Correspondence
Alpha-Tocopherol and Alzheimer's Disease
N Engl J Med 1997; 337:572-573August 21, 1997
- Article
To the Editor:
Sano et al. (April 24 issue)1 conclude that alpha-tocopherol and selegiline are effective in slowing the progression of moderately severe Alzheimer's disease. Did they prove a true effect of treatment, or can the result be reasonably attributed to bias arising in the design, performance, or analysis of the trial?
Presumably, they designed and performed their study as a double-blind, randomized, controlled trial in order to reduce bias. The verdict of the unadjusted analysis, which did not show a statistically significant advantage in any of the treatment groups, was not accepted. When they used the Cox proportional-hazards model to adjust for base-line differences in scores on the Mini–Mental State Examination, they found significant differences in outcome in the alpha-tocopherol group, the selegiline group, and the group receiving both treatments in combination. The Cox model2,3 introduced assumptions concerning the hazard functions2 of the compared groups and therefore concerning their survival functions. No attempt was made to justify these assumptions, which seem to have been made for mathematical convenience. If the assumptions do not reflect reality, the analysis is biased. In particular, if the survival curves for the compared groups cross, the assumption of proportional hazards is violated.3 The survival curves for the selegiline group and the combined-treatment group both cross the curves for the placebo group (Figure 1A and Figure 1C of the article). Although it is hard to tell, the survival curve for the alpha-tocopherol group may also cross the curve for the placebo group (Figure 1B). These facts suggest that the statistical adjustment is biased, that therefore no true effect of treatment has been proved, and that the internal inconsistencies noted may simply reflect the bias in the analysis.
3 ReferencesMartin M. Pincus, M.D., Ph.D.
New York Methodist Hospital, Brooklyn, NY 112151
Sano M ,Ernesto C ,Thomas RG , et al. A controlled trial of selegiline, alpha-tocopherol, or both as treatment for Alzheimer's disease. N Engl J Med 1997;336:1216-1222
Full Text | Web of Science | Medline2
Lagakos SW. Statistical analysis of survival data. In: Bailar JC III, Mosteller F, eds. Medical uses of statistics. 2nd ed. Boston: NEJM Books, 1992:281-91.
3
Concato J ,Feinstein AR ,Holford TR . The risk of determining risk with multivariable models. Ann Intern Med 1993;118:201-210
Web of Science | Medline
To the Editor:
Sano et al. claim that a high daily intake of vitamin E (alpha-tocopherol) is beneficial in slowing the progression of Alzheimer's disease, when institutionalization is used as an end point. However, their study fails to show any effects of alpha-tocopherol on measurements of cognitive function, nor did the combination of alpha-tocopherol with selegiline have any convincing positive effect.
We hypothesized that if vitamin E enhances neuronal survival, there would be a positive relation between serum concentrations and cognitive function in people without dementia. Among 70-year-old men living in the community, serum concentrations of alpha-tocopherol and gamma-tocopherol were measured in a consecutive sample of 654 subjects participating in a population-based cohort study of vascular risk factors and cognitive function, with 1221 subjects. Values were expressed as the ratio of the serum tocopherol concentration to the sum of the serum concentrations of cholesterol and triglycerides.1 The psychometric testing included the Mini–Mental State Examination 2 and the Trail Making Tests.3 Measurements of tocopherol and cognitive function were available for 547 men. We found no indication of a relation between the lipid-corrected tocopherol values and the cognitive-test results. Neither was there a subgroup of men with both low cognitive function and low serum tocopherol values, after we adjusted for socioeconomic factors.
There was a strong positive relation between educational level and the corrected serum alpha-tocopherol value. The mean (±SD) alpha-tocopherol value was 1.57±0.02 in 307 men with seven years of education or less, 1.65±0.02 in 170 men with a secondary education, and 1.72±0.03 in 70 men with a university education (P<0.001 for the comparison of low vs. high level of education). An inverse association between intake of alpha-tocopherol and mortality from coronary causes has been proposed.4 Intake of alpha-tocopherol may mediate the relation between socioeconomic status and cardiovascular risk. From our data, tocopherol concentrations have no relation to cognitive function in a general population of elderly persons. We conclude that alpha-tocopherol has no major effect on cognitive decline.
4 ReferencesLena Kilander, M.D.
Margareta Öhrvall, M.D., Ph.D.
Uppsala University, S-751 25 Uppsala, Sweden1
Ohrvall M ,Tengblad S ,Vessby B . Lower tocopherol serum levels in subjects with abdominal adiposity. J Intern Med 1993;234:53-60
CrossRef | Web of Science | Medline2
Folstein MF ,Folstein SE ,McHugh PR . “Mini-Mental State“: a practical method for grading the cognitive state of patients for the clinician. J Psychiatr Res 1975;12:189-198
CrossRef | Web of Science | Medline3
Lezak MD. Neuropsychological assessment. 3rd ed. New York: Oxford University Press, 1995.
