Abstract P4-05-02: Machine learning predicts rapid relapse in triple negative breast cancer

Abstract

Abstract Purpose: Cancers with short interval between diagnosis and metastasis are associated with aggressive clinical course. Specifically, metastatic relapse of triple-negative breast cancer (TNBC) within 2 years of initial primary diagnosis is associated with marked chemoresistance, rapid progression, and poor prognosis. We hypothesized that rapid relapse TNBCs (rrTNBC; distant metastatic relapse or death <2 years) reflect distinct clinical and genomic features vs. late relapse (lrTNBC; >2 years) or no relapse (nrTNBC; no distant relapse or death with at least 5 years follow-up). Patients and Methods: We identified 453 primary TNBCs from three publicly-available datasets and characterized each as rrTNBC, lrTNBC, or nrTNBC. We compiled primary tumor clinical and multi-omic data, including transcriptome (n=453), copy number alterations (CNAs; n=317), and mutations in 171 cancer-related genes (n=317), then calculated expression and immune signatures. Results: Patients with rrTNBC were higher stage at diagnosis (Chi-square p<0.0001) while lrTNBC were more likely to be non-basal PAM50 subtype (Chi-square p=0.03). Among 125 expression signatures, rrTNBC and lrTNBC had significantly lower immune signatures relative to nrTNBC suggesting an immune suppressed microenvironment. lrTNBCs were enriched for eight estrogen/luminal signatures (all FDR p<0.05). There was no significant difference in tumor mutation burden or percent genome altered across the groups. Among mutations, only TP53 mutations were significantly more frequent in rrTNBC compared to lrTNBC (Fisher exact FDR p=0.009). To develop an optimal classifier, we used 77 significant clinical and ‘omic features for training (n=214 patients) using six modeling approaches encompassing simple, machine learning, and artificial neural network (ANN), then evaluated performance to predict rrTNBc vs. lrTNBC vs. nrTNBC in validation cohort (n=90 patients) and independent testing cohort (n=81 patients). Among modeling approaches, support vector machine had the highest average receiver-operator characteristic area under curve (AUC) in both validation (AUC=0.79) and independent testing (AUC=0.72) cohorts. Conclusions: We provide a new approach to define TNBCs based on timing of relapse. We identify distinct clinical and genomic features that can be incorporated into machine learning models to predict rrTNBC. Citation Format: Yiqing Zhang, William Nock, Meghan Wyse, Zachary Weber, Elizabeth J Adams, Asad Sarah, Sinclair Stockard, David Tallman, Jasneet Singh, Junu Bae, Eric P Winer, Nancy U Lin, Yi-Zhou Jiang, Ding Ma, Peng Wang, Leming Shi, Wei Huang, Zhi-Ming Shao, Claire Verschraegen, Mathew Cherian, Maryam B Lustberg, Bhuvana Ramaswamy, Sagar Sardesai, Jeffrey VanDeusen, Nicole Williams, Wesolowski Robert, Daniel G Stover. Machine learning predicts rapid relapse in triple negative breast cancer [abstract]. In: Proceedings of the 2019 San Antonio Breast Cancer Symposium; 2019 Dec 10-14; San Antonio, TX. Philadelphia (PA): AACR; Cancer Res 2020;80(4 Suppl):Abstract nr P4-05-02.