Flowchart describing the patient population included in this study. There were 55 participants in each group, and all of them had a baseline electrocardiogram (ECG).
Comparison of the mean rate of Bazett's corrected QT interval (errors bars indicate confidence intervals). A, Comparison in the 3 study groups during the trial. Filled symbols indicate significant Tukey post hoc comparisons between treatment weeks and week 0 (P < .05). Asterisks indicate significant Tukey post hoc comparisons to buprenorphine hydrochloride group at that treatment week (P < .05). B, Comparison in the levomethadyl acetate and methadone hydrochloride groups receiving a fixed dose of the study drug for at least two 4-week intervals. Filled symbol indicates a statistically significant difference compared with the first steady state period (P < .05).
Percentage of study population exceeding the cutoff value for Bazett's corrected QT of 470 milliseconds for males and 490 milliseconds for females at the different “on-drug” points in the study. Overall trend is significant by general estimating equations (P < .001).
Wedam EF, Bigelow GE, Johnson RE, Nuzzo PA, Haigney MCP. QT-Interval Effects of Methadone, Levomethadyl, and Buprenorphine in a Randomized Trial. Arch Intern Med. 2007;167(22):2469-2475. doi:10.1001/archinte.167.22.2469
Copyright 2007 American Medical Association. All Rights Reserved. Applicable FARS/DFARS Restrictions Apply to Government Use.2007
Levomethadyl acetate, methadone hydrochloride, and buprenorphine hydrochloride are equally effective treatments for opioid dependence. Each blocks the human ether-a-go-go–related gene (hERG)-associated channel in vitro and represents a risk for QT prolongation. To compare the effects of 3 known hERG-associated channel blockers on the corrected QT (QTc), we conducted a randomized, controlled trial of opioid-addicted subjects.
We analyzed 12-lead electrocardiograms collected at baseline and every 4 weeks from 165 opioid-addicted participants in a 17-week randomized double-blind clinical trial of equally effective doses of levomethadyl, methadone, and buprenorphine at a major referral center. Analyses were limited to the 154 patients with a normal baseline QTc = (QT/√ R-R) who had at least 1 subsequent in-treatment electrocardiogram. Patients were randomized to receive treatment with levomethadyl, methadone, or buprenorphine (hereinafter, levomethadyl, methadone, and buprenorphine groups, respectively). The prespecified end points were a QTc greater than 470 milliseconds in men (or >490 milliseconds in women), or an increase from baseline in QTc greater than 60 milliseconds.
Baseline QTc was similar in the 3 groups. The levomethadyl and methadone groups were significantly more likely to manifest a QTc greater than 470 or 490 milliseconds (28% for the levomethadyl group vs 23% for the methadone group vs 0% for the buprenorphine group; P < .001) or an increase from baseline in QTc greater than 60 milliseconds (21% of the levomethadyl group [odds ratio, 15.8; 95% confidence interval, 3.7-67.1] and 12% of the methadone group [odds ratio, 8.4; 95% confidence interval, 1.9-36.4]) compared with the buprenorphine group (2% of subjects; P < .001). In subjects whose dosage of levomethadyl or methadone remained fixed over at least 8 weeks, the QTc continued to increase progressively over time (P = .08 for the levomethadyl group, P = .01 for the methadone group).
Buprenorphine is associated with less QTc prolongation than levomethadyl or methadone and may be a safe alternative.
