Abstract A034: Development of a novel class of genomic biomarkers through mathematical modeling of cancer genome contiguity

Abstract

Abstract Introduction: Chromosomal instability (CIN) is a hallmark of cancer aggressiveness. However, in lung adenocarcinoma (LUAD) the most common methods for measuring genomic instability, such as the weighted Genome Instability Index (wGII), are not reliable predictors of patient outcomes. The wGII measures the portion of the genome altered by copy number variation (CNV). The wGII quantifies aneuploidy but ignores other properties of CNVs such as their length distribution. We hypothesized that quantification of genome contiguity may overcome the limitations of wGII and help identify novel prognostic biomarkers. Method: We develop a novel measure of CIN based on mathematical modeling of genome breakage patterns. Utilizing segmented CNV profiles we infer two properties of the genome breakage process – its intensity and dispersion. Intensity is proportional to the number of genome breakpoints and is estimated by the total number of CNV segments. Dispersion quantifies spatial dependence of breakpoints over a wide range of genomic scales. It is computed by Ripley’s K-function which compares the distribution of breakpoints with a random Poisson point process. Jointly, intensity and dispersion, which we term the weighted Genome Contiguity Index (wGCI), estimate the contiguity of the genome. We contrast the wGCI with a heuristic score based on quantiles of segment lengths. Results: We sought to evaluate the association between the wGCI and patients’ survival in a proteogenomic cohort of 215 primary LUAD tumors profiled using whole-genome sequencing (WGS). In support of our hypothesis, we observed that patients had significantly better outcomes if their tumor genomes had longer segments compared to those with a high degree of fragmentation (P=0.014). Utilizing the wGCI, we found that tumor genomes with dispersed breakage (and low contiguity) were associated with worst outcomes (P=0.01). To identify mechanisms associated with dispersed breakage patterns we compared genomic features of the two groups. We found that LUAD tumors with highly fragmented genomes have a higher rate of TERT amplifications (P=0.005) and increased TERT RNA expression (P=2e-6). These tumors are also enriched for TP53 mutations (P=2e-16) and MYC amplifications (P=0.01). Given the cost and practical challenges of clinical WGS we leveraged mass-spectrometry proteomics to identify protein biomarkers. We found IGF2BP3 to be significantly upregulated (P=0.0006) in highly fragmented tumors and to be associated with worse outcomes (P=0.011). We use immunohistochemistry (IHC) to develop and validate IGF2BP3 as a prognostic biomarker in independent lung cancer cohorts. Conclusion: We introduced a novel quantitative measure of CIN that uniquely captures the impact of chromosome breakage on LUAD patients' clinical outcomes. Further, we identified upregulation of IGF2BP3 as a proxy to identify fragmented genomes using IHC. Our work underscores the value of mathematical modeling of genome breakage patterns to develop complex genomic biomarkers. Citation Format: Noshad Hosseini, Rahul Mannan, Alexey Nesvizhskii, Saravana Dhanasekaran, Marcin Cieslik. Development of a novel class of genomic biomarkers through mathematical modeling of cancer genome contiguity [abstract]. In: Proceedings of the AACR Special Conference in Cancer Research: DNA Damage Repair: From Basic Science to Future Clinical Application; 2024 Jan 9-11; Washington, DC. Philadelphia (PA): AACR; Cancer Res 2024;84(1 Suppl):Abstract nr A034.