Impact of genomic and clinical factors on therapeutic response to arly‑phase epigenetic therapies: An international cohort from two major phase I units

Introduction: Epigenetic agents-including inhibitors of BET bromodomains, PRMT5, EZH2, and other chromatin regulators-have emerged as promising therapeutic strategies across solid tumors. However, clinical activity remains variable, and robust predictive biomarkers are lacking. We conducted a retrospective two-center analysis to characterize outcomes of patients treated with these compounds in early-phase trials and to evaluate the contribution of next-generation sequencing (NGS) profiles to interpatient variability. Methods: A total of 262 patients treated within phase I trials of epigenetic agents at two academic early-phase drug development units were included. Clinical data (demographics, tumor type, ECOG status, prior therapies) and NGS results (pathogenic variants, VUS, short variants) were harmonized across centers into a unified dataset. Associations with overall survival (OS) were assessed using two-sided univariable Cox proportional hazards models. Hazard ratios (HR), 95% confidence intervals (CI), proportional hazards assumptions, Akaike Information Criteria (AIC), and C-statistics were reported. Results: Median age was 59.9 years (IQR 51-69), and 55% were female. Most patients had metastatic disease (88.2%), ECOG 1 (59.2%), prior chemotherapy (78.5%), and prior immunotherapy (70.9%), with a median of 3 prior systemic lines (IQR 1-4). The median number of metastatic sites was 2 (IQR 1-3). Patients received diverse epigenetic compounds, most commonly BET inhibitors (46.6%), PRMT5 inhibitors (22.5%), HDAC inhibitors (14.1%), and EZH2 inhibitors (8.4%), across tumor types including gastrointestinal (21.4%), thoracic (19.8%), CNS (12.6%), and sarcoma/GIST (12.6%). Best overall response included 10% CR/PR (n=21), 39.7% SD (n=104), and 44.6% PD (n=117), reflecting modest clinical activity.In the clinical Cox model, worse OS was associated with higher metastatic burden (HR 1.19; 95% CI, 1.08-1.32; p<0.001), prior chemotherapy (HR 1.64; 95% CI, 1.17-2.31; p=0.004), and increasing prior lines (HR 1.05; 95% CI, 1.00-1.11; p=0.044). Age was also statistically significant (HR 0.99; 95% CI, 0.98-1.00; p=0.029). In the genomic Cox model, short EGFR variants were significantly associated with worse OS (HR 2.74; 95% CI, 1.36-5.52; p=0.005). Other alterations, including pathogenic EGFR mutations (HR 1.83; p=0.065), ATM (HR 3.00; p=0.066), and ASPM (HR 0.29; p=0.087) variants showed non-significant trends. Conclusions: Epigenetic therapies showed modest activity in this heterogeneous early-phase cohort. Clinical factors-but not individual genomic alterations-were modestly prognostic for survival. These findings underscore the need for integrated biomarker approaches beyond single-gene predictors to optimize patient selection in epigenetic drug development.