Biopsy site-specific variation in immune signatures identified by RNA profiling in breast cancer

Background: Breast cancer is generally considered to be immune cold; however, emerging data indicate heterogeneity in its immune landscape. Triple negative and HER2-positive tumors typically show higher immune cell infiltration than hormone receptor-positive (HR+) tumors. Additionally, the local microenvironment at the biopsy site may influence the immune profile of the tumor, contributing to distinct immunophenotypes. In this study, we used gene expression profiling to evaluate biopsy site-specific patterns of immune activity in breast cancer. Methods: Tumor tissue from 1,009 patients with breast cancer (all-comers) underwent targeted RNA profiling using a laboratory-developed test, FoundationOne ® RNA, as part of routine clinical care. Scores for immune and stromal gene sets were generated using a single-sample gene set enrichment analysis (ssGSEA) using a research use algorithm. K-means clustering of Z-normalized scores identified two immunophenotypes: immune-high and immune-low. Samples with a negative silhouette width were classified as unknown phenotype. Results: The most common sample sites were breast (n=363), liver (n=142), lymph node (n=120), bone (n=81), and lung (n=55). Immune ssGSEA scores varied significantly by sample site. Brain biopsies (n=23) exhibited the lowest immune scores, with only 5% (1/23) classified as immune-high, whereas lymph node and lung biopsies showed higher immune scores, with 42% (51/120) and 47% (26/55) classified as immune-high, respectively (p<0.05). Among local breast samples, 26% (96/363) were immune-high. Although liver metastases are typically considered to be immune cold, 12% (17/142) in this cohort were immune-high. Consistent with reports of lower immune activity in HR+ tumors, ESR1 expression was significantly higher in immune-low tumors compared to immune-high tumors (median expression 4,872 vs. 986 transcripts per million; p <10 -4 ). Furthermore, distinct immune cell populations and gene signatures also showed site-specific patterns. IFN-Ɣ scores were above median in 60% of breast biopsies, compared with 17% of brain, 39% of liver, and 62% of lung metastases. T-cell and CD8+ T-cell scores were typically higher in lung (65%, 62% above median) and lymph node metastases (66%, 62%), intermediate in breast biopsies (53%, 56%), and lowest in liver (32%, 37%) and brain metastases (17%, 22%). In contrast, dendritic cell scores were highest in liver metastases (83% above median), compared with breast samples (48%). Conclusion: Immunophenotypes in breast cancer vary substantially by sample site, reflecting-heterogeneity of the local microenvironment. These site-specific immune signatures may have important implications for immunotherapy selection, prognostic assessment, and vaccine development strategies.