Predictive diagnosis of Major Depressive Disorder using integrated NMR based metabolomics and logistic regression model approach

The current diagnosis for Major Depressive disorder (MDD) is dependent on symptomatic clusters and resulting high error rates. The study identifies a panel of biomarkers using 1H NMR spectroscopy and logistic regression prediction modelling. The VIP score of >1.5 and S-plot based on OPLS-DA depicted 4 significant metabolites (phosphocreatine, phosphocholine, glycerophosphocholine and glutamine) indicating abnormalities in energy and lipid metabolism. Phosphocreatine showed the highest AUC of 0.875 with 90% sensitivity and specificity, while with a combination of 4 metabolites, the AUC increased to 0.927 with 96.3% sensitivity and 87.5% specificity, which may act as a supplementary diagnostic tool for MDD.