A bacterial ethanol model developed in nutrient broth did not retain the same performance in every human blood or plasma culture. Boumba and colleagues showed that changing the matrix could alter both ethanol production and the relationship with other alcohols, even when the starting glucose concentration was matched. A successful broth model was not a universal correction for a postmortem specimen.
What the researchers cultured
The study used four reference bacterial strains: Escherichia coli, Clostridium perfringens, Clostridium sporogenes and Enterococcus faecalis. They were cultured separately under controlled anaerobic conditions for 30 days, principally at 25°C, in brain-heart-infusion broth. Starting bacterial concentration was approximately 10⁴ cells/mL. No growth was observed in the 4°C comparison cultures.
Alcohol concentrations were tracked during incubation. Three species produced usable alcohol patterns; E. faecalis produced negligible ethanol and other alcohols under these conditions, so it did not supply a practical regression model.
The human-matrix work used blood from healthy living donors, not blood removed from decomposed bodies. The methods described sets of 16 tubes for sampling in pairs at eight time points through day 15. Sixteen tubes should not be described as sixteen independent donors; a donor count was not clearly reported.
Changing the medium changed the outcome
C. sporogenes produced the largest ethanol amounts in broth, but produced no ethanol or other measured alcohols in the tested plasma cultures during 15 days. Adding glucose to achieve 200 mg/dL did not restore that production.
E. coli produced more ethanol in glucose-supplemented plasma than in unsupplemented plasma, yet less than in broth with the same initial glucose concentration of 200 mg/dL. Glucose alone therefore did not account for the matrix effect.
The blood anticoagulant also mattered. In citrate-containing blood inoculated with C. perfringens, ethanol appeared after day five without detectable production of the other measured alcohols. The EDTA-blood experiment behaved differently. These observations concern defined inoculated cultures, not every preserved forensic sample.
Analytical measurement and model performance were separate
The laboratory quantified ethanol and other volatiles by headspace GC–FID. For ethanol, the reported detection limit was 0.01 mg/dL, quantification limit 0.03 mg/dL and calibration range 0.03–400 mg/dL. The validation reported precision below 6% relative standard deviation and accuracy errors from −1.0% to 2.5% across the tested analytes and controls.
These analytical characteristics did not guarantee that a regression fitted in one medium predicted another medium correctly. In EDTA whole blood inoculated with C. perfringens, the broth-derived model did not successfully predict the experimental values; a second-degree polynomial provided a better fit, R² = 0.79.
Other transfers performed better under particular conditions. The authors reported average prediction errors of approximately 16% for one C. perfringens plasma application and 22% for glucose-supplemented E. coli plasma. Those percentages were conditional experimental results, not universal uncertainty limits for casework.
What did the forty postmortem cases establish?
The investigators also applied models retrospectively to 40 selected case chromatograms with co-detected alcohols. Thirty-nine cases showed marked putrefaction; one had extensive trauma. This selected series was not a random validation cohort.
Critically, the amount of ethanol truly formed after death was not independently known. Comparison with the measured total ethanol could not separate predeath ingestion from microbial formation. The authors acknowledged that limitation, the small number of modeled microbes and the need for further work before routine extrapolation.
What follows for a case review?
Identify the model’s organism, medium, temperature, additives and validation population before applying it to blood. Accurate ethanol quantitation and accurate attribution of ethanol origin are separate achievements. Neither a fitted model nor detection alone establishes drinking, impairment or cause of death.
For the related measurement distinction, see analytical uncertainty versus postmortem ethanol formation.
Primary source
Boumba VA, Economou V, Kourkoumelis N, Gousia P, Papadopoulou C, Vougiouklakis T. Microbial ethanol production: Experimental study and multivariate evaluation. Forensic Science International. 2012;215:189–198. doi:10.1016/j.forsciint.2011.03.003.
Leave a Reply