Replimune's own filing shows the RP1 melanoma pivotal was a single-arm cohort — the randomized answer is dated 2030

Replimune's melanoma registrational path was a cohort, and the company put that on the record itself. An 8-K exhibit dated 3 June 2021 states that "Based on the initial data with RP1 in melanoma, the Company initiated a registration-directed 125-patient cohort of anti-PD1 failed melanoma which is expected to read out in 2022" ^[5]. The randomized study that would settle the efficacy question exists — IGNYTE-3 — and its primary completion date is 30 September 2030 ^[2]. Those two facts stand on primary records.

The third does not. Endpoints News reported on 28 July 2026 that FDA staff, in documents released ahead of an advisory committee meeting, said Replimune's pivotal study of RP1 in advanced melanoma failed to prove the drug works — and that the objection was aimed at how the study was designed rather than at a shortfall in the reported efficacy ^[1]. Prognyx could not obtain the agency's briefing document, and no regulator record, SEC filing or company release in the sources reviewed carries that account; it is carried here as one outlet's report and nothing more. The specific grounds — absence of a concurrent control, confounding by prior or subsequent therapy, response assessment, injected versus non-injected lesions — are not established in the sources reviewed ^[1].

That design-versus-data distinction, if it holds, is the whole story. A data problem can be re-analysed. A design problem cannot.

The design is on the record — in Replimune's own filings

The same 125-patient anti-PD-1-refractory cohort had already been listed as enrolling in the investor deck filed as an 8-K exhibit a year earlier, on 3 June 2020 ^[6]. Prognyx's reading is that this cohort is the study now at issue; the reporting reviewed does not name it.

The signal that set that strategy was thin at the time it was set. In the June 2021 filing, sixteen anti-PD-1-failed cutaneous melanoma patients had been enrolled into a 30-patient melanoma cohort; five met formal response criteria — four of them having failed both anti-PD-1 and anti-CTLA-4 — for a stated response rate of 31% ^[5]. That is a five-year-old interim number and it is emphatically not the dataset in the current BLA.

The more damaging comparison is internal. Replimune's first registration-directed programme was not melanoma and it was randomized: an 8-K of 15 October 2019 announced a registration-directed, randomized, controlled Phase 2 of RP1 plus cemiplimab versus cemiplimab alone in cutaneous squamous cell carcinoma, roughly 240 patients, with response rate as the primary comparative objective ^[7]. The company knew how to build a controlled registrational trial. In melanoma it chose not to.

The randomized answer exists. It is dated 2030.

IGNYTE-3 is the trial that would answer the question. It is a randomized, controlled, multicentre, open-label Phase 3 of vusolimogene oderparepvec plus nivolumab versus Physician's Choice in unresectable Stage IIIb–IV cutaneous melanoma that has progressed on an anti-PD-1 and an anti-CTLA-4 containing regimen, or in patients who are not candidates for anti-CTLA-4 therapy ^[2]. Primary endpoint: overall survival. Planned enrolment: 400 patients ^[2]. It started on 11 July 2024, it was recruiting as of its registry record of 15 July 2026, no results are posted, and its primary completion date is 30 September 2030 ^[2].

So the randomized evidence is more than four years out, while the filing under review is — per Dermatology Times, secondary reporting Prognyx could not verify against any primary record — a third resubmission. On Prognyx's reading, three resubmissions imply at least three prior complete response letters; only two have been reported, and the first is absent from the record reviewed.

Read the comparator arm as a regulatory statement in its own right. Physician's Choice in IGNYTE-3 comprises nivolumab, nivolumab plus relatlimab, pembrolizumab or single-agent chemotherapy ^[2]. That is four named active options in the setting where a single-arm response-rate dataset is being asked to carry an approval. Any sponsor arguing an unmet-need exemption in post-PD-1 melanoma is arguing against that list.

The sequence behind the adcomm — trade reports, not records

Everything known publicly about the regulatory steps of the past four months rests on trade reporting that Prognyx could not verify against a primary record, and that reached the engine only as news-aggregator redirects — links that do not resolve to a publisher page and therefore cannot be cited here. Named, dated, and carried on that basis alone: OncLive reported on 10 April 2026 that FDA issued a complete response letter for RP1 plus nivolumab in advanced melanoma; Targeted Oncology on 13 April 2026 and CancerNetwork on 10 April 2026 both described it as the second such letter; Pharmaceutical Technology carried the same complete response letter on 13 April 2026. The staggered 10 and 13 April datelines look like pickup of one event rather than two, though Prognyx could not confirm that. The Pharma Letter reported on 1 June 2026 that Replimune planned a resubmission, and Dermatology Times reported on 26 June 2026 that FDA accepted a third BLA resubmission for advanced melanoma — the filing that, per that reporting, sits behind the advisory committee meeting now being briefed. No Replimune press release, no 8-K and no fda.gov document in the sources reviewed corroborates any step in that chain.

