Adding a CD27 Agonist to Checkpoint Blockade Fails to Improve Response in Biliary Tract Cancer

Why This Study Matters

Immune checkpoint inhibitors targeting the PD-L1 axis are now standard first-line therapy for biliary tract cancer when paired with chemotherapy, but clinical benefit remains modest — as single agents, checkpoint inhibitors produce response rates below 10% in this disease. A prior randomized trial (ETCTN 10139) showed that adding the MEK inhibitor cobimetinib to atezolizumab significantly improved progression-free survival compared with atezolizumab alone in previously treated biliary tract cancer, but did not extend overall survival, leaving room to improve on that strategy.

This trial tested whether adding a CD27 costimulatory agonist antibody (varlilumab) could restore T-cell function that MEK inhibition can blunt, on the theory that combining immune agonism with MEK-driven tumor immunogenicity might finally translate into clinical benefit for a checkpoint-resistant tumor type.

Study Design

Patients with advanced, previously treated (second- or third-line), unresectable biliary tract cancer were randomized 1:1, stratified by tumor subtype (gallbladder cancer, intrahepatic cholangiocarcinoma, extrahepatic cholangiocarcinoma), across NCI Experimental Therapeutics Clinical Trials Network (ETCTN) sites in the United States. Prior anti-PD-(L)1 exposure was permitted, and 32% of enrolled patients were immunotherapy-experienced. No enrolled patients had deficient mismatch repair (dMMR) or microsatellite-instability-high (MSI-H) disease. Co-primary endpoints were objective response rate and progression-free survival; treatment-related change in CD8+ tumor-infiltrating lymphocytes was the primary correlative endpoint. Blinding was not specified in the source publication.

Patients in the doublet arm (Arm AV; n=28 randomized, 27 treated) received atezolizumab 840 mg IV on days 1 and 15, plus varlilumab (CDX-1127) 3 mg/kg IV on days 1 and 15, in 28-day cycles. Patients in the triplet arm (Arm CAV; n=29 randomized, 26 treated) received the same atezolizumab and varlilumab dose and schedule, plus cobimetinib 60 mg orally once daily on days 1–21 of each 28-day cycle (off days 22–28).

Patient Population

Of the 57 randomized patients, 67% had intrahepatic cholangiocarcinoma, 14% had extrahepatic cholangiocarcinoma, and 19% had gallbladder adenocarcinoma. A majority (58%) had received two or more prior lines of systemic therapy in the metastatic setting. Randomization produced some numerical imbalances between arms: the AV arm had a higher proportion of patients over age 70 (39% vs. 21% in CAV) and extrahepatic cholangiocarcinoma (21% vs. 7%), while the CAV arm had a higher proportion of patients who had progressed on prior PD-(L)1 therapy (38% vs. 25%) and a lower frequency of high tumor mutational burden (≥10 mut/Mb: 3% CAV vs. 14% AV).

Primary Endpoint Results

The single objective response in the trial was a partial response in an immunotherapy-naive patient in the doublet arm with intrahepatic cholangiocarcinoma carrying IDH1, BRAF, and TP53 mutations, lasting 3.68 months. At data lock (May 10, 2024), no patients remained on treatment; 85% had come off treatment for progressive disease. 36 of the 57 patients who initiated therapy (67.9%) had died at final analysis.

Subgroup Analyses

Patients with known RAS/RAF pathogenic mutations showed a PFS signal numerically favoring the MEK-inhibitor-containing triplet, though it did not reach statistical significance (HR 0.43; 95% CI, 0.15–1.22). Among patients with an elevated baseline CA19-9 who achieved a 30% or greater on-treatment decline, more CAV patients responded than AV patients (7 of 12, 58.3%, vs. 4 of 14, 28.6%), with a nonsignificant trend toward longer PFS (median 5 vs. 1.8 months; HR 0.53, 95% CI 0.23–1.21; P = 0.13). Tumor anatomic location, prior immunotherapy exposure, and tumor mutational burden (high vs. low) were not significantly associated with either PFS or OS in covariate-stratified analysis; the authors report that most covariate subgroups trended toward more favorable PFS with the triplet therapy, but caution that none reached significance given the trial's small sample size — a qualitative observation not accompanied by subgroup-specific numbers in the source.

Safety Profile

Both regimens were generally well tolerated, with no new or unexpected toxicity signals and no treatment-related deaths. Serious treatment-related adverse events occurred more often with the triplet: 38.5% with CAV versus 11.1% with AV. Three patients discontinued treatment because of treatment-related adverse events (two in CAV, one in AV) — one case of myasthenia gravis with myositis, and two cases of biliary obstruction/cholecystitis judged more likely related to the underlying cancer than to study treatment.

The dermatologic and gastrointestinal toxicity added by cobimetinib was consistent with its known adverse-event profile. The addition of the CD27 agonist did not appear to meaningfully increase immune-related adverse events, though immune versus non-immune events were not prospectively adjudicated in the safety data collection.

Interpretation and Broader Context

The authors frame this as a signal-seeking study that was negative for its primary, response-based activity endpoint, with the PFS and OS comparisons treated as descriptive and hypothesis-generating given the trial was not powered for those comparisons. Combined CD27 agonism did not translate the biological rationale into clinical benefit: despite confirming that the triplet regimen drove a significant increase in intratumoral CD8+ and CD4+ T-cell density after one treatment cycle compared with the doublet, that immunologic effect did not correspond to improved response, PFS, or OS.

These findings suggest that the addition of CD27 costimulation is not sufficient to overcome biliary tract cancer resistance to ICI-based regimens.
— Discussion, Heumann et al., Clinical Cancer Research, 2026

The authors also note operational limitations likely to have blunted any treatment effect: a substantial share of patients (34.6% CAV, 22.8% AV) discontinued after one or fewer treatment cycles, mostly because of disease progression or clinical deterioration in this aggressive, treatment-refractory population, and delays between baseline imaging, mandatory research biopsies, and treatment start may have further narrowed the window to observe benefit.

Limitations

This trial closed early after a preplanned interim analysis found neither arm cleared the objective-response continuation threshold, so the PFS and OS comparisons reported here are descriptive rather than confirmatory: the study was never powered to detect a difference in either endpoint, and the OS hazard ratio (0.96, P = 0.908) in particular should not be read as evidence of no difference. With only 57 patients randomized (29 CAV, 28 AV), subgroup analyses — including the RAS/RAF and CA19-9-decline signals favoring the triplet — are small and hypothesis-generating, not confirmatory, and the source publication does not report subgroup-specific patient counts for several of these comparisons. Blinding was not specified, which leaves open-label assessment bias as a possible contributor to investigator-assessed outcomes. Baseline imbalances between arms in age, tumor subtype, prior immunotherapy exposure, and tumor mutational burden were not adjusted for in the primary comparison. Finally, a substantial share of patients in both arms discontinued after one or fewer cycles, mostly for disease progression, which may have limited the window in which any treatment effect — including the confirmed increase in tumor-infiltrating T-cells with the triplet — could translate into measurable clinical benefit.