4
Knekt P ,Reunanen A ,Jarvinen R ,Seppanen R ,Heliovaara M ,Aromaa A . Antioxidant vitamin intake and coronary mortality in a longitudinal population study. Am J Epidemiol 1994;139:1180-1189
Web of Science | Medline
Author/Editor Response
The authors reply:
To the Editor: We agree that when factors such as the effect of treatment are studied for their impact on risk, the assumptions and limitations of the multivariable method become especially important for the accuracy of results and the validity of interpretations. We also agree that “when the Cox method is used, the independent variables should be evaluated for adherence to the assumption of proportional hazards.” 1 This evaluation can be performed by calculating Schoenfeld residuals.2 If the assumption of proportional hazards is met, the Schoenfeld residuals should be randomly distributed across time.3 Table 1Table 1
Tests of the Significance of the Proportional-Hazards Assumptions for the Variables Used in the Cox Model. shows the results of formal tests of the significance of the proportional-hazards assumptions for each variable, based on the correlation coefficient between the survival time and the scaled Schoenfeld residuals.3 For each of the three primary comparisons in our report,4 a correlation coefficient (R) between the Schoenfeld residual and time and a two-sided P value are given. Clearly, none of the correlations are significant.Pincus raises the issue of the crossing of the survival curves in Figure 1 of our article. It is important to distinguish between the true, underlying survival curves and the estimated Kaplan–Meier curves. The Kaplan–Meier curves must cross and their associated confidence bands must separate for there to be a violation of the proportional-hazards model. This was not seen in our data.4 Thus, the figure in our article and the statistical data provided here consistently support the conclusion that the assumptions of the proportional-hazards model were not violated.
The interesting retrospective, cross-sectional data of Kilander and Öhrvall support the notion that serum tocopherol levels in a cohort not receiving supplements do not relate to measures of cognitive function in healthy elderly subjects. But this finding does not speak to the issue of the effect of supplementation or to the likelihood of the development of cognitive decline. Since few populations have substantial exposure to alpha-tocopherol supplementation, it is difficult to assess the effect of this regimen through epidemiologic studies. We reiterate the need for double-blind, randomized clinical trials in appropriate populations (such as at-risk elderly persons) to determine whether there is evidence that cognitive deterioration is slowed.
4 ReferencesMary Sano, Ph.D.
Columbia University College of Physicians and Surgeons, New York, NY 10032Ronald G. Thomas, Ph.D.
Leon J. Thal, M.D.
University of California, San Diego, La Jolla, CA 920371
Concato J ,Feinstein AR ,Holford TR . The risk of determining risk with multivariable models. Ann Intern Med 1993;118:201-210
Web of Science | Medline2
Grambsch PM ,Therneau TM . Proportional hazards tests and diagnostics based on weighted residuals. Biometrika 1994;81:515-526
CrossRef | Web of Science3
Therneau T. Splus 3.3 guide to statistical and mathematical analysis. Seattle: Mathsoft, 1995:20-11–20-15.
4
Sano M ,Ernesto C ,Thomas RG , et al. A controlled trial of selegiline, alpha-tocopherol, or both as treatment for Alzheimer's disease. N Engl J Med 1997;336:1216-1222
Full Text | Web of Science | Medline
- Citing Articles (7)
Citing Articles
1
Goran Bjelakovic, Dimitrinka Nikolova, Lise Lotte Gluud, Rosa G Simonetti, Christian Gluud, Goran Bjelakovic. 2008. Antioxidant supplements for prevention of mortality in healthy participants and patients with various diseases. .
CrossRef2
Marie-Laure Ancelin, Yves Christen, Karen Ritchie. (2007) Is Antioxidant Therapy a Viable Alternative for Mild Cognitive Impairment? Examination of the Evidence. Dementia and Geriatric Cognitive Disorders 24:1, 1-19
CrossRef3
Jacqueline Birks, Leon Flicker, Jacqueline Birks. 2003. Selegiline for Alzheimer's disease. .
CrossRef4
Silvia Mandel, Edna Gr??nblatt, Peter Riederer, Manfred Gerlach, Yona Levites, Moussa B H Youdim. (2003) Neuroprotective Strategies in Parkinson???s Disease. CNS Drugs 17:10, 729-762
CrossRef5
Benoit I. Giasson, Harry Ischiropoulos, Virginia M.-Y. Lee, John Q. Trojanowski. (2002) The relationship between oxidative/nitrative stress and pathological inclusions in Alzheimer’s and Parkinson’s diseases1,2 11Guest Editors: Mark A. Smith and George Perry 22This article is part of a series of reviews on “Causes and Consequences of Oxidative Stress in Alzheimer’s Disease.” The full list of papers may be found on the homepage of the journal.. Free Radical Biology and Medicine 32:12, 1264-1275
CrossRef6
Ippolita Cantuti-Castelvetri, Barbara Shukitt-Hale, James A Joseph. (2000) Neurobehavioral aspects of antioxidants in aging. International Journal of Developmental Neuroscience 18:4-5, 367-381
CrossRef7
B BAIR. (1998) FREQUENTLY MISSED DIAGNOSIS IN GERIATRIC PSYCHIATRY. Psychiatric Clinics of North America 21:4, 941-971
CrossRef
