Abstract B094: Novel method for patient stratification in breast carcinoma based upon spatial analysis of tumor microenvironment

Abstract

Breast cancer consists of several intrinsic molecular subtypes, providing the basis for clinical treatment decisions. Lately, it is becoming increasingly recognized that factors other than the intrinsic cancer characteristics, such as immune components’ activity in the tumor microenvironment, have important effects on treatment choices and efficacy. bioSyntagma has developed a method, the Molecular Fingerprint (mPrint®), that enables multiplexed analysis of spatially defined regions in formalin-fixed, paraffin-embedded (FFPE) tumor samples allowing for analysis of the gene signatures unique to the tumor microenvironment. This method was applied to molecularly defined sets of breast cancers and used to evaluate four different tumor regions of interest (ROIs): 1) viable carcinoma proper (>90% cancer cells), 2) fibrotic tumor center (sparse cellularity), 3) interface between viable tumor and inflammatory component (tumor and inflammatory microenvironment) and 4) tissue away from the tumor (normal breast tissue). This was compared to the whole tissue scrapes from each patient block. Each ROI and tissue scrape was analyzed by high throughput qPCR for a panel of 248 genes using SmartChip technology (Takara Bio, USA). Sequential tissue slices from each patient were also analyzed using immunohistochemistry (IHC) for three targets and investigated for correlation with qPCR results for validation of the method. Overall, reasonable concordance was observed in general expression trends between selected IHC and RNA expression. qPCR data were further analyzed using hierarchical clustering analysis and showed that morphologically defined ROI’s cluster completely differently than traditional clustering of entire tissue scrapes. Notably, patient clustering based on morphological regions was independent of the intrinsic cancer subtype, as determined by molecular profiling of whole tissue scrapes, as well as independent of trends in Tumor Mutational Burden (TMB) and Microsatellite Instability (MSI). These findings suggest that current methods of patient stratification based on whole tumor molecular subtyping may be inferior to stratification based on molecular characteristics of the tumor microenvironment.