A new multivariate statistical approach, based on the novel combination of projection on latent structure analysis with an inbuilt orthogonal filter (OPLS-DA) coupled with a spectroscopic correlation method statistical total correlation spectroscopy (STOCSY), was used to characterize the in vivo metabolic pathway perturbations of a model renal cortical toxin HgCl2, in the rat, using urine as an indicator of metabolic homeostasis disruption. This method provided an unbiased, sensitive approach to biomarker extraction and identification, and showed potential for generating potential novel pathway connectivities.