Abstract 1892: Proteomic profiling identifies and validates a prognostic survival biomarker and pathways associated with novel subtypes of esophogeal adenocarcinoma

Abstract Introduction: Esophageal adenocarcinoma (EAC) is an aggressive malignancy with poor survival rates. Current management relies on tumor stage and grade, with neoadjuvant therapy (NT) showing improved outcomes over surgery alone. However, robust biomarkers are still required to improve staging and patient selection for therapy, including NT. Methods: A training cohort of fresh frozen samples from 64 patients with EAC, not treated with NT, was analyzed by data-independent acquisition mass spectrometry. Formalin-fixed paraffin-embedded samples from an independent cohort of 188 patients with EAC treated with NT were used as validation. Unsupervised consensus clustering (CC) was performed followed by differential abundance (DA) analyses (DAA) and pathway enrichment analyses (PEA) to identify the DA proteins and enriched pathways in each cluster. Survival analyses using Cox regression were performed to assess the prognostic power of the proteomic clusters for overall survival (OS). Finally, a prognostic signature for OS was built on the training cohort using multivariable Cox regression analyses with LASSO regularization followed by StepAIC to select the most prognostic proteomic combination. The predictive power of the proteomic signature was tested using the Area Under the Receiver Operating Characteristic (AUROC) analysis and via inclusion in a nomogram with other prognostic variables. Results: Proteomic analyses identified 8,730 and 5,214 proteins in the training and validation cohorts, respectively. The CC analyses revealed five clusters in the training cohort. Clusters 2 and 3 showed shorter OS compared to the other clusters (p=0.0025). Independent DAA and PEA revealed upregulation in extracellular matrix-related pathways in the poor prognostic clusters within both cohorts. A five-protein signature was developed in the training cohort and was associated with OS (p<0.001). This signature was independent of other prognostic variables, including N stage, T stage, grade, and age. The five-protein signature was also associated with higher odds of having positive nodes at the time of resection (Odd Ratio 6.3, p=0.009). The proteomic signature demonstrated an AUROC of 0.81 and was the most prognostic variable in a nomogram for survival risk. Finally, the proteomic signature maintained its prognostic power in the validation cohort (p=0.016) and was independent of other clinical variables, showing the highest prognostic value in the nomogram. Conclusion: Using two independent cohorts of EAC surgical samples, we were able to identify a prognostic proteomic-based classification and develop a proteomic-based signature that was prognostic for OS and independent of other clinical variables. This signature may be investigated further as a criterion to tailor treatment and select patients who can benefit from treatment de-escalation. Citation Format: AuAdel T. Aref, Daniel Bucio-Noble, Ahilya Singh, Jerresa Jabson, Myra Cheung, George Craft, Daniella Jaber, Rebecca Poulos, Erin K. Sykes, Jennifer M. Koh, Erin M. Humphries, Dylan Xavier, Peter G. Hains,V Phillip J. Robinson, Thejaani Udumanne, Sarah J. Lord, Yi Jin Liew, Jason Ross, Dan Falkenback, Jan Johansson, Duncan McLeod, David Whiteman, Reginald V. Lord, Roger R. Reddelthors. Proteomic profiling identifies and validates a prognostic survival biomarker and pathways associated with novel subtypes of esophogeal adenocarcinoma [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 1892.