Send email for updates
Palpacuer C., Laviolle B., Boussageon R. et al.
PLoS Medicine: 2015, 12(12), e1001924.
Unable to obtain a copy by clicking title? Try asking the author for a reprint by adapting this prepared e-mail or by writing to Dr Naudet at email@example.com. You could also try this alternative source.
‘A pill for every ill’ is the gist of the attacks levelled at nalmefene in the form of Selincro, a drug expected to extend the benefits of pharmacotherapy to drinkers not physically dependent or in need of detoxification – or for critics, to medicalise psychosocial dependence on shaky scientific grounds.
To reduce harm alcohol-dependent individuals are usually advised to abstain, but transitioning to controlled, moderate drinking may also be helpful. Marketed under the trade name Selincro, the European Medicines Agency recently approved the medication nalmefene to help reduce alcohol consumption among alcohol-dependent men drinking over 60g (7.5 UK units) alcohol per day or women drinking over 40g (5 UK units). Nalmefene blocks the body’s opioid receptors and reduces the craving for alcohol.
Under the trade name Selincro, in 2013 the medication nalmefene was granted European marketing authorisation to help reduce drinking among dependent – but not physically dependent – high-risk drinkers.
The featured analysis found that the drug could not be shown to affect health and had only minor effects on drinking, or none if it was assumed patients missing at follow-ups had continued with or resumed pre-trial drinking.
Despite on balance positive results, shortcomings in the manufacturer’s trials and the analyses on which authorisation was based have led to concerns that less severe forms of dependence are being inappropriately perhaps ineffectively medicalised, displacing psychosocial support.
Another concern is that there are reasons to believe that the parent drug, naltrexone, would be just as effective and much cheaper.
However, several expert bodies have concluded that nalmefene shows no benefit over naltrexone (an older treatment for alcohol dependence which also blocks opioid receptors) and do not recommend nalmefene’s use for the treatment of dependent drinking.
To clarify these issues, the authors of the featured review investigated the risks and benefits of nalmefene in the treatment of alcohol dependence in adults by undertaking a systematic review and meta-analysis of double-blind, randomised controlled trials of nalmefene for this condition. A systematic review uses predefined criteria to identify all the research on a given topic. A meta-analysis combines the results of several studies. A double-blind randomised controlled trial compares outcomes in people chosen at random to receive different treatments without the researchers or the participants knowing who received which treatment until the end of the trial, an accepted way to ensure the outcomes are not due to pre-existing differences between people given the treatment versus those not, and that the results are due to the drug’s effects, not due to expectations of patients or research staff about those effects.
The researchers identified five relevant trials which in total had involved 2567 participants, all of which compared the effects of nalmefene with a placebo (dummy drug). None of the trials specifically sampled participants drinking at the levels specified by the European Medicines Agency’s approval.
In respect of the health outcomes examined in the meta-analysis, there were no significant differences between participants allocated to nalmefene versus placebo in death rates after six months or one year of treatment, in the quality of life at six months, or in a summary score indicating mental health at six months. Trials included in the meta-analysis did not report other health outcomes such as accidents.
Nalmefene patients drank heavily on one to two fewer days per month six months and one year after treatment started than participants allocated to a placebo, and their total alcohol consumption was slightly lower, all statistically significant differences. Measures of the severity of dependence and alcohol problems also improved more among the nalmefene patients. However, more people allocated to nalmefene than to a placebo withdrew from the studies, often for safety reasons. This and the high overall rate of attrition from the studies makes the findings vulnerable to bias because differences between the remaining two sets of patients can no longer be assumed to have been eliminated by random allocation. When the analysts accounted for patients lost to the studies by assuming they continued to drink as they did at the start of the trials, alcohol consumption outcomes were no longer significantly better among patients allocated to nalmefene. However, this was a worst-case-scenario ‘conservative’ assumption.
Deaths were too few to register a statistically significant difference, but were in favour of nalmefene, totalling to one out of 991 patients prescribed nalmefene and four out of 901 prescribed placebo.
Among patients being treated for alcohol dependence, there is no high-grade evidence to support the use of nalmefene to reduce alcohol-related harm, and little evidence that it reduces alcohol consumption. The value of nalmefene in the treatment of alcohol dependence is not established. Importantly, these findings reveal a lack of information on clinically relevant outcomes in the evidence that led to nalmefene’s approval by the European Medicines Agency.
The value of nalmefene in the treatment of alcohol dependence is not established
It follows that they also call into question the decisions of this and other regulatory and advisory bodies which have approved nalmefene on the basis of the available evidence from randomised controlled trials, and highlight the need for further trials of the drug compared both to placebo and to naltrexone for the drinking disorders specified in the approvals.
