Epigenetic Boost Fails to Awaken Immune Therapy in Bile Duct Cancer Trial

A closely watched clinical trial testing whether an epigenetic drug could prime bile duct cancer for immunotherapy has returned a sobering answer: the combination did not work, but the detailed immune data collected along the way may prove just as important as the disappointing response rates. The phase 2 study, led by researchers at the Sidney Kimmel Comprehensive Cancer Center at Johns Hopkins, combined entinostat, an oral histone deacetylase inhibitor, with the PD-1 blocker nivolumab in patients whose advanced cholangiocarcinoma had already resisted standard treatment. The findings, published in Epigenetics Communications, offer a window into why this aggressive malignancy so stubbornly evades one of modern oncology’s most powerful weapons.

Cholangiocarcinoma, the second most common primary liver malignancy, carries a high metastatic potential and a poor prognosis, and its global incidence is rising. The addition of PD-L1 or PD-1 inhibitors such as durvalumab and pembrolizumab to first-line gemcitabine and cisplatin chemotherapy has produced statistically significant but modest survival gains in large phase 3 trials, including TOPAZ-1 and KEYNOTE-966. Beyond the first-line setting, patients with actionable alterations such as IDH1 mutations, HER2 amplification, or FGFR2 fusions can receive targeted therapies, and those with microsatellite instability-high or mismatch repair deficient tumors can benefit from immunotherapy, while FOLFOX chemotherapy remains an option for many others. For most patients, however, the immunosuppressive tumor microenvironment of cholangiocarcinoma blunts the impact of checkpoint inhibitors, which is precisely the barrier the Johns Hopkins team hoped to dismantle with epigenetic reprogramming.

The rationale for the trial rested on a striking preclinical and clinical track record. Histone deacetylase inhibitors can alter the immunogenicity of tumors by inducing expression of tumor-associated antigens, increasing tumor cell MHC class II display, and reducing immunosuppressive regulatory T cells and myeloid-derived suppressor cells. The ENCORE-601 trial showed that entinostat plus pembrolizumab had clinical activity in melanoma patients who had already progressed on pembrolizumab alone, and the PEMDAC phase 2 study in metastatic uveal melanoma demonstrated disease stabilization in a substantial fraction of participants. In mouse models of pancreatic ductal adenocarcinoma, entinostat increased infiltration by cytotoxic CD8-positive T cells and converted immunosuppressive myeloid-derived suppressor cells into a non-functional phenotype, translating into improved survival when paired with checkpoint blockade. Those findings carried into the clinic, where the same entinostat and nivolumab regimen produced durable radiological responses in a subset of pancreatic cancer patients.

Encouraged by the pancreatic results, the investigators enrolled 13 patients with unresectable or metastatic cholangiocarcinoma between November 2017 and November 2020. The trial design was deliberately mechanistic: every participant first received entinostat at 5 milligrams orally once weekly for a 14-day lead-in period, allowing the researchers to isolate the epigenetic drug’s effects on circulating immune cells before nivolumab, given at 240 milligrams intravenously every two weeks, was added. After four months, responders could transition to a maintenance regimen of weekly entinostat with higher-dose nivolumab every four weeks. The primary endpoint was the objective response rate by RECIST 1.1 criteria, assessed under a Simon’s two-stage minimax design that would halt the study for futility if no responses emerged among the first 13 patients.

The futility boundary was triggered. Among the 13 enrolled participants, whose median age was 56.7 years and whose tumors were mostly intrahepatic, none achieved a complete or partial response. Two patients, or 15 percent, had stable disease as their best response, while the remaining eleven, 85 percent, experienced disease progression. Median progression-free survival was just 2.2 months and median overall survival was 6.4 months, figures consistent with historical controls in this heavily pre-treated population. Because no responses were observed, the study did not advance to its planned second stage of enrollment. The result stands in sharp contrast to the pancreatic cohort studied under the same protocol, raising a fundamental question: why did a strategy that remodeled the immune landscape in one tumor type fail so completely in another?

Safety data provided a partial answer about tolerability, though not about efficacy. Twelve of the 13 patients, 92 percent, experienced an adverse event related to at least one study drug, with entinostat implicated in 12 patients and nivolumab in eight. Grade 3 or higher treatment-related adverse events occurred in five patients, or 38 percent, most commonly decreased blood cell counts, and no deaths attributable to treatment were recorded. More than half of the participants, seven patients, required a dose reduction, six because of fatigue and one because of thrombocytopenia. The profile echoed what had been seen in other entinostat combinations: manageable but not trivial, particularly in a population already worn down by prior chemotherapy.

