Abstract Breast cancer is considered to be a relatively immunologically “silent” tumor type. However, there is substantial heterogeneity in breast cancer, including differences in the immune microenvironment of tumors. For example, differences in the immune profile of primary and metastatic breast tumors have been observed. A deeper investigation of the immune profile of primary and metastatic breast tumors would enhance our understanding of late-stage breast cancer and provide insights that can guide the development of immunotherapies for primary or metastatic disease. In the current study, we explored differences in the immune microenvironment of primary and metastatic breast tumors in a clinico-genomic, real-world cohort (Tempus) of 420 subjects with breast cancer. Use of a real-world cohort for this study provided a diverse set of samples across subtypes, stages, and sites, and in particular a relatively large number of biopsies from metastatic sites as compared to other large breast cancer cohorts. Analyses were conducted using gene expression data generated by bulk RNA-sequencing of biopsies from lesions at primary (stage I-IV, N = 110) and metastatic (N = 310) sites, along with subject medical histories compiled from a variety of sources, including clinical records, pathology images, and lab reports. Differences in the immune microenvironment of assayed tumors were characterized by evaluating gene expression signatures representing immune-related processes as well as individual immune cell types, with emphasis on signatures known to be associated with response to checkpoint blockade, as well as differential expression analysis of individual genes. We observed higher expression of inflammation signatures in early stage and locally advanced primary tumors as compared to metastatic tumors (excluding lymph node biopsies), as well as enrichment of a number of immune cell types, including CD8+ T cells, regulatory T cells, B cells, and dendritic cells. We further examined individual metastatic sites (including lymph nodes, liver, brain, bone, and lung, among others), and noted variability in the immune contexture of metastatic tumors across sites. For example, we observed a trend to reduced inflammation and infiltration of certain cell types in brain and liver metastases as compared to primary tumors. Differential gene expression analysis of immune checkpoint molecules and other key immune-related genes was also conducted, revealing, for example, higher expression of IL-8 at certain metastatic sites; elevated IL-8 expression has been linked to reduced response to PD-1/PD-L1 agents and an immunosuppressive microenvironment. Finally, transcriptome-wide differential gene expression and gene set enrichment analyses were conducted using several gene-set and pathway databases (Gene Ontology, KEGG, Reactome, and Molecular Signatures Database hallmark collection). The results from these hypothesis-free analyses confirmed observations from the gene signature analyses described above; for example, inflammation-related pathways like the Hallmark “Interferon Gamma Response” and Gene Ontology “Regulation of Inflammatory Response” were enriched in primary tumors. Overall, the results of this study provide insights into the immune microenvironment of breast tumors and may be further used to inform the development of immune-oncology therapies in the space. Citation Format: Aparna Chhibber, Kim Van Naarden Braun, Celine Han, Han Chang, Mustimbo Roberts, Bin Li. Characterization of the immune landscape of primary vs metastatic breast cancer in a real-world data cohort [abstract]. In: Proceedings of the 2021 San Antonio Breast Cancer Symposium; 2021 Dec 7-10; San Antonio, TX. Philadelphia (PA): AACR; Cancer Res 2022;82(4 Suppl):Abstract nr PD6-05.