CYP2B6 allelic variants and non-genetic factors influence CYP2B6 enzyme function

Human CYP2B6 enzyme although constitutes relatively low proportion (1–4%) of hepatic cytochrome P450 content, it is the major catalyst of metabolism of several clinically important drugs (efavirenz, cyclophosphamide, bupropion, methadone). High interindividual variability in CYP2B6 function, contributing to impaired drug-response and/or adverse reactions, is partly elucidated by genetic polymorphisms, whereas non-genetic factors can significantly modify the CYP2B6 phenotype. The influence of genetic and phenoconverting non-genetic factors on CYP2B6-selective activity and CYP2B6 expression was investigated in liver tissues from Caucasian subjects (N = 119). Strong association was observed between hepatic S-mephenytoin N-demethylase activity and CYP2B6 mRNA expression (P < 0.0001). In less than one third of the tissue donors, the CYP2B6 phenotype characterized by S-mephenytoin N-demethylase activity and/or CYP2B6 expression was concordant with CYP2B6 genotype, whereas in more than 35% of the subjects, an altered CYP2B6 phenotype was attributed to phenoconverting non-genetic factors (to CYP2B6-specific inhibitors and inducers, non-specific amoxicillin + clavulanic acid treatment and chronic alcohol consumption, but not to the gender). Furthermore, CYP2B6 genotype–phenotype mismatch still existed in one third of tissue donors. In conclusion, identifying potential sources of CYP2B6 variability and considering both genetic variations and non-genetic factors is a pressing requirement for appropriate elucidation of CYP2B6 genotype–phenotype mismatch.


Materials and methods
Human liver samples. Human liver tissues (N = 119) were obtained from donated organs at the Department of Transplantation and Surgery, Semmelweis University (Budapest, Hungary). The study was approved by the Hungarian Committee of Science and Research Ethics, Medical Research Council (125/PI/2011, 4799-0/2011EKU), and was performed in accordance with the relevant guidelines and regulations (Act CLIV of 1997 on Health, decree 23/2002 of the Minister of Health of Hungary and the declaration of Helsinki). The subjects' demographic and clinical data (sex, age, cause of death, acute and chronic medication prior the explantation, smoking and alcohol consumption status) were recorded (Supplementary Table 1). Liver tissues from those subjects who were recorded chronic alcohol consumption [N = 11] were evaluated to be fibrotic due to alcohol related liver disease. Human livers were perfused with Euro-Collin's solution (Fresenius SE & Co. KGaA, Bad Homburg vdH, Germany) and excised. The tissues (approximately 1 g) were homogenized in 0.1 mM Tris-HCl buffer (pH 7.4) containing 1 mM EDTA and 154 mM KCl. Microsomal fraction was isolated by differential centrifugation and protein content of microsomes was determined by the method of Lowry et al. using bovine serum albumin as the standard 40,41 . Approximately 50 mg of liver tissue were homogenized in TRIzol reagent (Thermo Fisher Scientific, Waltham, MA) and total RNA was isolated according to the manufacturer's instructions. The hepatic RNA samples were stored in ultra-pure water containing 0.1% diethylpyrocarbonate at − 80 °C for further analyses.
CYP2B6 enzyme activity assay. The S-mephenytoin N-demethylation activity selective for CYP2B6 were performed in the incubation mixture containing NADPH-generating system (1 mM NADPH, 10 mM glucose-6-phosphate, 5 mM MgCl 2 and 2 units/ml glucose-6-phosphate dehydrogenase), human liver microsomes (0.8 mg/ml protein) and S-mephenytoin (1 mM). After 40-min incubation at 37 °C, the enzyme reactions were terminated by ice-cold acetonitrile, and the incubation mixtures were centrifuged at 10.000×g for 10 min. Formation of nirvanol was quantified by high-performance liquid chromatography according to Heyn et al. 5 . CYP2B6 enzyme assay for each donor was performed in triplicate, and the activity was expressed as pmol nirvanol*(mg protein) -1 *min -1 .
