Transcriptomic features of breast cancer locoregional recurrence (LRR) remain poorly understood. We therefore sought to investigate transcriptomic features associated with LRR in newly diagnosed invasive breast tumors from our institutional dataset.

Transcriptomic profiling was performed on 632 tumors from consecutive patients treated within our health system for newly diagnosed non-metastatic breast cancer. Univariable Cox models identified genes whose expression was associated with LRR (q-value < 0.05). Up-regulated (UR) genes were defined as hazard ratio (HR) > 1 and down-regulated (DR) genes were defined as HR < 1. Gene set enrichment analyses were performed for UR and DR gene sets and validated within two external cohorts of ER- tumors.

With a median follow-up of 7.6 years, we observed 38 LRRs: 28/481 (5.8%) in ER + and 10/151 (6.6%) in ER-. There were 43 UR and 7 DR genes associated with LRR in ER + tumors, while 417 UR and 1150 DR genes were associated with LRR in ER- tumors. UR genes in ER + tumors were enriched for roles in cell proliferation (q < 0.05). In contrast, LRR in ER- tumors was most strongly associated with DR genes enriched for MHC-II-mediated antigen presentation and T cell activation (q < 0.05). In external cohorts of ER- tumors, 97 significant DR genes (p < 0.05) were enriched for 18 pathways, including 5 pathways involved in MHC-II signaling, antigen presentation and T-cell activation.

Transcriptomic patterns associated with LRR appear distinct between ER + and ER- tumors. In ER + tumors, LRR appears predominantly associated with proliferation, whereas ER- LRR suggests a robust pattern of suppressed antigen presentation via MHC-II.