The Landscape of Immune Microenvironments in Racially Diverse Breast Cancer Patients

Abstract

Background: Immunotherapy is a rapidly evolving treatment option in breast cancer (BC); However, the BC immune microenvironment is understudied in Black and younger (<50 years) patients. Methods: We used histological and RNA-based immunoprofiling methods to characterize the BC immune landscape in 1,952 tumors from the Carolina Breast Cancer Study, a population-based study that oversampled Black (n=1,030) and young women (n=1,039). We evaluated immune response leveraging markers for 10 immune cell populations, compared profiles to those in the Cancer Genome Atlas Project [n=1095 tumors, Black (n=183), and young women (n=295)], and evaluated in association with clinical and demographic variables, including recurrence. Results: Consensus clustering identified three immune clusters in CBCS [adaptive-enriched, innate-enriched, or immune-quiet] that varied in frequency by race, age, tumor grade and subtype; however, only two clusters were identified in TCGA, which were predominantly comprised of adaptive-enriched and innate-enriched tumors. In CBCS, the strongest adaptive immune response was observed for basal-like, HER2+, TNBC, and high-grade tumors. Younger patients had higher proportions of adaptive-enriched tumors, particularly among estrogen receptor (ER)-negative cases. Black patients had higher frequencies of both adaptive-enriched and innate-enriched tumors. Immune clusters were associated with recurrence among ER-negative tumors, with adaptive-enriched showing the best and innate-enriched showing the poorest 5-year recurrence-free survival. Conclusions: These data suggest that immune microenvironments are intricately related to race, age, tumor subtype, and grade. Impact: Given higher mortality among Black and young women, more defined immune classification using cell-type specific panels could help explain higher recurrence and ultimately lead to targetable interventions.

Incorporating RNA-based Risk Scores for Genomic Instability to Predict Breast Cancer Recurrence and Immunogenicity in a Diverse Population

“…NanoString gene expression data in CBCS was collected and normalized using Remove Unwanted Variation ( 30 ) as described previously ( 31, 32 ), along with generation of PAM50 molecular subtypes, global immune classes and 10 immune cell scores. Briefly, a research version of the PAM50 molecular subtype predictor was used to classify tumors as luminal A, luminal B, HER2-enriched, basal-like or normal-like, and to generate risk of recurrence scores (ROR-PT) incorporating tumor size, proliferation, and subtype.…”

“…Briefly, a research version of the PAM50 molecular subtype predictor was used to classify tumors as luminal A, luminal B, HER2-enriched, basal-like or normal-like, and to generate risk of recurrence scores (ROR-PT) incorporating tumor size, proliferation, and subtype. Global immune classes were identified using a 48-gene panel of immune cell–related genes and hierarchical clustering analysis ( 31 ). We had previously identified three global immune expression classes reflecting adaptive-enriched, innate-enriched, and immune-quiet tumor microenvironments using consensus clustering ( 33 ).…”

“…CBCS global immune classes had high concordance with TCGA's pan-cancer immune subtypes ( 19 ) [see Supplementary Fig. S3 in Hamilton and colleagues, 2022 ( 31 )].…”

“…In addition, 10 major immune cell scores (B-cell, T-cell, CD8-T cell, Th cell, regulatory T cell, T follicular helper (Tfh) cells, eosinophil, neutrophil, natural killer, and macrophage) and a cytotoxic cell, adaptive cell, innate cell and PD-L1 score, were generated by computing the median cell-type specific gene expression as described previously ( 31 ). Markers for these immune cell types were selected on the basis of work from Bindea and colleagues, 2013 ( 35 ) and Danaher and colleagues, 2017 ( 36 ), which respectively utilized whole transcriptome and targeted RNA expression methods from immune cell and bulk tumor samples to establish marker panels, and subsequently validated against flow cytometry and IHC data.…”

“…RNA profiling was completed using tumor cores so the correlations are per tumor and not per specimen, but reflect strong concordance (e.g., RNA adaptive-cell score vs. TILs: rho = 0.29, P = 2.2e-16; ref. 31 ). Similarly high correlations were confirmed between the B-cell score and IF for CD19 [RNA B-cell score vs.…”