PMI estimation through ¹H NMR metabolomics on human pericardial fluid: a validation study

This study builds on a prior proof-of-concept metabolomic analysis of post-mortem pericardial fluid to assess its reproducibility and validate its utility for estimating the post-mortem interval. Sixty-five pericardial fluid samples were collected during medico-legal autopsies in two different Forensic Medicine Institutes with post-mortem intervals spanning 16 to 199 h. Samples underwent liquid-liquid extraction and ¹H NMR analysis, quantifying 50 metabolites. Multivariate statistical analyses were employed to develop post-mortem interval estimation models, controlling for age to minimize its confounding effects. Reproducibility was confirmed, with 92% of metabolites showing high similarity (cosine similarity ≥ 0.90) in 23 re-analyzed samples, demonstrating robust intra-laboratory consistency. For post-mortem intervals of 16 to 100 h, the regression model achieved presented a prediction error of 16.7 h, identifying nine key predictors, including choline, glycine, citrate, betaine, ethanolamine, glutamate, ornithine, uracil, and β-alanine. For intervals of 16 to 130 h, the prediction error was 23.2 h, and for 16 to 199 h, it was 42.1 h. A classification model distinguishing intervals below 48 h from those above 48 h showed high accuracy for detecting longer intervals, with key predictors including aspartate, histidine, and proline. These findings underscore the stability and reproducibility of pericardial fluid metabolomics, establishing its potential as a reliable forensic tool for post-mortem interval estimation, particularly beyond 48 h, with significant implications for forensic investigations.