Incyte closed the one trial that could have told it whether an oral PD-L1 inhibitor works

What was tried

Incyte ran NCT05888844 as a Phase 2, open-label, randomised study of its oral small-molecule PD-L1 inhibitor INCB099280 in advanced cutaneous squamous cell carcinoma. Part 1 randomised participants with metastatic or locally advanced disease 1:1:1 to three dose levels twice daily for up to two years. Part 2 was to expand one selected dose.

Eligibility was narrow in the way that matters. Participants needed confirmed disease not amenable to curative surgery or radiotherapy, measurable disease by RECIST 1.1, and ECOG status 0 or 1. Prior anti-PD-1, anti-PD-L1 or anti-PD-L2 therapy was an exclusion, as was any prior immune modulator. This was a checkpoint-naive population in a tumour where checkpoint blockade is standard of care.

The primary outcomes were objective response rate, treatment-emergent adverse events, and events leading to dose modification or discontinuation. Secondary outcomes included disease control rate, duration of response, progression-free survival and overall survival.

The study started on 9 October 2023, enrolled 63 participants, and completed on 28 September 2026. The registry records it as terminated and states: "This was a strategic business decision. There were no safety concerns contributing to this decision." No results have been posted, and the record was last updated on 20 October 2026.

The biological hypothesis

CD274 encodes PD-L1. Blocking the PD-L1 and PD-1 interaction with antibodies is established therapy across many tumours, and in advanced cutaneous squamous cell carcinoma the anti-PD-1 antibody cemiplimab produced a pooled objective response rate of 46.1 percent by independent central review across 193 patients, with an estimated 87.8 percent of responders still in response at 12 months.

The small-molecule thesis was not that the pathway was wrong but that the modality could be better. Characterisation of the lead compound, INCB086550, showed selective blockade of the PD-L1 and PD-1 interaction, induction of PD-L1 dimerisation and internalisation, and stimulation-dependent cytokine production in primary human immune cells, plus tumour growth reduction in humanised mice. The claimed advantages over antibodies were deeper tumour tissue penetration, distinct pharmacology, and the access and convenience of oral dosing.

Open Targets scores the CD274 association with cutaneous squamous cell carcinoma at 0.116, built from a literature datatype score of 0.302 and a clinical datatype score of 0.175, and lists 13 distinct drugs and clinical candidates against the target.

What actually happened

This study posted no results, so what is known about INCB099280 comes from the pooled Phase 1 report published in 2026, covering 182 patients across advanced solid tumours.

The safety profile was manageable. Treatment-related adverse events occurred in 144 of 182 patients, 79.1 percent, and treatment-related grade 3 or higher events in 25 patients, 13.7 percent. Two dose-limiting toxicities occurred, grade 2 vomiting and grade 2 maculopapular rash. The most frequent immune-related events were skin reactions and hepatitis. Target engagement was demonstrated, including PD-L1 binding.

The efficacy was the problem. The objective response rate for INCB099280 was 8.8 percent, 16 of 182, with a 95 percent confidence interval of 5.1 to 13.9 percent. The siblings performed no better, at 10.9 and 8.7 percent. Most patients had received at least two prior lines of therapy and between 9.6 and 16.5 percent prior immunotherapy, so these were not populations designed to flatter a checkpoint agent. INCB086550 was discontinued separately after 14 patients, 10.1 percent, had nervous system events for which an immune-mediated cause could not be excluded.

Incyte announced the discontinuation of INCB099280 and INCB099318 on 30 July 2024 in a strategic pipeline review. This study nonetheless continued for a further two years before its record was closed in October 2026. A companion Claidex claim covers the Phase 1/2 axitinib combination, NCT05949632, terminated in June 2025.

openFDA returns no adverse event reports naming INCB099280, as expected for a compound without marketing authorisation, and ChEMBL holds no record under that name.

Failure mechanism, best guess

The registry states a strategic business decision and rules out safety. Classified on the record, this is strategic_reprioritization.

The more useful reading is that the decisive experiment was abandoned before it reported. Phase 1 all-comers response rates of 9 to 11 percent in heavily pre-treated mixed tumours do not establish that a modality has failed, because antibodies in the same setting would not look much better. Checkpoint-naive cutaneous squamous cell carcinoma is the opposite situation. The antibody response rate is high and well characterised, so a small molecule matching antibody pharmacology should have produced responses near 40 percent, and one that did not would have been clearly inferior. That is the comparison NCT05888844 was positioned to make, and the one never published.

