Abstract 7558: Novel liquid biopsy based determination of lung cancer histology using a comprehensive epigenomic platform reveals enhancer activity as a key determinant for accurate classification

Abstract BACKGROUND: Accurate molecular characterization of lung cancer including histology classification and therapeutic target profiling is critical for treatment planning and resistance monitoring. For example, detecting lineage transformation from lung adenocarcinoma (LUAD) to small cell lung carcinoma (SCLC) following treatment, an established resistance mechanism, is challenging due to deteriorating patient health and complicated invasive procedures. Here, we describe an accurate lung cancer histological classification using an innovative, multimodal epigenomic liquid biopsy method that comprehensively profiles tumor-specific transcription activation from 1 mL of patient plasma. METHODS: 37 LUAD, 34 SCLC patient plasma samples collected from commercial biobanks, and reference cell lines were profiled for enhancer, promoter and DNA methylation activity to identify epigenomic loci characteristic of LUAD and SCLC. The resulting map of select quantified loci was used as input to a machine learning algorithm that distinguishes SCLC from LUAD, in a cross validated training and testing approach. RESULTS: The epigenomic maps derived from lung cancer patient plasma were highly informative of the regulatory programming of the tumor cells. The enhancers and promoters most strongly linked to lung cancer profiles across the patient population include key, histology-specific driver genes such as NKX2-1, FOXA1, FOXA2, ONECUT2, DLL3, and ASCL1. The resulting classifier was able to accurately predict histology with a cross validated AUC greater than 0.94. In silico simulation results suggest that accurate classification can be achieved with ctDNA content below 1%. Further correlative examination of the SCLC enhancer landscape revealed a partitioning of SCLC patients into molecular subtypes characterized by active enhancer marks at subtype-specific transcripts, including ASCL1, POU2F3, NEUROD1, and YAP1. The transcriptional regulation of ADC targets is apparent through comprehensive epigenomic profiling of plasma in both SCLC and LUAD, with the expression of Claudin 18.2, SEZ6, and others showing a strong dynamic range independent of ctDNA levels. This represents a novel liquid biopsy approach to distinguish LUAD from SCLC and provide simultaneous insight into expression of key targets at clinically relevant ctDNA levels. CONCLUSION: These data demonstrate the feasibility of epigenomic mapping from 1mL of plasma in lung cancer patients to distinguish lung cancer histology. The resulting data is biologically interpretable through the lens of transcriptional activation of key histology specific genes. This liquid-biopsy based approach may provide a non-invasive alternative to invasive and challenging tissue biopsies for biomarker investigation and therapeutic decision making. Citation Format: Jamey Guess, Aparna Gorthi, Anthony D'ippolity, Jonathan Beagan, Travis Clark, Michael Coyne, Tyrone Tamakloe, Baovy Tran, Hyun-Hwan Jeong, Kristian Cibulskis, Hathairat Sawaengsri, Juliann Chmielecki, Corrie A. Painter, J Carl Barrett, Matthew Eaton. Novel liquid biopsy based determination of lung cancer histology using a comprehensive epigenomic platform reveals enhancer activity as a key determinant for accurate classification [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(6_Suppl):Abstract nr 7558.