An enrichment-based approach to interpreting metabolomic data using differential metabolomic profiles within the iDMET framework

Pathway enrichment analysis is a crucial method for the biological interpretation of metabolomic data by identifying associations between altered metabolites and biological pathways. However, such traditional approaches often rely on a limited set of predefined metabolic pathways, resulting in a low likelihood of discovering pathways associated with a given metabolic profile. To overcome this limitation, we extended our previously developed iDMET methodology to incorporate a broader range of metabolite sets, including those derived from differential metabolomic profiles. This enhanced approach, termed iDMET+, significantly expands dataset diversity and size, increasing the likelihood of discovering associated metabolite sets for a given metabolic profile, thereby enables more biological insights to be obtained from the metabolic profile. We validated iDMET+ through case studies on three diseases: clear cell renal cell carcinoma, colorectal cancer, and small cell lung cancer. First, using a clear cell renal cell carcinoma study as input, iDMET+ correctly identified another study of the same disease that involved metabolomic analysis. This pair of studies was identified as relevant in our previous iDMET results, showing the consistency between iDMET+ and iDMET. Second, using the metabolomic profile of colorectal cancer as input, iDMET+ identified not only another metabolomic study of the same cancer but, surprisingly, also metabolomic studies on prostate cancer and a high-fat diet. These studies focused on MYC-driven metabolic reprogramming, which was also a major focus of the input study. In both case studies, related studies were enriched because the differential metabolomic profiles of directly associated studies were part of the metabolite set. In contrast, the small cell lung cancer study highlighted limitations in dataset coverage—the absence of directly relevant differential metabolomic profiles resulted in fewer enriched metabolite sets. Nevertheless, the analysis of commonly altered metabolites still yielded some meaningful results. Metabolite alterations associated with inhibition of the purine salvage pathway were observed, suggesting potential involvement in tumor metabolic reprogramming. These results demonstrate that iDMET+ offers broader biologically relevant information than the conventional pathway-based approaches and has the potential to uncover biologically significant findings by searching across diverse datasets. This work also identifies areas of improvements for iDMET+.