CYP2B6 genotyping. Genomic DNA templates were isolated from liver tissues using Quick-DNA Miniprep Plus Kit (Zymo Research, Irvine, CA). The following CYP2B6 polymorphisms were determined using validated TaqMan™ Drug Metabolism Genotyping Assays (Thermo Fisher Scientific) for g.-82 T > C (rs34223104, C_27830964_10), g.15631G > T (rs3745274, C_7817765_60) and g.25505C > T (rs3211371, C_30634242_40). Each reaction (in 5-µl reaction volume) contained Luminaris Probe qPCR Master Mix (Thermo Fisher Scientific), TaqMan™ Drug Metabolism Genotyping Assay (Thermo Fisher Scientific), 10-15 ng genomic DNA sample and nuclease-free water and incubated at 50 °C for 2 min and 95 °C for 10 min; and in 50 cycles of 95 °C for 15 s and 60 °C for 1 min. For identification of g.18053A > G nucleotide substitution (rs2279343), no validated PCR (polymerase chain reaction) primers and probes are commercially available; therefore, a two-step PCR assay based on the 'nested' PCR method with 'touchdown' thermal cycling protocol and the TaqMan PCR was developed ( Supplementary Fig. 1A) 42,43 . In the PCR reactions, two sets of primer pairs were applied and used consecutively to increase the CYP2B6 specificity of the SNV-discrimination and to avoid the amplification of CYP2B7P (Table 1). The first step was a 'nested' PCR containing PCRBIO VeriFi Master Mix (PCR Biosystems Ltd., London, UK), 400-400 nM forward and reverse primers (first set of primer pairs, Table 1) and 40-50 ng genomic DNA template. The first primer pair generated a relatively long (1275-bp) amplicon containing the whole CYP2B6 exon 5 and exon 6 with the intron 5 in between and with some short surrounding upstream and downstream intron region sequences ( Supplementary Fig. 1A). The homology of CYP2B6 intron sequences with CYP2B7P is somewhat lower than that of the exons 44 ; therefore, the first primer pair designed for the upstream and downstream intron regions (in introns 4 and 6) with the 'touchdown' PCR thermal cycling protocol was expected to provide the CYP2B6 gene specific hybridization of the primers. The thermal cycling protocol was performed according to the principles of 'touchdown' PCR: 95 °C for 1 min and 10 cycles of 95 °C for 15 s, 72-62 °C for 15 s (decreasing 1 °C/cycle), 72 °C for 1 min, and 10 cycles of 95 °C for 15 s, 62 °C for 15 s, 72 °C for 1 min. During the initial cycle, an annealing temperature (72 °C) higher than the targeted melting temperature of primers (62 °C) was used. Afterwards, the annealing temperature was decreased progressively over 10 cycles which made the reaction conditions more permissive. Theoretically, each cycle with decreasing annealing temperature by 1 °C produced four-fold exponential differences between correct and incorrect annealing resulting in the enrichment of the CYP2B6 specific over the non-target CYP2B7P specific product 42 . For testing CYP2B6 specificity of the 'nested' PCR reaction, the three verifying primers designed by Jacob et al. were modified: (1) a CYP2B6 specific (t5b6_2) forward primer, (2) a CYP2B7P specific (t5b7_2) forward primer and (3) a common (t5con_2) reverse primer (Table 1) 45 . The optimization of the length, GC-content and the melting temperatures of these verifying primers provided more suitable reaction conditions than those published by Jacob et al. 45 .  Fig. 1B). It confirmed that substantial amount of CYP2B6 specific amplicon (more than 5000-fold) was produced in the 'nested' PCR comparing to CYP2B7P specific amplicon, whereas for CYP2B6-selective amplification, the genomic DNA appeared to be not an appropriate template. The second PCR was carried out using Luminaris Probe qPCR Master Mix (Thermo Fisher Scientific), 300-300 nM forward and reverse primers (second set of primer pairs), 200-200 nM 'wild' and 'mutant' TaqMan probes (Table 1) and the 100-fold dilution of the 'nested' PCR product as the template. The incubation protocol was 50 °C for 2 min, 95 °C for 10 min and 50 cycles of 95 °C for 15 s, 66 °C for 1 min. The second primer pair generating a short, 137-bp amplicon with the rs2279343 mutation site in exon 5 and TaqMan probes designed for the wild-type and the mutant type of rs2279343 were used for SNV-discrimination (Supplementary Fig. 1C). The in silico design and validation of the primers and probes (  19 . Our aim was to detect only the full-length CYP2B6 mRNA and to distinguish it from the aberrant splicing variant; therefore, the primer pair for CYP2B6 expression assay was designed to the exons 3 and 4 (forward and reverse primers, respectively), and the intron-spanning amplicon detected exclusively the functional full-length mRNA variant ( Table 1). The quantity of the target mRNA relative to that of the housekeeping gene glyceraldehyde 3-phosphate dehydrogenase (GAPDH) was determined. The sequences of primers and probes used for the real-time PCR analyses of CYP2B6 and GAPDH expression are shown in Table 1.
