Memantine in Alzheimer’s disease: reliable analyses are required

In September 2009, the German Institute for Quality and Efficiency in Health Care (IQWiG) concluded that no benefit has been proven for memantine in the treatment of Alzheimer's disease. An assessment now presented of 2 previously unpublished studies and new analyses by the manufacturer Merz do not change this conclusion. However, IQWiG describes the deficits of the newly provided analyses and specifically illustrates how the analyses would have to be presented in order to be considered in a benefit assessment.

Merz provides additional data and analyses to the G-BA

Merz finally saw reason. Since 2006 IQWiG had repeatedly asked the pharmaceutical company Merz (from Frankfurt, Germany) to provide unpublished data from 2 studies on memantine. Although, at least for one study, Merz had repeatedly announced their intention to fulfil this request, the 2 clinical study reports requested were never provided. Therefore, IQWiG completed its final report without these studies in July 2009. However, at the beginning of 2010 the company submitted the results of both memantine studies to the German Federal Joint Committee (G-BA). At the same time, Merz presented additional analyses to the G-BA, which they had calculated themselves on the basis of several studies; these analyses had also not been published previously.

IQWiG assessed these new data in a working paper and published its conclusions on 2 August 2010. "Due to the way the additional analyses have so far been presented by Merz, they cannot be utilized,” says Thomas Kaiser, Head of IQWiG's Department of Drug Assessment. "However, it seems possible that a benefit of memantine might be shown in the area of cognition if the data were analysed in an appropriate manner.”

For the working paper the analyses of the final report were repeated including the additional studies. Patients with moderate to severe Alzheimer's disease had participated in both studies. However, the new data did not affect the conclusions of the final report published in 2009, i.e. that no benefit of memantine had been proven. The results of the final report are described .

In principle, responder analyses are well-suited

IQWiG still views critically the additional analyses presented by Merz. These are called "responder analyses”. In this type of analysis it is counted how many patients experience a noticeable improvement in health status after treatment ("responders”). For this purpose it is determined how large the difference in health status must be in a patient to be classified as an improvement. Then all patients in both the memantine and placebo group who achieved at least this degree of improvement are counted as responders. According to the analyses performed by Merz, statistically significantly more patients taking memantine showed an improvement in cognitive function compared to placebo.

"In principle, this type of analysis is well-suited for the benefit assessment of an intervention,” says Thomas Kaiser. However, the specific calculations in the documents presented by Merz leave essential questions unanswered. On the one hand, for example, once again not all of the relevant studies were considered in the analysis, and on the other, the general standards for statistical evaluations were not followed. The working paper includes information on how an adequate calculation should be performed. "If new, and then hopefully comprehensible data are presented, we will again review our conclusions,” says Thomas Kaiser.

Contact: Tel. ++49(0)221-35685-0, info@iqwig.de