Liver Samples and Tissue Fractions.

Human liver cytosols were isolated from liver samples obtained from pediatric and adult donors (n = 158) and were acquired from five sources: University of Minnesota (n = 34), The Liver, Tissue, and Cell Distribution System at the University of Pittsburg (n = 40), Vitron (Tucson, AZ; n = 4), Xenotech (now Sekisui Xenotech, Kansas City, KS; n = 23), and University of Maryland-Baltimore Brain and Tissue Bank (UMB, n = 57). Samples obtained from Minnesota were flash frozen within 1 hour of donor death/ischemia. Samples from Pittsburg, Vitron, and Xenotech were from livers harvested for transplantation, placed on ice, and perfused with preservation solution. UMB samples were obtained upon autopsy with various post-mortem intervals up to 15 hours.

Chemicals and Reagents.

Naltrexone, 6-beta-naltrexol, naltrexone-d3 and 6-beta-naltrexol-d3 were purchased from Sigma Aldrich (St. Louis, MO). Methanol and formic acid were also purchased from Fisher Scientific (Waltham, MA). All other chemicals were reagent or higher grade and purchased from either Sigma Aldrich or Fisher Scientific.

Human Liver Cytosol In Vitro Incubation Reactions.

Cytosolic fractions were isolated from pediatric and adult human liver samples according to the method of Lu and Levin (Lu and Levin, 1978). Pooled human liver cytosol (n = 200, mixed gender, lot #1410229) was purchased from Xenotech (Kansas City, KS) and was used as an internal assay control (minimal interday variability %CV = 5.5). Incubations were performed in triplicate at 37°C in a shaking water bath for 30 minutes. In a total volume of 400 µl, incubations contained 100 mM phosphate buffer (pH 7.4), nicotinamide adenine dinucleotide phosphate generating system [consisting of nicotinamide adenine dinucleotide phosphate (NADP⁺ 1 mM), glucose-6- phosphate (G-6-P 5 mM) and glucose-6- phosphate dehydrogenase (G-6-PDH 1 unit/ml)], therapeutically relevant concentrations of naltrexone (0.1, 1 µM) and 0.05 mg/ml of cytosolic protein. Incubation conditions were chosen based on linear product formation achieved between 0 and 45 minutes at 0.05 mg/ml of protein. Reactions were stopped by the addition of an equal volume (400 µl) of ice-cold methanol containing the internal standards, 6-β-naltrexol-d3 and naltrexone-d3 final concentration 10 nM). The samples were gently vortexed and then subjected to centrifugation at 10,000 g × 10 minutes. An aliquot of the supernatant (600 µl) was transferred to a collection plate and evaporated to dryness using heated nitrogen displacement. Samples were reconstituted in 150 µl of 1:1 methanol-water and 10µl was injected onto ultra-performance liquid chromatography tandem mass spectrometry (UPLC-MS/MS).

Naltrexone biotransformation was determined by UPLC-MS/MS quantification of the primary metabolite, 6-beta-naltrexol (6βN), and 6βN formation rates (pmol/mg protein/min) were calculated. All incubations were performed in triplicate and accepted if %CV <15. Results are reported as the mean of triplicate incubations.

Human Microsome In Vitro Incubation Reactions.

Pooled human liver microsomes (n = 200, mixed gender, lot #1710084) were purchased from Xenotech (Kansas City, KS). To screen for potential contribution to naltrexone biotransformation, incubations were performed in triplicate at 37°C in a shaking water bath for 30 minutes. In a total volume of 400 µl, incubations contained 100 mM of phosphate buffer (pH 7.4), nicotinamide adenine dinucleotide phosphate generating system (NADP⁺ 1 mM, G-6-P 5 mM, G-6-PDH 1 unit/ml), concentrations of naltrexone (1, 10 µM) and 0.5 mg/ml of microsomal protein. Reactions were stopped by the addition of an equal volume (400 µl) of ice-cold methanol containing the internal standard, 6-β-naltrexol-d3 (final concentration 10 nM). The samples were gently vortexed and then subjected to centrifugation at 10,000 g × 10 minutes. An aliquot of the supernatant (600 µl) was transferred to a collection plate and evaporated to dryness using heated nitrogen displacement. Samples were reconstituted in 150 µl of 1:1 methanol-water, and 10µl was injected onto UPLC-MS/MS. naltrexone biotransformation was determined as above.

UPLC-MS/MS Analysis and Analytical Method Validation.

