On 25 March 2021, at a Type B meeting, FDA told Replimune that it considered "the interpretation of responses in the setting of the intra-tumoral route of administration to be problematic in the lesions which have been injected." On 6 August 2026 the agency approved the product, and the approved response rate was calculated on 91 patients, being those who had at least one lesion that was never injected. Between the meeting and the approval sit a biologics license application, two complete response letters, an advisory committee meeting and five years, across which the objection stayed the same.
The advice given in 2021 and 2024
The April 2026 complete response letter, published in full under FDA's rejection-letter transparency programme, quotes the agency's earlier advice back to the company. At that March 2021 meeting FDA said the proposed single-arm study "will not enable identification of the contribution of each component of the combination to the overall response rate." It recommended the company "conduct a randomized controlled trial (RCT) to demonstrate efficacy and safety of the combination and isolate the contribution from each component." It added that a 30% to 40% target response rate "may not translate into unequivocally improved clinical benefit in this indication," and raised "potential issues of interpreting efficacy result due to baseline heterogeneities among study subjects." It also said it "would not recommend that the Sponsor submit a BLA based on the results of a single-arm study."
The same position appears again at the pre-BLA meeting on 3 September 2024, where FDA said the company must "provide evidence to demonstrate that there is a compelling reason that both RP1 and nivolumab are necessary to achieve the observed treatment effect." Summarising all of this in 2026, the letter says that "FDA advice has remained consistent as evidenced by our communications dating back to March 2021."
The application and the first rejection
The letter records that "FDA ultimately did not object to the submission of the BLA based on data from RPL-001-16, reflecting FDA's flexibility in disease settings with high unmet need." On review, though, "the study design concerns previously communicated were not addressed, and the contribution of vusolimogene oderparepvec to the observed response rate in RPL-001-16 could not be determined." The first complete response letter issued on 21 July 2025, and the company resubmitted that October. Multi-cycle programmes of this kind have come up here before, in Outlook's fourth attempt after three rejections, in Atara and Pierre Fabre's third submission, and in Regenxbio's decision to resubmit without running a new study.
Why the cohort was exposed
The Summary Basis for Regulatory Action, published on 13 August, describes how the pivotal dataset came to exist. IGNYTE "was initiated as an exploratory study" of the product alone or with anti-PD-1 therapy across a range of solid tumours, and the melanoma cohort was added later by amendment, "rather than being prospectively designed as a pivotal cohort from the study's outset." The memo adds that the protocol "underwent several subsequent amendments that materially changed the cohort's eligibility criteria and conduct." That is where the documentation burden shifts. A dossier built on a cohort that became pivotal by amendment has to carry that history, rather than present the result as though it had always been the plan. The reconciliation problem is the same one that appears whenever an analysis population changes late and every document downstream of it has to be brought back into line, and it is the reason what a study is expected to carry at each phase is worth settling before the cohort is the one being reviewed.
The second review cycle
For the resubmission FDA changed the reviewers, recording that "to maintain objectivity and account for potential bias, the review team members for this BLA resubmission were different than those who reviewed the initial BLA." That team, its supervisory leadership in the Office of Therapeutic Products and subject matter experts from the Oncology Center of Excellence "unanimously determined data presented are insufficient to conclude substantial evidence of effectiveness," and the second complete response letter issued on 10 April 2026. Who sits on a review team is not a fixed feature of a programme, whether the change is deliberate as it was here, or the result of a centre director's post falling vacant or of a chief reviewer leaving mid-cycle.
What the company had filed in October 2025 was an early unplanned analysis from its ongoing Phase 3 trial, covering 22 patients in the investigational arm and 18 in the control arm against a planned enrolment of 400, together with exploratory analyses of the original study. At a Type A meeting on 16 September 2025, FDA had told the company that the response criteria used "were not consistent with RECIST v1.1." It had also "provided recommendations regarding use of data from the ongoing Phase 3 trial to potentially support Accelerated Approval." The letter records the resubmission as having gone a different way: "Instead, you submitted data from an early unplanned analysis from RP1-104, representing 10% of the planned enrollment."
