GNLX / olvi-vec-platinum-resistant-ovarian — Olvi-Vec (olvimulogene nanivacirepvec) for platinum-resistant or platinum-refractory ovarian cancer

Program analysis · bpiq_drug_id 17406 · prepared 2026-08 · USD · framework v5.6.1 · Coverage: CLEARED

Company context, financials, ownership, options and the full pipeline: ../company.md.

How to read this. Written for a reader who has not studied pharmacology or finance. Terms specific to this program are in the Glossary below; general terms are defined once in framework/04-glossary.md. Tags: [VERIFIED — source] / [UNVERIFIED] / [WEB ESTIMATE — source, date].

Glossary

Only terms specific to this program. General terms — progression-free survival, overall survival, objective response rate, single-arm trial, open-label, Fast Track and so on — are defined once in framework/04-glossary.md.

Executive summary

• What it is (one sentence): A genetically engineered smallpox-vaccine virus, poured into the abdominal cavity through a temporary catheter over two days, intended to make an ovarian tumour that has stopped responding to platinum chemotherapy respond to it again.
• The event and when (as disclosed): Topline progression-free survival from the randomised Phase 3 OnPrime / GOG-3076 trial (NCT05281471), disclosed only as “H2 2026” — a six-month period, not a day. This refresh judges the readout window as 2026-10-01 to 2027-04-30, likeliest 2026-12-15, at PERIOD precision and LOW confidence. See Readout, below, for how that was built and why it extends past the company’s own guidance.
• The main reason it could work: The single-arm Phase 2 that preceded it produced a 54% response rate and a median progression-free survival of 11.0 months in women who had already had a median of four lines of treatment, which is far beyond anything that population normally achieves. An independent Phase 1 study at a different institution confirmed, with a blinded pathologist, that the virus does infect tumour cells and does pull immune cells in.
• The main risk: The trial gives platinum chemotherapy to patients whose disease is defined by not responding to platinum. If the virus does not reverse that resistance, the treatment arm receives a drug that cannot work plus its toxicity, and can finish worse than the control arm. That is exactly what happened to the closest comparable oncolytic virus, Pexa-Vec, which was stopped for futility in its Phase 3 with overall survival numerically worse than control.
• What it means for the stock: The arithmetic below puts the probability-weighted value about 50% above the current share price, but the single most likely outcome is a fall of 52% to 83%. Those two statements disagree because the payoff is lopsided, not because the result is likely to be good. The company is a one-molecule company with roughly one to two quarters of cash left at the readout and a $100M at-the-market facility that this refresh confirms is still essentially undrawn and therefore still fully available to be issued into a positive result. See ../company.md C.3 and C.4.

0. Program-tier coverage — CLEARED

Company-tier coverage is in ../company.md C.0. It is not repeated here.

Verdict: CLEARED. No mandatory row reads NOT CALLED, and — unlike the previous version — no row reads BLOCKED either. The Open Targets block that shaped the last analysis has cleared, and ChEMBL returned after the full retry policy rather than before it.

A. Scientific & clinical assessment

A.1 Mechanism of action and scientific rationale

Olvi-Vec is a live virus used as a drug. It is a modified strain of vaccinia, the virus that was used for a century as the smallpox vaccine, so its behaviour in humans was already well characterised before anyone tried to treat cancer with it [VERIFIED — BPIQ fetch_company_info company description; ChEMBL CHEMBL4594308, which records the laboratory name GLV-1h68 and the earlier clinical name GL-ONC1 as synonyms of the same agent, read 2026-08-11].

The virus has been engineered so that it grows much better inside tumour cells than inside healthy ones. When it grows inside a tumour cell, it bursts it. That is the first half of the idea and it is the straightforward half.

The second half is the reason this trial exists. When tumour cells burst, they spill their own proteins into the surrounding tissue at the same moment as the virus is setting off the body’s alarm signals for an infection. The immune system, which had been ignoring the tumour, arrives to deal with the infection and finds the tumour proteins while it is there. Killer T cells move in. The company’s claim is that this changes the tumour so much that platinum chemotherapy starts working again in patients whose disease had stopped responding to it [VERIFIED — the hypothesis as stated in the trial's own detailed description, ClinicalTrials.gov NCT05281471 read 2026-08-11, and in the design paper, Holloway et al., Int J Gynecol Cancer 2023;33(9):1458-1463, [DOI 10.1136/ijgc-2023-004812](https://doi.org/10.1136/ijgc-2023-004812), according to PubMed].

Delivery is unusual and matters commercially as well as scientifically. The virus is not injected into a vein. A catheter is placed into the abdominal cavity by keyhole surgery, the virus is poured in on two consecutive days at 3 × 10⁹ plaque-forming units per day, and the catheter is then removed [VERIFIED — NCT05281471 detailed description]. This makes sense for ovarian cancer specifically, because the disease typically spreads as many small deposits across the lining of the abdominal cavity, so a drug delivered into that cavity reaches the tumour directly rather than having to survive the bloodstream.

How well is the target validated? This refresh can finally answer the question rather than record a blocked connector. At the last sweep, Open Targets was BLOCKED across four attempts, so nothing was asserted either way. This sweep it answered. Searching olvimulogene nanivacirepvec returns exactly one entity — CHEMBL4594308, typed drug — and no target entity of any kind, while a control query returns three disease entities normally [VERIFIED — Open Targets search_entities, 2026-08-11].

That is a real finding, and it is not the same as the previous version’s silence. Olvi-Vec has no molecular target in the ordinary sense — there is no receptor it binds and no enzyme it blocks — so the genetic-validation evidence that Open Targets exists to supply does not exist for this agent, and now that has been checked rather than assumed. A.5 still scores target validation Low, but for a stated reason rather than a missing call.

