and posted 3 December 2024. The letter records that the agency could not verify the study data, and separately documents an undisclosed dosing error in which at least 19 subjects received 80 per cent of the protocol dose because investigational product was mislabelled, while the sponsor reported protocol doses rather than the doses actually administered. A Complete Response Letter followed in the same period, along with shareholder litigation.
Where control failed
The audit trail was held inside the system it was recording, and was deletable by the party operating that system. The record does not show a control that prevented bulk deletion of regulated records, nor an independent copy of the audit trail outside the vendor's control.
The authority question
FDA's letter does not establish who authorised or performed the deletion. The record that would ordinarily answer that question was the record deleted. Sponsor oversight of the tasks delegated to the vendor is a separate finding in the letter.
What could be proven afterward
This is the clearest documented case in recent FDA enforcement of evidence failing at the moment it is required. The consequence was not that the record was disputed but that verification became impossible: the agency states plainly that it could not verify the data underlying the trial's efficacy endpoints. Evidence held only by the party it could incriminate is evidence conditional on that party's continued cooperation.
Control state, before and after
Before the consequence
Electronic clinical outcome data and its audit trails held in a vendor-operated system, deletable by the vendor. No independent record of the study data outside that system.
After the consequence
FDA warning letter dated 27 November 2024, posted 3 December 2024. Complete Response Letter in the same period. Shareholder litigation followed, and the approval prospects for the product were materially affected.
Sources
- Primary: FDA, Warning Letter to Applied Therapeutics, Inc., CMS #696833, dated 27 Nov 2024, posted 3 Dec 2024 · https://www.fda.gov/inspections-compliance-enforcement-and-criminal-investigations/warning-letters/applied-therapeutics-inc-696833-12032024
Record history
Published 17 August 2026. Load-bearing facts re-verified against the cited sources on 17 August 2026. Corrections and material changes are appended here with their dates. To report an error in this record, write to hello@regulayer.com with the record slug and the source you believe is authoritative.
This record describes what sources establish about a consequence and the control state around it. It separates confirmed fact from source-stated cause and from architectural analysis, and it makes no claim that any control or product would have prevented the outcome. Gap codes identify a failure class, not a remedy.