The calendar — and where it stops

Two of the dates that matter come from the registry and are firm. Randomized overall-survival evidence: not before the 30 September 2030 primary completion of IGNYTE-3 ^[2]. One further pair brackets the platform — the non-interventional five-year follow-up of patients dosed with RP1, RP2 or RP3 (NCT06887348) opened on 12 December 2025 and runs to a primary completion in December 2030, enrolling 50 patients to detect delayed adverse events and systemic HSV-1 infection related to treatment ^[3].

The rest of the calendar rests on single-outlet press reporting and is not firm. The reviewers' briefing documents were reported released ahead of the panel on 28 July 2026, which places the meeting itself after that date ^[1]. The third resubmission was reported accepted by Dermatology Times on 26 June 2026; the acceptance date itself is not stated in the sources reviewed, and an acceptance letter routinely precedes trade pickup by days or weeks.

One date is missing, and it is the one competitors would price. The resubmission class — which sets the review clock and therefore the action date — is not stated in the sources reviewed, and no goal date has been published in them. Prognyx's range, not a published goal date: measured from the resubmission Dermatology Times reported accepted on 26 June 2026, a Class 1 resubmission would put FDA action around late August 2026 and a Class 2 resubmission around late December 2026. Four caveats travel with that arithmetic — the class is an assumption rather than a fact, the statutory clock runs from FDA receipt of the resubmission rather than from the report of its acceptance, neither the receipt date nor a goal date appears in the record, and the acceptance itself is unconfirmed in primary sources. What can be said without arithmetic is that FDA action falls after the advisory committee meeting, which falls after 28 July 2026 ^[1].

Who is exposed

Patients first: the pre-approval route is already shut. The expanded access programme providing RP1 plus nivolumab in advanced melanoma is no longer available on the registry record dated 18 December 2025, with no reason recorded ^[4].

The platform next. RP1 is the backbone of Replimune's Immulytic HSV-1 franchise — an enhanced-potency oncolytic immunotherapy armed with GM-CSF and the fusogenic protein GALV-GP R-, with RP2 additionally expressing anti-CTLA-4 and RP3 adding CD40L and 4-1BBL ^[6]. A design-level objection to the backbone's registrational path is not confined to one asset; it sets the evidentiary shape every follow-on filing has to take.

And then everyone else holding a single-arm registrational cohort in a setting with active alternatives. If the agency's position is that the design, not the dataset, is the defect, that standard travels — to intratumoral agents, to oncolytic viruses, and to any response-rate filing where a randomized comparison was deferred rather than run.

What we could not verify

Six things a reader should hold against this analysis: the FDA briefing document's own text, which reaches us only through the Endpoints News summary ^[1]; the advisory committee's identity, its meeting date and the voting questions put to it; the pivotal trial's own ClinicalTrials.gov record, whose absence is why the identification of the 125-patient cohort as the study at issue stays an inference; the efficacy figures actually contained in the BLA; the complete response letters and the third resubmission acceptance, none of which Prognyx could trace to a regulator record, an SEC filing or a company release; and the European status of vusolimogene oderparepvec, including whether a marketing authorisation application exists and at what stage.

The now-what

Option one — argue the confirmatory trial is already running. It is, and it is properly powered on overall survival ^[2]. The weakness is arithmetic: 30 September 2030 is the primary completion date, and a design objection raised in 2026 is not resolved by evidence that arrives in 2030 ^[2].

Option two — randomize earlier, not later. The single most transferable lesson is that the 2020–2021 decision to run a cohort rather than a controlled study is documented in the sponsor's own filings ^[5][6], while the same company's contemporaneous cutaneous squamous cell carcinoma programme was randomized from the start ^[7] — and no choice made inside a 2026 review cycle changes which of those two designs generated the melanoma dataset. Sponsors with a single-arm oncolytic or intratumoral cohort heading toward a filing should assume the burden is now a concurrent control, and convert while conversion is still cheap.

Option three — treat Europe as a separate question. Medscape reported on 22 May 2026 that EMA backed a breast cancer SERD after an FDA no vote — a single secondary account, on a different molecule in a different disease, which reached Prognyx only as an aggregator redirect and is therefore named here without a citable publisher link.

The falsifiable read: if the deficiency is design-level, no reanalysis of the existing cohort fixes it — only IGNYTE-3 does. Watch whether the agency's action turns on the addition of randomized data rather than on any new cut of the response data.

Verdict — threat level high for any single-arm registrational cohort in post-PD-1 melanoma or in oncolytic immunotherapy generally; the window to act closes when your own design locks, not when the briefing documents land.