Methadone dose in heroin-dependent patients: role of clinical factors, comedications, genetic polymorphisms and enzyme activity

Abstract

Aims

Methadone is characterized by wide intersubject variability regarding the dose needed to obtain full therapeutic response. We assessed the influence of sociodemographic, ethnic, clinical, metabolic and genotypic variables on methadone maintenance dose requirement in opioid-dependent responder patients.

Methods

Eighty-one stable patients (60 men and 21 women, 43.7 ± 8.1 years old, 63.1 ± 50.9 mg day⁻¹ methadone), divided into quartiles with respect to the median daily dose, were enrolled and underwent clinical examination, treatment history and determination of liver/intestinal cytochrome P450 (CYP) 3A4 activity measured by the midazolam test, R,S-methadone trough concentration and clinically significant polymorphisms of the OPRM1, DRD2, COMT, ABCB1, CYP2B6, CYP3A5, CYP2C19 and CYP2D6 genes.

Results

Methadone maintenance dose was correlated to the highest dose ever used (r² = 0.57, P < 0.0001). Fractioned methadone intake (odds ratio 4.87, 95% confidence interval 1.27–18.6, P = 0.02), bodyweight (odds ratio 1.57, 95% confidence interval 1.01–2.44, P = 0.04), history of cocaine dependence (80 vs. 44 mg day⁻¹ in never-addict patients, P = 0.005) and ethnicity (Asian > Caucasian > African, P = 0.04) were independently associated with high-dose methadone in multiple regression analysis. A modest correlation was observed between liver/intestinal CYP3A4 activity and methadone dose at steady state (Spearman rank correlation coefficient [r_s] = 0.21, P = 0.06) but not with highest dose ever used (r_s = 0.15, P = 0.18) or dose-normalized R,S-methadone trough concentrations (r_s = −0.05, P = 0.64). Concomitant CYP3A4 inhibitors only affected the relationship between methadone dose and R,S-methadone trough concentration. None of the genetic polymorphisms explored was predictive of the methadone maintenance dose.

Conclusions

Methadone maintenance dose was predicted by sociodemographic and clinical variables rather than genetic polymorphisms or liver/intestinal CYP3A4 activity in stable patients.

What is Already Known about this Subject

• Overall, 30–80% of patients on methadone maintenance treatment are still receiving doses that are too low to be effective and experience withdrawal symptoms or decreased methadone efficacy for part of the dosing interval and/or have persistent heroin use or dropouts because of relapses.

• The role of several genetic polymorphisms, including the CYP2B6, CYP2C19, CYP2D6 and MDR1 genotypes, on steady-state concentrations of methadone enantiomers was also controversial in clinical studies.

• These studies did not simultaneously address the influence of sociodemographic and clinical variables and concomitant medications on methadone maintenance treatment with a multivariate approach.

What this Study Adds

• Fractioned methadone intake, bodyweight, history of cocaine dependence and ethnicity were independently associated with high-dose methadone in multiple regression analysis.

• A modest correlation was observed between liver/intestinal CYP3A4 activity and methadone dose at steady state, but concomitant CYP3A4 inhibitors only affected the relationship between methadone dose and R,S-methadone trough concentration.

• None of the genetic polymorphisms explored was predictive of the methadone maintenance dose in the present French cohort.

Introduction

Methadone is a full agonist of the μ-opioid receptor used to treat opioid-dependent patients 1,2. Despite its therapeutic use for more than 60 years, with more than a million people treated worldwide, understanding of methadone pharmacokinetics, disposition and pharmacodynamics is surprisingly incomplete, but extensive inter- and intra-individual variability clearly exists 1.

Successful treatment that prevents opiate use, withdrawal and craving relies in part on individual dose optimization. Knowledge of the factors determining the methadone dose/plasma concentration–effect relationship, hence the optimal dose at steady state, is hard to target or anticipate. Even in programmes with liberal dosing policies, 30–80% of patients under methadone maintenance treatment (MMT) are still receiving doses that are too low to be effective and experience withdrawal symptoms or decreased methadone efficacy for part of the dosing interval and/or have persistent heroin use or dropouts because of relapses 3–5. Methadone dose is highly variable among countries, reaching 200 mg day⁻¹ or more in the USA, while ranging between 60 and 100 mg day⁻¹ in previous French cohort studies 5,6. For example, in order to obtain the recommended methadone plasma concentration of 250 ng ml⁻¹, doses as low as 55 mg day⁻¹ or as high as 921 mg day⁻¹ can be required in a 70 kg patient without any comedication 7. Hence, the optimal dosage is usually the result of several months of dose adjustment, and clinicians are in need of predictive biomarkers of individual target dose.

Methadone is administered as a racemic mixture of (R)- and (S)-methadone enantiomers. The (R)-methadone is the active enantiomer at the μ-opioid receptor 1,2. Oral methadone is rapidly absorbed, with peak plasma concentrations reached in 2–4 h and half-life in humans ranging from 16 to 28 h 2. It is metabolized in the liver and small intestine by N-demethylation, leading to an inactive major metabolite, 2-ethylidene-1,5-dimethyl-3,3-diphenylpyrrolidine, which can subsequently be converted by N-demethylation to the inactive metabolite 2-ethyl-5-methyl-3,3-diphenylpyrroline 2. Methadone metabolism in humans is attributed primarily to the cytochrome P450 (CYP) enzyme 2B6 and, to a lesser extent, CYP3A4/5, CYP2D6 and CYP2C19 2,4,8. Methadone is also a substrate of P-glycoprotein (P-gp), the multidrug resistance 1 (ABCB1) gene product, which shows a weak stereoselectivity towards the (S)-enantiomer 4. The association between small intestinal and blood–brain barrier P-gp activity and methadone concentrations and pharmacodynamics in vitro, in mice and healthy volunteers still remains controversial 9–12. Methadone may be subject to several clinically relevant drug–drug interactions with concomitant medications known to induce or inhibit CYP2B6, CYP2D6, CYP3A4 and P-gp that may affect methadone concentrations and clinical efficacy 13,14. It is also known from clinical practice that patients receiving MMT still use alcohol and various psychotropic drugs that may have an impact on methadone disposition, but this is rarely evaluated in clinical trials.

