Integrative analysis of metabolic subtypes in triple-negative breast cancer reveals new therapeutic strategies

Abstract

Background: Dysregulated metabolism is a hallmark of cancer and targeting specific metabolic pathways of cancer cells is a promising therapeutic strategy. Despite recent studies identified triple-negative breast cancer (TNBC) molecular subtypes, its metabolic pathway dependencies and metabolic classification has not been fully understood. Methods: We applied an integrative analysis on transcriptomic and genomic data of TNBC (n=360) from Fudan University Shanghai Cancer Center to explore the metabolic heterogeneity of TNBC. We conducted gene set variation analysis to calculate the relative enrichment score of 86 metabolic pathways in each sample. Furthermore, K-means clustering method was performed to classify the TNBC metabolic enrichment levels into heterogeneous subtypes. In addition, we systematically explored the metabolic differences and prognostic significance of the metabolic pathway based subtypes (MPS). A subset of TNBC cell lines were used to identify the different metabolites and metabolic phenotypes produced by the subtypes. Results: We identified three heterogeneous subtypes in TNBC: MPS1, the lipidic subtype, with upregulated lipid metabolism; MPS2, the glycolytic subtype, with upregu-lated carbohydrate and nucleotide metabolism; MPS3, the mixed subtype, with part of pathways dysregulated. We found that MPS2 patients have worse prognosis, high homologous recombination deficiency score and are characterized by an amplification in chr12p12-12p13 genomic region. This amplification resulted in the over-expression of GAPDH, TPI1, ENO2, LDHB and DERA. TNBC cell lines belonging to MPS1 and MPS2 showed striking differences in fatty acid uptake, oxygen consumption and glucose utilization, and corresponded to differences in cell sensitivity to inhibitors of gly-colysis, glutamine metabolism, lipid synthesis. Moreover, knockdown of TPI1 in MPS2 TNBC cells inhibited cell growth, migration and invasion. Conclusions: We identify three metabolic pathway based subtypes of TNBC with different prognosis outcomes which showed distinct sensitivities to metabolic inhibitors. Our study suggests opportunities for personalized therapies and TPI1 might serve as a potential therapeutic target.