Analytes were detected using a Waters TQD triple quadrupole mass spectrometer coupled to a Waters H-class Acquity UPLC system. Naltrexone was monitored to ensure baseline resolution of structurally similar naltrexone and 6βN and to ensure there was no bleed through. 6βN concentrations were quantified as the focus on this work was metabolite formation. Analytes were separated on a Raptor Biphenyl analytical column (1.8µm, 2.1 × 100 mm) with a matching precolumn held at 60°C. The mobile phase was composed of 0.1% formic acid in water (A) and 0.1% formic acid in methanol (B). The liquid chromatography gradient proceeded from 20–40% B over 3 minutes at a flowrate of 0.3 mL/min. The total chromatographic method time was 6 minutes. Samples were ionized in positive mode with electrospray ionization parameters as follows: capillary voltage, 1.25 kV; sampling cone voltage, 40 V; source temperature, 150°C; desolvation temperature, 400°C; desolvation gas flow, 800 L/h; cone gas flow, 10 L/h The following analyte transitions were monitored: m/z 342.3 → 324.1 for naltrexone, m/z 344.3 → 326.1 for 6β-naltrexol, m/z 345.3 → 327.1 for naltrexone-D3, m/z 347.3 → 329.1 for 6β-naltrexol-D3. The collision energy for all transitions was 20 V. Deuterium‐labeled analytes served as internal standards. Calibration curves were constructed by plotting analyte peak area divided by the area of the internal standard versus the nominal concentration of the analyte. Calibration curves were linear over the range 0.1–100 nM. Quality control (QC) samples were prepared at four concentration levels: 0.1, 0.3, 3, and 80 nM. Method accuracy and precision were within acceptable limits according to the Food and Drug Administration guidelines for a bioanalytical assay (accuracy 85–115% of nominal and coefficient of variation [CV] <15%; lower limit of quantitation accuracy 80–120% of nominal and coefficient of variation [CV] <20%).

Genotype Analysis.

The presence of non-synonymous AKR1C4 coding region single-nucleotide variants (SNPs) (rs3829125 and rs17134592) was determined by Taq-Man genotype assays in all liver samples (n = 158). AKR1C4 SNPs were selected if they met the following search criteria: dbSNP mean allele frequency >5% and evidence of impact on drug biotransformation. AKR1C4 (K_cat/K_m 660 ±113 mL/µmol/min) demonstrates ∼20-fold and ∼800-fold higher catalytic efficiency than AKR1C2 (K_cat/K_m 34 ± 3 mL/µmol/min) and AKR1C1 (K_cat/K_m 0.84 ± 0.08 mL/µmol/min), respectively, in the biotransformation of naltrexone (Breyer-Pfaff and Nill, 2004). DNA samples obtained from Coriell were used as positive controls. All assays were performed in triplicate.

Statistical Analysis.

6βN rates of formation were log transformed to reduce skew. Descriptive statistics and Spearman’s rho were calculated in SPSS version 24 (IBM, Armonk, NY), and statistical modeling was carried out in R (R Core Team, Vienna, Austria).

We initially assessed the potential for source-dependent differences in activity. UMB samples were noted to have consistently lower cytosolic enzyme activity assessed by 6βN formation rate compared with non-UMB samples, and recently published data by our group found reduced microsomal enzyme activity in UMB samples along with reduced tissue quality based on metrics like RNA quality and homogenate POR activity (Leeder et al., In Press). Values for these quality metrics were comparable among the non-UMB samples. Because of these source-specific differences between UMB and non-UMB samples, these cohorts were analyzed separately.

We examined the effects of age and S145C/L311V genotype on naltrexone biotransformation separately, using descriptive statistics and plots and together in a multiple regression model. Statistical modeling was limited to samples from donors through age 18 (n = 73 non-UMB, n = 51 UMB) because all non-UMB samples from donors aged 19 and above were from a single source (Pittsburgh). Source and age effects in the non-UMB data were disentangled for modeling as follows. First, we trimmed the data from each source to create a smaller data set (n = 59) in which the mean age was nearly the same (9.8–10.0 years) for each source. Next, we estimated source effects by fitting a linear model to these trimmed data, modeling log-6βN formation rate as a function of source, controlling for S145C/L311V genotype and a linear age effect. Finally, the resulting source effect estimates were used to adjust each log-6βN value in the untrimmed data set to remove, as much as possible, variability due to source-specific factors.

Using this source-adjusted data set, we used weighted ordinary least squares regression to model log-6βN as a function of age, sex, and S145C/L311V genotype. To allow for a non-linear effect of age, age was centered at 9 years (near its mean) and squared, and this quadratic term was included as an additional explanatory variable. Observations were weighted by the precision of their source effect estimate in the trimmed model. We explored natural spline models for age with 2–3 knots, but these provided negligible additional explanatory value. The same model (allowing for a quadratic effect of age) was fit to the UMB data for comparison.