How to prevent this next time

With no endpoint data from this trial, the tool the evidence supports is a precision calculation about what it could have concluded.

Illustrative: assume the 63 participants split evenly across three dose arms, roughly 21 each, and that one arm produced 6 responses. Assumptions are an even split, no expansion weighting, exact binomial intervals, and no adjustment for the disease mix. That arm would show 28.6 percent with a 95 percent confidence interval of 11.3 to 52.2 percent, an interval containing both a 15 percent null and the 46.1 percent cemiplimab benchmark. Pooled across all 63, 18 responses would give the same 28.6 percent with an interval of 17.9 to 41.3 percent, which would at least exclude the antibody benchmark.

Two design lessons follow. First, a three-arm dose-ranging Phase 2 on a 63-patient budget buys dose information at the cost of the efficacy answer, and in a tumour with a known benchmark that answer was the asset. Second, the decision to stop the oral PD-L1 program was announced in July 2024, yet this study ran until September 2026 and published nothing. A programme no longer funded should still be run to a reportable readout when that readout settles a modality question for the field.

The single highest leverage change would have been to run this study at one dose in a single larger cohort, powered against the published cemiplimab response rate, so that one trial could rule the oral modality in or out rather than leaving three underpowered dose arms unreported.

What this means for similar programs

Oral small molecules aimed at protein interfaces that antibodies already drug well carry a specific evidentiary burden. They must show not that they work but that they work as well, and that comparison requires a setting with a high, stable antibody benchmark. Testing them first in heavily pre-treated all-comers produces response rates that are uninformative in both directions.

CD274 now carries three documented programme failures in the Claidex graph and a Mechanism Risk Score of 62, in the orange band. Saturation contributes 14.09 of a possible 15 points across 14 distinct programmes, and the genetic deficit term contributes 13.26 on an association score of 0.116. The risk here is crowding and modality, not doubt that PD-L1 is a real target.

Open questions

What response rate did the 63 participants achieve, and was any dose active? Did the siblings show the same hepatitis and skin toxicity pattern in a checkpoint-naive population? Will Incyte post results for a study that completed in September 2026? Does oral PD-L1 inhibition retain a rationale where antibody access, not antibody efficacy, is the constraint?

Sources

1. • ClinicalTrials.gov, NCT05888844, A Phase 2 Study Evaluating INCB099280 in Participants With Advanced Squamous Cell Carcinoma. https://clinicaltrials.gov/study/NCT05888844 - Prenen H, Rottey S, Pinato DJ, et al. Phase I safety, efficacy, and biomarker response evaluations of three oral PD-L1 inhibitors: INCB086550, INCB099280, and INCB099318. J Immunother Cancer.

2. https://- Koblish HK, Wu L, Wang LS, et al. Characterization of INCB086550: a potent and novel small-molecule PD-L1 inhibitor. Cancer Discov. 2022, volume 12, issue 6, pages 1482-1499. https://- Rischin D, Khushalani NI, Schmults CD, et al. Integrated analysis of a phase 2 study of cemiplimab in advanced cutaneous squamous cell carcinoma: extended follow-up of outcomes and quality of life analysis. J Immunother Cancer. 2021, volume 9, issue 8, e002757. https://- Migden MR, Khushalani NI, Chang ALS, et al. Phase 2 study of cemiplimab in patients with metastatic cutaneous squamous cell carcinoma: primary analysis of fixed-dosing, long-term outcome of weight-based dosing. J Immunother Cancer. 2020, volume 8, issue 1, e000775. https://- Fierce Biotech. Incyte drops pair of PD-L1 drugs as part of 2nd pipeline clear-out this year. 31 July 2024. https://www.fiercebiotech.com/biotech/incyte-drops-pair-pd-l1-drugs-part-second-pipeline-clear-out-year - Open Targets Platform, target CD274 (ENSG00000120217), association with cutaneous squamous cell carcinoma (EFO_1001927), retrieved 23 October 2026. https://platform.opentargets.org/target/ENSG00000120217 - Claidex claim, INCB099280 with axitinib in advanced solid tumours, NCT05949632. https://app.claidex.com/claim/incb099280-cd274-oral-pd-l1-solid-tumors-phase1-strategic-discontinuation - openFDA drug adverse event endpoint, query for INCB099280, retrieved 23 October 2026. https://open.fda.gov/apis/drug/event/. Available from: https://clinicaltrials.gov/study/NCT05949632.