innovaTV 205 closed by sponsor decision after five years, with the best-responding arm also the least tolerable

What was tried

innovaTV 205, also registered as ENGOT-cx8 and GOG-3024, was an open label phase 1b/2 trial of tisotumab vedotin as monotherapy and in combination with other agents in recurrent or stage IVB cervical cancer. Tisotumab vedotin is an antibody-drug conjugate that binds tissue factor, the protein encoded by F3, and delivers monomethyl auristatin E. Seagen, now a wholly owned subsidiary of Pfizer, was lead sponsor, with Genmab, Merck Sharp and Dohme, ENGOT and the GOG Foundation as collaborators.

The study opened on 27 February 2019 and enrolled 214 participants across eight non-randomized parallel arms. Escalation paired tisotumab vedotin with bevacizumab, pembrolizumab or carboplatin in previously treated patients. Expansion covered first-line tisotumab vedotin with carboplatin (arm D), first-line and later-line tisotumab vedotin with pembrolizumab (arms E and F), monotherapy (arm G), and a first-line triplet or quadruplet with carboplatin and pembrolizumab, with or without bevacizumab (arm H). Primary endpoints were dose-limiting toxicities in escalation and objective response rate in expansion. The trial completed on 19 March 2026 and the registry status changed to terminated on 22 July 2026, with the reason "Sponsor has decided to stop the trial based on strategic decisions."

The biological hypothesis

Tissue factor initiates the extrinsic coagulation cascade by complexing with factor VII or VIIa to activate factors IX and X. It is also expressed at high levels on many solid tumours, including cervical carcinoma, which makes it usable as a delivery address rather than as a driver to be inhibited. Tisotumab vedotin exploits the address: the antibody binds tissue factor, the conjugate internalizes, and monomethyl auristatin E kills the cell from inside.

That distinction matters for scoring the target. Open Targets gives F3 an association score of 0.556 against cervical cancer, built from clinical evidence at 0.907, literature at 0.168, and genetic association at zero. The score is largely a record of tisotumab vedotin itself having been approved here. No human genetic argument says tissue factor drives cervical cancer, and the mechanism does not require one.

The combination hypothesis was separate. Carboplatin was chosen for cytotoxic synergy, pembrolizumab because immunogenic cell death from auristatin payloads should improve checkpoint response, and bevacizumab because it is part of the first-line standard.

What actually happened

The trial worked, and it was closed anyway. Five-year results published in April 2026 report 139 participants in the four dose-expansion arms at a data cutoff of 15 October 2025. Confirmed objective response rates across arms D, E, F and H were 54.5, 40.6, 35.3 and 65.8 percent. Median duration of response was 8.6 months in arm D, not reached in arm E, 18.2 months in arm F and 13.3 months in arm H. Median progression-free survival was 6.9, 5.3, 5.6 and 10.6 months, and median overall survival was 25.5, 30.7, 15.3 and 28.0 months.

Toxicity tracked the same ordering. Grade 3 or higher adverse events related to any treatment component occurred in 72.7, 45.5, 48.6 and 86.8 percent of those arms, and adverse events leading to discontinuation of tisotumab vedotin in 24.2, 24.2, 34.3 and 55.3 percent. Comparing the two first-line arms, arm H delivered 11.3 percentage points more response than arm D while adding 14.1 points of grade 3 or higher events and 31.1 points of treatment discontinuation. The authors reported no new safety signals.

None of these arms advanced to a randomized combination trial. Monotherapy did, and innovaTV 301 established an overall survival benefit for tisotumab vedotin as second or third-line therapy in a randomized setting, published in 2024. That is the study that carried the asset.

Failure mechanism, best guess

This is a strategic reprioritization, and the registry reason states it plainly. The scientific content sits in the gap between a strong single-arm response rate and a registrable claim.

Three factors plausibly kept that gap open. First, the first-line comparator moved during the trial. A regimen entering development in 2019 faced a different standard of care than one reading out in 2025, and a single-arm 65.8 percent response rate does not settle whether the conjugate added anything over the chemotherapy, bevacizumab and pembrolizumab already in the arm. Second, discontinuing the investigational agent in 55.3 percent of participants makes an intention-to-treat efficacy claim structurally difficult to carry into a randomized first-line trial. Third, the asset already had a randomized win in later-line monotherapy, so a large first-line combination program had to be weighed against a path that was already delivering.