No randomised trials were found which compared nalmefene versus placebo in the specific population for which the European authority approved Selincro – alcohol-dependent adults drinking over 60g a day for men or 40g for women. In this population, the only available data are pooled analyses derived from subgroups of the total samples in two or three of the trials, subgroups not specified before the results were known. The credibility of such analyses is usually low and they are not considered to confirm that the intervention is effective, only to identify avenues worth pursuing in trials designed from the start for this purpose.
No trials were found which compared nalmefene with another medication, but current evidence suggests there is little or no difference in efficacy in reducing heavy drinking between nalmefene and naltrexone. Although naltrexone is not authorised for reducing drinking as opposed to promoting abstinence, reduction is probably its main effect.
Clinicians must be aware that the value of nalmefene for the treatment of alcohol addiction is not established. At best, nalmefene has limited efficacy in reducing alcohol consumption.
commentary The story of Selincro’s entrance in 2013 into the European market and its subsequent approval for NHS patients in the UK has important ramifications for how heavy drinking is regarded and (not) treated. To corroborate the following commentary we analysed the story in detail, and provide this detail in the form of background notes. Links to relevant passages in the notes are included in the commentary. The notes are also available at the end of this commentary in the form of headings with an ‘eye-opener’ link which you can click on to unfold the text under that heading.
Under the trade name Selincro, nalmefene’s marketing authorisation by the European Medicines Agency in 2013 ( panel) paved the way to realising the hope that it will help tackle the bulk of dependent drinking lying below the iceberg-tip of physically dependent drinkers aiming for abstinence – and at the same time open up for Danish manufacturer Lundbeck a huge market previously all but closed to pharmaceutical solutions. The same prospect – but seen as a chimera – has also stimulated criticism from observers concerned at what in their eyes could prove an expensive and inappropriate medicalisation of lesser degrees of dependence based on unproven effectiveness.
Lundbeck may have carved out a potentially profitable marketing niche for Selincro, but this does not mean there is no substance to claims that the medication is effective and distinctive in helping dependent drinkers cut back. The questions on this front are twofold: whether in this role the drug has reliably been shown to be more effective than an inactive placebo; and if it has been, whether it has been shown more effective than established treatments, especially its less expensive parent drug naltrexone and ‘talking’ therapies, including inexpensive brief interventions.
Having reanalysed the available trials and been granted access to the data which underpinned authorisation, on both counts the featured analysis delivered a ‘not proven’ verdict, the trials neither offering statistically significant and reliable evidence of improved health nor (once patients lost to follow-up were accounted for) reduced drinking, while the absence of head-to-head trials against naltrexone left open whether the medication has the distinct advantages claimed for it.
Critical weaknesses in the case for Selincro are the unreliability of the analysis of outcomes from a selection of the patients in the trials which underpinned authorisation and subsequent approvals, and the apparently equivalent effects of naltrexone at a much cheaper price. Based on current evidence, the innovation in Selincro is not nalmefene nor the way it has been trialled in the treatment of alcohol dependence, but its market re-positioning by Lundbeck for a population of patients (not in need of detoxification, not physically dependent, wanting still to drink, probably not in specialist treatment) and for an objective (reducing consumption rather than sustaining abstinence) for which no medication had yet been specifically authorised. However, absence of reliable evidence might not mean there are no appreciable benefits, just that these have yet to be demonstrated. These conclusions are expanded on in the following text.
Their findings led the authors of the featured review to question whether the drug’s record warranted European authorisation. Highlighted as a major limitation of the Lundbeck-sponsored trials which led to authorisation was the substantial minority of patients lost to follow-up. Once this reaches high levels, there is no reliable way to be sure what the outcomes really were across all the patients in a trial.
Faced with this problem, the featured analysis assumed patients lost to follow-up were drinking the same as before joining the trials, an assumption which would be valid to the degree to which complete relapse was the reason for their non-response – conceivably the case for many drop-outs, but almost certainly not all. On this basis – one supported by the company and by the European Medicines Agency – it was estimated that placebo patients had reduced drinking by about the same as nalmefene patients, a result which must have been partly due to this pessimistic assumption having to be made for more nalmefene than placebo patients. But to base outcomes on just those patients who did complete final assessments would also have been unreliable.
