A foursome of papers in the February 17 Archives of Neurology continues the shifting debates about three epidemiological questions in Alzheimer’s research.

Let’s consider first the one bit of positive results. Martha Clare Morris and colleagues at Rush-Presbyterian/St.Luke’s Medical Center in Chicago report their analysis of the Chicago Health and Aging Project, an observational community study to identify risk factors for Alzheimer’s. They found that a diet high in saturated and trans-unsaturated fats (partially hydrogenated fats found in many processed foods) correlated with an increased risk of developing AD, whereas a diet low in those fats but high in vegetable fat, as well as mono- and polyunsaturated fat, correlated with a decreased risk in a cohort of 815 white and African American people 65 years and older. The researchers followed the study participants for a mean of 3.9 years and analyzed the contents of different types of fat in a food frequency questionnaire that most of the participants had filled out once before the study began.

The Chicago study used a questionnaire similar to those developed by Walter Willet and colleagues at the Harvard School of Public Health and used in the Nurses and Physicians Health Studies conducted by these investigators. Morris’s results also roughly follow that study insofar as it found a similar pattern of fat types associated with changes in AD risk as the Harvard group had previously found for coronary heart disease (Hu et al., 1997; Ascherio et al., 1996).

Mono- and polyunsaturated fats tend to reduce LDL cholesterol, while saturated fat and trans-unsaturated fats tend to increase it. Consumption of vegetable fat was protective, but intake of dietary cholesterol did not correlate with AD risk. That last finding may still fit into the developing hypothesis on cholesterol and AD, since saturated fat intake determines serum cholesterol levels more strongly than does dietary cholesterol, the authors note.

More broadly, findings on the fat connection are far from unanimous. Morris and colleagues quote a prior study from the Rotterdam group, drawn on a larger cohort than the Chicago study, which supported a link between dietary fats and AD risk (Kalmijn, et al., 1997). Yet, since the Morris paper was accepted for publication last August, the Rotterdam researchers have reported in their own follow-up study that they could not confirm this original link (see ARF related news story). On the other hand, a study by Jose Luchsinger, Richard Mayeux, and colleagues did find that a diet high in calories and fat might increase one’s AD risk, but that study did not analyze the contribution to risk of individual types of fat, but instead focused more on the selective vulnerability to dietary fat of people with the ApoE4 allele (see ARF related news story). Morris and colleagues did not find this link between total fat intake and AD risk, leaving this topic wide open for further investigation.

Vitamins: Case Weakens
So much for the good news. The bad news concerns vitamins and estrogen. Jose Luchsinger and colleagues at Columbia University in New York poured some cold water on the widely held notion that the antioxidant vitamins C and E can reduce AD risk by counteracting the contribution of oxidative stress to the underlying neurodegeneration in AD. Only last June, both Martha Morris’s group and Monique Breteler’s group from Erasmus University in Rotterdam had found support for vitamin E, though only Breteler’s study found support also for vitamin C, (see ARF related news story). Molecular and cell biology work also supports this idea, and many people already take these vitamins. Yet, when Luchsinger, Mayeux, and others analyzed 980 elderly participants of the Washington-Heights Inwood Columbia Aging Project, they did not see an association with decreased AD risk for either of these vitamins, regardless of whether the vitamins were taken as dietary supplement or in foods. None of the three recent studies confirmed a role for carotene.

Why the discrepant results? All these studies share the limitation that food frequency questionnaires, though validated previously and widely used in epidemiology, are relatively crude instruments. They depend on people’s recollection at a given point in time of foods and supplements they take over long periods of time; they don’t measure accurately what a person actually consumes every day, or how that changes. One of the differences between the New York and the Rotterdam cohorts was that people in the New York study were older, implying that the Rotterdam study may have been able to capture a preventive effect that the New York study didn’t. Furthermore, Luchsinger et al. caution that the numerous discrepancies among these three studies raise the question whether some of the results may be due to chance. They also point to past experience with cardiovascular disease, where observational studies had implicated vitamins C and E, but subsequent clinical trials did not confirm the benefit. Experiences with NSAID treatment so far have been similar. Eventually, primary prevention trials must settle the question.

