Abstract 4145: Genome-wide metabolic modeling identifies key modifiers of precancerous LUSC evolution

Abstract Background: Metabolic alterations drive tumorigenesis, yet few metabolic cancer drivers have been established. The squamous cell carcinoma subtype of lung cancer (LUSC) is particularly underserved, with few, potentially harmful treatment options and overall poor prognosis. Here, we aim to discover actionable modifiers of tumorigenesis through a systematic, genome-scale metabolic analysis of human lung samples. We identify metabolic genes whose knockout (KO) may drive carcinogenesis and, critically, those whose KO may revert (pre)cancerous states back to the healthy lung. Methods: We analyzed bulk transcriptomes from 122 normal, precancerous, and cancerous samples spanning the LUSC evolutionary spectrum. First, sample-specific metabolic states were modeled in-silico using iMAT (Zur et al., 2010) on the HumanGEM model. Second, gene KOs were simulated with rMTA (Valcárcel et al., 2019) to estimate their oncogenic risk and back-to-normal reversion potential. Third, top scoring genes in both categories were corroborated by independent and orthogonal validations: a mouse gene-null phenotype dataset (KnockOut Mouse Project), differential expression in a separate LUSC precancer progressors vs regressors cohort, the Dependency Map (DepMap), and an LLM/AI-enhanced literature review. Corroborated, top-scoring genes were finally mapped to small molecule inhibitors via the DrugBank database. Results: Reassuringly, top predicted risk genes (KO → progression) emerging from the metabolic modeling analysis were associated with greater cancer incidence in null allele mice (Wilcoxon p = 0.02), downregulated in LUSC precancer lesions that progressed to cancer (p 0.0094), and associated with tumor suppressive activity in literature (p 0.01), compared to background genes. Conversely, top reversion genes (KO → regression) were downregulated in LUSC precancer samples that regressed to normal (p 0.04), overlapped with genes whose KOs lowered cancerous growth in the DepMap (odds-ratio/OR 1.6, p 0.0083), and were associated with oncogenic activity in literature (p 0.01). Top validated reversion genes were enriched in glutamine and glucose usage pathways (OR 7, padj 0.09). Top risk-mitigating inhibitors included antiviral Lamivudine (CMPK1), antidepressant Phenelzine (AOC3, GPT1/2), and antihyperlipidemic Statins (ABCB1). Conclusions: Genome-scale metabolic modeling analysis of LUSC samples uncovered several gene KOs and drugs that may revert precancerous metabolic states to non-cancerous ones. Although these genes and drugs warrant further experimental testing, top candidates were prioritized through and supported by multiple independent computational analyses. This work paves the way for a new class of treatments that may bring tumors closer to a non-cancerous homeostasis instead of the standard-of-care cell killing that too often leaves resistant residual disease. Citation Format: Neel Sanghvi, Thomas Cantore, Chi-Ping Day, Sanna Madan, Nishanth Ulhas Nair, Eytan Ruppin, . Genome-wide metabolic modeling identifies key modifiers of precancerous LUSC evolution 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 4145.