Patients' polymorphisms in genes encoding for methadone metabolizing enzymes and transporters have been actively studied, but the clinical relevance in MMT outcome is still unclear. The recent study by Fonseca et al. failed to find an influence of allelic variants of genes encoding CYP3A5, CYP2D6, CYP2C9, CYP2C19 and P-gp in 105 opioid-dependent Caucasians, except for CYP2D6 metabolizing phenotype displaying higher dose requirements in the ultrarapid metabolizers and more frequent withdrawal symptoms 4. The role of the CYP2B6 genotype on steady-state concentrations of methadone enantiomers was also controversial in clinical studies 14–16. Conversely, ABCB1 genotype frequencies were significantly different between the ‘higher’ (>150 mg day⁻¹) and ‘lower’ (<150 mg day⁻¹) methadone dose groups for the c.1236C>T (rs1128503, rs1045642 and rs2032582) single nucleotide polymorphisms (SNPs) in 98 Israeli drug-free patients taking a stable methadone dose and no other treatment 17.

These studies did not simultaneously address the influence of sociodemographic and clinical variables and concomitant medications on methadone maintenance treatment and methadone trough plasma concentrations with a multivariate approach. Indeed, clinically significant interactions may occur when methadone is taken concomitantly with other drugs and lead to precipitation of withdrawal symptoms, relapse in the use of heroin or increase in maintenance dose 13. We therefore aimed to assess the influence of sociodemographic and clinical factors, including the use of concomitant medications, liver and intestinal CYP3A4 activity and clinically relevant genetic polymorphisms, on methadone dose requirement in ‘real-life’ opioid-dependent responder patients taking a stable dose for at least 3 months. We hypothesized that, taken together in a multivariate model, clinical variables and comedications would be more predictive of methadone maintenance dose than the expected, relevant CYP2B6, CYP3A5 and MDR1 genetic polymorphisms identified in selected patients from clinical studies 2,4,14–17.

Methods

Study design and ethics statement

METHADOSE was a French prospective multicentre study, in which heroin-dependent patients, aged 18 years and older, receiving a MMT, were enrolled at the Therapeutic Research Unit at Lariboisière Hospital, Paris, France. Patients were considered to be taking MMT if they received a stable methadone dose, had no clinical symptom of heroin overdose or withdrawal, no intravenous drug abuse, no opiate or illicit drug intake as confirmed by urine testing and no cocaine, benzodiazepine or alcohol dependence as defined by the DSM-IV criteria for the last 12 months. The overnight-fasted patients were admitted on the morning of the study day and underwent urinalysis to detect illicit drug or opiate abuse. After complete clinical examination, during which medical history and concomitant medications were recorded, patients received 2 mg of oral midazolam (MDZ) in solution and their usual methadone dose. Blood samples were taken before and 30 min and 4 h after midazolam and methadone intake. Written informed consent was obtained from each subject. The study was approved by the appropriate ethics committee (Ref #2008/227, CPP Île-de-France VI, Pitié-Salpétrière), registered on the Agence Nationale de Sécurité du Médicament (ANSM, EudraCT #2007–007662-37) and on the clinicaltrials.gov website (NCT00894452).

Genotyping

DNA was extracted from whole blood with a Promega Maxwell 16 extractor, as recommended by the manufacturer (Promega France, Charbonnières-les-Bains, France). All patients were genotyped for polymorphisms of the following genes (OPRM1, c.118A>G rs1799971; DRD2 Taq1 A rs1800497/ANKK1; DRD2 c.957C>T rs6277; COMT p.Val158Met rs4680; ABCB1 c.3435C>T rs10456; CYP2B6*4 rs2279343, CYP2B6*6 rs3745274; CYP3A5*3 rs776746; and CYP2C19*2 rs11188072 and *17 rs12248560) by real-time PCR (StepOne Plus; Applied Biosystems, Saint-Aubin, France) with predesigned kits (Life Technologies, Saint-Aubin, France). For CYP2D6 Copy Number Variants (CNV) determination, real-time PCR was used using relative quantification as recommended by the manufacturer, using a predesigned kit (Life Technologies) and RNAse for standardization.

Methadone assay

Enantioselective separations of (R)- and (S)-methadone were performed on a liquid chromatography mass detection system (TSQ Quantum Ultra; Fisher Scientific, Illkirsch, France) using a chiral stationary phase Chiralcel® OJ-RH (150 mm × 2.1 mm, 5 µm) with a guard column (10 mm × 4.6 mm, 5 µm; Daicel Chemical Industries, Tokyo, Japan). The mobile phase consisted of acetonitrile and 0.02% triethylamine buffer (35:65, v/v) with an isocratic mode. The flow rate was 0.25 µl min⁻¹, the injection volume was 10 µl and the column temperature was kept at 25°C. Data were acquired in the positive ion mode with an electrospray (ESI) source. Multiple reaction monitoring (MRM) was used for data collection. For methadone, transitions 310.18/265.02 (collision energy 15 V) for quantification and 310.18/105.00 (24 V) for confirmation were applied, and 313.18/268.03 (15 V) and 313.18/104.96 (28 V) were selected for methadone-d3 Internal Standard (IS). Plasma samples (100 µl) were treated 1:2 (v/v) with the IS solution in acetonitrile in order to precipitate the proteins. After centrifugation and evaporation, the mobile phase (100 µl) was added to residues before injection. Calibration was carried out daily based on a six-point curve between 2 and 500 ng ml⁻¹. The external control was used at 115.3 ng ml⁻¹ (ACQ Science, Rottenburg-Hailfingen, Germany). The limit of quantification was b1 ng ml⁻¹ for each enantiomer. Precision and accuracy were estimated from the analysis of homemade controls (1 and 200 ng ml⁻¹). Inter- and intra-assay precision was below 8.5%, and accuracies ranged from 82.8 to 104.7% for the two enantiomers. No matrix effect was observed.

