Abstract 2040: βAlt score provides independent and significant clinical value in multiple tumor types

Abstract We reported that a functional relationship between loss of TGFβ signaling and gain of error-prone alternative end-joining DNA repair in cancer cells is evident in RNA sequencing data as a negative correlation between the transcriptomic signatures of TGFβ targets and genes necessary for alternative end-joining DNA repair. A score, βAlt, was designed to represent this relationship, and was shown to predict response to genotoxic chemotherapy and radiotherapy in multiple cancers. High βAlt, indicative of loss of TGFβ signaling competency and gain of error-prone repair, is associated with better response to genotoxic therapy (DOI;10.1126/scitranslmed.abc4465). As previously reported (DOI: 10.1158/1078-0432.ccr-21-2846), we used quartiles to analyze human papilloma virus (HPV)-positive versus HPV-negative head and neck squamous cell carcinoma (HNSCC) using specimens from TCGA and the HPV-negative HNSCC GSE41613 data. The βAlt score predicted response in HNSCC TCGA specimens (p=0.0015) and HPV-negative HNSCC GSE41613 patients (p=0.0033). To demonstrate the independent and significant clinical value of βAlt score and to justify the potential clinical utility of βAlt, we optimized the cut-off of βAlt score. We defined the maximized overall survival difference between patient stratification (survminer package in R, version 0.4.9), and thereafter partitioned patients in each cohort of study into βAlt Low and βAlt High groups. We determined that the value of βAlt score was independent from other clinical factors revealed by multivariate Cox-proportional hazard regression analysis (survival package in R, version 3.3.1) in all cohorts: HPV-positive HNSCC TCGA, p=0.006, HR 5.586 (1.655, 18.851); HPV-negative HNSCC TCGA, p=0.029, HR 1.654 (1.052, 2.601); HPV-negative HNSCC GSE41613 patients, p=0.019, HR 2.068 (1.125, 3.803). The C-Index predictive power calculated using a bootstrapping strategy at an 80% sampling rate with 1, 000 iterations of βAlt combined with these clinical factors was significantly improved (p<2.22e-16). We also analyzed βAlt using the TCGA glioblastoma (GBM) cohort. High βAlt in GBM also predicts longer survival (p<0.0001), remains significant after adjusting for important clinical factors, i.e., stage, age, and gender (p<0.001), and significantly improves C-Index predictive power of overall survival (p<2.22e-16). These data underscore that loss of TGFβ signaling, which underlies the marked sensitivity of HPV-positive HNSCC to standard of care chemoradiation, is predictive in non-HPV associated cancers. These results provide robust support that βAlt is a high-quality biomarker for the precision management of cancer patients. Citation Format: Mary Helen Barcellos-Hoff, Hang Chang. βAlt score provides independent and significant clinical value in multiple tumor types [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2025; Part 1 (Regular Abstracts); 2025 Apr 25-30; Chicago, IL. Philadelphia (PA): AACR; Cancer Res 2025;85(8_Suppl_1):Abstract nr 2040.