Lundbeck’s trials themselves depended on a different strategy to account for missing patients: using all the available data including any interim follow-ups to estimate final outcomes. However, in the end this too relies on extrapolation from patients who complete final assessments, and also on the unlikely assumption that researchers have accounted for all the factors related both to drop-out and to the assessed outcomes. On this basis, relative to placebo the primary measures of drinking – total consumption and heavy drinking days at the 24-week follow-ups – had been reduced to a statistically significant degree by nalmefene in three of the tests made across the three studies, but not in another three. Across the three trials, most ways of assessing the primary drinking outcomes left nalmefene without a statistically significant advantage over a placebo.
Faced with patchily positive results, Lundbeck and their research associates conducted ‘sub-sample’ analyses which excluded over half the participants in the trials, including medium-risk drinkers and those at higher risk who even before treatment started had rapidly remitted to a lower risk level on joining the trials. From an earlier analysis and because this selection was made after trial results were in, it would have been known that the rapid remitters tended to stay that way. The effect was to exclude patients who left any treatment little to improve on. What remained were a higher risk sub-sample who remained at high risk when nalmefene treatment started. Among these patients, the drug had greater scope to prove it could reduce drinking, and the results were more consistently positive, but at the cost of reliability.
Sub-sampling deprived the resulting findings of the assurance of a level playing field created by randomisation of the original samples, and left the outcomes vulnerable to manipulation and bias – the reasons why they are considered suitable only for suggesting possibilities, not proving effectiveness. Particularly worrying is the statement in the European Medicines Agency’s assessment that sub-sampling had been “proposed” by Lundbeck “in order to define a population where the benefit of Selincro would be greatest”; it seems that not just the effect but the intention was to find a way of dividing the sample which would be advantageous to Selincro. Sub-sampling also left in the analysis a small and probably atypical set of drinkers, recruited on average just once every four months at each research site. Together with the multiple reasons for excluding applicants from the trials, it meant the results could not be relied on as an indication of nalmefene’s likely impact among the generality of drinkers considered eligible for the medication.
Lundbeck’s trials and European authorisation were based on assessment of whether nalmefene was better than (pharmacologically) nothing in the form of an inactive placebo when both were accompanied by a low-intensity psychosocial programme whose main aim was to enhance compliance with treatment. But that is not enough to justify publicly funding the drug’s supply to National Health Service patients. To capitalise on the authority to market Selincro, its manufacturers had to persuade authorities like Britain’s National Institute for Health and Care Excellence (NICE) that it offered acceptable levels of health-improving and lifesaving benefit per £ of public resources, and that either there were no alternatives, or that these offered less net benefit.
The comparator in Lundbeck’s trials amounted to a psychosocial programme whose main aim was to get patients to take inactive pills – not a real-world alternative to nalmefene. More realistic alternatives were an active medication accompanied by psychosocial support, or a standalone psychosocial therapy focused on reducing drinking and related risks.
Pharmacologically, the obvious alternative was naltrexone. On that front, the company’s argument was that the remit specified in its approvals meant it was not an alternative for the patients and uses envisaged for Selincro. Despite not being licensed for Selincro’s role, in fact naltrexone has successfully been used for essentially the same types of drinkers and taken in the same way – ‘as needed’ in anticipation of drinking. That naltrexone has not been specifically authorised for these purposes seems mainly down to no pharmaceutical company having pursued such authorisation; little would be gained from promoting new uses for a medication exposed to generic competition from products no longer protected by patents.
On psychosocial alternatives, the argument was that in practice these were often similar to the approach whose effects nalmefene augmented in the Lundbeck trials. However, the comparator in the original remit for the decision to be made by NICE was not a brief package to promote the taking of inactive pills, but the fully-fledged, standalone psychological and social therapies NICE had previously recommended for the types of drinkers targeted by Selincro.
In respect both of naltrexone and psychosocial therapies, no direct comparisons were available in the form of studies which had randomly allocated patients to nalmefene versus these alternatives. It left a crucial blank in the data available to the national authorities which held the keys to releasing public funds for Selincro. In the end, they largely based their decisions on the data they did have from Lundbeck: that according to an analysis unsuited to demonstrating effectiveness, among a sub-sample of high risk and perhaps also highly atypical drinkers, paired with a basic psychosocial programme intended to get patients to take pills as directed, it was slightly better if those pills were an active medication in the form of nalmefene than an inactive placebo.