Midazolam assay

Midazolam and its metabolites, 1-OH (1-OH-MDZ) and 4-OH midazolam (4-OH-MDZ), were separated and quantified on a liquid chromatography mass detection system (TSQ Quantum Ultra; ThermoFisher, Illkirsch, France) using an Atlantis DC18 (100 mm × 2.1 mm, 3 µm) column (Waters, Guyancourt, France). Mobile phases consisted of water/methanol (30/70% v/v) with formic acid (0.1% v/v), and 2 m m ammonium acetate for phase A and water/methanol (70/30 % v/v) with formic acid (0.1% v/v) and 2 m m ammonium acetate for phase B. Elution of MDZ and its metabolites was performed using gradient phases starting with 40% of phase B and finishing with 95% of phase B at 4 min. The flow rate was 0.45 ml min⁻¹, and the column temperature was kept at 25°C. Midazolam-d4 was used as internal standard (EI). Data were acquired in the positive ion mode with an electrospray (ESI) source. Multiple reaction monitoring (MRM) was used for data collection. Transitions selected for quantification were 326.1/243.97 for MDZ, 342.2/203.18 for 1-OH-MDZ, 342.2/234.18 for 4-OH-MDZ and 330.0/247.97 for midazolam-d4. Plasma calibrations were based on six-point curves between 3.1 and 50, 0.31 and 5 and 0.16 and 2.5 ng ml⁻¹ for MDZ, 1-OH-MDZ and 4-OH-MDZ, respectively. Homemade controls were prepared at two levels, 8 and 35, 0.8 and 3.5 and 0.4 and 1.7 ng ml⁻¹ for MDZ, 1-OH-MDZ and 4-OH-MDZ, respectively. Plasma (100 µl; calibration, controls or samples) was precipitated by acetonitrile containing EI (500 ng ml⁻¹) to precipitate proteins. After centrifugation, decantation and evaporation at room temperature, dry residues were dissolved in 150 µl of mobile phase A, and 15 µl was injected into the system. Limits of quantifications were 3.1, 0.31 and 0.16 ng ml⁻¹ for MDZ, 1-OH-MDZ and 4-OH-MDZ, respectively. Precision and accuracy (bias) were estimated from the analysis of homemade controls. Inter- and intra-assay precision was below 14.6%, and accuracies ranged from 82 to 108% for any analytes. No matrix effects were observed. The 1-OH and 4-OH-MDZ/MDZ ratios were determined at 30 min and 4 h after midazolam administration, in order to determine liver and intestinal CYP3A4 activity, as previously described 18.

Statistical analysis

As a clinician-oriented approach, our primary aim was to determine the interindividual variability in methadone maintenance dose and the effect of sociodemographic, clinical, genetic and phenotypic variables on this variability tested by means of simple and multiple regression analysis. However, as methadone dose explains <50% of R-methadone plasma concentrations at steady state even in patients not receiving concomitant medications 19, we also used R- and R,S-methadone as dependent variables in two subsequent multiple regression analyses in order to evaluate the contribution of any available covariates. A sample size of at least 80 patients was calculated to detect any clinically relevant correlation, as defined by a correlation coefficient of 0.27 or higher with a significance level (α) of 0.05 and a power of 80%. Descriptive statistics were presented as means ± SD (quantitative variable) and absolute and relative frequencies (categorical variable), as appropriate. Due to the wide variability observed in methadone maintenance dose, subjects were enrolled and then divided into quartiles with respect to the median daily dose, as follows: Q1 = 20 mg day⁻¹ (range 5–30), Q2 = 40 mg day⁻¹ (range 30–55), Q3 = 60 mg day⁻¹ (range 60–80) and Q4 = 105 mg day⁻¹ (range 80–320). Differences in demographic and clinical characteristics between quartiles were examined using anova followed by Fisher's least significant difference adjusted for multiple comparisons, the non-parametric Kruskal–Wallis or Mann–Whitney U tests, as appropriate. Differences in genotype and phenotype frequencies were assessed by the χ² or Fisher's exact tests, as appropriate. Holm's correction was used for multiple comparisons. Simple and multiple regression analyses were performed to determine whether sociodemographic, clinical, genotypic or phenotypic variables were independently predictive of methadone maintenance dose and, subsequently, R- and R,S-methadone trough plasma concentrations. Each independent variable significantly correlated to the dose was then tested by means of a polytomous (multinomial) logistic regression model, and the relationship was expressed by the odds ratio (OR) and 95% confidence interval (95% CI). Statistical analyses were implemented in Statview v9.0 (SAS Institute, Cary, NC, USA). A P value of 0.05 or less was considered statistically significant.

Results

Demographics

Among the 81 patients enrolled [60 men (74.1%), 21 women (25.9%), sex ratio, 3:1], 21 received a median maintenance dose of 20 mg day⁻¹ (Q1), 20 patients were treated with 40 mg day⁻¹ (Q2), 20 with 60 mg day⁻¹ (Q3) and 20 with 105 mg day⁻¹ (Q4). In this last subset, 15 patients received 120 mg day⁻¹ of methadone or less and five patients received >120 mg day⁻¹ of methadone (mean daily dose 214 ± 67.7 mg day⁻¹). Seventy-five patients (92.6%) were smoking (mean daily cigarettes smoking 16.8 ± 8.7, median 17, range 3–60); 50 patients (61.7%) smoked cannabis at the time of the study. Sixty-nine patients (85.2%) were Caucasians, eight (9.9%) were Africans and four (4.9%) were Asians. Demographic and clinical characteristics, with respect to each quartile, are displayed in Table1. There was a significant correlation between methadone maintenance dose received at the time of the study and the highest methadone dose ever used since the patient entered the programme (r² = 0.57, P < 0.0001; Table1). Twenty-two patients (27.2%) had occasional heroin use and were evenly distributed across the dose quartiles (P = 0.56; Table1), 31 (38.3%) had a past history of cocaine dependence, 43 (53.1%) had a past history of alcohol abuse and 38 (46.9%) of alcohol dependence.