What does exist is direct human tissue evidence, and one piece of it is genuinely independent:

• In a Memorial Sloan Kettering-sponsored Phase 1 study of the same virus given into the chest cavity (n=18), vaccinia was detectable in tumour cells two to five days after treatment, and a pathologist who did not know which patients had been treated recorded a fall in tumour-cell density and a rise in immune-cell density. Killer T cells, natural killer cells, dendritic cells and neutrophils all increased — and so did regulatory T cells, which suppress immune responses [VERIFIED — Chintala et al., Front Immunol 2023;14:1112960, [DOI 10.3389/fimmu.2023.1112960](https://doi.org/10.3389/fimmu.2023.1112960), according to PubMed]. The authors are from Memorial Sloan Kettering and NYU, not from Genelux. This is the single most trustworthy piece of mechanistic evidence in the file.
• A German academic Phase 1 study of intraperitoneal delivery (n=9) confirmed infection, replication and oncolysis in eight of nine patients — but only in cycle 1. Every patient developed neutralising antibodies against the virus [VERIFIED — Lauer et al., Clin Cancer Res 2018;24(18):4388-4398, [DOI 10.1158/1078-0432.CCR-18-0244](https://doi.org/10.1158/1078-0432.CCR-18-0244), according to PubMed].
• New since the last version, and now formally published: Genelux’s own translational analysis of the Phase 2 appeared in print on 2026-06-19 and is indexed in PubMed Central. It reports preclinical synergy between Olvi-Vec and cisplatin in a mouse model of platinum-resistant ovarian cancer, viral infection and eradication of tumour cells in patients’ ascitic fluid, a large influx of CD8+ tumour-infiltrating lymphocytes on multiplex staining of paired biopsies, and RNA profiling identifying four expression patterns the authors associate with immune activation, viral infection, chemotherapy sensitisation and anticancer activity [VERIFIED — Yu et al., Gynecol Oncol Rep 2026;66:102145, [DOI 10.1016/j.gore.2026.102145](https://doi.org/10.1016/j.gore.2026.102145), PMID 42383154, according to PubMed]. This paper is not independent: its first author is a Genelux employee in Clinical Research & Development, and the second author is the trial’s national principal investigator. It is the most direct published support the unproven step has, and it comes from the sponsor.

The single-cycle limit is not a flaw in the design; the trial gives exactly one cycle, which is consistent with the biology. But it does mean the entire therapeutic effect has to come from one hit.

The exact scientific step the next readout must prove. Not that the virus infects tumours — that is already shown by a blinded independent pathologist. Not that immune cells arrive — that is shown too. The step under test is the arrow marked in yellow on the diagram above: that the immune activation translates into the tumour becoming sensitive to platinum again, enough to extend progression-free survival against physician’s choice of chemotherapy plus bevacizumab. Everything between the virus entering the abdomen and a longer time to progression is established; that one link is not. The 2026 translational paper argues for that link from mouse pharmacology and human biopsies — but the argument runs from mechanism to plausibility, not from randomised evidence to benefit, and its authors are the sponsor.

The honest scientific risk, stated plainly. The trial gives platinum to patients who are defined by not responding to platinum. Outside a trial, re-challenging a platinum-resistant patient with platinum is regarded as futile and is avoided. So this design is not a free ride: the experimental arm is not simply “control arm plus an extra drug”. If the virus does not do what is claimed, those patients receive a chemotherapy that cannot work, together with its toxicity, instead of a chemotherapy that can. The arm can finish worse than control, not merely no better.

That is not a hypothetical failure mode. Pexa-Vec (pexastimogene devacirepvec) is also an engineered vaccinia oncolytic virus, and was also tested in a sequential design alongside a standard therapy. Its Phase 3 PHOCUS trial in liver cancer was stopped on the advice of its own monitoring committee after a planned futility analysis, with median overall survival of 12.7 months in the virus arm against 14.0 months in the control arm, and time to progression of 2.0 months against 4.2 months — the virus arm was worse on both [VERIFIED — PHOCUS results as reported, Liver Cancer 2024;13(3):256, PubMed PMID 38756145] [WEB ESTIMATE — GEN, "Pexa-Vec/Nexavar Combination Fails Phase III Trial in Liver Cancer"].

In the entire history of the field, one oncolytic virus has been approved in the United States and Europe: talimogene laherparepvec, a herpes virus injected directly into skin lesions in melanoma, approved on a durable-response endpoint [VERIFIED — OPTiM Phase 3, PMC5129499]. That is the base rate this programme is arguing against.

A.2 Clinical development plan, timeline, feasibility, resourcing

Trials

Comparator for the pivotal trial’s claims. The comparison that matters is Olvi-Vec’s own Phase 2 against the two agents approved in this disease since OnPrime opened. VIRO-15 reported a median progression-free survival of 11.0 months in a single arm of 27 patients. Relacorilant plus nab-paclitaxel reported 6.5 months against 5.5 months in a randomised trial of 381 patients, and was approved on it [VERIFIED — ROSELLA results as reported; see B.1]. An 11.0-month single-arm figure sitting at nearly double a randomised winner’s result is the shape that most often does not replicate.

First and last patient in, last patient out, database lock and topline. First patient in was 2022-08-31 [VERIFIED — ClinicalTrials.gov]. Last patient in has never been disclosed, and this remains the most important gap in the whole document. The database lock date has never been disclosed either — it appears in no ClinicalTrials.gov field, in none of the PubMed records read, and in none of the 137 press items read for ../company.md C.0. Topline is guided to “H2 2026”; the window this analysis judges is in Readout, below.

