Metabolomic and Microbial Profiling of Rat Internal Organs for Postmortem Interval Estimation: A Multi-Omics Analysis

Abstract Accurate postmortem interval (PMI) estimation remains a major challenge in forensic medicine. Here, we tested whether full-length 16S rRNA gene profiles of internal organs add predictive information to untargeted LC−MS/MS metabolomics for PMI estimation in rats. Metabolomics-only, microbiome-only, early-fusion, and stacking models were compared under leakage-controlled leave-one-animal-out (LOAO) validation, with parsimonious fixed-K models selecting the top 10 metabolites and top 5 microbial features within each training partition. 16S amplification failed in all 30 day-0 and 25 of 30 day-1 internal-organ samples, in both the full-length assay and an independent V3−V4 assay, indicating that the microbial signal of internal organs is effectively undetectable in the first 0−24 h; modelling therefore covered a 2−7 day window. Metabolomic profiles tracked PMI closely (LOAO RMSE 0.467−0.677 days across tissues), whereas microbiome-only models yielded substantially higher RMSEs than the metabolomics models, although their point estimates remained below the intercept-only baseline (RMSE approximately 2.0 days). The added value of multi-omics integration was limited and tissue dependent: stacking yielded a significantly lower RMSE than metabolomics alone only in the spleen after within-tissue correction, and this difference did not remain significant after correction across the full set of comparisons. All performance estimates are internal, derived from a single cohort without independent validation, and no forensic applicability is claimed; confirmation in independent animal cohorts, broader environmental conditions, and ultimately human forensic samples is required.