Table 1.

Demographic and clinical characteristics of the METHADOSE cohort (mean ± SD or n and percentage, as appropriate)

Characteristic	Q1* (n =21)	Q2* (n =20)	Q3* (n =20)	Q4* (n =20)	Total* (n =81)
Age (years)	44.9 ± 9.3	43.9 ± 7.1	40.8 ± 7.5	45.2 ± 7.9	43.7 ± 8.1
Age of opioid addiction start (years)	20.5 ± 4.7	21.8 ± 4.7	21.0 ± 5.8	18.9 ± 3	20.6 ± 4.7
Current dosing regimen (mg day−1)†	20.2 ± 8	41 ± 8.4	65.5 ± 6.9	128 ± 61	63.1 ± 50.9
Age of methadone start (years)	31.6 ± 7.7	33.3 ± 8.1	32.8 ± 7.3	33.5 ± 7.1	32.8 ± 7.5
Time in programme (years)	8.9 ± 5.5	6.3 ± 4.3	5.8 ± 4.3	9.4 ± 6.9	7.6 ± 5.4
Time since last dose intake (h)	22.5 ± 6.3	22.2 ± 4.9	23 ± 5.8	20.3 ± 6.2	22 ± 5.8
Highest methadone dose (mg day−1)†	49.8 ± 12.4	73 ± 8	93 ± 11.7	170.5 ± 58.8	97.5 ± 55.4
Occasional heroin use [n (%)]	7 (33)	4 (20)	7 (35)	4 (20)	22 (27.2)
Bodyweight (kg)	67.5 ± 12.7	72.6 ± 15	75.7 ± 17	77.4 ± 17.2	73.2 ± 15.7
Patients with fractioned dose [n (%)]	4 (19)	2 (10)	2 (10)	9 (45)‡	17 (21)
Sex (men/women)	14/7	14/6	16/4	16/4	60/21
Human immunodeficiency virus-positive patients [n (%)]	2 (9.5)	1 (5)	1 (5)	6 (30)	10 (12.3)
HCV-positive patients [n (%)]	10 (47.6)	11 (55)	12 (60)	13 (68.4)	46 (57.5)
Cigarette smoking [n (%)]	19 (90.5)	19 (95)	18 (90)	19 (95)	75 (92.6)
Caucasians/others (n)§	17/4	16/4	19/1	17/3	69/12

Abbreviation is as follows: HCV, Hepatitis C virus.

^*Q1 (median 20 mg day⁻¹, range 5–30), Q2 (median 40 mg day⁻¹, range 30–55), Q3 (median 60 mg day⁻¹, range 60–80), Q4 (median 105 mg day⁻¹, range 80–320) and total population (median 55 mg day⁻¹, range 5–320).

^†P < 0.03 between the respective quartiles, anova and Fisher's least significant difference.

^‡P = 0.02 vs. Q1, Q2, Q3 and the total population, Fisher's exact test.

^§P = 0.03, Fisher's exact test.

Effect of demographic or clinical variables on methadone maintenance dose and R,S-methadone trough plasma concentrations

Although the time in programme tended to be longer in Q1 and Q4 in comparison to Q2 or Q3, this trend was not significant (P = 0.08; Table1). Likewise, age at the start of opioid addiction and age at the start of methadone were not predictive of methadone dose (P > 0.2). Seventeen patients (21%), including 45% of those belonging to the highest dose quartile (nine of 20) were fractioning their dose. Although patients taking a fractioned methadone dose had a higher dose than those taking a single methadone dose, the difference was not statistically significant (79.1 ± 49 vs. 58.9 ± 50.9 mg day⁻¹, P = 0.07). Likewise, methadone dose did not differ between men and women (66.2 ± 52.5 vs. 54.4 ± 45.9 mg day⁻¹, respectively, P = 0.18). In this cohort, human immunodeficiency virus (HIV)-positive patients [n = 10 (12.3%); Table1] were treated with almost twice as much methadone as HIV-negative patients (109.5 ± 99 vs. 56.6 ± 36.7 mg day⁻¹). However, probably due to the small number of HIV-infected patients, the difference did not reach statistical significance (P = 0.09). More than half the patients were Hepatitis C virus (HCV)-positive [n = 46 (57.5%); Table1]. The methadone dose was similar between HCV-positive and HCV-negative patients (69.9 ± 59 vs. 52.6 ± 35.9 mg day⁻¹, P = 0.17). There was only a trend towards higher methadone maintenance dose in the 69 Caucasians compared with the eight Africans (64.5 ± 51.7 vs. 35.9 ± 18.3 mg day⁻¹, P = 0.06). A blood sample for R-, S- and R,S-methadone plasma trough concentration was taken 22 h after last dose intake (median 24 h, range 4.4–29.8 h), and time elapsed between the last methadone dose and blood sampling did not differ between the respective quartiles (Table1). The R-, S- and R,S-methadone plasma trough concentration increased significantly and proportionally with the dose, except between Q3 and Q4, where the dose almost doubled while the trough concentrations increased by only <25% (r² = 0.29, P < 0.0001; Table2). As expected, there was a significant correlation between R- (the active methadone enantiomer at the μ-opioid receptor) and S-methadone and between R- and R,S-methadone trough concentrations (r_s = 0.80 and 0.95, respectively, P < 0.0001). However, R- and R,S-methadone trough concentrations displayed only a small but significant correlation with methadone dose (r_s = 0.48 and 0.52, respectively, P < 0.001 for the respective correlations), as previously observed 19.

Table 2.