Feasibility matrix

The timeline does not obviously close, and the case has weakened since the last version. A trial that is still enrolling in August 2026 needs its last patient to be followed long enough to contribute a progression event before the database can lock. Progression-free survival in the control arm of this population runs roughly 3.5 to 5.5 months on the two randomised precedents in B.1, so events do accrue quickly once patients are in — which is the argument that a H2 2026 topline is still achievable, and it is a real argument: the 127 events the trial is powered for come mostly from patients enrolled years ago, not from the last one in. The counter-argument has strengthened. At the 2026-08-05 analysis the trial was 47 months old and recruiting; it is now 48 months old and still recruiting, six weeks into the guided half-year, with no enrollment-completion announcement and no Q2 update to explain it (see ../company.md C.3 on the overdue results release). [UNVERIFIED — judgement, built on the verified enrollment history, the verified registry status and the verified control-arm PFS benchmarks]

Resourcing sufficiency. Genelux does not have the cash to reach approval and does not claim to. Cash was $26.209M at 2026-03-31, guided to last “into the first quarter of 2027”, against the readout window judged below that runs to 2027-04-30 — see ../company.md C.3. It has 32 sites running, which it can plainly fund to the readout on either burn route, and a $100M at-the-market facility live since 2026-03-19 which is how anything after the readout will be paid for. This refresh establishes that the facility is essentially undrawn through 2026-05-03, so the company has not been funding the trial by quietly selling stock — the whole $100M is still ahead of shareholders rather than behind them. A new chief medical officer started in January 2026, five months into the final year of a pivotal trial [VERIFIED — company release 2026-01-05]. The company has the people and sites to finish the trial; it does not have the money to file and launch without raising several times what it is currently worth.

A.3 Target label, target product profile, strategic questions, designations

A.3a Target label. Adult women with platinum-resistant or platinum-refractory, non-resectable high-grade serous, endometrioid or clear-cell ovarian, fallopian tube or primary peritoneal cancer who have received at least three prior lines of systemic therapy including bevacizumab, treated with one cycle of intraperitoneal Olvi-Vec followed by platinum-doublet chemotherapy and bevacizumab [VERIFIED — NCT05281471 eligibility criteria and arm descriptions, read 2026-08-11].

A.3b Target product profile

A.3c Strategic Go/No-Go questions.

Pre-Phase-II (Go-to-Phase-II) — answered retrospectively, because this decision was taken in 2022:

Pre-Phase-III (Go-to-Phase-III / registration) — this is the live set, because the next decision is a regulatory filing:

A.3d Regulatory designations.

• Fast Track designation, granted 2023-11-27 for Olvi-Vec in platinum-resistant/refractory ovarian cancer. It grants more frequent meetings and written communication with the FDA, eligibility for accelerated approval and priority review if the criteria are met, and the ability to submit a marketing application in sections as they are completed rather than all at once. It does not imply any view on whether the drug works [VERIFIED — company release 2023-11-27; BPIQ historical catalyst id 1079].
• FDA alignment on the approval pathway, announced 2025-03-25 following a Type D meeting. The FDA stated that an interim analysis of overall survival should be planned at the time of the primary progression-free-survival analysis, and that a clinically meaningful PFS advantage in the absence of a decrement in overall survival could potentially support traditional approval. This is not a designation and grants nothing, but it is the clearest public statement of what the trial has to produce, and it puts an explicit overall-survival tripwire on the result [VERIFIED — company release 2025-03-25, GlobeNewswire and BioSpace].
• Orphan Drug designation: not confirmed, and now corroborated as absent by a second source. The exclusivity web search again found no orphan designation, none appears in the press feed, and ChEMBL’s record for CHEMBL4594308 carries orphan: 0 [VERIFIED — ChEMBL 2026-08-11, for the field value only] [UNVERIFIED — the underlying regulatory fact]. Still not claimed here.
• Breakthrough Therapy designation: none announced [VERIFIED — absent from 137 press items].

A.4 Target product profile valuation matrix

Calibration examples, not asset claims: in oncology each incremental month of overall survival is worth roughly $150–300M in peak-sales potential; oral formulations command roughly 15–25% price premiums over injectables. The second of those cuts against this asset harder than usual, because Olvi-Vec is not merely injectable but surgically delivered.

A.5 Evidence quality and endpoints

Endpoints

A.5b Key opinion leaders.

Panel as of. 2026-08-11 — the date the investigator and independent-voice searches below were run.

Investigators

From the OnPrime registry record and the programme’s own publications. The registry lists no overall official, so the national principal investigator is taken from the trial’s own detailed description, which names him explicitly. Being paid by the trial is a relationship with the sponsor by construction; Conflicts below records relationships beyond that one.

The registry lists named contacts at all 32 sites; six are recorded here — the national principal investigator, the two site investigators who are also authors on the programme’s own publications, and three further named site principal investigators. The remaining site contacts were not individually conflict-checked, and this table does not imply they are clean.

Independent voices

Named commentators on this program’s endpoint who are not investigators on its trial and hold no disclosed relationship with the sponsor.

One prominent voice was found and deliberately not recorded as independent. The most directly on-point recent commentary on oncolytic viruses in platinum-resistant ovarian cancer is an editorial by Akseli Hemminki and colleagues, “Oncolytic adenoviruses in platinum-resistant ovarian cancer: a new era in immunotherapy?” [VERIFIED — Expert Opin Biol Ther 2025;25(6):561-564, [DOI 10.1080/14712598.2025.2501731](https://doi.org/10.1080/14712598.2025.2501731), according to PubMed]. Four of its six authors give TILT Biotherapeutics Ltd as an affiliation — a company developing competing oncolytic viruses. Under rule 40 that is a disclosed competitor relationship, so the piece cannot be presented as an independent read on this endpoint, and under 03b’s own rule a voice carrying a conflict does not belong in the table above. It is named here rather than dropped, because “the loudest advocate for this mechanism in this indication works for a competitor” is itself a finding about how thin genuinely independent enthusiasm is.

Judgement

Dissent

B. Commercial assessment

B.0 Current treatment algorithm

Ovarian cancer is usually found late and is usually treated first with surgery and a platinum-based chemotherapy, often with bevacizumab and sometimes followed by a PARP inhibitor as maintenance. Most patients respond well the first time.

