Oncometabolite signatures from tumor-stroma crosstalk as potential non-invasive biomarkers

Abstract Here, we show, in vitro, predominant support of patient-specific tumor associated fibroblasts (TAFs) isolated from non-metastatic colorectal adenocarcinomas on cancer cell proliferation, whereas TAFs isolated from metastatic adenocarcinomas facilitated cancer cell migration. An in vitro analysis revealed that variations in mitochondrial activity and the secretion of specific metabolites correlated with patient diagnoses. Among the oncometabolites identified, we validated amino acid expression in urine samples from 19 colon cancer patients to assess their potential as diagnostic biomarkers. Our results suggested patient-specific alterations in oncometabolite expression, which were significantly different from those of healthy controls. The specificity, sensitivity, and accuracy analysis indicated 93–100% specificity, 74–82% sensitivity, and 84–89% accuracy of single metabolites in distinguishing cancer patients from controls. While no false negatives were observed, urine samples from nine patients with various inflammatory conditions (including diverticulitis, appendicitis, and chronic gastritis) yielded false positives. Sensitivity analysis and the t-Distributed Stochastic Neighbor Embedding (t-SNE) nonlinear dimensionality reduction technique revealed distinct metabolite profiles for healthy controls, colon cancer patients, and patients diagnosed with inflammation. Our findings suggest that the identified oncometabolites hold a promise as a novel non-invasive tool for screening individuals at risk of colorectal cancer.