There are estimated to be more than 2 million users of opioid drugs in the United States, with 800 000 individuals with long-term opioid addiction.1 The mortality rate associated with opioid dependence has been estimated to be as high as 1% per year in some study populations.2,3 Levomethadyl acetate, methadone hydrochloride, and buprenorphine hydrochloride are agents proven to be effective in opioid addiction treatment4- 8; however, serious adverse events, such as lethal arrhythmias, have been reported with use of levomethadyl9 and methadone.10- 15 Sufficient concern was raised with regard to levomethadyl that in 2001 the European Medicines Evaluation Agency withdrew the medication from the market and the US Food and Drug Administration included a black box warning in its labeling, which led to withdrawal of the product in the United States by the manufacturer in 2003. In vitro studies of levomethadyl, methadone, and buprenorphine have each demonstrated considerable blockade of the human ether-a-go-go-related–gene (hERG) channel activity,16 a property that is strongly associated with prolongation of the QT interval and the induction of torsades de pointes ventricular tachycardia (TdP).17,18 Buprenorphine, however, blocked the hERG channel with notably less potency than either methadone or levomethadyl. To our knowledge, no randomized control trial has compared the effects of these agents on the QT interval. We hypothesized that the reduced in vitro potency of hERG blockade associated with buprenorphine would result in a notably lesser degree of QT prolongation in vivo. We tested this hypothesis using electrocardiographic data obtained during a randomized, blinded study of levomethadyl, methadone, and buprenorphine.
Electrocardiographic data obtained from a randomized, blinded evaluation of the efficacy of levomethadyl, methadone, and buprenorphine in an outpatient opioid-dependence maintenance treatment program were analyzed. The design, execution, and results of the original study have been published previously.4 In a 17-week randomized study of 220 patients, levomethadyl acetate (75-115 mg), buprenorphine hydrochloride (16-32 mg), and high-dose (60-100 mg) and low-dose (20 mg) methadone hydrochloride were compared as treatments for opioid dependence. Levomethadyl and buprenorphine were administered 3 times a week. Methadone was administered daily. Dosages were individualized except in the group assigned to low-dose methadone. Patients with poor responses to treatment were switched to methadone. The study had 3 phases: dose induction (weeks 1 and 2), maintenance (weeks 3-17), and disposition (weeks 18-28). On days of clinic attendance, all the patients received 3 solutions, only 1 of which contained active medication. Missed doses were treated according to preestablished blinded protocols developed by the investigators. Patients were discharged from the study if they were absent for 5 consecutive calendar days.
The trial found that levomethadyl, buprenorphine, and high-dose methadone were equally effective in reducing the use of illicit opioids (hereinafter, levomethadyl, buprenorphine, and methodone groups), and all 3 treatments were notably more effective than low-dose methadone. During the study period, electrocardiograms (ECGs) were obtained at study enrollment and at 4-week intervals until the participant discontinued assigned treatment or study completion at week 17.
The ECGs were reviewed by an experienced cardiologist (E.F.W.) blinded to treatment allocation. Standard 12-lead ECGs were obtained at a paper speed of 25 mm/s and voltage of 10 mm/mV. For each ECG, the PR, R-R, QRS, and QT intervals were manually measured with a digital micrometer. Lead II was used unless the T wave morphologic characteristics were indistinct, in which case lead V5 or V2 was used. The same lead was used for each subject. The QT interval was defined as the time between the onset of the QRS interval to the intersection point of the baseline with the tangent line to the steepest down slope of the intervening T wave with exclusion of U waves.19 The R-R interval was measured from the preceding QRS complex to the measured QRS complex. Heart rate, rhythm, axis as well as conduction, QRS, ST, and T wave abnormalities were also measured.
The QT interval shortens as the heart rate increases (R-R interval decreases). To compare QT measurements made at different heart rates, heart rate “correction” (QTc) formulas are used. The measured QT intervals were corrected for heart rate (R-R interval) using the Bazett20 (QTc = QT/√ R-R) and the Fridericia21 (QTf = QT/R-R1/3) equations, but because Bazett's correction formula is better established in the literature, it was prespecified for the primary end point. The thresholds for abnormal prolongation of the QTc were based on the Long QT Syndrome Registry, in which a QTc of 450 milliseconds for males and 470 milliseconds for females represented the top 1% of QTc.19 To better define the risk for QT prolongation, we prespecified a more stringent threshold that was 20 milliseconds longer (ie, 470 milliseconds for males and 490 milliseconds for females) and identified subjects whose QTc exceeded 500 milliseconds. In addition to absolute thresholds for abnormal QTc, a change in QTc greater than 60 milliseconds from baseline was designated as a measure of clinically relevant QTc prolongation. Finally, the main effects were compared using a third QTc correction formula taken from the Framingham longitudinal study,22
QTFramingham = QT + 0.154 × (1000 − R-R).