What Pfizer is still running supports that reading. The active company-sponsored tisotumab vedotin trial as of 2026 is a phase 4 single-arm study of ocular assessments in patients on the approved product, opened in May 2025. Investment moved from expanding the label toward managing the ocular toxicity that carries the boxed warning.

How to prevent this next time

The data do not support a Bayesian recalculation or a futility analysis, because there was no futility. The arms met their response endpoints. The question is why a program with positive single-arm data never converted, so the applicable levers are design levers rather than statistical ones.

Randomize the increment, not the regimen. Arm D and arm H differ by two agents and both are single-arm. A small randomized comparison of the conjugate added to a fixed backbone would have produced an interpretable increment at roughly the same patient cost.

Cost the tolerability up front. The 55.3 percent discontinuation rate in arm H was observable well before the five-year analysis. A pre-specified rule treating an investigational-agent discontinuation rate above a threshold as a program-stopping signal, on par with an efficacy futility boundary, would have redirected resources years earlier.

Fix the comparator to a moving target. Where the first-line standard is under active revision, an expansion cohort with no concurrent control is dated before it reads out.

The single highest leverage change would have been to randomize the first-line expansion arms against the contemporaneous standard of care backbone from the outset, rather than accumulating 139 participants of single-arm combination data that could not answer whether the conjugate contributed to the responses observed.

What this means for similar programs

F3 in cervical cancer enters the Claidex graph with a modelled risk score of 24 out of 100 in the green band, reflecting one documented phase 1/2 record, a strategic rather than mechanistic archetype, an Open Targets score of 0.556, and a single clinical program against the target. Nothing here says tissue factor is a bad address for a conjugate, and the approved monotherapy indication says the opposite.

The transferable lesson concerns combination expansion in antibody-drug conjugates. Auristatin payloads produce single-arm response rates that look like signal and often are, but the cost of stacking them onto full-dose chemotherapy and checkpoint blockade shows up in discontinuation rates rather than response rates.

Open questions

Would arm H have shown a real increment over its own backbone in a randomized comparison? The 65.8 percent response rate is uninterpretable without one, and none was run.

Did the 15.3 month median overall survival in arm F reflect the population or the regimen? Arm F enrolled previously treated patients and arm E did not, so the 15.4 month gap between them is confounded by line of therapy.

Sources

1. • ClinicalTrials.gov, NCT03786081, innovaTV 205 / ENGOT-cx8 / GOG-3024, full record, status terminated, last update posted 22 July 2026. https://clinicaltrials.gov/study/NCT03786081 - Van Nieuwenhuysen E, Vergote I, Randall LM, et al. Tisotumab vedotin plus carboplatin or pembrolizumab in recurrent or metastatic cervical cancer: 5-year results from the innovaTV 205/ENGOT-cx8/GOG-3024 study. Gynecol Oncol. 2026;207S:3-13. https://- Vergote I, et al. Tisotumab Vedotin as Second- or Third-Line Therapy for Recurrent Cervical Cancer. N Engl J Med. 2024;391(1):44-55. https://- Coleman RL, et al. Efficacy and safety of tisotumab vedotin in previously treated recurrent or metastatic cervical cancer (innovaTV 204). Lancet Oncol. 2021;22(5):609-619. https://- ClinicalTrials.gov, NCT06952660, phase 4 single-arm study of ocular assessments in patients treated with TIVDAK, recruiting, start date 7 May 2025. https://clinicaltrials.gov/study/NCT06952660 - Open Targets Platform, F3 (ENSG00000117525) and cervical cancer (MONDO_0002974), association score 0.556 with clinical evidence 0.907 and genetic association 0, accessed 27 July 2026. https://platform.opentargets.org/target/ENSG00000117525 - ChEMBL, TISOTUMAB VEDOTIN, CHEMBL4297841, antibody drug conjugate, maximum phase 4, first approval 2021, black box warning flagged. https://www.ebi.ac.uk/chembl/compound_report_card/CHEMBL4297841/ - openFDA drug adverse event and drug label APIs, tisotumab vedotin queries, accessed 27 July 2026. https://open.fda.gov/apis/drug/event/.