Intratumor molecular tumor heterogeneity in low ER-expressing primary breast tumors

Abstract Background: Change in estrogen receptor (ER) status between primary breast cancer and distant recurrence are seen in the clinic in up to 15-20% of patients. This is difficult to reconcile with the current understanding of breast cancer molecular subtypes, the large-scale molecular differences between ER-positive (ER+) and ER-negative (ER-) cancers suggest different cellular origins, luminal versus basal breast epithelium, respectively. A potential explanation for ER status switch is the presence of mixed molecular subtypes at diagnosis. ER+ cancers with less than 100% positivity may be composed by both luminal and basal-like cancer cells, and a recurrence might arise from one subtype that survived adjuvant therapy. The goal of this study was to test if molecular subtype heterogeneity exists at diagnosis in less than 100% ER+ breast cancer using the NanoString GeoMx™ platform. Methods: GeoMx™ is a highly multiplexed assay that quantifies RNA expression from spatially discrete regions of interest (ROIs) within formalin fixed paraffin embedded (FFPE) tissue sections. We identified 4 cancers with 30-40% ER-positivity on routine immunohistochemistry (IHC) staining (clone SP1). Eight ROIs per case were selected to represent both ER-high and -low regions of the section using ER IHC (clone 1D5). Invasive tumor cells were defined as PanCK positive cells. 1,825 mRNA species were measured separately in each ROI. mRNA expression results (unique molecular index [UMI] counts) were quantile normalized and differential expression analyses for all genes, ESR1, ER-regulated gene set, and PAM50 gene set were performed between ROIs within the same cancer. We also assessed PAM50 subtypes of each cases on bulk RNA using NanoString BC360 panel. Results: Three of the 4 cases had heterogenous ER IHC (1D5) staining; the fourth case was uniformly ER-negative during ROI selection. For this last case, ROIs were selected based on PanCK staining from different locations in the cancer. Bulk PAM50 subtyping indicated 1 basal-like, 2 HER2-enriched (HER2-E), and 1 luminal (lum)A cancer. We also attempted ROI-level subtyping; however, Pearson correlations to the nearest centroid were low (<0.5) for all ROIs, possibly due to low expression of PAM50 genes or to the exploratory methodology used. There was substantial within cancer, between ROI, expression heterogeneity for many genes including ESR1 and sensitivity to endocrine therapy (SET) index genes. Conclusions: ER- and ER+ cells are intermixed which makes defining low and high ER regions challenging. We observed region to region variation in ESR1 and SET index gene expression. ROI-level PAM50 subtyping was unstable. The novel NanoString GeoMx™ DSP technology can be used to study intratumor molecular heterogeneity in regions of interest on FFPE breast cancer tissue sections. We found substantial region-to-region gene expression differences within tissue sections. Case ID1234ER IHC SP130%40%40%30%ER IHC 1D50%40%40%30%Bulk PAM50 SubtypeBasalHER2-ELumAHER2-EESR1 UMI average (range across ROIs)28.45 (22.4-34.6)64.9 (18.8-173)179.9 (114-245)34.8 (25.1-45.4)SET Index (genes positively correlated with ESR1) UMI average (range across ROIs)41.5 (39.1-44.8)120.1 (86.9-182.9)94.9 (72.9-118.4)143.5 (124.9-150.3)SET Index (genes negatively correlated with ESR1) UMI average (range across ROIs)57.6 (45-75.3)43.4 (38.8-52.6)33.7 (31.8-36.7)44.8 (41.4-50.6) Citation Format: Julia Foldi, Emily Reisenbichler, Liuliu Pan, Krsitina Sorg, Sarah E. Church, Lajos Pusztai. Intratumor molecular tumor heterogeneity in low ER-expressing primary breast tumors [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 P1-05-02.