Recent studies23 suggest that linear correction may be superior to traditional exponential formulas. Selection of clinically relevant QTc categorical cutoff points was based on a position paper released by the Committee for Proprietary Medicinal Products of the European Agency for the Evaluation of Medicinal Products24 and a concept paper produced jointly by the US Food and Drug Administration and Health Canada's Therapeutic Products Directorate.25
Between-group comparisons on baseline measures were conducted using χ² tests for dichotomous variables and analysis of variance for continuous variables. Interpretation of group main effects and treatment interactions relied on Tukey post hoc tests. The QTc values were analyzed using multilevel analyses and SAS Proc Mixed statistical software (SAS Inc, Cary, North Carolina) with treatment group and week as factors.26 Proc Mixed was selected for its ability to handle missing observations.
Categorical QTc values (prolongation threshold of 470 milliseconds for males and 490 milliseconds for females) were assessed repeatedly over treatment weeks and analyzed using general estimating equations (GEEs),27 which are particularly suited for analyses of longitudinal data and allow for correlations among observations within an individual subject, for the presence of missing data, for subjects measured at different time points, and for covariates. Results are reported as odds ratios (ORs) with 95% confidence intervals (CIs) surrounding the OR with Bonferroni corrections applied to group comparisons.
Certain risk factors that predispose to QTc prolongation (eg, sex, renal failure, and hypokalemia) were considered as covariates to the change in QTc greater than 60 milliseconds from baseline categorical GEE analysis. All analyses were conducted using SAS statistical software (version 9.1 for Windows; SAS Inc) with P < .05 indicating statistical significance.
Of the initial 165 patients enrolled, 151 patients had ECG data collected at week 4. By week 17, 102 patients remained enrolled in the study for analysis (Figure 1). No statistically significant differences between groups were evident at baseline, as shown in Table 1. In the original study there were 4 treatment groups to include therapy with levomethadyl, buprenorphine, high-dose methadone, and low-dose methadone. Analysis of data from the low-dose methadone treatment group was not included owing to a near 80% attrition rate resulting in rescue with high-dose methadone. Analysis of heart rate within and between groups demonstrated no notable group difference or change over time during drug treatment (mean [SD] beats/minute, 66.3 [0.90], 67.3 [0.81], and 65.5 [0.89] for the groups treated with levomethadyl, methadone, and buprenorphine, respectively).
Treatment resulted in a notable increase in the mean QTc from baseline through the course of therapy as shown in Figure 2. By the fourth week, significant QTc increases from week 0 were observed in the levomethadyl (mean [SD] increase, 27.0 [4.8] milliseconds; P < .001), methadone (17.3 [5.0] milliseconds; P = .003), but not buprenorphine (5.9 [3.0] milliseconds; P = .99) groups. By week 16, the mean QTc of the levomethadyl, methadone, and buprenorphine groups increased an additional 21.7 (5.3) milliseconds (P = .04), 17.0 (5.3) milliseconds (P = .18), and 5.4 (4.1) milliseconds (P > .99), respectively. Comparing groups at each week revealed significant differences between the levomethadyl and buprenorphine groups at weeks 8, 12, and 16 (P = .03). In addition, a significant difference was observed between the methadone and buprenorphine groups at week 16 (P = .02). The progressive prolongation of the QTc over time was unanticipated given that steady state drug levels should have been achieved by week 4. The continued rise in QTc is partially explained by the fact that dosages of study medications could be increased every 2 weeks under blind conditions according to the study protocol and clinical criteria. However, in those participants receiving the same dosage at multiple time points, a considerable progressive increase was still present (Figure 2). This trend was not significant for levomethadyl (P = .08) but was significant for methadone (P = .01). When the 2 groups were combined, the trend was significant (P = .003). Finally, when the data were reanalyzed using the Framingham linear correction, the between-group differences remained significant (P < .001), and the progressive prolongation of the QTc was still seen in the levomethadyl and methadone groups.