Estrogen: Serum Levels Don’t Predict Memory or Size of Hippocampus
The other set of disappointing news deals a blow to earlier, hopeful results on estrogen replacement therapy to improve cognition and mood in women with AD. Abundant epidemiological and preclinical work over the past decade had suggested a protective effect for estrogen. Then, in the year 2000, three clinical trials failed to substantiate this claim, instead showing that estrogen therapy did not increase the performance of a total of 210 treated women on global cognitive or functional scales. And yet, two small pilot trials led by Sanjay Asthana, at the University of Washington in Tacoma, did find tantalizing improvements in attention and verbal memory in six of 12 women treated with a transdermal estrogen patch (Asthana et al., 1999; Asthana et al., 2001).

In the present study, Leon Thal and Mike Grundman of the University of California, San Diego, with colleagues there and elsewhere, report that they further analyzed their original trial data of 120 women with mild to moderate AD who received Premarin for one year. This multicenter ADCS trial had been published previously (Mulnard et al., 2000). In light of the pilot trials’ success, however, Thal et al. wondered if perhaps some women in the ADCS trial who had higher estradiol levels in their blood had indeed responded, but their improvements had gone unnoticed in the overall results. Maybe it was just a question of dose?

To answer this question, the researchers correlated these women’s change in plasma estradiol levels with their change in seven neuropsychological measures, including specific tests of memory, attention, and language. They found that baseline estradiol levels were not linked to test performance. Estradiol levels increased to means of 20 (in women on low-dose Premarin) and 40 (on high-dose Premarin) picograms/milliliters, but these elevated estradiol levels still did not correlate with test performance, or cause any significant improvements. The only exception, a negative link to one test, disappeared when the researchers removed three outliers with extremely high estradiol levels from the analysis.

What to make of this? Most cognitively healthy women who currently receive HRT across the country and abroad have levels of supplementation around 6 pg/mL, the authors write. The mean estradiol concentrations achieved in this ACDS trial are higher than that, but well below the 75 to 130 pg/mL achieved in the Asthana et al. trials. The estradiol concentration range in treated ACDS patients did reach up to peak levels of 140 mg/mL, and even these did not predict improved test scores. "We conclude that the results obtained by Asthana et al. are most likely secondary to the use of the small sample sized and the presence of a few outliers," the authors write. While this study thus cools hope for estrogen as a treatment for women with mild to moderate Alzheimer’s, it does not address the separate question of estrogen as a preventive measure in younger women (see ARF related news story).

Finally, another downer for the estrogen hypothesis comes in a study led by Monique Breteler and colleagues in Rotterdam and elsewhere. These scientists asked whether people with higher estrogen levels had larger hippocampal volume and better memory than those with low estrogen levels. This notion had grown, in part, out of findings that estrogen receptors are especially densely strewn throughout the hippocampus, and out of animal studies on the benefits of estrogen on synapse formation and survival of hippocampal neurons. Perhaps ERT prevents atrophy of the hippocampus, one of the brain areas affected early on in AD? Indeed, William Jagust of the University of California, Davis, had reported at the 2001 Society for Neuroscience Conference in San Diego that the hippocampus was larger in postmenopausal women who were taking estrogen replacement therapy than in postmenopausal women not taking hormones or a group of elderly men. This study has not yet appeared in print but the abstract, No. 550.2, is accessible at http://sfn.ScholarOne.com/itin2001/.

First author Tom den Heijer and colleagues approached the question of estrogen, hippocampal size, and memory performance by analyzing subjects in the Rotterdam Scan Study, an imaging spinoff of the ongoing 13 year-old population-based Rotterdam Study. The Rotterdam Scan Study took MRI brain scans, including a hippocampal volume measurement, of 563 elderly people without dementia. Participants underwent a Mini-Mental State examination and a delayed recall test of verbal memory, and donated blood for estradiol serum measurement. The present analysis included 412 study subjects with data on total estradiol, for 311 of whom the scientists also had data on bioavailable and free estradiol levels.

