Abstract Background: Metastatic breast cancer (MBC) patients often have short survival, and successful treatment represents one of the most challenging aspects of cancer care. This poor prognosis is likely multifactorial, including increased clonal heterogeneity, drug resistance mechanisms, and alterations of the tumor microenvironment. Methods: The primary data source was multi-platform data coming from the AURORA US Project that includes RNAseq, DNAseq, and DNA methylation arrays, which assayed 55 MBC patients representing 51 primary cancers and 102 linked metastatic specimens. In addition, RNA sequencing data from two other datasets of primary tumor-metastasis pairs were also used (i.e. UNC Tumor Donation Program/RAP dataset (24 primary and 74 metastasis specimens), and the GEICAM/2009-03 ConvertHER trial dataset (102 primaries and 102 metastatic pairs)). In total, this combined RNAseq dataset contained 177 primary tumors and 278 metastases, including 28 liver, 18 lung, 12 brain, and 24 lymph node metastases. We used these data to address two pressing questions, namely: 1) do gene expression features vary in primary tumors vs metastases according to PAM50 expression subtype, and 2) do gene expression features vary according to site of metastasis. Results: Using the AURORA multi-platform data, we determined that 17% of metastatic tumors (mainly TNBC/Basal-like) showed reduced expression of HLA-A that was associated with DNA methylation and/or focal DNA deletions near the HLA-A locus; these methylated tumors also showed concomitant lower immune cell infiltrates. Reduced expression of HLA-A gene and immune cell infiltrates were also validated at the RNA level in RAP dataset. Next, RNA expression differences were examined using the combined data set and varied according to tumor subtype. ER+/Luminal metastases had lower fibroblast and endothelial cell content, while triple negative (TNBC)/Basal-like metastases showed a dramatic decrease in T cell and B cell signatures/features. Comparative analyses between primary and site-specific metastasis (i.e., primary vs liver metastasis) or between sites of metastases (i.e., liver vs lung metastasis) revealed that both liver and brain, on average, had low immune cell features regardless of the primary tumor phenotype. Even within the same patient, we detected low immune cell features in brain and liver metastases compared to lung and lymph node metastases. Lastly, liver metastases showed a gain of Luminal B/HER2E gene expression features and MYC targets, and brain TNBC metastasis showed a gain of cell differentiation/Luminal-related gene signatures. Conclusions: These findings could have direct implications for the treatment of MBC patients with immune-based therapies and suggest new therapeutic avenues depending upon the tumor metastasis phenotype, and site of metastasis. Citation Format: Susana Garcia Recio, Toshinori Hinoue, Gregory L. Wheeler, Benjamin J. Kelly, Justin M. Balko, Katherine A. Hoadley, Peter W. Laird, Elaine R. Mardis, Charles M. Perou. Multiplatform analysis of matched primary and metastatic breast tumors from the AURORA US Network identifies microenvironment features as drivers of metastasis [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2022; 2022 Apr 8-13. Philadelphia (PA): AACR; Cancer Res 2022;82(12_Suppl):Abstract nr LB176.