The most scientifically revealing component of the study came from its translational arm. Using high-dimensional mass cytometry, or CyTOF, the team analyzed cryopreserved peripheral blood immune cells collected from every patient at baseline and again after the two weeks of entinostat monotherapy. Separate myeloid- and lymphoid-oriented antibody panels, processed through a computational pipeline built on the diffcyt framework and FlowSOM unsupervised clustering, allowed an unusually granular view of how the epigenetic drug reshaped circulating immunity. What emerged was a mixed picture: entinostat did reprogram the immune system in cholangiocarcinoma patients, but not in the ways that had correlated with benefit in pancreatic cancer.

In the myeloid compartment, entinostat significantly reduced the abundance of plasmacytoid and conventional dendritic cells while classical monocytes trended upward, and the drug increased expression of the co-stimulatory molecule CD40 on monocytes and precursor-like dendritic cells, along with CD103 on the latter, changes consistent with enhanced antigen-presenting capacity. Yet the drug also raised arginase-1 expression on precursor-like dendritic cells, an enzyme associated with immunosuppression, and upregulated PD-L1 across multiple myeloid subpopulations, a possible sign of an interferon-gamma-driven T cell response being met with a compensatory checkpoint shield. In the lymphoid compartment, naive T cell fractions increased, HLA-DR rose across T cells, and CCR5 and OX40 appeared on memory and effector subsets, but the drug simultaneously elevated exhaustion markers including PD-1, CTLA-4, and TIM-3, increased CD25 on regulatory T cells, and significantly reduced circulating B cells and their CD40 expression.

The authors interpret these findings cautiously. The activation and exhaustion signals may reflect not functional immune remodeling but a broad re-expression of epistemically silenced genes, since histone deacetylases maintain transcriptional repression through chromatin remodeling and their inhibition can derepress previously silent immune-regulatory genes. Critically, the phenotypic changes seen in cholangiocarcinoma differed from the pancreatic experience, where entinostat expanded effector memory and central memory T cell populations and improved dendritic cell maturation. In this cohort, the immune-activating effects of entinostat appear to have been insufficient to overcome compensatory suppressive and exhaustion signals, even with anti-PD-1 therapy on board, suggesting that cholangiocarcinoma’s immune biology is genuinely distinct and more refractory to epigenetic reprogramming than that of other solid tumors.

The study’s limitations are acknowledged by its authors: the sample was small, the population heavily pre-treated, and the lack of paired tumor biopsies prevented multiplex immunohistochemistry or RNA sequencing that could have compared the tumor microenvironment directly across the pancreatic and biliary cohorts. The two-week lead-in of entinostat monotherapy, designed to isolate mechanistic effects, may also have allowed rapidly progressive disease to advance before checkpoint blockade began, a trade-off the authors flag for future trial designs in aggressive malignancies. Even so, the message is clear: epigenetic modulation alone is not enough to convert cholangiocarcinoma into a tumor that responds to immune checkpoint inhibition. Overcoming its uniquely immunosuppressive barriers, the team concludes, will require innovative combinations of novel immunomodulatory agents with precision-based strategies tailored to the individual immune landscape of each patient’s disease.

Subject of Research: Entinostat plus nivolumab immunotherapy for advanced cholangiocarcinoma

Article Title: Entinostat in combination with nivolumab for patients with advanced cholangiocarcinoma: a phase 2 clinical trial

Article References: Hallab, E., Durham, J. N., Walker, R., Cope, L., Charmsaz, S., Gross, N. E., Sidiropoulos, D. N., Hernandez, A., Leatherman, J., Mitchell, S., Yarchoan, M., Ho, W. J., Azad, N. S., & Baretti, M. (2025). Entinostat in combination with nivolumab for patients with advanced cholangiocarcinoma: a phase 2 clinical trial. Epigenetics Communications, 5(1), Article 6. https://doi.org/10.1186/s43682-025-00037-z

Image Credits: AI Generated

DOI: 10.1186/s43682-025-00037-z

Keywords: cholangiocarcinoma, entinostat, nivolumab, HDAC inhibitor, immune checkpoint inhibitors, epigenetics, phase 2 trial, CyTOF, tumor microenvironment, immunotherapy, biliary tract cancer, PD-1