Why the response rate could not be relied on
The letter sets out the problems with the reported response rate, and each of them concerns how the evidence was captured rather than whether the product has activity. Almost half of responders, 49%, had all their target lesions injected, and two further responders had no target lesions at baseline on independent review, so "over half (53%) of patients with objective response did not have noninjected target lesions to assess systemic anti-tumor activity." Patients with new or enlarging lesions were selected for re-injection before the independent review committee could determine progression, which the letter says "obscures the determination of response." Surgical procedures including excisional biopsies were performed on target lesions and "in some cases, immediately preceded the next response assessment that determined a partial or complete response." Histopathology was used to change radiologic response calls on specimens that "were not centrally reviewed." Taken together, FDA concluded that the reported results "were not reliably assessed, could be artifactually increased, and may not be reproducible." None of that concerns the compound, only whether the underlying record supports the number, which is the same question that surfaced when pivotal data was retracted after publication.
The approval and the analysis behind it
The Summary Basis for Regulatory Action sets out two analyses of the same cohort. Replimune reported a 33.6% response rate and a 24.8-month median duration of response in 140 evaluable patients. FDA's analysis, restricted to the 91 patients with at least one non-injected lesion, gave 24.2% and 14.1 months. The agency's figures are the ones in the public record of the approval. That restriction follows directly from the 2021 objection, since a lesion that was never injected is the only place a systemic effect can be observed in an intratumoral therapy, and the clinical benefit question moves into the confirmatory trial. It is the agency's numbers that the prescribing information carries, which is the label-writing consequence of a review that lands on a different population from the sponsor's.
The advisory committee vote on 30 July has the same shape. The committee voted 10 to 3 that the efficacy results were evaluable and clinically meaningful, but it was not asked to endorse the sponsor's analysis and did not vote on approval.
Manufacturing and site inspections
The CMC review team concluded at the original application that the process and its controls "can yield a product with consistent quality characteristics." It recorded that "there was no additional information submitted in the two resubmissions," and recommended approval throughout. Site inspections "did not reveal substantive issues that impact the data submitted." Across five years and two rejection cycles nothing turned on manufacturing readiness or on data integrity at the sites, and the whole argument sat in which patients could support a claim of systemic activity.
What this means for the submission record
The objection was written down in the March 2021 meeting minutes in FDA's own words, repeated at the pre-BLA meeting, restated in the first complete response letter and set out again in the second. The exploratory analyses filed in October 2025 included one on response in patients who had both injected and non-injected tumours, so a related cut of the data already existed in the dossier, filed as exploratory rather than as the primary case.
For anyone drafting a single-arm submission the reading is narrow and fairly unglamorous. The written record of agency interactions is not background material sitting behind a dossier. It is the list of questions the dossier will be marked against, and a submission that presents around a stated objection rather than answering it stays legible as such to a review team years and cycles later. Holding that record so it can be reached five years on is a matter of where the source of truth actually sits and of who defined the provenance of each claim, rather than of how carefully the final document was written.
What is required next
Continued approval depends on IGNYTE-3, a randomised open-label trial of roughly 400 patients against physician's choice with overall survival as the primary endpoint, with the final protocol due 31 August 2026, study completion set for September 2030 and the final report for March 2031. A required pediatric assessment will be met through the same trial, and the memo records no risk evaluation and mitigation strategy and no agreed postmarketing commitment safety study. That obligation is a writing surface in its own right rather than a milestone sitting outside the dossier, and Sarepta's route to traditional approval after a confirmatory trial missed shows how much rests on how that trial is documented.
None of this reconstruction would have been possible two years ago. The April 2026 letter is readable because FDA now publishes its rejection letters, a practice that paused and restarted in July and is now the subject of rulemaking. Earlier coverage here set out what the agency's briefing documents said before the advisory committee, the position after the second rejection, and how response-rate approvals have changed since Project Optimus.