99P Exploratory analysis of differential gene expression (DGE) and non-negative matrix factorisation (NMF) clustering in KATHERINE: Adjuvant trastuzumab emtansine (T-DM1) vs trastuzumab (H) in patients with HER2-positive residual invasive breast cancer after neoadjuvant treatment (NAT)

This exploratory biomarker analysis aimed to identify prognostic gene sets in the. T-DM1 and H arms of the phase III KATHERINE study (NCT01772472). RNA sequencing was performed on post-NAT surgical samples. Genes pathways associated with prognosis were identified using DGE, comparing pts invasive disease-free survival (iDFS) events censored at a 3-year cut-off, set enrichment (GSEA), Hallmark, KEGG, xCell, selected signatures. NMF used transcriptional subgroups; their association iDFS assessed by Cox regression. Association analyses adjusted for tumour content (TC) stratification factors. Eight hundred fifteen samples included analysis. GSEA showed that cell cycle, oxidative phosphorylation DNA repair poor both arms; arm, metabolism-related signatures while immune good prognosis; apoptosis epithelial mesenchymal (EM) transition (EMT) fibroblast, stroma endothelial scores prognosis. Trends seen malignant-specific EM arms. clusters are described table. Table: 99PCluster, % prevalenceGene signature expression, TC prognosisiDFS hazard ratio,T-DM1 vs. (95% confidence interval)CL1, 24.0%Cell cycle repair-related genes. High TC, higher HER2 lower ESR1 levels other clusters, poorest prognosis0.42 (0.24, 0.75)CL2, 8.3%Metabolism keratinisation-related genes0.58 (0.16, 2.19)CL3, 40.6%Focal adhesion, TGFβ, Wnt β catenin, EMT extracellular matrix-related Low best prognosis0.25 (0.11, 0.57)CL4, 18.5%Oestrogen- cilium assembly-related highest expression0.49 (0.22, 1.08)CL5, 8.5%Immune-related genes0.65 (0.18, 2.38) Open table new tab . Both DGE approaches pathway-related genes as Stromal high stromal The advantage over across all clusters.