Of the 9 genetically 'poor' metabolizer subjects with two loss-of-function alleles (CYP2B6*6/*6 or CYP2B6*6/*9 genotypes), only one displayed poor metabolism of S-mephenytoin, whereas the CYP2B6 activity    www.nature.com/scientificreports/ however, only 2 displayed extensive S-mephenytoin N-demethylation, and in the medical history of one subject, chronic alcohol consumption appeared to be responsible for high-intermediate metabolism.
In conclusion, CYP2B6 genetic polymorphisms explained S-mephenytoin N-demethylase activity in not more than 29 tissue donors (27.6%), and considering non-genetic factors improved the phenotype prediction by further 33.3% (35 tissue donors); however, in the medical history of 41 subjects, no relevant information on the non-genetic factors for altered CYP2B6 metabolism was found. It also means that phenoconverting non-genetic factors in these 41 subjects could not confirm altered CYP2B6 phenotype predicted from genotype, and genotype-phenotype mismatch still existed in 39.1% (41/105) of the subjects.

Effect of genetic and non-genetic factors on hepatic CYP2B6 mRNA expression. The nucleo-
tide change g.-82 T > C (rs34223104) has been reported to be associated with increased expression of CYP2B6 mRNA, whereas g.15631G > T (rs3745274) is associated with an aberrant mRNA splicing variant lacking exons 4 to 6 and entailing reduced CYP2B6 function or reduced expression of the full-length CYP2B6 mRNA variant 11,19 .

Multivariate analysis of CYP2B6 activity and mRNA expression. Multiple linear regression
analysis was performed to estimate the influence of genetic (CYP2B6 SNVs or haplotypes) and non-genetic covariates (sex, medication with CYP2B6 inducers or amoxicillin + clavulanic acid, chronic alcohol consumption) on CYP2B6 activity and on CYP2B6 mRNA expression (Table 4). Significant associations were observed between S-mephenytoin N-demethylation activity and the nucleotide substitution g.15631G > T (P = 0.034) or the CYP2B6 inducer therapy (P = 0.005). When the CYP2B6 haplotypes were involved in the analysis, the impact of none of the haplotypes containing g.15631 T was significant (g.-82 T/15631 T/18053G/25505 T P = 0.383; g.-82 T/15631 T/18053A/25505 T P = 0.427), whereas CYP2B6 activity-reducing non-genetic factors (chronic alcohol consumption and amoxicillin + clavulanic acid therapy) appeared to be associated with CYP2B6 activity (P = 0.050). Furthermore, hepatic CYP2B6 mRNA expression was found to be significantly associated with the nucleotide substitution g.15631G > T (P = 0.004) and also with g.18053A > G (P = 0.038). Involving CYP2B6 haplotypes in the multiple regression model, the association between CYP2B6 mRNA expression and the g.-82 T/15631 T/18053G/25505 T haplotype present in CYP2B6*6 allele became significant (P = 0.025). Both the CYP2B6 inducer therapy and the expression reducing non-genetic factors displayed significant association with hepatic CYP2B6 expression (P < 0.001 and P = 0.021, respectively). However, sex appeared to have no influence on either CYP2B6 activity or mRNA expression.

