Urinary metabolomics of gastric cancer.

Abstract

22 Background: Gastric cancer (GC) has 70-75% mortality, attributable to delayed diagnosis. There is no standard screening in North America. Metabolomics is a systems biology approach to measure low molecular weight chemicals (metabolites) in body fluids or tissues to provide a phenotypic “fingerprint” of disease etiology. In this preliminary study it was hypothesized that metabolic profiling of urine samples using 1H-NMR spectroscopy could discriminate between resectable gastric adenocarcinoma (GC), benign gastric disease (BN), and healthy (HE) patients (pts). Methods: Midstream urine samples were collected, processed, and biobanked at -80°C, from 30 BN, 30 HE and 16 of 29 GC pts visiting three Edmonton clinics from August 2013 – January 2014. Thirteen of 29 samples were retrieved from a 2009-13 GC biobank. Samples were matched on age, gender and BMI. Using a validated standard operating procedure each sample was analyzed using high resolution 1H-NMR spectroscopy. Resulting spectral traces were converted into annotated and quantified metabolite profiles of 58 metabolites. Univariate and multivariate statistical analysis uncovered a disease specific biomarker profile. Partial Least Squares Discriminant Analysis (PLS-DA) developed a GC vs. HE discriminative model. A Receiver Operator Characteristic (ROC) curve was constructed. Results: There was no significant difference in metabolite profiles between GC and BN pts. However, univariate analysis revealed 13 metabolites that differed significantly between GC and HE (p<0.05). Correlation analysis, followed by PLS-DA produced a discriminative model with an area under ROC curve of 0.996, such that for a specificity of 100% the corresponding sensitivity was 93%. Conclusions: GC pts have a distinct urinary metabolite profile compared to HE controls; however in this study metabolic profiling was unable to discriminate GC from BN pts. This was probably due to sample size and phenotypic heterogeneity of BN patients. This preliminary study shows clinical potential for metabolic profiling for early GC detection.