Molecular classification with NanoString nCounter system in triple-negative breast cancer.
e23198 Background: Previous studies have shown that several distinct subtypes identified by gene expression profiling (GEP) consisted of triple-negative breast cancer (TNBC). Compared with the subtypes defined by GEP, we developed molecular classification with NanoString nCounter system in TNBC. Methods: GEP was conducted on 188 FFPE containing chemotherapy-naïve TNBC tumors collected at Gangnam Severance Hospital. To select core genes for classification, other 120 samples from public GEP database were used. Correlation between nCounter system and GEP using identical RNA was done. In a part of tumors, BRCA1 methylation, homologous recombination deficiency (HRD) assay, and drug response assay with ATP was comprehensively assessed. Results: To classify TNBC into 4 major subtypes (Basal-like: BL, Luminal androgen receptor: LAR, Mesenchymal: M, and Immune-modulatory: IM) according to the Vanderbilt classification, we selected 110 genes in 220 samples with GEP (100 from Gangnam Severance Hospital and 120 from public database). In other 88 samples, the classification with the 110 genes were validated. In 149 tumors excluding UNS subtype by the Vanderbilt, a correlation between 110 genes-classification and the Vanderbilt system was 74.4% (111 of 149, Pearson’s R = 0.726). In 180 tumors with GEP and nCounter assay, a correlation between them was 85.0% (153 of 180, Pearson’s R = 0.827). Compared with tumors of the IM type by 110-genes, the recurrence-free survival was significantly reduced in tumors of the non-IM type by 110-genes ( P= 0.049). In cases with BRCA1 methylation test (n = 147), a significant higher rate of BRCA1 promoter methylation was found in BL type by nCounter system (BL: 41.0% vs. non-BL: 19.4%). In cases with HRD test, a significant lower rate of HRD was found in nCounter-identified LAR type (LAR: 0% vs. non-LAR: 79.3%). In patients with in vitro drug response assay with cisplatin (n = 36), tumors with nCounter-BL had a significant higher responsiveness than tumors with others ( P= 0.028). Conclusions: Our work shows the feasibility of molecular classification with nCounter system in TNBC. Future study warrants the clinical utility of this classification to guide the subtype-specific therapy in patients with TNBC.