Blood samples preserved on FTA cards offer unique opportunities for genetic research. DNA recovered from these cards should be stable for long periods of time. However, it is not well established as how well the DNA stored on FTA card for substantial time periods meets the demands of forensic or genomic DNA analyses and especially so for from post-mortem (PM) samples in which the quality can vary upon initial collection. The aim of this study was to evaluate the time-dependent degradation on DNA quality and quantity extracted from up to 16 years old post-mortem bloodstained FTA cards. Four random FTA samples from eight time points spanning 1998 to 2013 (n=32) were collected and extracted in triplicate. The quantity and quality of the extracted DNA samples were determined with Quantifiler(®) Human Plus (HP) Quantification kit. Internal sample and sample-to-sample variation were evaluated by comparing recovered DNA yields. The DNA from the triplicate samplings were subsequently combined and normalized for further analysis. The practical effect of degradation on DNA quality was evaluated from normalized samples both with forensic and pharmacogenetic target markers. Our results suggest that (1) a PM change, e.g. blood clotting prior to sampling, affects the recovered DNA yield, creating both internal and sample-to-sample variation; (2) a negative correlation between the FTA card storage time and DNA quantity (r=-0.836 at the 0.01 level) was observed; (3) a positive correlation (r=0.738 at the level 0.01) was found between FTA card storage time and degradation levels. However, no inhibition was observed with the method used. The effect of degradation was manifested clearly with functional applications. Although complete STR-profiles were obtained for all samples, there was evidence of degradation manifested as decreased peak heights in the larger-sized amplicons. Lower amplification success was notable with the large 5.1 kb CYP2D6 gene fragment which strongly supports degradation of the stored samples. According to our results, DNA stored on FTA cards is rather stable over a long time period. DNA extracted from this storage medium can be used as human identification purposes as the method used is sufficiently sensitive and amplicon sizes tend to be <400 bp. However, DNA integrity was affected during storage. This effect should be taken into account depending on the intended application especially if high quality DNA and long PCR amplicons are required.

Cytochrome p450 family 2, subfamily D, polypeptide 6 (CYP2D6) may be used to infer the metabolizer phenotype (MP) of an individual as poor, intermediate, extensive/normal, or ultrarapid. Metabolizer phenotypes may suggest idiosyncratic drug responses as contributing factors to cause and/or manner of death in postmortem investigations. Application of CYP2D6 has used long-range amplification of the locus and restriction enzyme digestion to detect single-nucleotide variants (SNVs) associated with MPs. This process can be cumbersome and requires knowledge of genotype phase. Phase may be achieved using long-read DNA sequencing and/or computational methods; however, both can be error prone, which may make it difficult or impractical for implementation into medicolegal practice. CYP2D6 was interrogated in postmortem autopsied Finns using supervised machine learning and feature selection to identify SNVs indicative of MP and/or rate of tramadol O-demethylation (T:M1). A subset of 18 CYP2D6 SNVs could predict MP/T:M1 with up to 96.3% accuracy given phased data. These data indicate that phase contributes to classification accuracy when using CYP2D6 data. Of these 18 SNVs, 3 are novel loci putatively associated with T:M1. These findings may enable design of small multiplexes for easy forensic application of MP prediction when cause and/or manner of death is unknown.

No abstract

Predicting metabolizer phenotype (MP) is typically performed using data from a single gene. Cytochrome p450 family 2 subfamily D polypeptide 6 (CYP2D6) is considered the primary gene for predicting MP in reference to approximately 30% of marketed drugs and endogenous toxins. CYP2D6 predictions have proven clinically effective but also have welldocumented inaccuracies due to relatively high genotype-phenotype discordance in certain populations. Herein, a pathwaydriven predictive model employs genetic data from uridine diphosphate glucuronosyltransferase, family 1, polypeptide B7 (UGT2B7), adenosine triphosphate (ATP)-binding cassette, subfamily B, number 1 (ABCB1), opioid receptor mu 1 (OPRM1), and catechol-O-methyltransferase (COMT) to predict the tramadol to primary metabolite ratio (T:M1) and the resulting toxicologically inferred MP (t-MP). These data were then combined with CYP2D6 data to evaluate performance of a fully combinatorial model relative to CYP2D6 alone. These data identify UGT2B7 as a potentially significant explanatory marker for T:M1 variability in a population of tramadol-exposed individuals of Finnish ancestry. Supervised machine learning and feature selection were used to demonstrate that a set of 16 loci from 5 genes can predict t-MP with over 90% accuracy, depending on t-MP category and algorithm, which was significantly greater than predictions made by CYP2D6 alone.