Pattern recognition analysis of proton nuclear magnetic resonance spectra of postmortem cerebrospinal fluid from rats with drug-induced seizure or coma.

Cerebrospinal fluid (CSF) is routinely subjected to gross evaluation in postmortem investigations; however, its use in chemical evaluations has not been fully realized. Analysis of nuclear magnetic resonance (NMR) spectra with pattern recognition methods was applied to CSF samples. Rats were treated with pentylenetetrazol (PTZ) to induce seizure or pentobarbital (PB) to induce coma, and postmortem CSF was collected after CO2 gas euthanization. Pattern recognition analysis of the NMR data was performed on individual postmortem CSF samples. The aim of this study was to determine if pattern recognition analysis of NMR data could be used to classify the rats according to their drug treatment. The applicability of NMR data with pattern recognition analysis using postmortem CSF was also assessed. Partial Least Squares-Discriminant Analysis (PLS-DA) score plots indicated that the PTZ, PB, and NS (control) groups were clustered and clearly separated. PLS-DA correlation loading plots showed respective spectral and category variances of 41% and 42% for factor 1, and 17% and 27% for factor 2. Thus, factors 1 and 2 together described 58% (41%+17%) and 69% (42%+27%) of the variation, respectively. NMR study of postmortem CSF has the potential to be utilized as both a novel forensic neurochemistry method and in the clinical setting.