Discussion
Genetic variability of CYP2B6 has been reported to be associated with significant interindividual variations in pharmacokinetics of several clinically important drugs (antiretroviral, anticancer, antidepressant, antimalarial drugs) 58 . Moreover, preliminary pharmacogenetic testing is highly recommended for patients on efavirenz therapy for proper therapeutic efficacy and for limitation of adverse reactions 23 . Identification of SNVs in CYP2B6 gene and haplotype estimation constitute a major challenge, because for CYP2B6 genotyping, reliable and CYP2B6-selective assays are required that can distinguish CYP2B6 sequences from the highly homologous pseudogene CYP2B7P. TaqMan PCR assays offer accurate, sensitive, cost-efficient and fast SNV-discrimination method. Validated TaqMan CYP2B6 genotyping assays are commercially available for g.-82 T > C, g.15631G > T and g.25505C > T polymorphisms, but not for g.18053A > G, which is present in many allelic variants including CYP2B6*4 and the most frequent and widely studied CYP2B6*6; therefore, for identification of g.18053A > G, we have developed a novel, two-step genotyping assay. In the 119 liver samples, the frequencies of CYP2B6 alleles and genotypes in liver tissue donors were demonstrated to be similar to those in Caucasian populations (Table 2) (https:// www. pharm gkb. org/ page/ cyp2b 6RefM ateri als, access date: 24.01.2022) 8 . Although genetic polymorphisms of CYP2B6 can elucidate the substantial interindividual variability in CYP2B6 expression and activity to some extent, non-genetic factors can significantly modify the CYP2B6 phenotype predicted from genotype. The present study investigated the contribution of CYP2B6 genetic and non-genetic factors to CYP2B6-selective S-mephenytoin N-demethylation and CYP2B6 mRNA expression as well as the CYP2B6 genotype-phenotype mismatch in human liver tissues. S-Mephenytoin as the probe substrate and its N-demethylation reaction is frequently used for characterization of hepatic microsomal CYP2B6 activity 59 ; however, only a few studies have applied this CYP2B6-selective reaction in genotype-phenotype analysis 21,30,60 . In CYP2B6 expression analysis, the primer pair was designed to the exons 3 and 4 for the quantification of the full-length CYP2B6 mRNA. The strong association between CYP2B6 mRNA and S-mephenytoin N-demethylase activity in human liver tissues proved that the amplicon produced in the quantitative PCR was appropriate for the identification of the functional CYP2B6 mRNA and did not detect the truncated mRNA variant. Several in vitro and in vivo studies indicated that some CYP2B6 allelic variants may have substrate-specific effect on CYP2B6 function that further complicates the CYP2B6 phenotype estimation based on PharmVar www.nature.com/scientificreports/ criteria system 8,9 . PharmVar classification is appropriate for CYP2B6 genotype guided efavirenz therapy, whereas CYP2B6 pharmacogenetics appear to have an opposite impact on cyclophosphamide bioactivation. The most common CYP2B6*6 is associated with the expression of an mRNA variant lacking exons 4-6 due to aberrant splicing, and consequently with decreased hepatic activity in efavirenz and bupropion metabolism 18,19,22 . Efavirenz exposure has been reported to increase in patients with CYP2B6*1/*6 or CYP2B6*6/*6; therefore, substantial dose reduction has been recommended for better tolerability 15,23 . Although the presence of CYP2B6*6 allele appeared to have minor or negligible effect on bupropion exposure, hydroxylation of both enantiomers was lower in patients carrying CYP2B6*6/*6 17,61 . Contradictory results have been reported on metabolic activation of the prodrug cyclophosphamide in hepatic microsomes from CYP2B6*6 carriers 18,62,63 ; however, lower 4-hydroxy-cyclophosphamide production and worse treatment response to cyclophosphamide was observed in patients with CYP2B6*6 allele than in CYP2B6*6 non-carriers 64 . The g.18053A > G nucleotide change in CYP2B6*4 allelic variant has been reported to significantly alter the enzyme structure leading to a functionally different