Bladder cancer has a high incidence and is marked by high morbidity and mortality. Early diagnosis is still challenging. The objective of this study was to create a metabolomics-based profile of bladder cancer in order to provide a novel approach for disease screening and stratification. Moreover, the study characterized the metabolic changes associated with the disease. Serum metabolomic profiles were obtained from 149 bladder cancer patients and 81 healthy controls. Different ensemble machine learning models were built in order to: (1) differentiate cancer patients from controls; (2) stratify cancer patients according to grading; (3) stratify patients according to cancer muscle invasiveness. Ensemble machine learning models were able to discriminate well between cancer patients and controls, between high grade (G3) and low grade (G1-2) cancers and between different degrees of muscle invasivity; ensemble model accuracies were ≥80%. Relevant metabolites, selected using the partial least square discriminant analysis (PLS-DA) algorithm, were included in a metabolite-set enrichment analysis, showing perturbations primarily associated with cell glucose metabolism. The metabolomic approach may be useful as a non-invasive screening tool for bladder cancer. Furthermore, metabolic pathway analysis can increase understanding of cancer pathophysiology. Studies conducted on larger cohorts, and including blind trials, are needed to validate results.
A serum metabolomic signature for the detection and grading of bladder cancer
Troisi J.;Colucci A.;Cavallo P.;Landolfi A.;Maiorino F.;Fabiano M.;D'Antonio A.;Altieri V.
2021-01-01
Abstract
Bladder cancer has a high incidence and is marked by high morbidity and mortality. Early diagnosis is still challenging. The objective of this study was to create a metabolomics-based profile of bladder cancer in order to provide a novel approach for disease screening and stratification. Moreover, the study characterized the metabolic changes associated with the disease. Serum metabolomic profiles were obtained from 149 bladder cancer patients and 81 healthy controls. Different ensemble machine learning models were built in order to: (1) differentiate cancer patients from controls; (2) stratify cancer patients according to grading; (3) stratify patients according to cancer muscle invasiveness. Ensemble machine learning models were able to discriminate well between cancer patients and controls, between high grade (G3) and low grade (G1-2) cancers and between different degrees of muscle invasivity; ensemble model accuracies were ≥80%. Relevant metabolites, selected using the partial least square discriminant analysis (PLS-DA) algorithm, were included in a metabolite-set enrichment analysis, showing perturbations primarily associated with cell glucose metabolism. The metabolomic approach may be useful as a non-invasive screening tool for bladder cancer. Furthermore, metabolic pathway analysis can increase understanding of cancer pathophysiology. Studies conducted on larger cohorts, and including blind trials, are needed to validate results.I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.
