Abstract 2489: A cell-free DNA methylation biomarker outperforms irRECIST in predicting treatment outcome

Abstract Immune checkpoint inhibitors (ICI) are established as first-line treatment for inoperable esophageal adenocarcinomas (EAC), in combination with chemotherapy (CTX). Unfortunately, response to immunochemotherapy is highly variable, despite most EAC cases featuring a high mutation burden. EAC response to ICI is assessed using radiological criteria such as irRECIST, but its ability to predict long-term survival is poor. Consequently, early recognition of a failing therapy is challenging. More accurate early biomarkers of response to ICI are therefore of critical importance. Here, we show that a classifier based on epigenetic features in cfDNA taken 4 weeks after treatment initiation accurately predicts treatment benefit after one year. Strikingly, this blood-based biomarker predicts OS better than radiological criteria. We collected cell free DNA (cfDNA) from patients enrolled in the LUD2015-005 trial, in which ICI was administered for four weeks before adding standard-of-care chemotherapy (ICI+CTX). For each patient, we collected cfDNA, tumor and adjacent normal tissue at diagnosis, before ICI+CTX, during and at the end of treatment, in addition to radiological clinical imaging. Tissue and cfDNA samples were sequenced using TET-Assisted Pyridine-borane Sequencing (TAPS), a method recently developed in Oxford which provides high-depth, base-resolution DNA methylation information from low-input samples. Based on tissue data, we identified a genome-wide hypomethylation signature specific to tumor cells. By accounting for sample-to-sample variability in stromal composition, we show that this signature is present in most EACs, allowing us to model the admixture of tumor and normal methylation patterns to quantify the amount of tumor methylation detected (TMD) in cfDNA. 16 patients had increased TMD (TMD-up) after 4 weeks of ICI monotherapy compared to pre-treatment levels, while 15 patients had decreased TMD (TMD-dn). Median OS for the two groups were 37 and 112 weeks, respectively (p 0. 0001). We evaluated the radiological response according to irRECIST at similar timepoints. 17 of 32 patients (53. 1%) had no definitive evidence of progression nor response and were deemed to have radiologically stable disease. 8 of these patients went on to develop progressive disease within 1 year - all 8 patients belonged to the TMD-up group. Contrastingly, 8 out of 9 patients who had PFI greater than 1 year belonged to TMD-dn. In all 32 patients, TMD predicted radiological disease progression within 1 year with 83. 3% sensitivity and 92. 3% specificity. Taken together, we report the discovery of a circulatory biomarker that measures molecular response to ICI in EAC and identifies poor prognosis patients that appear stable from radiological imaging. Furthermore, patients with favorable molecular response were found to have a median OS almost 3 times that of patients with unfavorable molecular response. Citation Format: Phil Xie, Masato Innoue, Paulina Siejka-Zielińska, Robert Amess, Jaeho Chang, Sam Wilding, Gareth Griffiths, Mark R. Middleton, Richard P. Owen, Chunxiao Song, Xin Lu, Benjamin Schuster-Boeckler. A cell-free DNA methylation biomarker outperforms irRECIST in predicting treatment outcome abstract. In: Proceedings of the American Association for Cancer Research Annual Meeting 2024; Part 1 (Regular Abstracts) ; 2024 Apr 5-10; San Diego, CA. Philadelphia (PA): AACR; Cancer Res 2024;84 (6Suppl): Abstract nr 2489.