protein variant and to increased drug-metabolizing activity 12,13,[15][16][17]65 . The CYP2B6*4 allele contributed to increased activity in efavirenz 8-hydroxylation and to reduced plasma concentration of efavirenz in HIVinfected patients 15,18 ; however, dose modification was not required for efficient efavirenz therapy 23 . Intrinsic clearance of both bupropion enantiomers was minimally increased by CYP2B6*4, whereas pharmacokinetic studies demonstrated significantly high bupropion clearance in vivo and consequently high hydroxy-bupropion exposure in CYP2B6*4 carrier subjects 17,61,66 . Interestingly, CYP2B6*4 displayed lower cyclophosphamide 4-hydroxylation activity in vitro than CYP2B6*1 18,63 ; however, CYP2B6*1/*4 genotype appeared to have no impact on 4-hydroxy-cyclophosphamide formation in vivo comparing to CYP2B6*1/*1 64 . In the liver tissues of the present study, CYP2B6 mRNA expression was significantly associated with the g.15631C > T and g.18053A > G SNVs and even more with the g.-82 T/15631 T/18053G/25505 T haplotype designated as CYP2B6*6. S-Mephenytoin N-demethylase activity appeared to be significantly influenced by g.15631C > T, and marginally significant impact of g.18053A > G was demonstrated; however, the haplotype g.-82 T/15631 T/18053G/25505 T (CYP2B6*6) did not affect CYP2B6 activity at all. Besides CYP2B6 genetic variations, phenoconverting non-genetic factors, such as sex, age, co-medication and co-morbidities, have been considered to contribute to the interindividual variability in CYP2B6 activity and expression 9,29,58 . Al Koudsi et al. attributed 10% of variations in CYP2B6 protein expression to CYP2B6 genotype, 14% to gender and 21% to exposure to hepatic CYP inducers 32 . Several studies indicated that females displayed significantly higher CYP2B6 expression and activity than males 30,32,67 . It was explained by estradiol induction through an estrogen response element in the regulatory region of CYP2B6 gene 68,69 . However, other studies, including the present work, demonstrated no association between gender and CYP2B6 phenotypes 19,31,33,34 . Environmental non-genetic factors, such as CYP2B6-specific or non-specific medication and the consequences of chronic alcohol consumption are expected to contribute to the high interindividual variability in CYP2B6 expression and activity. Of less than 30% of liver tissue donors, CYP2B6 genetic variability influenced the CYP2B6 phenotype, whereas of more than 35% of tissue donors, non-genetic factors were reported in the medical history that significantly altered S-mephenytoin N-demethylase activity and/or CYP2B6 mRNA expression. In addition to genetic variations, the impact of both CYP2B6-selective inhibitors and inducers is highly recommended to be taken into account during CYP2B6 phenotype prediction 8,9,29 . The function of several CYP enzymes have been demonstrated to be inhibited by 1,4-dihydropyridine calcium-channel antagonists, including amlodipine that was found to strongly inhibit in vitro activities of CYP2B6 and CYP3A4; however, the clinical significance of the interaction between amlodipine and CYP2B6 might be minor because of the relatively high IC 50 values towards CYP2B6 substrates 49,70 . In two subjects, one with CYP2B6*1/*1, the other with CYP2B6*5/*6 genotype, and predicted to be 'normal' and 'intermediate' metabolizers, respectively, the low S-mephenytoin N-demethylation was attributed to the antihypertensive amlodipine therapy that might have transiently evoked poor CYP2B6 activity. Furthermore, the exposure to CYP2B6 inducers, including the antibiotics rifampicin, the corticosteroid derivative prednisolone, cortisone, hydrocortisone and dexamethasone, the benzodiazepine diazepam and midazolam, and the calcium channel blocker felodipine induces transcriptional expression of CYP2B6 gene via nuclear receptors (PXR, CAR) 25,50,52,53,55,57,71 . Rifampicin treatment has been demonstrated to substantially increase the clearance of bupropion and the formation of hydroxy-bupropion metabolite 72 . Furthermore, selective activation of CAR has been found to lead to increased