Using the categorical definition for QTc prolongation of more than 470 milliseconds for males and more than 490 milliseconds for females, 28% of the levomethadyl group, 23% of the methadone group, and none in the buprenorphine group exceeded this threshold during treatment (P <.001). The proportion of each treatment group exceeding this threshold over the course of the study is shown in Figure 3. The OR of QTc prolongation for levomethadyl compared with methadone was not significantly increased whether the Bazett correction formula (OR, 1.7 [95% CI, 0.8-4.0]; P = .19) or Fridericia correction formula (OR, 2.2 [95% CI, 0.7-6.4]; P = .17) was used. Computing the OR for QTc prolongation for levomethadyl or methadone compared with buprenorphine required altering a single QTc value to abnormal in the buprenorphine group (also done with the levomethadyl and methadone groups) to allow convergence of the statistical model (Table 2); this is a conservative approach because it makes the groups seem more similar. For the levomethadyl-buprenorphine comparison, the OR for QTc prolongation was 25.1 (95% CI, 3.3-193.3; P = .002); for the methadone-buprenorphine comparison, the OR for QTc prolongation was 14.4 (95% CI, 1.9-109.5; P = .01). Finally, an absolute QTc value of more than 500 milliseconds is thought to be associated with significant risk for TdP; none in the buprenorphine group but 6 of 52 individuals in the methadone group and 5 of 46 in the levomethadyl group exceeded this value (P < .001) at some point during the trial.
Using the stringent standard of an increase in QTc from baseline of more than 60 milliseconds at any time during the study, significantly more subjects demonstrated an increase from baseline if treated with levomethadyl (21% of subjects; OR,15.8 [95% CI, 3.7-67.1]) or methadone (12% of subjects; OR, 8.4; 95% CI, 1.9-36.4) compared with buprenorphine (2% of subjects; P < .001). With respect to comparing levomethadyl and methadone, there was no significant difference when using Bazett's correction (P = .06), but using the Fridericia correction levomethadyl was again associated with a significantly greater risk of QTc prolongation (>60 milliseconds compared with methadone (OR, 2.5 [95% CI, 1.2-5.2]; P = .01).
Female sex has been associated with considerably higher risks for QT prolongation and TdP,28 but in the present study there was no significant difference between men and women with respect to prolongation more than 60 milliseconds from baseline (P = .68). When the GEE was repeated using sex as a covariate, there was no change in the direction or magnitude of the QT effects. Concurrent medications that prolong the QT interval could also have affected our analysis. We compared the participants' medications with a list of drugs thought to potentially prolong the QT interval.29 There were only 2 individuals prescribed an agent potentially prolonging the QT who manifested a QTc of 470 milliseconds or greater: 1 individual in the buprenorphine group (erythromycin; maximum QTc, 470 milliseconds) and 1 in the methadone group (erythromycin; maximum QTc, 483 milliseconds). With respect to hypokalemia, there were 2 participants in the levomethadyl group who had potassium levels lower than 3.5 mEq/L, 1 in the buprenorphine group, and none in the methadone group. The mean (SD) intake potassium level was similar in the 3 groups (4.30 [0.06] mEq/L in the levomethadyl group, 4.23 [0.04] mEq/L in the methadone group, and 4.33 [0.06] mEq/L in the buprenorphine group; P = .93). No participants manifested a creatinine level higher than 1.5 mg/dL. The differences in QTc prolongation among the groups, therefore, could not have been caused by these competing factors. (To convert potassium to millimoles per liter, multiply by 1.0; to convert creatinine to micromoles per liter, multiply by 88.4.)
