Nalmefene (Selincro) is an opioid antagonist medication that was approved for the treatment of alcohol dependence by the European Medicines Agency (EMA) in 2013 and recommended by the National Institute for Health and Care Excellence in the UK in 2014 (NICE, 2014). Its approval was controversial, because some doctors were skeptical about the quality of the evidence (see Braillon, 2014; Spence, 2014).
Two recent articles are likely to revive this debate, and they raise broader issues about why the drug was licensed and recommended as a treatment for alcohol dependence:
- The first paper (Palpacuer et al., 2015) is a systematic review and meta-analysis of all published and unpublished double-blind randomised controlled trials (RCTs)
- The second paper (Fitzgerald et al., 2016) is a position piece that highlights a number of limitations with those RCTs and their interpretation.
Palpacuer et al. (2015) identified nine double-blind RCTs through searches of the published literature, trial registries, and by direct contact with pharmaceutical companies. Seven of these trials were published and two were unpublished. Their primary meta-analysis was limited to five RCTs that compared nalmefene to placebo and had a follow-up period of at least six months. The other studies were included in additional sensitivity analyses.
Fitzgerald et al. (2016) discussed these same trials, plus one additional unpublished trial.
Palpacuer et al.’s (2015) primary meta-analyses on five RCTs that had a follow-up period of at least six months indicated that there was no difference between nalmefene and placebo on mortality or quality of life.
At 6 month follow-up, there were equivalent numbers of serious adverse events in participants who received nalmefene or placebo (RR = 0.89, 95% CI = 0.55 to 1.43. However, there were more minor adverse events in participants who received nalmefene (RR = 1.18, 95% CI = 1.11 to 1.24) and, possibly related to this, participants were more likely to withdraw from the study if they had received nalmefene (RR = 1.27, 95% CI = 1.01 to 1.59).
The most important outcomes are those related to alcohol consumption: did nalmefene actually reduce the amount of alcohol that people drink? The primary analyses suggest a small but robust effect: participants who received nalmefene had fewer heavy drinking days each month (Mean Difference (MD) = -1.65, 95% CI = -2.41 to -0.89) and they also reduced their total alcohol consumption (Standardised Mean Difference (SMD)= -0.20, 95% CI = -0.30 to -0.10). There were similar beneficial effects of nalmefene on other indices of alcohol consumption such as scores on the Alcohol Dependence Scale.
However, the devil is in the detail. These values do not take account of the fact that participants who received nalmefene were more likely to report minor adverse events and to withdraw from the study. When this was taken into account using Baseline Observation Carried Forward (BOCF) analyses, the beneficial effects of nalmefene were no longer robust (heavy drinking days MD = -0.22, 95% CI = -1.12 to 0.68; total alcohol consumption SMD = -0.01, 95% CI = -0.11 to 0.08). It was only possible to perform this analysis on three of the five RCTs (data was not available for the other two).
Fitzgerald et al. (2016) make a number of important observations about the conduct and reporting of the three published industry-funded trials of nalmefene, as well as highlighting the results from Palpacuer’s sensitivity analyses (described above).
First, it appears that some of the primary outcome measures (heavy drinking days and total alcohol consumption) were altered after data collection had started.
Second, secondary analyses of trial data were based on a subsample of participants (approximately two thirds of the total sample) who did not reduce their alcohol consumption after 1-2 weeks of monitoring their drinking, prior to randomisation to nalmefene or placebo. This is not necessarily a problem in itself (after all, some medications may only work for some people), but it does raise a red flag because these analyses were post-hoc rather than planned before data collection started.
Third, the existing RCTs only compared nalmefene with placebo. However, nalmefene has a very similar biochemical profile to naltrexone, another opioid antagonist drug that is generic (and therefore cheap), already licensed for the treatment of alcohol dependence, and is recommended by NICE for this purpose. Indeed, the absence of this comparative data was highlighted by the German equivalent of NICE.
Finally, the industry-funded RCTs provided a novel psychosocial intervention (‘BRENDA’) to all participants. This intervention appears to be less intensive than the psychosocial interventions that are recommended by NICE for the treatment of alcohol dependence, so it remains to be seen whether the superiority of nalmefene compared to placebo would still be evident if both were offered alongside a ‘gold standard’ psychosocial intervention.
The findings from the Palpacuer meta-analysis are clear: the beneficial effects of nalmefene (versus placebo) on alcohol consumption do not appear to be robust. One could argue that the analysis that they applied to correct for study dropouts is too conservative but, even without this correction, the effect size for nalmefene was small.
One of Fitzgerald et al.’s observations is that the industry-funded trials of nalmefene emphasised findings from sub-groups of participants who did NOT reduce their drinking after 1-2 weeks of simply monitoring how much they drank. At face value this was reasonable because findings from trials of other treatments have shown that problem drinkers tend to reduce their alcohol consumption as soon as they start to monitor it, and this drop in alcohol consumption could obscure any beneficial effects of medication over placebo. Indeed, the power of simply monitoring one’s own drinking behaviour, and of the placebo effect and ‘non-specific’ responses to treatment, are very important topics for further study. But, importantly, the problem with the industry trials is that the decision to focus on these sub-groups may have been made after results were known, rather than in advance of data collection starting. As Fitzgerald and colleagues note, “(this) post-hoc sample refinement should not be regarded as anything other than hypothesis-generating”. In other words, these post-hoc data from subgroups of patients should have been followed up with an additional clinical trial in which the effect of nalmefene in this specific population was evaluated. To my knowledge, no such trial has been conducted.
These subgroup analyses are particularly important because they appear to have influenced the decisions of the EMA and NICE to license and recommend the drug, respectively. For example, the wording of the EMA approval and NICE guidance makes it clear that nalmefene should only be offered to alcohol-dependent patients who do not spontaneously reduce their drinking two weeks after assessment.
Finally, although nalmefene may bring about a small reduction in alcohol consumption in at least some people with alcohol dependence, we don’t know if the drug is better than a similar generic (and therefore cheaper) drug that is already recommended for the treatment of alcohol dependence. We also don’t know if nalmefene would help people to reduce their drinking if combined with (or directly compared with) an appropriate psychosocial intervention.
In conclusion, alcohol dependence is very difficult to treat and there is an urgent requirement for novel treatments that are more effective than those that are currently available. Unfortunately the findings from RCTs of nalmefene are not particularly compelling, and further trials are required to determine if the drug should have been recommended by NICE and the EMA for the treatment of alcohol dependence.
Palpacuer C, Laviolle B, Boussageon R, Reymann JM, Bellissant E, Naudet F. (2015) Risks and benefits of nalmefene in the treatment of adult alcohol dependence: a systematic literature review and meta-analysis of published and unpublished double-blind randomized controlled trials. PloS Medicine, doi:10.1371/journal.pmed.1001924.
Fitzgerald N, Angus K, Elders E, de Andrade M, Raistrick D, Heather N, McCambridge J. (2016) Weak evidence on nalmefene creates dilemmas for clinicians and poses questions for regulators and researchers. Addiction, in press, doi:10.1111/add.13438
NICE (2014) Nalmefene for reducing alcohol consumption in people with alcohol dependence. NICE technology appraisal guidance [TA325] Nov 2014.
Braillon, A. (2014). Nalmefene in alcohol misuse: junk evaluation by the European Medicines Agency. BMJ, 348: g2017. Doi: dx.doi.org/10.1136/bmj.g2017
Spence, D. (2014). Bad medicine: nalmefene in alcohol misuse. BMJ, 348: g1531. doi: dx.doi.org/10.1136/bmj.g1531.