Mean (±SD) R-, S- and R,S-methadone trough concentrations and midazolam test in the METHADOSE cohort

	Q1* (n =21)	Q2* (n =20)	Q3* (n =20)	Q4* (n =20)	Total* (n =81)
Dosing regimen (mg day−1)†	20.2 ± 8	41 ± 8.4	65.5 ± 6.9	128 ± 61	63.1 ± 50.9
R-Methadone (ng ml−1)‡	30.5 ± 24.5	56.5 ± 26	80.4 ± 31.1	109.8 ± 51.6	68.8 ± 44.9
S-Methadone (ng ml−1)‡	28.9 ± 26.9	56 ± 32.2	74.1 ± 36.1	90.9 ± 53.6	62.1 ± 44
R,S-Methadone (ng ml−1)‡	59.4 ± 50.6	112.5 ± 56.3	154.5 ± 66.4	200.7 ± 99.8	130.9 ± 86.6
T0.5h 1-OH-midazolam/midazolam ratio	0.92 ± 0.59	0.95 ± 0.81	1.06 ± 0.48	1.13 ± 1.65	1.01 ± 0.97
T4h 1-OH-midazolam/midazolam ratio§	1.2 ± 0.68	1.33 ± 0.97	1.96 ± 1.54	6.12 ± 15.65	2.64 ± 7.95

^*Q1 (median 20 mg day⁻¹, range 5–30), Q2 (median 40 mg day⁻¹, range 30–55), Q3 (median 60 mg day⁻¹, range 60–80), Q4 (median 105 mg day⁻¹, range 80–320) and total population (median 55 mg day⁻¹, range 5–320).

^†P < 0.03 between the respective quartiles, anova and Fisher's least significant difference.

^‡P < 0.0001 between the respective quartiles, anova and Fisher's least significant difference.

^§P < 0.04 (Q4 vs. Q1 and Q2), P = 0.07 (Q4 vs. Q3), anova and Fisher's least significant difference. Blood samples were taken before (T₀), 30 min (T_0.5h) and 4 h (T_4h) after midazolam and methadone intake.

Logistic regression analysis incorporating all sociodemographic and clinical variables revealed that fractioned methadone intake (P = 0.09), bodyweight (P = 0.02), history of cocaine dependence (P = 0.002) and ethnicity (P = 0.01) were independently associated with a high dose of methadone. Using polytomous logistic regression analysis, each 10 kg increase in bodyweight was associated with a 57% @increase in the probability of receiving high-dose methadone at steady state (OR 1.57, 95% CI 1.01–2.44, P = 0.02). Fractioned dosing was associated with a methadone dose >60 mg day⁻¹ at steady state (OR 3.48, 95% CI 0.86–14.1, P = 0.09), and a history of cocaine dependence was predictive of a high steady-state methadone dose (OR 4.87, 95% CI 1.27–18.6, P = 0.002) in comparison to those taking a lower dose. Patients who never had DSM-IV criteria for cocaine dependence received 44 mg day⁻¹ methadone at steady state vs. 80 mg day⁻¹ in those who were former cocaine addicts (P = 0.01).

Further univariate analysis showed that R-methadone trough concentrations decreased when the time since the last dose increased (P = 0.005) and that R-methadone trough concentrations increased with bodyweight (P = 0.018). Multivariate analysis showed that these two variables were independent predictors and explained 15% of the variability of R-methadone trough concentrations in the present study (r² = 0.15, P = 0.014). The time since the last dose of methadone and bodyweight were also the two independent variables explaining 20% of the variability of R,S-methadone trough concentrations (r² = 0.20, P = 0.005).

Influence of liver/intestinal CYP3A activity on methadone maintenance dose

Liver and intestinal CYP3A4 activity, as measured by the 1-OH-MDZ/MDZ ratio at 0.5 and 4 h, was highly variable, especially in patients treated with the highest methadone dose (i.e. Q4; Table2). The T_0.5h 1-OH-MDZ/MDZ ratio ranged between 0.01 and 7.38 and the T_4h 1-OH-MDZ/MDZ ratio ranged between 0.02 and 65.52. One patient displaying the lowest 1-OH-MDZ/MDZ ratio received 15 mg day⁻¹ methadone, while the patient displaying the highest metabolic ratio was treated with 190 mg day⁻¹ methadone. Another HIV- and HCV-positive patient had a T_4h 1-OH-MDZ/MDZ ratio of 32.50 and received the highest dose of methadone observed in the present cohort, i.e. 320 mg day⁻¹. When these outliers were removed from the analysis, the mean T_0.5h and T_4h 1-OH-MDZ/MDZ ratio dropped down to 0.90 ± 0.61 and 1.46 ± 1.13, respectively. Overall, a modest correlation was observed between liver/intestinal CYP3A4 activity, measured by the T_4h 1-OH-MDZ/MDZ ratio and methadone dose (r_s = 0.21, P = 0.06), but not with the highest dose ever used (r_s = 0.15, P = 0.18) or the dose-normalized R,S-methadone trough concentrations (r_s = −0.05, P = 0.64). Similar results were observed when correlations were tested with the T_0.5h 1-OH-MDZ/MDZ ratio, T_0.5h and T_4h 4-OH-MDZ/MDZ ratio and T_0.5h and T_4h 1- + 4-OH-MDZ/MDZ ratio, respectively (P > 0.1 for the respective correlations, data not shown).

Effect of concomitant medications on methadone maintenance dose or R,S-methadone trough concentrations

Fifty-seven per cent of patients (46 of 81) had used sedatives for the last 12 months, and 44.4% (36 patients) met the lifetime criteria for sedative dependence. Thirty-seven per cent (30 of 81) of patients were prescribed benzodiazepines, 22.2% (18 of 81) neuroleptic drugs, 3.7% (three of 81) efavirenz and nevirapine (well-known liver CYP3A4 inducers), 8.6% (seven of 81) nelfinavir and ritonavir-boosted HIV protease inhibitor (well-known liver/intestinal CYP3A4 inhibitors), all of whom were infected with HIV, and 11.1% (nine of 81) received antidepressants known to interact with liver CYP2D6. In patients receiving nelfinavir or ritonavir-boosted HIV-protease inhibitor, the T_4h 1-OH-MDZ/MDZ ratio was significantly decreased vs. those who did not (0.07 ± 0.05 vs. 2.88 ± 8.28, P < 0.0001). Similar results were obtained with the T_0.5h 1-OH-MDZ/MDZ ratio, T_0.5h and T_4h 4-OH-MDZ/MDZ ratio and T_0.5h and T_4h 1- + 4-OH-MDZ/MDZ ratio, consistent with a significant inhibition of CYP3A activity by nelfinavir or ritonavir-boosted HIV-protease inhibitor (not shown).

