Abstract 5344: Variant allele frequency machine learning model identifies unique TP53-mutant phenotypes with worse post-operative survival in pancreatic cancer.

Abstract Introduction: Even after neoadjuvant therapy (NAT) and resection for localized pancreatic ductal adenocarcinoma (PDAC), overall survival (OS) varies greatly. Clinical factors such as positive lymph nodes (LNs) can stratify risk, but greater precision is needed; comprehensive genomic profiling (CGP) can bridge this gap. We present a novel method of combining machine learning (ML) with variant allelic frequency (VAF) to identify drivers of OS. Methods: We identified all localized PDAC patients who completed NAT, resection, and had CGP data. Stratifying OS from surgery by median, an XGBoost ML framework was utilized to extract features of importance using clinicopathologic variables, as well as VAFs for any present pathogenic mutations (with VAF=0% for wildtype [wt]). Model performance was evaluated using area under the curve (AUC) and Shapley additive explanation plots (SHAP), with Kaplan-Meier curves for clinical validation. In patients with available whole transcriptome data, DESeq2 was utilized for differential expression profiling. Results: Among 110 patients with CGP data, 89 (81%) were KRAS mutated (G12D 33%, G12V 23%, and G12R 18%), and 67 (61%) had pathogenic TP53 mutation (mut). Initial models using known pathogenic mut VAFs identified TP53 as the highest impact feature. Addition of TP53 VAF, when combined with time-of-surgery clinical variables (age, comorbidity index, pathologic LN and T stage, lymphovascular or perineural invasion), improved prediction of OS (AUC = 0.81), vs. clinical variables alone (AUC = 0.73). SHAP analysis showed high TP53 VAF and LN+ status as the two highest contributing features for poor OS. When categorized by TP53 and LN status, TP53 mut/LN+ patients had significantly worse OS than all other groups (median 11.0 mo [95%CI 7.4-15.3] vs. 23.0 mo [20.4-32.1], p<0.0001); there was no difference in OS between TP53 wt/LN+, TP53 mut/LN-, and TP53 wt/LN-. When stratified into high vs. low VAF by median (6.5%), only high VAF patients had worse OS (10.7 mo [6.7-22.0]) compared to wt(25.9 mo [16.5-38.7], p=0.05.)78 of the 110 patients had bulk transcriptomic data available on the same specimens. TP53 mut tumors had significantly higher expression of LYPD2 (one of the human lymphocyte antigen-6 proteins associated with worse outcomes; log10 fold 4.4, p<1e-9), as well as keratin genes KRT13 (log10 fold 2.3, p <0.0001) and KRT15 (log10 fold 1.3, p <0.0001) - associated with basal subtypes and worse outcomes. Conclusion: VAF analysis from CGP can uncover novel predictive targets for post-surgical outcomes. TP53 mut tumors express higher levels of genes associated with worse prognosis (e.g. LYPD2, KRT genes). When present with LN+, TP53 mut confers worst OS and may be a clonally dependent process (based on VAF). Post surgical TP53 mut/LN+ cohorts should be stratified as high risk and be considered for adjuvant treatment and clinical trials. Citation Format: Imaad Said, Eugene Chen, Megan Zeller, Mohammed Aldakkak, Matthew Sochor, Bhabishya Neupane, Kshitij Gaur, Mandana Kamgar, Alexandria Phan, Janice Zhao, Samih Thalji, Beth Erickson, Christina Small-Tom, Callisia Clarke, Kathleen K. Christians, Nikki K. Lytle, Thomas McFall, William A. Hall, Anai N. Kothari, Douglas B. Evans, Yongwoo David Seo. Variant allele frequency machine learning model identifies unique TP53-mutant phenotypes with worse post-operative survival in pancreatic cancer [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 5344.