Abstract Novel biomarkers that can be utilized for clinical decision support to improve patient outcomes are critically needed for ovarian cancer. At present, there are no standard of care biomarkers to inform the first-line treatment regimen (neoadjuvant chemotherapy vs. upfront surgical debulking) best suited for each patient or which patients are at the highest risk of recurrence. To address this unmet need, we identified pre-treatment computed tomography (CT) image-based radiomic features predictive of outcomes and primary treatment response among 298 women diagnosed with serous ovarian cancer from 2008 to 2019. A decision tree analysis identified a single volumetric feature, region of interest (ROI) volume center of mass (CoM) in the X direction, as the most informative radiomic feature that stratified patients with upfront surgery into high- and low-risk groups for overall survival. The high-risk group consists of patients with higher values of the radiomic feature. In the training (N=91) and test (N=91) cohorts of women treated with upfront surgery, high-risk patients had worse survival compared to low-risk patients (HR=2.01, 95% CI=1.07, 3.77 and HR=2.23, 95% CI=1.19, 4.17, respectively). This radiomic feature was not associated with survival among women treated with neoadjuvant chemotherapy (N=116; HR=1.23, 95% CI=0.73, 2.07). To reveal potential underlying biology of this radiomic feature, we performed RNAseq gene expression profiling on 47 formalin-fixed paraffin-embedded tumor specimens and correlated gene expression with the radiomic feature using DESeq2 from RSEM estimates. Dichotomized analysis (high- vs. low-risk) yielded 7 significant genes and continuous analysis yielded 15 significant genes (adjusted p<0.05), with both approaches identifying AKT2 and PSMC4. Gene set enrichment analysis (GSEA) was used to identify enriched gene sets from continuous associations using the MSigDB Hallmarks pathways (adjusted p<0.05). GSEA identified the oxidative phosphorylation (OXPHOS) pathway as one of the gene sets most negatively associated with the predictive radiomic feature (Hallmark Normalized Enrichment Score=-2.28, adjusted p<0.001). We derived a PCA-based gene signature from significantly associated OXPHOS genes (p<0.01) resulting in a negative correlation with the radiomic feature (R=-0.65, p<0.001). Prior studies have shown that high OXPHOS ovarian tumors are associated with an increased response to conventional chemotherapy, suggesting that OXPHOS may be a key pathway for chemoresistance in ovarian cancer. In summary, we identified an OXPHOS-associated radiomic feature predictive of survival among women with serous ovarian cancer treated with upfront surgery. Further research is needed to elucidate the biologic and mechanistic underpinnings of the identified radiomic feature and to validate these findings in a larger cohort of women with ovarian cancer. Citation Format: Christelle Colin-Leitzinger, Jaileene Perez-Morales, Steven Eschrich, Jamie K. Teer, Sweta Sinha, Melissa J. McGettigan, Daniel K. Jeong, Olya Stringfield, Mahmoud A. Abdalah, Natarajan Raghunand, Robert J. Gillies, Jing-Yi Chern, Matthew Schabath, Lauren Cole Peres. Oxidative phosphorylation-associated radiomic feature and survival of women with serous ovarian cancer [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2023; Part 1 (Regular and Invited Abstracts); 2023 Apr 14-19; Orlando, FL. Philadelphia (PA): AACR; Cancer Res 2023;83(7_Suppl):Abstract nr 6485.