Vidutolimod in skin cancers: a TLR9 agonist whose response collapsed after checkpoint failure

What was tried

Regeneron ran a multicenter, open-label Phase 2 study of intratumoral vidutolimod (CMP-001) combined with intravenous cemiplimab in selected advanced or metastatic cancers (NCT04916002). The trial used confirmed objective response rate as its primary endpoint and split patients into cohorts by tumor type and prior therapy: cutaneous squamous cell carcinoma, Merkel cell carcinoma, and basal cell carcinoma, each separating patients with no prior systemic therapy from those who had progressed on prior PD-1 blockade. The study enrolled 77 patients and was terminated by the sponsor for a reason recorded as related to study drug supply.

Vidutolimod is a Toll-like receptor 9 (TLR9) agonist built as a virus-like particle that packages a CpG-A oligodeoxynucleotide inside a Qbeta bacteriophage capsid. Cemiplimab is an anti-PD-1 antibody. The combination was injected into tumor lesions to drive a local innate-immune response intended to convert checkpoint-resistant tumors into responders. Vidutolimod entered Regeneron through the 2022 acquisition of Checkmate Pharmaceuticals (Regeneron investor disclosures, 2022).

The biological hypothesis

TLR9 senses unmethylated CpG DNA and, when engaged on plasmacytoid dendritic cells, triggers type I interferon and a cascade of antigen presentation and T-cell priming. Delivering a CpG-A agonist directly into a tumor was meant to inflame the local microenvironment, recruit and activate dendritic cells, and generate a systemic antitumor T-cell response that anti-PD-1 could then sustain. Early data supported the premise. In anti-PD-1 refractory melanoma, intratumoral vidutolimod plus pembrolizumab produced durable responses in roughly 25 percent of patients, including regression of non-injected lesions, which is the signature of a systemic abscopal effect (Cancer Discov 2021, 10.1158/2159-8290.CD-21-0425).

The genetic anchor to the specific tumor types tested is weak. The Open Targets association score for TLR9 and Merkel cell carcinoma is 0.0032, literature-only, which is expected for an innate-immune target whose rationale is pharmacologic rather than tumor-genetic.

What actually happened

The posted results show a sharp split by prior checkpoint exposure. In treatment-naive cohorts, response was meaningful: cutaneous squamous cell carcinoma reached 44.0 percent and Merkel cell carcinoma reached 46.2 percent. In the cohorts of patients who had progressed on prior PD-1 therapy, response collapsed to 5.3 percent in cutaneous squamous cell carcinoma and 6.3 percent in Merkel cell carcinoma. The basal cell carcinoma cohort, enrolled without prior systemic therapy, reached 25.0 percent (NCT04916002 results).

The trial was then terminated for drug supply rather than for a formal efficacy or safety analysis. The FAERS footprint for vidutolimod is minimal and uninformative, with single-digit term counts (openFDA FAERS). The class context is unfavorable. The other late-stage intratumoral TLR9 agonist, tilsotolimod, failed its Phase 3 ILLUMINATE-301 study in anti-PD-1 refractory melanoma, with no improvement in response or overall survival despite a coherent mechanistic story in earlier work (J Clin Oncol 2025, 10.1200/JCO.24.00727; Cancer Discov 2021, 10.1158/2159-8290.CD-20-1546; Clin Cancer Res 2022, 10.1158/1078-0432.CCR-21-4486).

Failure mechanism, best guess

The recorded termination reason is operational, but the posted data carry the more important signal. In the setting the combination was designed to rescue, anti-PD-1 refractory disease, response rates of 5 to 6 percent indicate the local innate stimulus did not reliably restore systemic antitumor immunity once checkpoint resistance was established. The treatment-naive responses are encouraging but harder to attribute, because anti-PD-1 alone already produces substantial response in treatment-naive cutaneous squamous cell and Merkel cell carcinoma, so the cohorts that worked are also the cohorts where the TLR9 contribution is most confounded by the checkpoint backbone.

