Comparing Nalmefene and Naltrexone in Alcohol Dependence: Are there any Differences? Results from an Indirect Meta-analysis – Comment to Naudet

Florian Naudet has drawn up a letter to the editor [1] to comment on our paper “Comparing Nalmefene and Naltrexone in Alcohol Dependence: Are there any Differences? Results from an Indirect Meta-analysis” [2] in this journal. We would like to comment on this manuscript.

The aim of our paper was to compare nalmefene and naltrexone in efficacy and safety parameters. Due to a lack of a direct clinical comparison of both drugs, we performed an indirect comparison of nalmefene and naltrexone. The methodological approach is generally accepted in the scientific community as discussed in the publication.

Naudet listed some results of the meta-analysis and concluded that the superiority of nalmefene over naltrexone shows no statistically significant difference in any endpoint. It is relevant to differentiate the relevance of the patient-relevant outcome parameters. We analyzed the endpoints that are recommended by the EMA as primary endpoints in the “Guideline on the development of medicinal products for the treatment of alcohol dependence” (EMA, 2010) [3]. Primary endpoints in clinical trials with the goal of intermediate harm reduction are change to baseline in total consumption of alcohol as well as reduction in number of heavy drinking days. Both are considered primary endpoints (chapter 4.2.1, EMA 2010) [3]. Both endpoints can be measured in multiple ways as listed in Table 1 of the publication. Thus we decided to perform the meta-analysis as a cluster analysis. This is a generally accepted approach (see “Discussion” section of our paper). The “quantity of drinking” endpoint represents the more important endpoint.

Table 1 Effect sizes

Cluster	Week	Nalmefene	Naltrexone	Difference
Frequency	24	–0.336	–0.145	–0.191
	12	–0.275	–0.071	–0.204
Quantity	24	–0.372	–0.108	–0.264
	12	–0.371	–0.227	–0.144

The EMA recommends a treatment duration in confirmatory trials with the goal of harm reduction of at least 3 months (chapter 4.3.3, EMA 2010) [3]. That is why the result “change from baseline in the cluster quantity of drinking” after 24 weeks of treatment is the most relevant result (p=0.022). The 12-week results are data from sensitivity analyses.

Naudet cites as problematic that we selected RCTs according the Summary of Product Characteristics of the drugs. This selection criterion was carefully used because we wanted to compare both drugs against the background of real daily practice by physicians and on the basis of the authorization from both drugs.

All included RCTS with nalmefene were performed in the specific population defined by EMA. It is correct that the trials with nalmefene were initially performed in patients suffering from total spectrum of total alcohol consumption. The EMA published the “Guideline on the development of medicinal products for the treatment of alcohol dependence” during performance of the clinical research program of nalmefene. This guideline defines the inclusion criteria: “Patients included in the main trials should, besides fulfilling criteria for alcohol dependence, have (harm reduction study) or had (relapse prevention study) a high or very high level of total alcohol consumption at baseline (total consumption per month, presented on an average daily basis.” Thus the nalmefene studies were re-calculated in this patient population. This patient population corresponds to the approved indication of nalmefene.

Further statistical issues: Naudet discussed a clinical study performed for 4 endpoints in order to keep a global significance level of 0.05, and he recommended an α-adaption according to Bonferroni to 0.0125 for 4 separate significance tests. However, this was a substantially wrong approach for 2 very different reasons:

1. If we accept the approach of a significant discrimination of both compounds, then we must refer to the “Efficacy” section of the publication where the evaluation was explicitly based on 2 clusters of criteria quantified at the end of the planned maximum treatment period of 24 weeks. As mentioned above, the 12-week results are data from sensitivity analyses that were performed with interim results after 12 weeks of the nalmefene studies and results at study end of several naltrexone studies originally performed over 12 weeks (interim data of longer-lasting naltrexone studies were not available). The12-week analysis was a hierarchically subordinated assessment. But if only 2 tests were countable, the significance of the quantity of drinking could not be cancelled by the ‘Bonferroni trick’.

2. We do not accept Naudet’s approach concerning the significant discrimination of 2 groups in the sense of randomised clinical trials; we presented meta-analytically synthesised effect sizes within both groups which were clinically very relevant for both clusters and both times of control. These effect sizes were (see [Table 1]):

Effect sizes of approx. 0.2 were generally regarded as small but unneglectable. Differences of effect sizes can be assessed accordingly.

The message of our paper is:

a) Nalmefene showed (slightly) moderate effects with regard to frequency and quantity of drinking after 24 (and already after 12) weeks of treatment.

b) Naltrexone showed a relevant effect size regarding quantity of drinking only after 12 weeks, which was not confirmed in the analysis of longer-lasting studies.

c) In the main assessment of the nalmefene effect sizes, the difference vs. naltrexone was higher than the naltrexone effect vs. placebo. This can be regarded as an additional benefit of nalmefene over naltrexone.

In conclusion, we want to repeat our point of view stated in the “Discussion” section of the paper: “Data from our indirect meta-analysis may give the first indication that nalmefene is superior to naltrexone in the treatment of alcohol dependence, at least on the basis of a harm reduction”.

• References

• 1 Naudet F. Comparing nalmefene and naltrexone in alcohol dependence: Is there a spin?. Pharmacopsychiatry 2016; 49: 260-261
• 2 Soyka M, Friede M, Schnitker J.. Comparing nalmefene and naltrexone in alcohol dependence: are there any differences? Results from an indirect meta-analysis. Pharmacopsychiatry 2016; 49: 66-75
• 3 EMA (2010). Guideline on the development of medicinal products for the treatment of alcohol dependence. EMA/CHMP/EWP/20097/2008