**3.2 Interpretations of toxicological and genetic analyses**

Different studies tried to correlate concentrations of TR and its metabolites with different genotyping of CYP2D6. The best correlation was obtained for M2/M1 ratio and PM. The M2/M1 ratio in PM is almost above 7 [48].

Levo et al. calculated the TR/M1 and TR/M2. When the number of functional alleles increased, the median TR/M1 decreased. They also showed that median TR/M2 also correlated with the number of functional alleles, but in the reverse direction, as can be expected based on the complementary nature of two pathways [49].

The literature reported the value of M1/M2 ratio in case of massive TR intoxication. A quick death has been expected in case of M1/M2 is more than 1 [50].

The metabolic profiles of CYP2D6 PM showed large area under curve (AUCs) of the M2 enantiomers with low corresponding values of the M1 and M5 enantiomers. The (+)-enantiomers of M1 and M5 were affected to a larger extent than the (−)-enantiomers. The observed reduced levels of the M5 enantiomers were expected. M5 is formed from both M1 and M2. Since PMs only form low amounts of M1 due to the abolished (canceled) function of CYP2D6, the amounts of the M2 enantiomers will accumulate. Another metabolic profile of CYP2D6 IM showed the AUCs of both the M2 and M5 enantiomers exceeded the ones of the M1 enantiomers [10].

The general hypothesis in literature regarding adverse effects following TR administration is that the frequency and intensity is related to the concentrations of (+)-M1. The higher the concentration of (+)-M1, the higher the risk of side effects and toxicity. In Haage et al. study, the individual experiencing most drug reaction symptoms (both fainting and vomiting during the experimental day) was the one with the second lowest maximal concentration and AUC of (+)-M1 in the 100 mg dosage group [10].

A recent work reported that the TR/M1 ratio may not accurately reflect the rate of TR *O*-demethylation in clinical patients. This study accounted that due to (i) *postmortem* redistribution (PMR) phenomenon, and (ii) various interval and/ or time between TR administration and death. In addition, TR is metabolized by multiple enzymes (e.g., TR and/or its metabolites are metabolized by CYP2D6, CYP3A4, CYP2B6, UGT1A8, or ABCC2) (**Figure 1**) and may be metabolized at different rates based on body size, liver function, chronicity of TR, general opioid, and/or other drug use [39]. Surprisingly, the first abovementioned cause is not accurate as PMR is not a problem in case of interpretation of fatalities related to TR. Indeed, many studies revealed that cardiac-to-femoral blood ratios obtained for TR and M1 are close to 1 [51–54].

It is suggested that the monogenic model likely introduces error, particularly for samples at the extremes of CYP2D6 activity; a prediction using multiple genes may reduce these discrepancies although this comparison has not been performed. The forensic community has not yet leveraged the power of machine learning for such studies. However, the forensic DNA community has begun developing methods of individualizing humans [39].