This analysis of electrocardiographic data from a randomized controlled trial compared the effects of 3 equally efficacious treatments for opioid addiction on cardiac repolarization. On average, levomethadyl and methadone use were associated with a notable increase in mean QTc of more than 30 milliseconds over the course of the study, whereas those in the buprenorphine group experienced no notable prolongation. Despite similar QTc levels at baseline, the incidence of a QTc greater than 470 milliseconds (for males) and greater than 490 milliseconds (for females) with levomethadyl or methadone was considerably greater than that of buprenorphine. A QTc prolongation of more than 30 or 60 milliseconds was considerably more common in the levomethadyl and methadone groups. Ten percent of those in both the levomethadyl and methadone groups experienced a QTc of more than 500 milliseconds at some point during the study; none in the buprenorphine group exceeded this value. Levomethadyl and methadone, therefore, seem to be associated with a significant risk for QT prolongation compared with buprenorphine (see Table 2 for summary data and P values). Finally, unlike those in the buprenorphine group, there was evidence of progressive prolongation over time in individuals receiving fixed doses of levomethadyl and methadone.
The present data are consistent with prior data regarding the cardiac effects of levomethadyl and methadone. Preclinical investigations found that levomethadyl and methadone had a 100-fold greater tendency to block the hERG-associated potassium channel than buprenorphine at presumed therapeutic dosages.16 Prior clinical evaluation of QTc effects of methadone induction showed a mean QT increase of 10.8 milliseconds.30 During the course of therapy, levomethadyl and methadone each produced increases from baseline in QTc of approximately 7.5% to 11.0% while demonstrating peak QTc increases greater than 450 milliseconds in approximately 69% of males and greater than 470 milliseconds in approximately 72% of females.31 The mean QTc increases at 4 weeks in our study are similar to those reported previously.30,31 The present study, however, provides insight into the effects of these agents 8, 12, and 16 weeks into therapy. These data are important because these drugs are used long term and often require up-titration of dosages that could result in greater risk for arrhythmia over time. The manifestation of progressive QTc prolongation over time is not explained by conventional models of hERG blockade, however, and may point to remodeling effects of these agents.32 Recent work has found that agents such as pentamidine isethionate33 and the sphingolipid metabolite ceramide34 cause decreased expression of hERG channels on the cell membrane through increased channel degradation or decreased trafficking to the membrane. Whether levomethadyl and methadone share effects on hERG expression, the present data raise concerns about the long-term effects of these agents that suggest a need for repeated electrocardiographic screening over months.
Although official guidelines are lacking, European and North American pharmaceutical regulatory bodies have provided suggestions regarding electrocardiographic indicators of clinically significant drug-related cardiac effects. In particular, a QTc increase from baseline is considered potentially clinically significant if more than 30 milliseconds and clearly drug-related, and a clinically significant concern for risk of TdP if more than 60 milliseconds.22,23 A peak QTc greater than 500 milliseconds, or a group mean QTc increase greater than 20 milliseconds, is also recognized as conferring an increased risk for TdP. Levomethadyl and methadone each exceeded all of these thresholds. These findings are particularly notable because other agents, such as terfenadine, cisapride, moxifloxacin hydrochloride, and thioridazine, had similar or lesser degrees of QTc prolongation and were subsequently restricted in their use, specially labeled, or removed from the market.25,35 Buprenorphine, however, manifested rare increases of more than 60 milliseconds, did not notably increase the mean QTc, and was not associated with an absolute QTc of more than 470 milliseconds in males or 490 in females. Buprenorphine was associated with frequent increases in the QTc of more than 30 milliseconds, however, which is consistent with some evidence of effect on cardiac repolarization.