Mean R,S-methadone concentrations did not differ between patients taking CYP3A4 inducers and those who did not (174.5 ± 93.5 vs. 129.2 ± 86.5 µg l⁻¹, P = 0.37, Mann–Whitney U test). Likewise, mean R,S-methadone concentrations did not differ between patients taking CYP3A4 inhibitors and those who did not (92.7 ± 64 vs. 134.6 ± 87.9 µg l⁻¹, P = 0.32, Mann–Whitney U test). The highest methadone dose ever used was also similar between patients taking CYP3A4 inducers or inhibitors and those who did not (156.7 ± 141.5 vs. 95.2 ± 50.3 mg day⁻¹, P = 0.56 and 117.1 ± 64.7 vs. 95.6 ± 54.6 mg day⁻¹, P = 0.34, respectively). Surprisingly, concomitant liver/intestinal CYP3A4 inhibitors significantly decreased the slope of the relationship between methadone dosing regimen and R,S-methadone trough concentration (0.679 vs. 1.735, P = 0.011; Figure1), i.e. for the same methadone maintenance dose, patients taking concomitant liver/intestinal CYP3A4 inhibitors displayed lower R,S-methadone trough concentrations than those without concomitant medications known to decrease liver/intestinal CYP3A4 activity.

Figure 1.

Relationship between current methadone maintenance dose and R,S-methadone trough concentration, with respect to the concomitant intake of liver/intestinal CYP3A4 inhibitors. One patient treated with the highest methadone dose (i.e. 320 mg day⁻¹) concomitantly received liver/intestinal CYP3A4 inducer. •, with CYP3A4 inhibitor (a); ○, without CYP3A4 inhibitor (b);, with CYP3A4 inducer

Association between methadone maintenance dose and genotypes for polymorphisms in genes encoding various enzymes and transporters involved in methadone clearance

Table3 displays the allelic frequencies of the respective genetic polymorphisms studied and the effect of each polymorphism on methadone dosing regimen. None of the genetic polymorphisms explored was significantly predictive of methadone maintenance dose. Given the significant correlation between dose and, respectively, highest methadone dose ever used and R,S-methadone trough concentration, it was not surprising to find similar results when trying to correlate these two variables with the respective genotypes tested (P > 0.29 for the respective comparisons, data not shown).

Table 3.

Association between methadone maintenance dose and genotypes for polymorphisms in genes encoding various enzymes and transporters involved in methadone clearance

Variant	SNP alleles(A/a)	Genotypes	n (%)*	Mean(±SD)methadone dose(mg day−1)	P Value†
CYP2B6*6	rs3745274 G/t	GG	47 (58)	61.2 ± 43	0.93
		GT	26 (32.1)	64.6 ± 63.2
		TT	5 (6.2)	51.4 ± 26.2
CYP2B6*4	rs2279343 A/g	AA	39 (48.1)	63.5 ± 45.7	0.89
		GA	33 (40.7)	65.2 ± 61.1
		GG	7 (8.6)	53.9 ± 27.8
CYP3A5*3	rs776746 G/a	AA	3 (3.7)	40 ± 20	0.63
		GA	10 (12.3)	51 ± 22.7
		GG	66 (81.5)	66.3 ± 54.8
CYP2C19*2	rs11188072 G/a	AA	0	-	0.47
		GA	25 (30.9)	67.1 ± 50.7
		GG	53 (65.4)	59.2 ± 49.1
CYP2C19*17	rs12248560	AA	3 (3.7)	60 ± 20	0.2
		GA	28 (34.6)	47.8 ± 29.8
		GG	46 (56.8)	69.7 ± 58.6
CYP2D6 CNV	2	1	4 (4.9)	30 ± 18.7	0.19
		2	67 (82.7)	64.9 ± 52.5
		3	8 (9.9)	64.6 ± 51.2
		4	1 (1.2)	40
ABCB1 c.3435C>T	rs1045642 C/t	CC	23 (28.4)	62.8 ± 64.6	0.79
		CT	40 (49.4)	60.5 ± 35.6
		TT	17 (21)	68.2 ± 63.4

Abbreviation is as follows: CNV, Copy Number Variants; SNP, single nucleotide polymorphism.

^*Patients with missing data were integrated in the analysis but not displayed in the table.

^†Kruskal–Wallis test.

Association between methadone maintenance dose and genotypes for polymorphisms in genes encoding dopamine and opioid receptors

The OPRM1 c.118A>G, COMT p.Val158Met, DRD2 Taq1 A and DRD2 c.957C>T genetic polymorphisms were determined, and frequencies of the respective allelic variants are displayed in Table4. Although a trend towards lower methadone daily dose was observed in patients carrying at least one DRD2 Taq1 A₁ allele, the difference was not statistically significant.

Table 4.

Association between methadone maintenance dose and, respectively, Catechol-O-methyl transferase, dopamine and opioid receptor genetic polymorphisms

Gene	SNP alleles(A/a)	Genotypes	n (%)*	Mean (±SD) methadone dose (mg day−1)	P Value†
OPRM1 c.118A>G	rs1799971 A/g	AA	55 (67.9)	64.8 ± 56.7	0.86
		AG	23 (28.4)	57.8 ± 38.1
		GG	2(2.5)	65 ± 7.1
COMT p.Val158Met	rs4680 Val/met	VV	28 (34.6)	73.1 ± 65.8	0.26
		VM	36 (44.4)	51.2 ± 35
		MM	16 (19.6)	70.9 ± 50.1
DRD2 Taq1 A	Rs1800497 A2/a1	A1A1	7 (8.6)	47.9 ± 37.4	0.17
		A1A2	20 (24.7)	49.3 ± 36
		A2A2	53 (66.7)	69.9 ± 56.4
DRD2 c.957C>T	Rs6277 C/t	CC	26 (32.1)	47.2 ± 30.4	0.18
		CT	33 (40.7)	76.4 ± 67.1
		TT	21 (25.9)	60.8 ± 36.6

Abbreviation is as follows: SNP, single nucleotide polymorphism.

