Recently, there has been a proliferation of advertising about memantine, a new adjunct in the treatment of Alzheimer's Disease. The advertising is primarily in the distribution of two articles, one from the New England Journal of Medicine*, one from the Journal of the American Medical Association.2 Both studies that are reported were reasonably well done, and both produced statistically significant improvements in several measures of behavior and mentation in the study groups.
The graphs of those improvements are impressive. With their presentation of the average scores and standard errors of the study and control groups, the graphs demonstrate the statistical significance of the study data. Two of the graphs from the JAMA study are shown below. (Figure 1. Used by permission.) These graphs represent the data from the Severe Impairment Battery (SIB), and from the 19-item Alzheimer Disease Cooperative Study-Activities of Daily Living Inventory (the ADCS-ADL19). The graphs demonstrate the data for the 24 weeks of the study and include the endpoint scores with the Last Observation Carried Forward, or LOCF, score. (The LOCF score averaged the last score reported from each study subject whether he or she finished the study or not.)
A comment on the graphs is in order: They are misleading in representing the effect of memantine. The SIB graph shows a total span on the vertical axis of eight points. The actual SIB scale is from zero to 100. The ADCS-ADL19 graph in the JAMA article has a total range of six points. The ADCS-ADL19 scale has a total of 54 points. Both graphs have expanded vertical axes and compressed horizontal axes. These factors both exaggerate the differences between the study group and the control group curves. They do make the standard error bars easier to see, but they limit the perspective by abbreviating the actual ranges of the tests.
The SIB score evaluates cognitive dysfunction in the realms of memory, orientation, attention, visuo-spatial ability, and construction. It is based on a scale that goes from 0 to 100. The highest and lowest average scores from the study varied by just about five points. When the data is regraphed on a scale that represents the total 100point range of the battery, they appear as in Figure 2. (On this graph, the data are normalized so that they can be compared more easily. This is done by giving both the study and control groups a starting score of 80, even though the real starting average scores were 80.0 for the control group and 78.0 for the study group.)
The ADCS-ADL 19 Inventory measures later stages of dementia and independence in everyday tasks. It is based on a scale of 0 to 54. The greatest difference between average scores from the study is just about three points. When this data is re-graphed on a 0 to 54 point graph, it appears as in Figure 3.
Although the differences in the study groups and the control groups are indeed statistically significant, this re-graphing of data makes the results much less dramatic and might lead to a different interpretation. Whether these differences are clinically significant, that is, whether treatment is going to make a real improvement in a patient's life, is a much more difficult issue to address. Many other factors need to be considered. For example, what actual improvements are the patients making that are leading to these small improvements in their scores? Are these changes differences that are really improving their lives and the lives of their caretakers? This study was conducted on people who had Mini Mental Status Exam (MMSE) scores from 5 to 14 (out of a total of 30); in short, people who, as the article title suggests, are moderately to severely demented. The researchers did not stratify the improvements made by MMSE scores, so we do not know whether the results were similar in patients throughout the range of MMSE scores or whether the improvements varied as some function of the original MMSE score.
An important clinical difference is a very subjective measure. A two point improvement on the SIB measure might improve one person's quality of life, but a three point difference on the ADCS-ADL 19 measure might not make a difference for another person. The burden on the caretaker should also be considered. In assessing the differences suggested by this study one should consider the pre-study level of function of the subjects as well as their relative changes.
Another crucial issue that the study does not address is whether the difference in functioning between the two groups continues to enlarge over time or whether it stays at the distance measured at the end of the study. A static two-point difference in either of the measures, as shown in Figure 4, might not be of clinical significance; however, if the treated group has an overall slower rate of cognitive and functional decline, a clinical difference might almost certainly be achieved. Figure 5 demonstrates this latter relationship.
The study lasted 24 weeks. One could argue that this is a small time period for the medicine to make a clinical improvement for the patients. However, if a person has a MMSE score of 5 due to Alzheimer's type dementia, then that person has a short life expectancy, and twenty-four weeks may be a large percentage of that person's remaining life expectancy. Neither the JAMA study nor the New England Journal of Medicine study, which compared memantine to placebo without concomitant treatment with donepezil and which lasted 28 weeks, determined the effects of a longer duration.
Also, memantine, as prescribed in the study (two 5-mg tabs BID), would cost $358.36 per month at my local pharmacy, or at least $157.98 per month if more reasonably prescribed (one 10-mg tab BID). Even though we don't like to put monetary values on mental status, the first set of graphs would suggest that this would be money well-spent. Judging by the second and third graphs, the issue is less clear.
This is not to say that memantine should not be used. In interpreting the data, however, we should be careful in interpreting graphic data which may be exaggerated, and we should also realize that it is clinical improvement that we want our medications to effect not just statistical significance.
REFERENCES
1. Reisberg, B, et al, Memantine in moderate to severe Alzheimer's Disease. NEJM 2003; 348: 1333-41.
2. Tariot PN. Memantine treatment in patients with moderate to severe Alzheimer (sic) Disease already receiving donepezil. JAMA 2004; 291:317-24.
STEPHEN DAVIS, MD
Stephen Davis, MD, is a Clinical Associate Professor of Family Medicine, Brown Medical School/Memorial Hospital of Rhode Island.
CORRESPONDENCE:
Stephen Davis, MD
Department of Family Medicine
Memorial Hospital of RI
11 Brewster Street
Pawtucket, RI 02860
e-mail: AbuNoah@aol.com
Copyright Rhode Island Medical Society Jan 2005
Provided by ProQuest Information and Learning Company. All rights Reserved
Contact
Home
A
Macrodantin