The disease then comes back, and each time it comes back the interval before it returns gets shorter. When it returns within six months of the last platinum dose, the patient is called platinum-resistant, and platinum is no longer used. This is the point at which options narrow sharply, and it is where this drug is aimed.

What a platinum-resistant patient receives today, in order:

1. If the tumour is folate-receptor-alpha positive: Elahere (mirvetuximab soravtansine), an antibody-drug conjugate approved for patients who have had one to three prior regimens. Median overall survival 16.5 months against 12.7 months for chemotherapy in the MIRASOL trial [WEB ESTIMATE — MIRASOL as reported by Targeted Oncology and OncLive, 2026].
2. Regardless of biomarker, since March 2026: Lifyorli (relacorilant) plus nab-paclitaxel, approved for one to three prior regimens including bevacizumab, on the ROSELLA Phase 3 [WEB ESTIMATE — FDA approval as reported by OncLive and CancerNetwork, March 2026].
3. Emerging: pembrolizumab plus weekly paclitaxel with or without bevacizumab, which reported significant progression-free and overall survival benefit in ENGOT-ov65/KEYNOTE-B96 in both the PD-L1-selected and all-comer populations [WEB ESTIMATE — ASCO 2026 reporting].
4. After all of the above: single-agent chemotherapy — gemcitabine, a taxane, or pegylated liposomal doxorubicin — with or without bevacizumab. This is where response rates fall into the low teens and progression-free survival is measured in weeks.

Where Olvi-Vec would fit: at step 4, displacing single-agent chemotherapy. Its trial requires at least three prior lines, which is deeper than the one-to-three-regimen labels of both approved competitors. It would be a fourth-line-or-later option, given once, in patients with peritoneal disease who are still well enough for a laparoscopic procedure and platinum chemotherapy. That positioning is a genuine advantage in one respect — it is not competing head-on with Elahere or relacorilant for the same patient at the same moment — and a genuine disadvantage in another: the population is smaller, sicker and harder to enrol commercially.

B.1 Competitive landscape and positioning

Where this asset wins, and the single fact the thesis rests on.

It wins on line of therapy and on the absence of a biomarker. A patient who has failed Elahere and relacorilant has essentially nothing left, and Olvi-Vec is aimed exactly there, with no test to pass first. It wins on treatment burden in one narrow sense: a single two-day course, rather than indefinite infusions until progression.

The single fact the whole thesis rests on is the VIRO-15 result: a 54% objective response rate and 11.0 months median progression-free survival in 27 women with a median of four prior lines of therapy. If that number is real, this drug is transformative in a setting where nothing works. If it is the product of a small, unrandomised, two-site cohort selected by investigators who then went on to lead the Phase 3, it is the most ordinary kind of illusion in oncology.

The landscape moved underneath this trial while it was running, and that cuts two ways. The control arm — physician’s choice single-agent chemotherapy plus bevacizumab — was the correct comparator in 2022. It is now a weaker comparator than what many patients would actually receive, because relacorilant plus nab-paclitaxel and Elahere have both taken patients out of that pool. Mechanically that makes the trial easier to win, because the control arm’s expected performance is what it always was. Commercially it makes a win worth less, because approval would arrive into a setting with three recent entrants rather than none, and because payers and clinicians will ask whether a control arm from 2022 still describes practice in 2027.

One competitive fact moved in Genelux’s favour since the last version, and it is small. At ASCO 2026 Elahere failed to extend its franchise into platinum-sensitive disease [WEB ESTIMATE — Pharmaceutical Technology / Yahoo Finance, ASCO 2026 coverage]. That does not touch Elahere’s platinum-resistant label, which is the part of the market Olvi-Vec sits behind, so it changes the ceiling on the competitor rather than the floor under this asset. It is recorded because it is the only competitive development of the period that does not cut against this programme.

Calibration example, not an asset claim: oncology first-movers in novel mechanisms have shown roughly 3.2× higher peak-sales potential but about 1.8× higher development risk.

B.2 Addressable market

Launch markets. The United States first — the trial is entirely US-based, at 32 sites, and the FDA pathway is the one that has been discussed publicly. Europe would follow and is not addressed here. China, Taiwan, Hong Kong and Macau are out-licensed to Newsoara BioPharma and are not Genelux’s market [VERIFIED — BPIQ fetch_company_info company description].

The delivery method is the binding commercial constraint, and it is independent of whether the drug works. To give Olvi-Vec, a centre needs a gynaecologic oncologist who can place an intraperitoneal catheter laparoscopically, an institutional biosafety committee that has approved handling a live replicating virus, and staff trained in that handling. Community oncology practices, which treat most American cancer patients, have none of those things. Penetration is therefore capped by the number of capable centres long before it is capped by efficacy. The 32 trial sites are a reasonable first approximation of how many US centres could deliver it on day one [UNVERIFIED — judgement, built on the verified delivery method and the verified site list].

B.3 Value and feasibility

B.3a Expected peak sales.

This refresh can attempt the bottom-up build the previous version abandoned, and it is still the weaker of the two routes. Last time no verified eligible-patient count was obtained at all, so the estimate was anchored purely to a comparator’s revenue. A top-of-funnel epidemiology anchor now exists, so the chain is written out — with every link tagged, and the links that are not defensible named as such rather than smoothed over.

The attrition chain, US only:

Two of the four links that matter carry no source, so no bottom-up point estimate is produced from this chain, and rule 9 forbids inventing one. What the chain does establish is a ceiling: a US funnel that starts at 20,000 patients a year cannot support a fourth-line, procedure-limited product at blockbuster scale, whatever the response rate is. That is worth having, and it corroborates the comparator route below rather than replacing it.

The comparator route, updated:

The anchor is now an actual figure rather than a forecast. Elahere’s global sales reached approximately $750M in full-year 2025, with $182M in the fourth quarter alone, up from about $480M in 2024 [WEB ESTIMATE — AbbVie results as reported by pharmaphorum and Pharmaceutical Technology, 2026]. The previous version of this document used ”>$470M in 2024 actual, ~$750M modelled for 2025”; the 2025 figure has since printed at that level, so the anchor is firmer than it was. Elahere serves platinum-resistant patients at one to three prior regimens who are folate-receptor-alpha positive.

