Urine metabolomics based prediction model approach for radiation exposure

The radiological incidents and terrorism have demanded the need for the development of rapid, precise, and non-invasive technique for detection and quantification of exposed dose of radiation. Though radiation induced metabolic markers have been thoroughly investigated, but reproducibility still needs to be elucidated. The present study aims at assessing the reliability and reproducibility of markers using nuclear magnetic resonance (NMR) spectroscopy and further deriving a logistic regression model based on these markers. C57BL/6 male mice (8–10 weeks) whole body γ-irradiated and sham irradiated controls were used. Urine samples collected at 24 h post dose were investigated using high resolution NMR spectroscopy and the datasets were analyzed using multivariate analysis. Fifteen distinguishable metabolites and 3 metabolic pathways (TCA cycle, taurine and hypotaurine metabolism, primary bile acid biosynthesis) were found to be amended. ROC curve and logistic regression was used to establish a diagnostic model as Logit (p) = log (p/1 − p) = −0.498 + 13.771 (tau) − 3.412 (citrate) − 34.461 (α-KG) + 515.183 (fumarate) with a sensitivity and specificity of 1.00 and 0.964 respectively. The findings demonstrate the proof of concept and the potential of NMR based metabolomics to establish a prediction model that can be implemented as a promising mass screening tool during triage.