Evaluation of EpiSwitch in predicting immunotherapy response in hepatocellular carcinoma and gastrointestinal tumors.

Abstract

623 Background: The EpiSwitch Checkpoint Inhibitor Response Test (CiRT) is a blood-based assay that analyzes DNA conformations in immune cells to predict responses to immune checkpoint inhibitors (ICIs) targeting PD-L1/PD-1. CiRT has demonstrated superior predictive accuracy compared to traditional biomarkers, such as tumor mutational burden (TMB) and PD-L1 immunohistochemistry (IHC), for predicting responses in urothelial cancer (ESMO 2022). However, its role in hepatocellular carcinoma (HCC) and other gastrointestinal (GI) tumors remains unclear. As immunotherapy becomes more common in these cancers, identifying reliable predictive biomarkers is essential to optimize treatment outcomes. Methods: This retrospective study evaluates CiRT-predicted immunotherapy responses and clinical outcomes in patients with HCC and GI tumors, including cholangiocarcinoma, pancreatic adenocarcinoma, and gastric cancer. All patients received one or more lines of immunotherapy. CiRT responses were categorized into high probability (HP) and low probability (LP). Treatment response was assessed per RECIST 1.1 criteria. Sensitivity, specificity, positive predictive value (PPV), negative predictive value (NPV), overall survival (OS), and progression-free survival (PFS) were analyzed. Results: 43 patients (24 HP, 19 LP) were included. HCC patients (n = 33) received varied immunotherapy regimens, such as combinations of bevacizumab with atezolizumab, tremelimumab with durvalumab, and nivolumab with ipilimumab. Non-HCC patients (n = 10) received combinations of immunotherapy and chemotherapy such as durvalumab with cisplatin and gemcitabine and nivolumab with oxaliplatin and capecitabine. Across all patients, CiRT revealed a sensitivity of 73.91%, a specificity of 65%, a PPV of 70.83%, and an NPV of 68.42%. In HCC, sensitivity was 70.59% and specificity was 68.75%. In non-HCC, sensitivity was 83.33% and specificity 50%. HP responders showed better treatment outcomes compared to LP responders, with higher rates of complete response (CR: 12.5% vs 0%), partial response (PR: 29.2% vs 21.1%), and stable disease (SD: 29.2% vs 10.5%), and a lower rate of progressive disease (PD: 29.2% vs 68.4%) (p = 0.0467). Overall, 70.83% of HP achieved CR, PR, or SD vs. 31.58% of LP (p = 0.0098). PFS was significantly better in HP (p = 0.044, Log-rank test; median PFS = 2.0 months for the LP group, mPFS not reached for the HP group). No statistically significant OS difference was noted due to the lack of events in the HP group. Conclusions: CiRT HP status was associated with improved response rates to immunotherapy, longer PFS, and better treatment outcomes in HCC and GI tumors. HP responders experienced notably better outcomes, suggesting that CiRT could serve as a promising predictive biomarker for immunotherapy response, addressing a critical unmet need for reliable biomarkers to treat HCC and other GI tumors.