The adjustment, argued: Olvi-Vec’s population is later (at least three prior lines, so a smaller pool after attrition) but unselected by biomarker (so a larger share of the patients who are in that pool). The two effects push in opposite directions and the later-line effect is the stronger of them. Against that, delivery through a laparoscopic catheter caps site penetration in a way Elahere’s intravenous infusion does not.

The dollar figures are unchanged from the previous version; the percentages behind them fell, because the anchor grew. Last time the same $70M/$140M/$280M were 15%/30%/60% of a $470M anchor. Against a $750M anchor the identical dollar outcomes are 10%/19%/37%. Nothing about Olvi-Vec’s own prospects improved — the comparator got bigger. Holding the dollars and letting the percentages fall is the honest way to record that, because the dollars were never derived from the percentage in the first place: they were derived from a fourth-line, procedure-limited addressable population that has not changed.

Every figure above is conditional on approval, which has not happened and which requires a successful readout first. All three exclude China, Taiwan, Hong Kong and Macau, which are out-licensed. All three exclude the two lung programmes entirely. The input that is not defensible remains the patient count, and the attrition chain above shows exactly which two links break.

B.3b Feasibility

B.4 Product-development risk

Framing questions.

1. Does the development plan support the target product profile claims? Yes, for the claim that matters. The trial is randomised, active-controlled, adjudicated by blinded independent central review, uses the endpoint the FDA agreed to, and is powered for it. A positive result would support a label in the population described in A.3a. What the plan does not support is any claim about the virus’s contribution as distinct from the regimen’s, because there is no arm that receives platinum-doublet chemotherapy without the virus.
2. Will the identified risks affect the target product profile? Yes, one of them decisively. The overall-survival tripwire the FDA described means a safety or tolerability problem that shortens survival would remove the approval path even if progression-free survival were positive. Every other identified risk affects the size of the opportunity rather than its existence.
3. If a risk cannot be mitigated, is the asset still differentiated from competitors? The unmitigable risk is the delivery method, and yes — the asset remains differentiated, because no competitor treats fourth-line patients with a single two-day course and none claims to restore platinum sensitivity. It is differentiated into a smaller market than an intravenous version of the same drug would reach.

Readout

The judgement every timing call below reads instead of the BPIQ placeholder.

Sources

The modelled estimate. The registry’s primary_completion_date is 2026-12. data/benchmarks/readout-lag.json holds no observation for a comparable trial, so rule 34’s stated default applies: primary completion plus two to four months to database lock and analysis gives 2027-02 to 2027-04, tagged [UNVERIFIED — modelled, default lag]. The arithmetic is stated so it can be argued with, and the argument against it is real: the primary analysis is event-driven (127 progression-free-survival events), and an event count can be reached before the last patient’s last primary-endpoint assessment, so the lock could precede primary completion rather than follow it. No event count is public, so no independent event-side arithmetic is possible. This row therefore sets the window’s latest edge and nothing else.

Window

Basis. Earliest is 2026-10-01 because the registry still read RECRUITING on 2026-08-11 with all 32 sites individually recruiting, and a database cannot lock while a trial is still enrolling — a September topline is not credible, while an October one is if the event count is already near target from patients enrolled years ago. Likeliest sits at 2026-12-15 because the guidance is deadline-shaped (“H2 2026”, reiterated three times without ever narrowing) and a deadline-shaped guidance tends to land near its deadline; the registry’s own 2026-12 primary completion agrees with that end of the range. Latest is 2027-04-30, taken from the modelled row: it absorbs the plain reading that a trial whose primary-endpoint data collection ends in December cannot announce in December. Precision is PERIOD because no source names a month for the readout itself — “H2 2026” is a half-year, and the registry’s 2026-12 dates a different event (last data collection), not the announcement. Confidence is LOW, one notch below where a thrice-repeated guidance would otherwise sit, because this catalyst has already slipped twice, enrollment completion has never been announced, last patient in and database lock have never been disclosed, and the most recent company statement is three months old with the quarterly update overdue.

Disagreement. UNRESOLVED.

Which sources conflict and which way each points. The company and BPIQ both say H2 2026, ending 2026-12-31. The registry says primary-endpoint data collection runs to 2026-12, and the modelled arithmetic on top of it points to 2027-02 to 2027-04. These are not compatible on their face: a topline announced inside H2 2026 would have to be announced essentially simultaneously with the last primary-endpoint assessment, with no time for database lock or analysis. The reconciliation that would make them agree — that the event-driven analysis triggers well before primary completion, so the registry field is tracking a last-visit date the analysis does not wait for — is plausible and is an inference, not a disclosure; no source states it. The opposite reading, that the readout slips into 2027, is equally available and is what the window’s latest edge absorbs. Neither is picked and neither is averaged (rule 24). The previous version of this analysis recorded these same sources as agreeing (“all three sources agree; none names a day”); that reading looked only at whether the dates overlapped, not at whether the sequence they imply is physically possible, and it is corrected here.

Date slippage. Two slips across five dated statements, oldest first.

The previous version recorded one slip, counting only the transition captured inside the BPIQ note field. Counting the peer-reviewed design paper’s own 2025 projection as the first dated statement gives two. The two most recent statements have held the date unchanged for nine months, which is the mildly reassuring part; that no sixth statement exists, because the quarterly update is overdue, is not.

Attribution

Status. INDETERMINATE

Conflicts

Computed 2026-08-11 with lib/clustering.mjs’s attributionFor over this ticker’s pipeline for bpiq_drug_id 17406, CATALYST_CLUSTER_MIN_MONTHS = 6, using this program’s own readout window; transcribed, not estimated.