bioactivation of cyclophosphamide in hepatocytes and to enhanced cytotoxicity in leukemia cells 73 . Metabolic activation of cyclophosphamide is primarily catalyzed by CYP2B6 with minor contribution of CYP3A4, whereas CYP3A4 is responsible for the inactivation pathway 74 . Since CAR preferentially mediates transcriptional induction of CYP2B6 over CYP3A4, CAR activation resulted in an increase in the active metabolite formation and in elevated antitumor activity 73 . In the liver tissue donors, the CYP2B6 inducer therapy in their medical history (dexamethasone, methylprednisolone, prednisolone, hydrocortisone, cortisone, midazolam, felodipine, diazepam) was significantly associated with increased CYP2B6 mRNA expression and activity. Particularly in those with one or two loss-of-function alleles predicted to be 'intermediate' or 'poor' metabolizers, high or high-intermediate activity and expression were observed. The loss-of-function alleles (CYP2B6*6 and CYP2B6*9) display some residual expression and activity; therefore, it was logically assumed that the exposure to a CYP2B6 inducer ameliorated the reduced function of CYP2B6 predicted from genotype. The clinical study by Loboz et al. involving healthy volunteers demonstrated that rifampicin induction increased bupropion clearance even in those carrying the loss-of-function CYP2B6*6 allele 72 . As a consequence of rifampicin treatment (PXR activation), hydroxy-bupropion formation was enhanced in subjects with CYP2B6*1/*6 as well as in CYP2B6*1/*1 carriers 72 . In the liver tissue donors of the present study, increased S-mephenytoin N-demethylation activity and CYP2B6 mRNA expression were associated with CYP2B6 inducer therapy, and 83% of the subjects exposed to CYP2B6 inducers (19/23) carried one or two copies of CYP2B6*6 allele; however, for the clear evidence for the CYP2B6*6-dependent susceptibility to induction, further study involving a large population is required. www.nature.com/scientificreports/ genotype-dependent susceptibility to rifampicin induction due to increased recruitment of PXR to the promoter region in g.-82C carriers 10 . Genotype-dependent susceptibility to CYP2B6 inhibitors has also been reported by Talakad et al. demonstrating an increase in inhibitory constant (K i ) values of the CYP2B6*4 and CYP2B6*6 variants with sertraline or clopidogrel compared to the wild-type enzyme 75 . However, CYP2B6*6 was found to be more susceptible to voriconazole inhibition than CYP2B6*1 76 , indicating inhibitor-dependent susceptibility of CYP2B6*6. Duration of phenoconversion and the recovery after CYP2B6 inhibition or induction have been reported to depend on the elimination rate of the inhibitor or inducer drugs and/or on enzyme turnover 77 . In the liver tissue donors, the putative time-course of drug-induced phenoconversion might have influenced the hepatic CYP2B6 activity and expression after discontinuation of the CYP2B6 inhibitor amlodipine or CYP2B6 inducer drugs. Considering the fact that the information on the drug therapy (both chronic and acute) applied prior the brain-death was recorded in the clinical histories, and the time of tissue explantation never exceeded 3 h, we assumed that no loss of altered enzyme activity or expression occurred in liver samples. Amlodipine and the CYP2B6 inducers are drugs with relatively long half-lives (12-60 h); furthermore, the de-induction and the recovery of CYP enzyme after inducer discontinuation has been calculated to require 3-7 days 78,79 ; therefore, we considered that phenoconversion evoked by these drugs still existed at the time of explantation. Furthermore, reduced CYP2B6 activity and/or expression in liver tissues was also associated with the nonspecific amoxicillin + clavulanic acid therapy and/or chronic alcohol consumption. Amoxicillin, effective against a wide range of bacterial infections, is often used in combination with clavulanic acid that prevents bacterial metabolism of amoxicillin 80 . Moderately severe hepatotoxic side effect of this combination has been reported; severe hepatic dysfunction however rarely occurs 56,81 . Chronic