Metabolomics investigation of post-mortem human pericardial fluid

Background: Forensic science requires precise tools for estimating the post-mortem interval (PMI) to aid legal investigations. This PhD thesis focuses on enhancing PMI estimation through a metabolomic analysis of post-mortem pericardial fluid (PF), aiming to address the limitations of current methodologies by providing novel insights into post-mortem metabolomics changes. Materials and Methods: The research was initiated with a proof-of-concept study using 24 PF samples from judicial autopsies performed in Cagliari Legal Medicine Institute, covering a PMI range of 16 to 170 hours. Samples were processed using two methods, ultrafiltration and liquid-liquid extraction, for analysis via 1H NMR spectroscopy to compare metabolomic profiles. Subsequently, these samples were also analysed with UHPLC-QTOF-MS to validate findings across analytical platforms. The study was then expanded to include 65 samples, with PMIs extended to 199 hours, collected from two different Legal Medicine Institutes to evaluate reproducibility, sampling site effects, and sample integrity during transport. Regression models were constructed for different PMI windows (16-100, 16-130, and 16-199 hours) using multivariate analysis techniques like PLS and stability selection. Additionally, a classification model was developed to differentiate between PMIs below and above 48 hours using a subset of samples within the 16–100-hour range. Results: The proof-of-concept phase identified 50 metabolites, with both extraction methods yielding similar metabolomic profiles. Regression analysis indicated a moderate correlation between metabolite levels and PMI, with accuracy diminishing at higher PMIs. The UHPLC-QTOF-MS analysis corroborated these findings, showing distinct performance between ESI+ and ESI- modes for different PMI windows. With the expanded dataset, high reproducibility was confirmed on the proof-of-concept dataset, with 92% of metabolites showing consistent levels across experiments. Regression models for the extended PMI ranges identified key metabolites associated with PMI, particularly in shorter windows, where prediction errors were significantly reduced. The classification model effectively distinguished between short and long PMIs, with notable accuracy for identifying deaths occurring after 48 hours. Discussion: This study demonstrates the potential of metabolomics in PMI estimation, highlighting the importance of choosing the right biological matrix for the right PMI window to obtain an accurate model. The consistency across different analytical platforms and the high reproducibility of the 1H NMR approach suggest that metabolomics can be a reliable forensic tool, despite the challenges posed by individual biological variability and PMI range. The classification model's ability to differentiate PMI around the 48-hour threshold offers practical legal applications. The results suggest that while animal models have lower error rates due to controlled conditions, human PF metabolomics offers a practical, if more variable, approach for real-world forensic applications. Future research should aim at integrating different biological matrices for a more comprehensive PMI estimation, considering the complex biological changes in PF over time.