This study specifically focused on a surrogate marker of arrhythmia potential, QTc prolongation, and was not powered to evaluate “hard” end points such as sudden death or the incidence of TdP. As has been recently pointed out,18 a drug causing TdP in 1 of 1600 individuals might not be detected in a study with 5000 participants. Given that the mortality of TdP has been estimated to be as high as 17%,35- 38 regulatory agencies are particularly vigilant with regard to surrogate indicators that an agent might increase the risk of arrhythmia. Evidence of considerable QT prolongation is one important marker suggesting increased risk for arrhythmia. In addition, postmarket reporting of adverse events plays an important role in risk assessment, but failure to appreciate the true cause of death in individuals receiving treatment for addiction may result in considerable underrecognition of drug-associated adverse events. Levomethadyl and methadone each have had reports of notable clinical adverse events, including TdP.9- 15 A recent study of 167 hospitalized heroin users who received methadone found a 16% incidence of QTc prolongation greater than 500 milliseconds and a 3.6% incidence of TdP.39 We know of no published cases of TdP with the use of buprenorphine.
Physicians must use their judgment in choosing the appropriate therapy for opioid dependence; failure of therapy results in considerable mortality. However, given that buprenorphine has previously been proven to be equally efficacious in the treatment of opioid addiction, buprenorphine may be a safe alternative for treatment of this common and life-threatening problem.
The most important limitation in this study is the absence of a placebo arm to assess the random incidence of QT prolongation in this population. In our view, a placebo could not be ethically substituted in this life-threatening condition. In addition, the participants in this trial were outpatients and could have consumed other agents that prolong the QT interval. Specific data regarding serum drug levels of each agent in our study were not available. Finally, the ECGs in this study were recorded on paper at 25 mm/s, which reflects the standard of the time and is typical clinical practice. Some investigators have suggested that digitized ECGs may increase precision given the ability to make computerized measurements of intervals.
Correspondence: Mark C. P. Haigney, MD, Division of Cardiology, Department of Medicine, Uniformed Services University of the Health Sciences, A3060, USUHS, 4301 Jones Bridge Rd, Bethesda, MD 20814 (email@example.com).
Accepted for Publication: July 18, 2007.
Author Contributions: Dr Haigney had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis. Study concept and design: Bigelow, Johnson, and Haigney. Acquisition of data: Wedam, Bigelow, Johnson, and Haigney. Analysis and interpretation of data: Wedam, Bigelow, Nuzzo, and Haigney. Drafting of the manuscript: Wedam, Bigelow, Nuzzo, Johnson, and Haigney. Critical revision of the manuscript for important intellectual content: Bigelow and Haigney. Statistical analysis: Nuzzo. Obtained funding: Bigelow and Johnson. Administrative, technical, and material support: Wedam, Bigelow, and Johnson. Study supervision: Bigelow, Johnson, and Haigney.
Financial Disclosure: Dr Johnson is currently an employee of Reckitt-Benckiser Pharmaceuticals Inc, the manufacturer and distributor of buprenorphine. Dr Bigelow has received, or anticipates receiving, research support, through his institution, from Purdue Pharma LP, Biotek Inc, and Titan Pharmaceuticals Inc for studies of other buprenorphine formulations.
Funding/Support: This study was supported by National Institutes of Health/National Institute on Drug Abuse grants No. P50 DA05273, K02 DA 00332, K05 DA 00050, and R01 DA 08045.
Disclaimer: The views expressed in this article reflect the opinions of the authors only and not the official policy of the US Navy, Uniformed Services University, or the Department of Defense.
Additional Information: Dr Johnson was the principal investigator at the time of the original study (1996-1998).
Additional Contributions: Tim Mudric, BA, Jenna Shultz, BA, John Yingling, BA, and the staff of the Behavioral Pharmacology Research Unit at Johns Hopkins University provided data collection and technical support.