On this unreliable foundation substantially rested the case for allowing the company to market Selincro in Europe and subsequent decisions to mandate Britain’s National Health Service to make it available to patients. In self-justifying circles, during the European authorisation process Lundbeck conducted the sub-sampling analysis in order to leave nalmefene performing more convincingly than in the full samples,
Each link in the chain of approvals retained the vulnerability to bias of the initial analysis
which in turn justified authorisation for these kinds of drinkers, which then justified a published analysis focused on these types of drinkers, all of which justified cost-effectiveness analyses (1 2) limited to this sub-sample, leading to the decision that the NHS must make it available for those purposes – a chain of decisions, each link of which retained the vulnerability to bias and questionable applicability to the generality of patients introduced by the initial sub-sampling.
For Selincro and for Lundbeck, the unsatisfactory result is that relative both to a placebo and to alternative treatments, an unproven verdict is the only scientifically defensible one. Even assuming a real effect, this generally small advantage will have been bought at the cost of the unpleasant side effects experienced by a substantial minority of patients. In the three Lundbeck-funded trials on which European authorisation was based, nausea afflicted around a fifth to a quarter and vomiting from 6% to 18%, incidents much rarer on placebos.
Aggravating uncertainties about the science is the deep involvement of the manufacturers in all the main trials on which marketing authorisation and subsequent NICE approval were based. For Lundbeck, it was very important to profit from this new product as revenue fell from older products whose patents had expired, leaving them exposed to generic competition. Highlighting these “vested interests” has been a feature of criticisms of how marketing and approval decisions were made, fuelled partly by concern that heavy drinking will become medicalised rather than dealt with by psychosocial support or self-help, and that in particular the space in public policy hard-won for brief interventions will become occupied by a relatively costly medication. For people who see dependent drinking as fundamentally a psychological and social problem, it represents an unwelcome extension down the severity range of an opposing vision of the condition as due to neurochemical processes correctable by a drug which targets opiate receptors in the brain.
The opposing risk is of throwing the baby out as well as the bathwater
The opposing risk is of throwing the baby out as well as the bathwater – the risk that despite scientific imperfections and strongly vested interests, Selincro really can help fill an important gap in the reduction of alcohol-related harm. Absence of reliable evidence might not mean there are no appreciable benefits, just that these have yet to be demonstrated. Randomised trials are not easy to conduct; imperfections are inevitable, as usually is questionable applicability to the intended population. Failing to implement interventions found effective in these trials risks depriving sufferers of valid sources of relief. Some of Selincro’s marketing advantages for Lundbeck are also reasons why dependent drinkers who do not see themselves as ‘alcoholics’ in need of treatment might be prepared to give it a try. Small effects there may be, but if they are real and are multiplied over many thousands of people drinking at risky levels, they could cumulate to a worthwhile public health gain.
Fears too about a retrograde influence of the introduction of Selincro may not materialise. Though it could happen, there is no inevitability about Selincro gaining ground at the cost of brief interventions, which are targeted mainly at people not seeking treatment, nor at the cost of psychosocial therapies for treatment seekers, which perhaps appeal to a different kind of patient treated in different circumstances to that envisaged for Selincro. Nor is the fact that nalmefene is a medication necessarily a challenge to psychological understandings of addiction. Though the manufacturers portray alcohol dependence as a brain disease caused by repeated drinking, the proposed mechanism of action of nalmefene – diminishing the rewarding effects of alcohol – is compatible with understandings of dependence as an acquired habit, theories which underpin the cognitive-behavioural approaches popular in the addictions. In this scenario, nalmefene just makes it easier to lose the habit. A bonus for many will be that the medication’s approvals explicitly endorse controlled drinking as a treatment objective, a step towards extending the legitimacy of this contested route out of dependent drinking.
These themes in the story of what has become a highly controversial product attracting great expectations for extending health gains and correspondingly major criticism are expanded on under the headings below, which duplicate the background notes. For technical reasons, in the text below internal links to parts of the text which are still hidden will not work, but these links should work as expected in the background notes file.
Thanks for their comments on this entry in draft to research author Florian Naudet of Stanford University in the USA, and to Matt Stevenson of the University of Sheffield in England, Alain Braillon of the Department of Medicine at University Hospital in Amiens, France, and James Morris of the Alcohol Academy based in England. Commentators bear no responsibility for the text including the interpretations and any remaining errors.
Last revised 01 September 2016. First uploaded 10 August 2016
Background notes to this entry
Comment/query to editor
Give us your feedback on the site (one-minute survey)
Open Effectiveness Bank home page
Add your name to the mailing list to be alerted to new studies and other site updates
REVIEW 2017 Pharmacologically controlled drinking in the treatment of alcohol dependence or alcohol use disorders: a systematic review with direct and network meta-analyses on nalmefene, naltrexone, acamprosate, baclofen and topiramate
HOT TOPIC 2017 Overdose deaths in the UK: crisis and response