14 June 2013

Kao study: can adverse event reports inform medical practice? Cardiac safety/danger imbalance.

Trends in reporting methadone-associated cardiac arrhythmia, 1997-2011: an analysis of registry data. Kao D, Bucher Bartelson B, Khatri V, Dart R, Mehler PS, Katz D, Krantz MJ. Ann Intern Med. 2013 21;158(10):735-40
Dear Colleagues, 
Following the lead of Pearson and Woosley (see summary below*) in 2005, Kao, Krantz and colleagues have made a search of adverse event reports to the FDA in America regarding methadone and cardiac status. 

From 2003 to 2012 they report a mean of 3.5 cases per week of torsade de pointes or QT prolongation with a mortality of 11% across these two groups.  Mean age was 44 years and 57% were male.  Methadone was the primary suspect in only 57% of reports.  Median daily methadone dose was 150mg.  Table 1 states that the outcome required intervention in 9.5% of cases, making it appear that over 90% of cases were self terminating. 

Consistent with other reports, these authors find that many or most cases are due to more than one factor, lamivudine being the commonest co-prescribed drug (11% of reports).  Other drugs involved were ritonavir and AZT (each at 10%), lorazepam, morphine, apomorphine, tenofovir, lopinavir, ceftriaxone (a cephalosporin antibiotic) and oxazepam (5-7% of reports). 

By failing to separate the groups the reader is denied any real view of the extent of torsade de pointes tachycardia or its mortality (if any).  Torsade is a rare but serious arrhythmia while QT prolongation is a common and usually asymptomatic finding in methadone patients (Stimmel 1973; Wedam 2007; Fareed 2013).  From 2000 to 2002 Pearson reported 8.5% mortality from the combined group (5/59) but no deaths due to torsade itself. 

Clouding matters somewhat, Kao et al. give another combined figure for cardiac arrest and ventricular arrhythmia (1646, 81% mortality, 41% with methadone as the primary suspect).  Neither of these conditions is known to be associated with drug-related QT prolongation and furthermore, both could be terminal events due to overdose or other causes.  About half of the methadone in America is estimated to be used in for cancer and non-cancer pain purposes so many deaths have methadone as a bystander (consistent with 57% where it was not considered the prime suspect). 

If Kao et al. had a similar proportion of torsade cases to Pearson, it would represent approximately 2.5 reports of torsade per month in America.  Considering there are reportedly around 550,000 patients on methadone at any one time in America (FDA 2010) this is not consistent with Kao et al. who contend that the condition is a major factor in methadone deaths.  Furthermore, the figure is consistent with both French adverse reaction reports and a Norwegian mortality survey of methadone patients (Perrin-Terrin 2010, Anchersen 2009). 

