Metabolic reprogramming is a cancer hallmark important to tumor initiation, progression and drug response. However, our knowledge of the tumor metabolome is limited since only a small number of metabolites can be reliably measured at a time. To enhance our understanding of tumor metabolism, we integrate metabolomics with transcriptomics to reveal potential metabolism-related transcriptional drivers. Our novel method, METRI (for MEtabolomic-TRanscriptomic Integration), identifies (1) metabolic alterations between two conditions, before and after treatment in our case and (2) interactions between signaling pathways and metabolic alterations. First, to identify metabolic alterations, METRI builds a matrix of gene and metabolite expressions for each metabolic pathway obtained from Kyoto Encyclopedia of Genes and Genomes. METRI computes a distance before and after treatment, weighted by the pathway coherence, where coherence is defined as the average absolute correlations across the genes and metabolites, resulting in a score for pathway ranking. Bootstrap is employed to obtain a p-value to assess each pathway's ranking significance. Second, to build an interaction network between metabolic alterations and signaling pathways, METRI identifies signaling genes correlated with each metabolic pathway and obtains enriched signaling pathways of each altered metabolic pathway. We applied METRI to two public non-small cell lung cancer datasets (GSE49644 and GSE17708/ST000010) to study the metabolic alterations associated with the Epithelial-Mesenchymal Transition (EMT), a phenotypic change related to cancer invasion and drug resistance. Both studies generated metabolomic and transcriptomic data and induced EMT with TGF-β treatment on cell lines (A549, HCC827 and NCI-H358) with three replicates for each condition. Using METRI, we found glycerophospholipid metabolism (p-value=0.03), purine metabolism (p-value=0.03), pyrimidine metabolism (p-value=0.04), aminosugars metabolism (p-value=0.04) and sphingolipid metabolism (p-value=0.04) were consistently altered after EMT induction in both datasets. In particular, we identified that several genes of the hexosamine biosynthesis pathway, a major pathway producing aminosugars for protein glycosylation, were upregulated in EMT. In addition, we identified that MAPK (p-value=0.004) and TGF-β (p-value=0.01) signaling pathways were associated with aminosugars metabolism. We also identified p53 signaling (p-value=0.002) was associated with sphingolipid metabolism, which plays a critical role in cancer cell adhesion and migration. In summary, METRI is a new method for data integration to study metabolic alterations and signaling-metabolism interactions and has the potential to reveal new insights to cancer biology and suggest novel targets for therapeutic intervention.

Citation Format: Weiruo Zhang, Sylvia Plevritis. Studying tumor metabolic reprogramming through integration of metabolomics and transcriptomics [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2018; 2018 Apr 14-18; Chicago, IL. Philadelphia (PA): AACR; Cancer Res 2018;78(13 Suppl):Abstract nr 1297.