^*Patients with missing data were integrated in the analysis but not displayed in the table.

^†Kruskal–Wallis test.

Discussion

The relationships between dose, plasma levels and effects are still not clearly defined, presumably due to a lot of unexplained variability from as yet unidentified sources, and an optimal range of therapeutic concentrations is still warranted for MMT, as shown in this and other studies. Thus, in the course of long-term MMT, the daily dose must be personalized in an arbitrary manner, ideally based on clinical evaluation of global clinical improvement, craving suppression, absence of withdrawal symptoms, management of other drug use and taking into account concomitant medications. Besides, a clinician-oriented approach is warranted, because clinicians expect from pharmacologists a practical guideline for drug prescription in order to optimize their methadone oral dosage in milligrams, rather than the methadone plasma concentrations, barely used in the clinical setting.

In the present multicentre study conducted in French stable patients, R-, S- and R,S-methadone plasma trough concentration increased significantly and proportionally with the dose, except between Q3 and Q4, where the dose almost doubled while the trough concentrations increased by only <25% (r² = 0.29, P < 0.0001; Table2). Patients were interviewed for around 1 h, with specific questions on what would be the optimal dose for them, residual withdrawal symptoms, soft signs of methadone overdose, general wellbeing with the treatment, supplementary heroin use, fractioned methadone intake and also methadone dose lowering or habit not to take the total dose. Bad treatment adherence as well as diversion of excess methadone onto the black market cannot be ruled out as possible explanations of such observations; however, these were not specifically addressed in the present study.

We have found that bodyweight, past history of cocaine dependence, ethnicity and fractioned methadone intake were independently associated with methadone maintenance dose. Due to the small number of Asian and African patients, the ethnic variable would require further confirmation. When R- or R,S-methadone was chosen as the dependent variable, only time since last dose and bodyweight were independently predictive of 15–20% of the interindividual variability of methadone trough concentrations. Such discrepancy is not surprising, because the methadone maintenance dose was poorly correlated to plasma trough concentrations in this and other studies 19. To the best of our knowledge, this is the first clinical study simultaneously addressing the influence of sociodemographic, ethnic, clinical, phenotypic and genetic variables on methadone maintenance dose or concentrations in a representative cohort of stable patients enrolled in real-life conditions of treatment and follow-up, including those with concomitant medical conditions that are usually excluded from pharmacological studies. In comparison to previously published findings, usually obtained in highly selected populations, the present results highlighted the major role played by clinical and sociodemographic over genetic variables in predicting methadone dose during maintenance programmes.

Clinical studies designed to identify the determinants of methadone maintenance dose usually focus on genetic polymorphisms of enzymes, transporters or receptors involved in methadone pharmacokinetics and pharmacodynamics but do not simultaneously address the influence of sociodemographic and clinical variables and concomitant medications on methadone maintenance dose in a multivariate approach. In a 23 year national study of methadone dose levels, D'Aunno et al. found that older and unemployed patients were more likely to receive higher methadone doses, while programmes that served a higher proportion of African-American or Hispanic patients were more likely to report low-dose care, as observed in the present study 3. A recent population pharmacokinetic/pharmacodynamic study conducted in 88 patients on MMT revealed that 33% of the overall variation in unbound R-methadone (the active enantiomer) was independently explained by CYP3A4 activity (9%), age (16%) and sex (8%), with respect to the Total Mood Disturbance Score, used to assess the pharmacodynamics effect of methadone 20.

Although the allelic frequencies observed in the present study were in agreement with a previously published study for each tested gene, we were unable to replicate previous observations, such as effects of CYP2B6 polymorphisms 2,8,15,16. The lack of relationship between CYP2B6 polymorphisms and methadone maintenance dose was unexpected but not surprising, because CYP2B6 exhibits stereoselectivity toward the S-enantiomer of methadone, which is almost inactive as a μ-opioid receptor agonist 15,21. The recent association between methadone dose requirement and the CYP2B6*4 and *6 SNPs published in the study by Levran et al. may be explained by the fact that all patients taking concomitant medications were excluded from their study 2. Hence, our findings may be due to the effect of concomitant medications, especially CYP3A4/5 inhibitors, on the relationship between methadone dose and R,S-methadone trough concentrations observed in the present population (Figure1).

We also failed to find any association between the functional ABCB1 (MDR1) c.3435C>T synonymous SNP and methadone maintenance dose, consistent with a previously published study conducted in 98 Jewish former severe heroin-dependent patients from Israel treated with similar doses of methadone, in which only the c.1236C>T SNP was reported to be associated slightly but significantly with a steady-state methadone dose requirement higher than 150 mg day⁻¹ 17. This result, obtained among nine ABCB1 common SNPs tested, was intriguing, because the ABCB1 c.1236C>T SNP located at exon 12 was previously associated with decreased P-gp activity in vitro 22. One would therefore expect that such an SNP would be associated with a lower dose of methadone in order to block the heroin effect effectively, reduce drug craving and prevent relapse 17. Finally, the weak stereoselectivity previously observed in vitro in methadone P-gp-mediated transport towards the inactive S-enantiomer may also explain the lack of influence of ABCB1 genetic polymorphisms on MMT in clinical practice 23.

The lack of correlation observed between methadone dose and, respectively, CYP2C19, CYP2D6 and CYP3A5 genetic polymorphisms was not surprising, because these enzymes may make only a marginal contribution to methadone first-pass effect, in comparison to CYP2B6 and, to a lesser extent, CYP3A4 and P-gp 1,24.