The pattern repeats the tilsotolimod experience. Intratumoral TLR9 agonism produces visible local inflammation and occasional abscopal responses in small uncontrolled series, then fails to convert that biology into a controlled benefit in the refractory population. The likely mechanistic reasons are that injected-lesion inflammation does not guarantee a durable systemic T-cell response, that checkpoint-refractory tumors carry resistance mechanisms downstream of priming, and that intratumoral delivery limits which patients and lesions can be treated. A drug-supply termination ended this particular trial, but the refractory-cohort numbers suggest the harder biological question was already trending negative.

How to prevent this next time

Two quantitative tools would have sharpened the decision. First, biomarker enrichment should have been built into the refractory cohorts rather than treating prior PD-1 progression as a single stratum. Measuring baseline intratumoral T-cell infiltrate, interferon signature, and plasmacytoid dendritic cell density would have defined who could plausibly respond and would have raised the expected response rate in an enriched population, improving the number-needed-to-treat. Second, a Bayesian predictive-probability framework with a futility boundary would have allowed early stopping of the refractory cohorts on their own data:

PP(success)=∫θ​P(final success∣θ,interim data)⋅π(θ∣interim data)dθ

With historical anti-PD-1 refractory response near 10 percent for novel combinations, a cohort observing 5 to 6 percent would drive the predictive probability of a meaningful effect below a pre-set futility threshold well before full enrollment. A historical base-rate adjustment also matters for the treatment-naive cohorts, where the checkpoint backbone alone explains much of the observed response, so the design should have included a contribution-of-components arm or a randomized cemiplimab-alone reference to isolate the TLR9 effect.

The single highest leverage change would have been to enrich the refractory cohorts on interferon and T-cell infiltration biomarkers and run a randomized cemiplimab-alone reference to isolate the TLR9 contribution.

What this means for similar programs

For intratumoral innate-immune agonists, local pharmacodynamic activity is not a sufficient endpoint, and uncontrolled treatment-naive response can flatter a drug whose checkpoint partner is doing the work. Programs in this class should design from the start to separate the agonist contribution from the backbone and to enrich the refractory population that defines the unmet need. The Claidex Mechanism Risk Score for TLR9 is 34 out of 100, in the yellow band, driven by a genetic-deficit component of 14.9 out of 15, which reflects that the rationale is pharmacologic and not anchored in tumor genetics.

Open questions

Does intratumoral TLR9 agonism add measurable benefit over an anti-PD-1 backbone in any biomarker-defined population, or is the observed activity largely the checkpoint antibody? Which baseline features, for example interferon signature or plasmacytoid dendritic cell content, predict abscopal response? Can systemic or improved-delivery TLR9 agonists overcome the access limits of intratumoral injection? And after the tilsotolimod Phase 3 failure and this termination, does the intratumoral TLR9 strategy justify further controlled trials in checkpoint-refractory skin cancers?

Sources

1. • ClinicalTrials.gov, NCT04916002: cohort design, enrollment, objective response rate by cohort, termination reason. https://clinicaltrials.gov/study/NCT04916002 - Cancer Discov 2021, overcoming PD-1 blockade resistance with the CpG-A TLR9 agonist vidutolimod in metastatic melanoma. https://- J Clin Oncol 2025, randomized Phase 3 ILLUMINATE-301 study of tilsotolimod plus ipilimumab in anti-PD-1 refractory melanoma. https://- Cancer Discov 2021, tilsotolimod with ipilimumab tumor responses and mechanism. https://- Clin Cancer Res 2022, tilsotolimod exploits the TLR9 pathway to promote antigen presentation. https://- Open Targets Platform, TLR9 (ENSG00000239732) to Merkel cell carcinoma (EFO_1001471) association score 0.0032. - openFDA FAERS, adverse-event term counts for vidutolimod. - Regeneron investor disclosures, 2022, acquisition of Checkmate Pharmaceuticals and vidutolimod.