Abstract PS16-10: SPATIAL IMMUNE CELL DISTRIBUTION CAN REFINE PROGNOSIS IN EARLY-STAGE ER+/HER2-/LN- BREAST CANCER

Abstract

Abstract Lymphocytic infiltrate is a known prognostic biomarker in estrogen receptor(ER)-negative breast cancers (BCs). Comparatively ER+ disease is putatively cold, however, there exists an infiltrate-rich subset of ER+ tumours with significant spatial heterogeneity and unknown clinical impact. Using serial sections taken from early-stage, ER+/HER2- breast tumours of Irish patients enrolled in the TAILORx trial (n=450), we aimed to investigate the prognostic potential of markers encompassing tumour architecture. Immunohistochemistry for Ki67 and CD45 (leukocyte common-antigen), and staining for haematoxylin and eosin was applied to all sections [1]. Classifiers for stromal fraction (SF), stromal-infiltrate (sTIL), Ki67-LI, and their spatial relationships within the tumour were generated using QuPath [2] and R Studio. Trained marker classifiers were validated against an expert pathologist (R2 of M-Score to: CD45%: 0.968, Ki67-LI: 0.814, sTIL: 0.864) to define observed lymphocytes as tumour or stromal-infiltrating, retain only tumour Ki67+ density, and to investigate SF. Cohort mean SF was 67.69% (range 16.35 – 98.29%), with marginally significant differences in recurrence prediction for cohort high v low SF by mean (c-index = 0.58, p< 0.032), and luminal A disease (p=0.037) only. sTILs did not differ significantly between high/low mean SF (Mann-Whitney p=0.6201), though sTIL was prognostic in the OncotypeDx intermediate Recurrence Score (RS) category overall (p=0.0076). This trend appeared strongest in intermediate RS patients receiving chemo-endocrine therapy (HT+CT) (p < 0.0001) vs endocrine therapy (HT) alone (p=0.86). Investigating the effects of prescribed therapy on sTIL-derived recurrence risk also revealed significant trends in those patients receiving HT+CT only (p < 0.00001) vs HT alone (p=0.26). Spatial analysis of sTILs suggest that tumours become more immune excluded as Oncotype Dx RS increases - particularly from intermediate to high RS (Wilcoxon Intermediate v High: p=0.0077), with significant differences in survival for high/low tumour-immune hotspots observed in Intermediate RS, HT+CT treated patients (p=0.0052). Further spatial analysis suggests that high normalised frequency of infiltrate [method adapted from 3] (within 7um of tumour cells) have negative prognostic impact for premenopausal patients (p=0.017), and inter-sectional analysis identifies niche tumour environments where high Ki67-LI colocalize with immune infiltrates[methods adapted from 4, 5], most notable in Intermediate RS tumours.. Overall these data suggest immune-spatial subsets can be used to refine risk stratification of ER+ breast cancer.