Note. Not required for INDETERMINATE, and not written. The status is what it is for the reason 02-connectors.md gives: both colliding rows carry only the period-end placeholder 2026-12-31 derived from the bare text “2026”, so neither side rests on a disclosed date and no conflict between them can be confirmed. This is evidence of not knowing rather than a finding. What the reader does need is stated in the stock-direction call below, as rule 36 requires.

One thing about this particular INDETERMINATE is worth saying plainly, because it is unusual. On most tickers a placeholder collision is an artifact. Here the two colliding rows are the same molecule as this program, in two other cancers. If the readout is negative in a way that impugns Olvi-Vec itself rather than this indication, the two lung rows do not merely fail to cushion the result — they are re-priced by it. The clustering pass cannot see that, because it measures date proximity and not mechanistic correlation. It is recorded here so the stock call below is read with it in view.

Market and timing for this event

• Plain takeaway. The stock is sitting near the bottom of its 52-week range going into the most important event in the company’s history, with no fast money positioned either way, an options market that cannot express a view, and a treasury that runs out around the same time as the result arrives. Nothing has been priced in, in either direction, because there is nobody there to price it. It made a new recent low of $2.51 six days before this analysis on three times normal volume, with no explanation in any source.
• Months to this catalyst. Between 1.7 and 8.6 months, measured from readout.window.earliest (2026-10-01) and latest (2027-04-30) against the 2026-08-11 lock date; 4.1 months to the likeliest date of 2026-12-15. readout.precision is PERIOD, so none of these is a disclosed date: they are the edges of a judged window, and the BPIQ catalyst_date of 2026-12-31 is a period-end placeholder that is not used for timing anywhere in this document (rule 23).
• Expected move around this event. ../company.md C.6 records the options chain as unusable, and more so than at the last analysis: no at-the-money strike exists at any expiry, total open interest across the whole chain is 527 contracts, day volume was zero, and all four expiries now carry the documented implied-volatility floor or ceiling artifact where one contract still produced a real number last time. No market-implied move can be quoted. On the scenario ranges below, the implied one-way move is at least 50% and plausibly far more in either direction [UNVERIFIED — judgement, built on the anchors in the scenario table; the chain can neither confirm nor contradict it]. The chain also still has no expiry between 2026-10-16 and 2027-01-15, which brackets the likeliest readout date.
• Nearest comparable past reaction. There is no comparable row, and that is the finding. The four events in ../company.md C.7 are a Fast Track designation (+14.4%), a first-patient-dosed milestone (+11.5%), a bundled data-plus-regulatory-plus-offering day (+16.0% intraday against a −10.0% opening gap) and an interim lung release whose recorded −24.8% is attributable to an equity offering priced three days later rather than to the data. Not one is a randomised efficacy readout. The closest in kind — the January 2026 lung interim — is exactly the row that C.7 shows to be contaminated. C.7 establishes that this stock pays low double digits for secondary news and says essentially nothing about what it does on a pivotal one.
• Materiality. Dominant, as recorded for bpiq_drug_id 17406 in ../company.md C.2. This is the only randomised trial in the company, the only registration-intended readout, and the reason Fast Track was granted. The other two pipeline rows are early-stage studies of the same molecule, so a failure here would re-price them rather than cushion it. Essentially all of the roughly $96M recomputed enterprise value sits on this result.
• Date slippage. Two slips across five dated statements — see Readout, above.

Spot. $2.72, cited from ../company.md C.1 and C.4, as of 2026-08-10. The BPIQ price feed carries no date; C.4 establishes by cross-check against the committed price cache that BPIQ’s last_price is the previous session’s close, which makes this the 2026-08-10 close. Previous close $2.66 (2026-08-07).

Scenario prices

The share count underlying the miss-scenario cash-per-share anchor is now the EDGAR-filed figure rather than a figure implied from the connector’s market capitalisation. The two agree to two shares, so the range is unchanged; the anchor is simply better sourced than it was.

Expected value. The modelled probability below (35%) applied to the midpoint of each range: 0.35 × $10.00 + 0.65 × $0.875 = $4.07, which is +49.6% against the spot of $2.72. Method: probability_pct applied to the midpoint of each scenario range, rounded to the cent.

This is arithmetic, not advice, and it is not a price target. It is positive only because the positive range is wide, not because the positive outcome is likely. The single most probable outcome in this table is a fall of 52% to 83%.

Run-up

A run-up call is permitted here — readout.precision is PERIOD, not UNKNOWN, so the date_confidence driver does not floor at zero and rule 39 does not withhold the block. It is permitted, not attractive: the priority score below is 9 out of 100.

Entry and exit

Predicted move and peak

Priority score drivers

Priority score. Priority score 9 · formula_version 1.0.0

Nine out of a hundred. Six of the seven drivers are unremarkable-to-good — unmet need 90, priced-in 93, squeeze 64 — and the seventh overrides all of them, because date_confidence at 15 falls below the formula’s untradeable threshold and caps the result at 15 regardless of what the rest say. That is the machinery working as designed: a readout that could land anywhere in a seven-month window cannot be traded on a run-up, however cheap and however unloved the stock is.

Settlement. Left null at lock time. exit is also null: resolveExit cannot resolve T-5 trading days before 2026-10-01 yet, because the committed price cache ends 2026-08-05 and the series has not reached the window. It resolves once the cache carries bars through October 2026.

Verdict

What I would do. Avoid.

Why. The expected value sits about 50% above the share price, and the most likely single outcome is still a loss of more than half. Both are true because the payoff is lopsided, and a lopsided payoff is worth owning only through an instrument that limits the downside — which ../company.md C.6 shows does not exist here, and is now less available than at the last analysis: the chain traded zero contracts on the sweep date and all four expiries carry implied-volatility artifacts. That leaves common stock, and common stock in a one-molecule company whose pivotal trial is still recruiting six weeks into its own guided readout half-year, with a quarterly update overdue, roughly one to two quarters of cash at the readout, a $100M at-the-market facility now confirmed fully loaded rather than partly spent, no open-market insider purchase in three years, and a twice-repeated habit of selling equity within 72 hours of good news, is the wrong instrument for this bet. The science is more interesting than the setup, and the setup is what determines the outcome for a shareholder.