Dr Krantz is the corresponding and senior author of this paper.  His seminal paper on this subject in 2002 failed to separate addiction from pain management cases.  This omission may have contributed to confusion about the dose level at which torsade de pointes may be expected to occur due to the extremely high doses of his pain management cases (which were only revealed in a subsequent paper).  The mean reported dose across the groups was 397mg daily yet their follow-up paper allowed a calculation of very different mean daily dose levels between the two groups (541mg for 8 pain patients and 269mg for 9 dependency cases).  
Kao, Krantz et al. make comparisons with dofetilide and sotalol, antiarrhythmic drugs which can cause QT changes and torsade de pointes.  Yet it can hardly be a useful or valid comparison when one drug is being used in a person already known to have a cardiac problem.  Furthermore there were more than twice as many reported QT/torsade cases with these drugs (478) than with methadone (211) as primary suspects (no indication is given of the total patient years of the former, nor the mortality). 
Despite the conceded limitations of reports of this nature, the authors nevertheless speculate (their word) that there may be an increasing incidence and that the arrhythmia may represent an important but underappreciated contributor to methadone-related morbidity and mortality.  Yet the authors provide no evidence for this and they still cannot cite any group of confirmed deaths (or even one as far as I can find) due to torsade de pointes. 
I am surprised that Krantz allows his colleagues to cite Chugh’s study from Portland, Oregon to justify the claim that arrhythmia contributes to a substantial proportion of methadone deaths.  Close examination of Chugh’s very crude and indirect protocol finds that it is severely flawed and indeed if all his deductions were valid, torsade reports should number in the thousands. 
To try to place this item into some perspective I did a search of the MedWatch web site for torsade de pointes which showed up warnings for citalopram, arsenicals, droperidol, clozapine, azithromycin, LAAM, clarithromycin, thioridazine and ondansetron, on just their first search page.  Further searches for FDA adverse event reports on these drugs from the E-health-me web site reveals 35 reports for citalopram, 33 reports for droperidol, 2 for clozapine, 55 for azithromycin, 82 reports for clarithromycin and none for thioridazine.  There are even torsade reports from alcohol (10 reports), diazepam (41 reports) and salbutamol (14 reports).  When methadone is entered the number of reports is 26 plus another 23 under the trade name for methadone liquid (Methadose) making 49 total over a reporting period of about 15 years.  There may be a simple explanation for what appears to be an inconsistency with Kao’s paper which reports over 300 such cases from the FDA reporting system.
Even allowing for under-reporting, neither is a large number when compared with the number of torsade cases reportedly due to other drugs, nor in view of the very large numbers of patients now taking methadone.  
Pearson and Woosleys report included a detailed table with age, sex, methadone dose, type of report (QT or torsade), outcomes, other contributors, etc.  By contrast, Kao et al. leave the reader in the dark regarding how many actual torsade cases there were and how many, if any, were fatal.  I could find none in the English literature (there may be one from France). 
This Annals paper is one of several recent examples of what one commentator has called “unrepentant fear-mongering” about methadone and cardiac safety.  Despite the deficiencies of the present report, far more important, as the US Center for Disease Control points out, methadone is involved in thousands of overdose deaths every year, with the increasing number of such deaths correlating with the increasing use of methadone for pain in America.  After single-figure reports of torsade and examinations of ‘possible torsade’ (sudden death) from national surveys in France and Norway, it is clear that the increasing deaths are not significantly contributed to by QT prolongation/TdP problems as proposed again by Kao and colleagues.  Indeed, we know that being on supervised methadone treatment reduces deaths dramatically (Caplehorn reported by a factor of four). 
Hence, these patients are less in need of protection from the supposed hazards of QT prolongation or TdP than from irrational alarmism.  Torsade de pointes is an extremely rare adverse event which is complex and multifactorial.  Ignorant speculation needlessly frightens the field about the safety of the methadone treatment and is especially egregious when unproven across-the-board measures are recommended to ‘protect’ against the supposed ‘problem’, even before its scope is defined.  Krantz has still not addressed the anomaly of his large number of cases of torsade (17) in two small catchments in 2002 when compared with large national surveys which have found even fewer cases in far larger populations. 
Concluding comment on cardiac safety in methadone patients:
Torsade de pointes tachycardia is an unpleasant and unusual consequence of a series of events in a person predisposed to QT prolongation (but whose resting QT length is often normal).  Most cases of torsade do not involve methadone but a proportion occur in long term methadone recipients, often involving substantial doses (>100mg daily) with the addition of another drug.  These include common antibiotics, anti-virals, antidepressants and other drugs, each of which can be responsible for torsade in a predisposed individual.  Other factors which contribute to the risk levels are genetic QT prolongation, female sex, older age, pre-existing heart disease, alcohol use, low potassium and HIV infection. 
While torsade de pointes is extremely uncommon, it will still be seen occasionally in dependency and pain practice.  The arrhythmia needs to be considered in someone with fainting, fitting, palpitations, shortness of breath or occasionally, chest pain.  Treatment involves the use of urgent paramedic treatment and transfer to cardiac intensive care for monitoring.  Some patients will need intravenous magnesium, temporary pacemaker and/or cardioversion.  Withdrawal or replacement of the suspected drugs and/or reduction in doses may be useful.  There is no single agreed protocol for this condition but its treatment should be directed by cardiac experts, just as dependency should be directed by dependency experts and pain by pain management experts. 
There is no proven preventive strategy but it would seem prudent to order a cardiograph on patients who are prescribed high dose methadone (>150mg) even in the absence of other risk factors. 
Notes by Andrew Byrne ..
* Pearson found 59 reports in a 7 year period, 43 (73%) had torsade de pointes.  There were five deaths (8.5%), of whom 2 used methadone intravenously (360mg and 1680mg daily), one was 78 years old and took cisapride as well as the methadone, at unknown dosage; the fourth was a 40 year old female who took 29mg methadone daily (none of these had developed torsade but only had QT prolongation); the fifth and only fatal event in a torsade case was a 47 year old female who died from a myocardial infarction in addition to the use of droperidol, azithromycin, and methadone. 
Other references:
Pearson EC, Woosley RL. QT prolongation and torsades de pointes among methadone users: reports to the FDA spontaneous reporting system. Pharmcoepidemiol Drug Saf. 2005 14;11:747-753 [see summary below]
Stimmel B, Lipski J, Swartz M, Donoso E. Electrocardiographic changes in heroin, methadone and multiple drug abuse: a postulated mechanism of sudden death in narcotic addicts.  Proceedings of  Fifth National Conference on Methadone Treatment, March 1973
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;22:2469-2473
Fareed A, Vayalapalli S, Scheinberg K, Gale R, Casarella J, Drexler K. QTc interval prolongation for patients in methadone maintenance treatment: a five years follow-up study. Am J Drug Alcohol Abuse. 2013 Jun 28.
Perrin-Terrin A, Pathak A, Lapeyre-Mestre M. QT interval prolongation: prevalence, risk factors and pharmacovigilance data among methadone-treated patients in France. Fundam Clin Pharmacol. 2010 Sep 6
Anchersen K, Clausen T, Gossop M, Hansteen V, Waal H. Prevalence and clinical relevance of QTc interval prolongation during methadone and buprenorphine treatment: a mortality assessment study. Addiction 2009 104;6:993-999
Krantz MJ, Lewkowiez L, Hays H, Woodroffe MA, D. Robertson AD, Mehler PS. Torsade de Pointes Associated with Very-High-Dose Methadone. Ann Intern Med. 2002 137:501-504
Caplehorn JRM, Dalton MSYN, Cluff MC, Petrenas A. Retention in methadone maintenance and heroin addicts' risk of death. Addiction 1994 89:203-7