alcohol consumption is one of the major causes of liver diseases, the progression of which is explained by several pathological processes (e.g. inflammation, oxidative stress); however, the exact pathomechanism is not clearly understood 54 . No information is available about CYP2B6-selective inhibition or downregulation induced by amoxicillin + clavulanic acid or chronic alcohol consumption; however, hepatotoxicity and the associated inflammatory processes were assumed to influence CYP expression and function 82 . Clear evidence has been provided that the release of proinflammatory cytokines during inflammation downregulates both protein or mRNA expression of several CYPs, including CYP2B6 83,84 . IL-6 (interleukin 6) and IFNγ (interferon γ) proinflammatory cytokines have been demonstrated to downregulate the expression of CYP2B6 mRNA and protein as well as CYP2B6 activity 51,[85][86][87][88] . In the liver tissue donors with amoxicillin + clavulanic acid therapy and/or chronic alcohol consumption, decreased CYP2B6 expression and S-mephenytoin N-demethylase activity were attributed to a non-specific impact of inflammatory processes on CYP function rather than to a CYP2B6-selective suppression.
The present work indicated that CYP2B6 genetic polymorphisms influenced the expression and activity of CYP2B6 enzyme to some extent; however, the significance of phenoconverting non-genetic factors in enzyme function was comparable with that of genetic factors or even phenoconversion masked the effect of CYP2B6 allelic variants. Ing Lorenzini et al. evaluated the predictive values of CYP genotypes on CYP-mediated drug metabolism in patients, and observed relatively good CYP2B6 genotype-phenotype concordance for poor and rapid metabolizers (67% and 100%), but more variable for intermediate and normal metabolizers (0% and 38%) 89 . It should be noted that the number of patients involved in the interpretation of CYP2B6 genotype-phenotype concordance was limited (N = 36). In a retrospective study involving patients taking analgesic drugs for chronic pain, genetic variability and non-genetic factors influencing drug-metabolizing enzyme activities were associated with the occurrence of adverse drug reactions and/or non-response to the therapy in 40% and 28% of the cases 90 .
Some limitations of the present work should be discussed. First, we assessed the impact of CYP2B6 alleles most common in Caucasian populations, and some other, functionally relevant allelic variants were not identified (CYP2B6*7 or CYP2B6*12); however, their prevalence is low in Caucasian populations. Second, the medical history of some tissue donors may be assumed to be incomplete, and some information about relevant non-genetic factors was missing that might have influenced the interpretation of CYP2B6 phenotypes. Certain co-morbidities have been demonstrated to impact the function of drug-metabolizing enzymes; however, in the clinical histories of the tissue donors, the information of relevant pathological conditions was scarcely or incompletely recorded.
Although pharmacogenetic testing of drug-metabolizing CYP enzymes is an effective approach towards optimization and personalization of drug therapy 23,91 , the assessment of patients' drug-metabolizing capacity is far more complex than a simple prediction from the genotype. The present work has demonstrated that both CYP2B6 genetic and non-genetic variations were important to be taken into account in CYP2B6 phenotype interpretation. However, in approximately one third of the subjects, a CYP2B6 genotype-phenotype mismatch still existed. Identifying potential factors (both CYP2B6-specific and non-specific factors) in CYP2B6 phenotype variability and considering both genetic variations and non-genetic factors is a pressing requirement for appropriate elucidation of CYP2B6 genotype-phenotype mismatch that may improve prediction of pharmacokinetic variations and clinical outcome of a drug that is primarily or exclusively metabolized by CYP2B6 enzyme (e.g. cyclophosphamide, efavirenz, bupropion, ketamine, methadone). www.nature.com/scientificreports/