What would change this. A financing that funds the company through the readout with a year of runway on the far side of it — announced before the topline. That single fact would remove the forced-raise dynamic that caps the positive scenario and deepens the negative one, and would make the asymmetry ownable. A partnership that puts a third party’s capital and commercial organisation behind the ovarian programme would do the same thing more decisively. An enrollment-completion announcement with a last-patient-in date would not change the verdict on its own, but it would convert readout.precision from PERIOD toward MONTH, which is the single change that would most raise the run-up priority score above.

What to watch.

• On or before 2026-08-14, and now overdue against the company’s own precedent — second-quarter 2026 results. The 2025 equivalent came 2025-08-07; nothing had appeared on EDGAR or the press feed as of 2026-08-11, and the statutory 10-Q deadline is 2026-08-14. Four things to read: the cash balance, the share count on the cover (which extends the at-the-market answer past 2026-05-03), whether the runway guidance still says “into the first quarter of 2027”, and whether the enrollment language changes from “remains active”.
• Any date, and still the highest-value single signal — a release or 8-K announcing completion of enrollment in OnPrime, with a last-patient-in date. It has never been announced. Until it is, a H2 2026 topline is a guide rather than a countdown.
• Any date — the ClinicalTrials.gov status for NCT05281471 changing from RECRUITING to ACTIVE_NOT_RECRUITING. Same signal, independently sourced, updated without a press release. It had not changed as of 2026-08-11, with all 32 sites still individually listed as recruiting.
• Ongoing — Form 4 filings. Any open-market purchase by an officer or director would be the first in the entire recorded history of the company and would be genuinely new information.
• 2026-10-23 to 2026-10-27 — ESMO Congress 2026, Madrid. Watch the abstract titles for an Olvi-Vec ovarian entry, which would date the readout precisely. Genelux has not said it intends to present there, so this is a watch item and not a readout source (see Readout, above).
• Through the window — any equity offering, and specifically any at-the-market drawdown visible in the next 10-Q share count. On the C.7 pattern, one arriving within 72 hours of a positive topline should be treated as the base case rather than as a surprise.

Locked prediction

• Outcome-direction (will the readout succeed on its primary endpoint?): uncertain, leaning to a miss — probability 35%, band 20–55% [UNVERIFIED — modelled]. The probability is that OnPrime reports a statistically significant progression-free-survival advantage in the intention-to-treat population without a disclosed overall-survival decrement. Lowered from 40% at the 2026-08-05 analysis. Reasoning for the move: no new independent evidence supports the efficacy claim, while two things cut against it. The trial was still recruiting on 2026-08-11, 48 months in and six weeks into its own guided readout half-year, with enrollment completion never announced and the quarterly update overdue. And the most recent immunotherapy platform trial in exactly this population, IPROC, reported zero responses across two sub-studies with no change in inferred immune-cell frequencies on treatment. The standing reasons are unchanged: the entire efficacy case is a single-arm cohort of 27 patients at two sites led by the same investigators who now run the Phase 3; the design gives platinum to platinum-resistant patients, so a null virus effect produces a worse arm rather than an equal one; and Pexa-Vec in PHOCUS was stopped for futility with the virus arm worse on both survival measures. Against that, the February 2026 IDMC review recommended continuation without modification, the endpoint is FDA-agreed for traditional approval, the control arm’s expected performance is genuinely poor, scans are adjudicated by blinded independent central review, and the sponsor’s translational analysis is now formally published with preclinical platinum synergy. No third party has published a probability on this event; the published analyst views are price targets ($16–$31, Buy or Speculative Buy) that implicitly assume success rather than quantify it, and this view sits well below what those targets imply.
• Stock-direction (which way do the shares move?): no-edge — confidence low — window 2026-08-11 to 2027-05-31, basis: the readout window judged above runs 2026-10-01 to 2027-04-30 at PERIOD precision, and the stock window extends one month past its late edge to cover the reaction to a readout landing at the end of it. Materiality is dominant per ../company.md C.2, so the move will be large in whichever direction it goes. Attribution is INDETERMINATE: it could not be determined whether a price move in this window is attributable to this program alone. Two other has_catalyst rows on this ticker — the same molecule in non-small cell and small cell lung cancer — collide with this window, but both carry only period-end placeholder dates, so the collision is two guesses touching rather than a measured one. Direction is declined deliberately: the expected value below sits 49.6% above spot while the single most likely outcome is a 52% to 83% fall, and those two answers to the same question point opposite ways. What follows from that is an asymmetric, limited-downside position rather than a directional one — and ../company.md C.6 records that the options chain cannot express it.
• Scenario prices: positive $6.00–$14.00 · miss $0.45–$1.30
• Expected value: $4.07, +49.6% against spot $2.72 (arithmetic, not advice)
• Run-up: entry $2.72 on 2026-08-11, exit rule T-5 trading days before readout.window.earliest — predicted move −10% to +35%, predicted peak 0% to +45% around 2026-09 — priority score 9, formula_version 1.0.0
• Settles on the public disclosure of OnPrime / GOG-3076 topline progression-free survival. The pre-registered positive definition: NCT05281471 reports topline progression-free survival by RECIST 1.1 in the intention-to-treat population that is statistically significant in favour of the Olvi-Vec arm at the trial’s prespecified alpha, as stated in Genelux’s own topline release or the ClinicalTrials.gov results record. A result that is not statistically significant, a trial stopped for futility, or a statistically significant progression-free survival accompanied by a disclosed overall-survival decrement in the Olvi-Vec arm all count as a miss. The overall-survival condition is part of the definition because the FDA stated that a progression-free-survival advantage supports traditional approval only in the absence of an overall-survival decrement, so a win on one and a loss on the other is not a positive outcome in any economically meaningful sense.
• Locked: yes · Settled: no

