Abstract PO4-06-04: Understanding metastasis mixed-treatment responses through genomic analyses

Abstract Background Early-stage and metastatic breast cancer (MBC) exhibit genomic heterogeneity, with inter-patient variability between primary tumors, intra-patient variability of primary tumors versus metastasis, and even between different metastatic sites within the same patient. Response to therapy in MBC can also be heterogeneous, termed "mixed response", with different degrees of responsiveness to the same drug(s) across metastatic sites within the same patient. Whether this treatment response variability is influenced by factors such as intrinsic characteristics of metastatic lesions and/or the microenvironment is unknown. Through genomic analysis, we aim to identify genomic features that may explain mixed clinical responses across metastatic sites, which will also provide valuable insights into treatment sensitivity and resistance. Methods Eligible patients with MBC were identified from the UNC Rapid Autopsy Program (RAP) with primary tumors and multiple metastases (≥ 2 different sites) with RNA sequencing data. An independent radiologist retrospectively reviewed imaging studies to identify patients with mixed radiographic responses to a defined line of therapy. The response to treatment was categorized for each metastasis, defined as a complete or partial response, as stabilization ≥ 6 months. Tumor measurements were performed for each patient at different time points. We performed supervised learning with linear regression incorporating "RNA sequencing-based gene expression signatures" as a fixed effect and "patient" as a random effect. We performed two different analyses: (1) Using all tumors and (2) using Basal-like tumors only. We finally selected significant signatures with a p-value >0.05. Results Ten patients with mixed response were identified, and 33 metastases were used for the analysis; 3 had de novo stage IV disease, 5 had triple-negative (TNBC), 4 HER2+, and 4 hormone receptor+ (HR+/HER2-) disease. Of 33 metastases, 6 were brain, 8 liver, 7 lung, and 12 “other”. Regarding intrinsic subtype, 15 metastatic tumors were Basal-like, 3 HER2-Enriched (HER2E), 6 Luminal A, and 9 Luminal B. In this dataset, 6 metastatic lesions showed response and 27 non-response. When analyzing all 33 metastases, KRAS high expression, KRAS amplicon, 16q23 amplicon, and apocrine features correlated with response independently of metastatic site. Higher T reg and CDKN2A gene expression levels were associated with non-response in metastasis. Within Basal-like subtype metastases only (N=15), CD8 T cells, PI3K pathway, KRAS gene and amplicon, 16q23 amplicon, and Basal-like-associated 5q11 loss correlated with response. However, T reg cells, higher levels of CDKN2A gene, and stem-cell-like features were associated with treatment non-response. Conclusions By examining matched metastases with discordant responsiveness to treatment within individual patients, this genomic analysis provides valuable insight into treatment sensitivity and resistance. Higher T reg cells and CDKN2A gene expression values correlate with non-response, while the KRAS gene, KRAS amplicon, and the 16q23 amplicon were associated with response. These results suggest that the molecular tumor and microenvironment features of metastases may contribute to sensitivity and resistance to therapy in MBC. Further validation and exploration are needed in larger multi-metastatic cohorts to fully understand the clinical implications of these findings. Citation Format: Susana Garcia-Recio, Paola Zagami, Amy Wheless, Kerry Thomas, Lisa Carey, Charles M Perou. Understanding metastasis mixed-treatment responses through genomic analyses [abstract]. In: Proceedings of the 2023 San Antonio Breast Cancer Symposium; 2023 Dec 5-9; San Antonio, TX. Philadelphia (PA): AACR; Cancer Res 2024;84(9 Suppl):Abstract nr PO4-06-04.