Abstract 1190: Decoding the multiomic signatures of oral cancer progression.

Abstract Oral cavity squamous cell carcinoma (OCSCC) is frequently preceded by oral epithelial dysplasias (OEDs), yet current clinical tools provide limited ability to identify which OPMLs will progress. This study applied an integrated multiomics approach to identify molecular features of progression and define biomarkers that may improve risk assessment. We analyzed well-annotated flash-frozen oral tissues collected at surgery, with concurrent blood samples for germline DNA. Tissue samples were pathologically reviewed and microdissected to be labelled as “premalignant” or “malignant”. Short-read whole-genome sequencing (≥80× tumor, ≥30× germline) and RNA sequencing (≥200 million reads per library) were used to characterize somatic mutations, copy number alterations, and transcriptional differences. A subset of samples underwent long-read nanopore sequencing (20× coverage) to refine methylation patterns and their relationship to expression. Data were integrated using established pipelines, and candidate features were evaluated through machine learning-based classification. AI was used to enhance abstract clarity. A total of 305 patient samples were submitted of which 96% (138 OED, 154 OSCC) were successfully sequenced. Mutational landscapes were similar across premalignant and malignant groups, sharing alterations in TP53, CDKN2A, FAT1, NOTCH1, and CASP8. HLA-A emerged as the top differentially mutated gene, pointing to altered immune-related pathways. Differential expression analysis showed clear separation of sample types, with broad increases in expression in OSCC. TPM values from 961 genes were used to train a proof-of-concept random forest classifier after iterative feature elimination, yielding a model with 92% confidence and 96% AUC. The most informative features aligned with top differentially expressed genes, and pathway analysis indicated enrichment in signaling, adhesion, and structural organization. Differential methylation analysis (75 OED, 40 OSCC) showed higher CpG island methylation in OSCC samples. Top differentially methylated genes included ARHGEF17 and PHLDB1, which participate in cell cycle regulation and signaling, and MIR27B, previously described as a regulator of cellular growth and movement in oral epithelial models. Using one of the largest, most comprehensively profiled OED cohorts to date, this work shows that lesions harbor distinct genomic, epigenomic, and transcriptional changes detectable before malignant transformation. Convergent key markers identified support development and validation of clinically actionable tests to be applied to formalin-fixed paraffin embedded samples for early identification of higher-risk lesions. Ongoing efforts integrate these molecular signatures with longitudinal clinical outcomes and map them to biological pathways and molecular oral cancer subtypes to advance screening, early intervention, and individualized patient management. Citation Format: Maple Lei, Kelly Yi Ping Liu, Steven John Jones, Catherine FY Poh. Decoding the multiomic signatures of oral cancer progression abstract. In: Proceedings of the American Association for Cancer Research Annual Meeting 2026; Part 1 (Regular Abstracts); 2026 Apr 17-22; San Diego, CA. Philadelphia (PA): AACR; Cancer Res 2026;86(7 Suppl):Abstract nr 1190.