Program data-quality flags

• ChEMBL compound_search required the full retry policy to answer. Three consecutive Tool 'compound_search' exceeded 30.0s server timeout errors and one HTTP error in tool compound_search: 500 from upstream API preceded a successful fourth attempt. A fifth call, on the name variant “Olvi-Vec”, returned a legitimate empty (count 0), so that alias is simply not indexed. The row is recorded CALLED because it ultimately returned; the instability is recorded here because a shorter retry policy would have recorded it BLOCKED and lost nothing but would have looked like a different finding.
• Open Targets has moved from BLOCKED to CALLED since the last analysis, and this changes a claim. The previous version recorded four consecutive Rate limit exceeded for client: global errors and asserted no genetic validation either way. This sweep the connector answered: olvimulogene nanivacirepvec resolves to CHEMBL4594308 typed drug, with no target entity, and a control query returns disease entities normally. A.5’s “Low” target-validation score is now backed by a positive finding rather than by a missing call. One follow-up probe during this sweep did hit the global throttle again, so the connector remains intermittent.
• ChEMBL returns no usable characterisation for this agent. CHEMBL4594308 carries molecule_type “Gene”, structure_type NONE, no molecular properties, no SMILES, no InChI and no bioactivity records, because Olvi-Vec is a live virus. The call confirms identity, the three-name synonym problem, max_phase 3 and orphan 0, and nothing else. No selectivity or off-target claim is made anywhere in this document.
• The PubMed hit count is not comparable to the previous version’s. That analysis recorded 9 hits; the broad query used here returns 57, because it includes the laboratory name GLV-1h68 and its long preclinical literature. Metadata was pulled on the clinically relevant subset rather than on all 57, which is a departure from the “≤15, pull all” rule and is disclosed rather than glossed.
• The pivotal design paper carries a correction whose content still could not be retrieved. Int J Gynecol Cancer 2023;33(10):1672, DOI 10.1136/ijgc-2023-004812corr1, is indexed in PubMed with the title “Correction:” and no abstract, confirmed again this sweep. What was corrected in the OnPrime design publication is unknown. Every design figure quoted here — 2:1 randomisation, n=186, 127 events, PFS primary — is taken from the original abstract and independently corroborated against the ClinicalTrials.gov record, which agrees on enrollment, randomisation and the primary endpoint.
• Last patient in and database lock have never been disclosed. Neither appears in the ClinicalTrials.gov record, in any PubMed record read, or in the 137 press items read for ../company.md C.0. The H2 2026 guidance therefore cannot be checked against the trial’s own arithmetic, only against the 2026-12 primary completion date — which is what makes the readout disagreement UNRESOLVED.
• The registered status and the guidance are now in open tension, not merely in tension “in spirit”. The record read RECRUITING on 2026-08-11 with all 32 sites individually recruiting, six weeks into the half-year the company has guided topline for. Both facts are carried; neither is chosen. The previous version recorded these sources as agreeing, which this refresh corrects.
• The ovarian efficacy literature is not independent, and the newest paper does not change that. The Phase 2 primary publication, the Phase 1b publication and the 2026 translational analysis are all authored by Genelux personnel and the trial’s own site investigators. Robert Holloway is an author on all four and is the Phase 3 national principal investigator; the translational paper’s first author is a Genelux employee. The one genuinely independent mechanistic study remains the Memorial Sloan Kettering intrapleural Phase 1, and it addresses the virus’s behaviour in tissue rather than its efficacy.
• The most on-point independent commentary on this mechanism in this indication comes from a competitor. The Hemminki editorial on oncolytic viruses in platinum-resistant ovarian cancer carries four TILT Biotherapeutics affiliations. Named in A.5b and excluded from the independent panel rather than counted in it.
• Investigator conflict checks are authorship-level only. Journal disclosure appendices were not opened for any named investigator, and 26 of the 32 named site contacts were not conflict-checked at all. An empty Conflicts cell in A.5b means the recorded search found nothing, never that the person is clean.
• The 60% and 33% lung figures in ../company.md C.2 are quoted on three of five and three of nine patients respectively. They are recorded because the company press-released them, not because they support a conclusion.
• The Newsoara royalty rate is not disclosed anywhere this sweep could read, on a second attempt, so the value of the Greater China rights is not estimated. B.3b records the rights split and stops there.
• No composition-of-matter patent expiry could be verified for Olvi-Vec, on a second attempt. A third-party database reports a 2036 expiry for a combination patent covering a different, out-licensed product (V2ACT Immunotherapy), which is not the same thing and is not used.
• No orphan-drug designation was found, and ChEMBL’s orphan: 0 corroborates the absence in one more database. Absence of evidence in a web search and a chemistry database is still not evidence of absence, so this remains unverified rather than a negative finding.
• Competitor efficacy figures are taken as reported and the primary publications were not opened. The ROSELLA numbers (PFS 6.5 vs 5.5, OS 16.0 vs 11.9), the MIRASOL numbers (OS 16.5 vs 12.7), the KEYNOTE-B96 description and the ASCO 2026 Elahere platinum-sensitive result all come from secondary oncology-press reporting and are tagged [WEB ESTIMATE] throughout. The Elahere revenue anchor underlying B.3a carries the same caveat.
• The B.2 epidemiology anchor is a market-research figure, and two links in the chain it feeds have no source at all. The ~20,000 US and ~61,000 7MM incident-case figures come from a DelveInsight press release, not from a registry. B.3a states plainly which links break and produces no bottom-up point estimate from them.
• No quality-of-life or health-economic data exists for this agent in any source read. B.2 records the patient-journey row as unverified rather than inferring it from the two-day dosing schedule.