NMR-based metabolomics study of canine bladder cancer

Abstract

Bladder cancer is one of the leading lethal cancers worldwide. With the high risk of recurrence for bladder cancer following the initial diagnoses, lifelong monitoring of patients is necessary. The lack of adequate sensitivity and specificity of current noninvasive monitoring approaches including urine cytology, other urine tests, and imaging, underlines the importance of studies that focus on the detection of more reliable biomarkers for this cancer. The emerging area of metabolomics, which deals with the analysis of a large number of small molecules in a single step, promises immense potential for discovering metabolite markers for screening and monitoring treatment response and recurrence in patients with bladder cancer. Since naturally-occurring canine transitional cell carcinoma of the urinary bladder is very similar to human invasive bladder cancer, spontaneous canine transitional cell carcinoma has been applied as a relevant animal model of human invasive transitional cell carcinoma. In this study, we have focused on profiling the metabolites in urine from dogs with transitional cell carcinoma and healthy control dogs combining nuclear magnetic resonance spectroscopy and statistical analysis methods. (1)H NMR-based metabolite profiling analysis was shown to be an effective approach for differentiating samples from dogs with transitional cell carcinoma and healthy controls based on a partial least square-discriminant analysis of the NMR spectra. In addition, there were significant differences in the levels of six individual metabolites between samples from dogs with transitional cell carcinoma and the control group based on the Student's t-test. These metabolites were selected to build a separate partial least square-discriminant analysis model that was then used to test the classification accuracy. The result showed good classification between transitional cell carcinoma and control groups with the area under the receiver operating characteristic curve of 0.85. The sensitivity and specificity of the model were 86% and 78%, respectively. These results suggest that urine metabolic profiling may have potential for early detection of bladder cancer and of bladder cancer recurrence following treatment, and may enhance our understanding of the mechanisms involved.