PATTERN-RECOGNITION CLASSIFICATION OF THE SITE OF NEPHROTOXICITY BASED ON METABOLIC DATA DERIVED FROM PROTON NUCLEAR-MAGNETIC-RESONANCE SPECTRA OF URINE

The computer-based pattern recognition procedures of nonlinear mapping and principal-component analysis have been applied to analyze H-1 NMR -generated metabolic data on the biochemical effects of 15 acute nephr otoxin treatments affecting the renal cortex and/or renal medulla in r ats. The H-1 NMR signal intensities for 16 urinary metabolites represe ntative of several major intermediary biochemical pathways were estima ted using either a simple semiquantitative scoring system or complete peak intensity quantitation. NMR-derived data were treated as input co ordinates in a multidimensional metabolic space and were analyzed by p attern recognition methods through which the dimensionality was reduce d for display and categorization purposes. Different nephrotoxin treat ments were initially classified using semiquantitative metabolite scor es on the basis of their H-1 NMR-detectable biochemical effects, and a good separation of renal cortical toxin treatments from renal medulla ry toxin treatments was achieved. The refinement of using exact peak h eights rather than metabolic data scores utilized the available metabo lic information more fully and provided a unique classification of eac h type of toxin according to its pattern of biochemical effects and si te of toxic action. Principal-component analysis provided consistently better results than did nonlinear mapping in terms of discrimination between different sites of toxicity, and maps generated from correlati on matrices gave improved discrimination, compared with those based di rectly on the original metabolic data. A comparison between the use of an added internal quantitation standard (3-trimethylsilyl-[2,2,3,3-H- 2(4)]-1-propionate) and independently determined glucose excretion rat es for scaling to the NMR-detected urinary glucose levels demonstrated that the consistent classification of site-specific nephrotoxicity wa s independent of the quantitation standard used. This study has provid ed a rigorous assessment of data processing, relative quantitation, an d pattern recognition methods, and the utility of applying these metho ds to the classification of NMR-derived toxicological data. The consid erable potential of the NMR-pattern recognition approach in the assess ment of nephrotoxicity has also been confirmed with the discovery of n ew combinations of molecular markers of renal cellular damage.-