Drug Treatment Outperforms Placebo in Advanced Alzheimer's

I applaud Howard et al. for attempting a clinical trial that has great potential public health significance. I and surely others also appreciate the support provided by the Medical Research Council. As part of a team committed to caring for and studying people with Alzheimer's disease at all stages, I am particularly grateful that there is continued interest in trying to improve the welfare of our fellow citizens who are experiencing more advanced dementia.

Trying to comment within the rapid news cycle means I am doing so before having had a chance to track down the paper's supplement. That said, there are a number of initial points and a potential concern that I would like to raise. It seems the headline goes something like this: Patients in this trial with moderate to severe Alzheimer's disease who were previously treated with donepezil did better in terms of simple measures of cognition and daily functioning if they stayed on donepezil or started on memantine. In contrast, those individuals who had been on donepezil and were switched to placebo declined... Read more

While the limited instruments used in the trial are not particularly good at informing clinical decision-making, the results were nonetheless fairly stark and, therefore, in my opinion, highly clinically relevant. Any active treatment was beneficial: continuing donepezil, switching to memantine, or adding memantine. I note, however, that despite statements in the article to the contrary, there is no actual evidence of lack of effect of adjunctive memantine therapy. There are several reasons for saying this. First, the sample sizes at the study's end were small, ranging from 20 to 38 participants per group. Further, dropout rates between the groups were markedly different. There were also important baseline differences between the groups. It is important to be aware that, under these circumstances, the failure to demonstrate a "statistically significant" effect does not mean or even imply that there is no effect. Furthermore, inspection of the data indicates that there were differential effects by treatment groups for a large part of the treatment period for both co-primary endpoints; in my view, this is a clinically relevant finding.

I further note that the decline in measures of cognition and daily functioning among individuals treated with placebo was precipitous in this trial. Individual, let alone group, declines of this magnitude would typically have prompted concern during the course of a blinded trial that harm may have been done to patients, with possible implications for trial continuation. This issue is important. It is not addressed in the paper. The reader is left to wonder how patients were monitored and what the data safety monitoring board’s specific charter and role were in this trial, noting that the ethics of withdrawing treatment and putting patients on placebo should be considered carefully in designing and conducting such a trial.

There are some other methodological and operational considerations that can be raised. First, despite a two-year enrollment period, only 295 of an intended sample size of 800 were enrolled. The reasons for the shortfall are not explained. Second, an interim analysis was performed for unclear reasons in a manner not specified in the manuscript, indicating that the sample size could be reduced from 800 to 430. Interestingly, the funding agency then requested that the study be stopped before being completed according to the revised plan. The details of this unusual decision are unclear, although it was stated that it was important to report results as quickly as possible. Such a decision comes at a cost: Several goals of the study could not be achieved, yet the paper suggests that they were.

Alzforum requested a comment specifically on the statement in the NEJM that, "The findings of the study showing that combination therapy with memantine and a cholinesterase inhibitor was more effective than treatment with a cholinesterase inhibitor alone (Tariot et al., 2004) have not been replicated (Porsteinsson et al., 2008)." First, neither the present study, nor the JAMA paper referred to, for which I was an author and investigator, address "combination" therapy, but adjunctive therapy. The two are quite distinct. Second, a multicenter trial of adjunctive memantine in patients with moderate to severe Alzheimer's disease has been completed, and it replicates the findings reported in the 2004 JAMA paper. Its results have been presented publicly at international scientific conferences and are available at ClinicalTrials.gov. The statement quoted here is thus incorrect. Third, the citation used to indicate lack of replication (Porsteinsson et al.) refers to a study in patients with milder, rather than moderate to severe AD, and hence is not comparable.

Finally, there are some other differences between these two studies: design, sample size and power, state of equipoise, duration (six vs. 12 months), outcome measures (MMSE vs. SIB), dropout rates, and interim analyses. Interestingly, despite these differences, both studies still showed a positive effect on emerging behavioral symptoms measured by the NPI.

Taken together, the studies are very different, and lead to different conclusions. The bottom line, in my view, is that this was a very important undertaking. It achieved some important results that are likely to affect clinical practice and may influence considerations by NICE. Hopefully, those results that are less clear can be reframed so as to minimize confusion and avoid incorrect impact on best medical practice.

View all comments by Pierre Tariot