The involvement of methadone in significant drug–drug interactions is underdeveloped in general, because most references found are case reports or case series, focusing on inpatients receiving MMT, aged between 20 and 60 years and mostly taking 200 mg day⁻¹ or less, consistent with our study population 1,6,13,24. Drug interactions associated with methadone, summarized in previous publications, and their clinical significance are still poorly understood in general and are mostly theoretical and not necessarily verified in clinical practice 13,24. In the present study, only three patients were taking liver CYP3A4 inducers, one of whom was receiving the highest methadone dose in our patient population (i.e. 320 mg day⁻¹). Seven patients received ritonavir, a well-known liver/intestinal CYP3A4 inhibitor (confirmed by the significantly decreased midazolam metabolic ratio) and displayed lower R,S-methadone trough concentrations in comparison to patients treated with the same methadone dose and who did not receive concomitant CYP3A4 inhibitors (Figure1). Decreased methadone plasma concentrations have already been observed in previous clinical pharmacokinetic studies where patients concomitantly received lopinavir/ritonavir 25,26. The mechanism of this drug–drug interaction is not yet fully understood. Low-dose ritonavir has been shown previously to enhance levels of CYP promoter activity in white and sub-Saharan African patients with HIV, thus leading to CYP3A4 and CYP2B6 induction, which may explain the present observation 27. However, the lower midazolam metabolic ratio in patients who received ritonavir, in comparison to those who did not, does not support this hypothesis, at least for CYP3A4 in our study, and reinforces assumptions that CYP2B6, rather than CYP3A4, may be the most clinically relevant CYP involved in the metabolism of methadone, in agreement with previous studies 8,26,28. Besides, concomitant medications were not predictive of methadone maintenance dose, nor was liver/ intestinal CYP3A4 activity, consistent with previous findings 19,29. A correlation between CYP3A4 activity and, respectively, low (<99 mg day⁻¹) and very high methadone dose (>200 mg day⁻¹) was observed in a previous study conducted in patients stabilized for only one week prior to enrolment 29. The discrepancy between this and our study may be explained by the fact that our patients were stabilized for at least 3 months prior to enrolment.

The μ-opioid receptor, encoded by the μ-opioid receptor gene (OPRM1) is the pharmacological target of methadone, and the nonsynonymous c.118A>G SNP of OPRM1, which is associated with the loss of a putative N-glycosylation site, has been previously associated with a decreased potency of several opioids, including methadone, although this has been observed in healthy volunteers 30,31. In the present study, the nonsynonymous c.118A>G SNP of OPRM1 was not predictive of methadone maintenance dose, consistent with a recent study conducted on 238 patients in MMT with a higher median daily dose (125 mg day⁻¹) compared with the present study 30.

Dopamine receptors appear to play a large role in the rewarding effects of drugs of abuse, especially the DRD2 Taq1 A₁ allele, which had been widely studied in relationship to dependence and other psychiatric disorders, but also the synonymous DRD2 c.957C>T SNP, which was shown to be in linkage disequilibrium with the latter SNP 31. Homozygosity for the C allele of this SNP was twice as frequent in nonresponders to a MMT and associated with a fourfold shorter duration since the last positive urine screening for cocaine in 238 Caucasian patients 31. We did not identify the DRD2 c.957C>T SNP as a predictive variable of MMT, despite a trend towards lower mean methadone dose in CC vs. CT and TT carriers. However, a strong and positive association between lifetime history of cocaine dependence and methadone dose was observed in the present study, which may be due to as yet unknown genetic polymorphisms of the dopamine receptors genes or changes in dopamine receptor second messenger pathways by repeated cocaine use.

In conclusion, by using a multivariate approach including sociodemographic, clinical, phenotypic and pharmacogenetic variables together with measurement of the methadone trough plasma concentration immediately prior to the next dose, which proved to be reasonably consistent over a 2 month period 2, we have observed that simple and easily accessible clinical and sociodemographic variables (i.e. bodyweight, fractioned dose, past cocaine dependence and ethnicity) were independently predictive of methadone maintenance dose (which was strongly correlated to the R,S-methadone trough concentration) in unselected routinely managed responder patients and should be taken into account by physicians involved in methadone maintenance programmes when trying to target the dose on an individual basis. We failed to find an influence of any of the genetic polymorphisms involved in the pharmacokinetics or the pharmacodynamics of methadone in our patients enrolled in real-life conditions, including the CYP2B6 isoform, which has been established previously as a major contributor of the variability in methadone systemic exposure in humans 26,28,32.

Competing Interests

All authors have completed the Unified Competing Interest form at http://www.icmje.org/coi_disclosure.pdf (available on request from the corresponding author) and declare: support by Grants from the Délégation Régionale à la Recherche Clinique & au Développement – Assistance Publique Hôpitaux de Paris (PHRC National 2007, grant #OST07013) and from the Mission Interministérielle de Lutte contre la Drogue & la Toxicomanie (MILDT)-Institut National de la Santé & la Recherche Médicale (INSERM; grant #ASE07082KSA) for the submitted work; no financial relationships with any organizations that might have an interest in the submitted work in the previous 3 years; no other relationships or activities that could appear to have influenced the submitted work.

We are indebted to the following centres that actively participated in patient screening and enrolment: Dr E. Avril, Gaïa-Paris; Dr B. Badin de Montjoye, CSAPA Cassini, Cochin, APHP; Dr P. Coeuru, Aurore-Association EGO; Dr X. Laqueille, Centre Moreau de Tours, CH Sainte-Anne; Dr C. Orizet and B. Belforte, CSAPA Monte Cristo, HEGP, APHP; Dr P. Polomeni, 110-Les Halles, PSA 75; Dr A.-M. Simonpoli, ECIMUD Louis Mourier, APHP; and Dr Didier Touzeau, Clinique Liberté, Hôpital Paul Guiraud, Villejuif. We also thank Dr Beatrice Saint-Salvi very much for her precious comments and help in improving the quality of the manuscript.

Contributors

All authors have contributed to the manuscript. SM, VB, J-FB, J-MS, J-LL, J-PL and FV designed the project. KK, GS, SM and FV enrolled the patients. KP, PH, LL and XD performed the laboratory analyses. SM wrote and edited the manuscript. All authors read, improved and approved the final version of the manuscript.