Contrary to their expectations, the scientists found that women with higher estradiol levels had smaller hippocampal volumes and poorer memory performance than did women with lower estradiol levels, while in men there was no link between estradiol levels and hippocampal size, and only a trend for poorer memory with higher estradiol levels.

The authors point out that while the community-based nature and large size of MRI sample size are a strength of this study, it also had limitations. Incomplete data on free and bioavailable estradiol reduced the sample size, and the blood samples were drawn several years before the MRI, though the influence of that delay in normal people without dementia is unclear. The authors conclude a brief review of the literature on estrogen, cognition, and AD by writing: "The biological plausibility of the estrogen hypothesis in dementia is its strongest plea, whereas studies in humans are far from conclusive."—Gabrielle Strobel


  1. I’d like to comment on the Luchsinger and the follow-up Rotterdam studies on fats, and the apparent discrepancies thus far. This is a very new area of research, and it is difficult to assess at this point what is truth and what are chance findings. I believe that it is possible that none of these dietary components may be associated with Alzheimer's disease.

    On more specific points, I do not see a discrepancy between the Luchsinger/New York findings and the Chicago or Rotterdam studies. We did not observe a protective effect from vitamin E intake around 7 IU/d, nor did the Rotterdam study. This was the mean of the highest quartile of intake from food in the Luchsinger study. So, actually all three studies are consistent in that none of them find protection against Alzheimer's disease from supplements, or from low intake of vitamin E in food.

    On the differences between the Rotterdam six-year follow-up and the Chicago study on dietary fats, I am not sure these findings were discrepant, either. We also did not observe much association when the data were analyzed without adjustment for other types of fat. Intakes of the different types of fat are correlated and sometimes are in opposite directions, so this can confuse the observed associations. Also, most of the effects we observed for saturated and trans-unsaturated fats were at the extremes of the distribution, and the associations with the individual fats of poly and mono fats were marginally significant. We hope that we can study the population further and with more subjects to test these associations more conclusively. I think it will be critical in the longer follow-up of the population to have updated dietary intake. Those most likely to change their diets are people with disease, and disease is frequent in old age, so updated measurement of diet is crucial in these older populations especially.

    I hope this area of research continues so we can learn more about the complex associations between diet and neurodegenerative decline.

  2. I basically agree with the conclusions. Intake of vitamin E and C really made little to no difference. The same was true with fat, though we have not yet separated the types of fats. The highest fat intake was among individuals who developed AD.

    Most agree that dietary interviews in the form of food-frequency questionnaires are semiquantitative methods to allow investigators to rank intake of macro and micro nutrients. We have validated our interviews in New York using the 24-hour food intake recall method (actually 72 hours, because it is done three times). The validity was modest, but acceptable for these types of questionnaires.

    However, this does not address the major issues of these studies, which are timing and direction. If we accept that AD has a long prodromal period (studies from Framingham and other places suggest that the disease may start 10 to 20 years before it is clinically recognized), then whatever is observed may be an effect, not a cause. To be more specific, dietary patterns may change during the prodromal phase of AD. If this is the case, then what is being observed is a manifestation of the disease.

    I am not sure that this is correct, but it certainly could be. I think longer follow-up over a longer period will be the only way to figure out whether or not vitamin E or C, or a lean, low-fat diet can help.

  3. It is difficult to understand why the results of the Rotterdam study and ours are different. The analyses were different, the measures of estrogen were different (in our study we did not measure levels, but only looked at estrogen replacement or not), and it is not clear how different the subjects might have been. So I think we will need more data.

  4. Estrogen may be helpful after menopause, but less so once there is neuronal stress. Two randomized controlled trials of estrogen treatment for AD showed worse results in the estrogen-treated groups, and the recently published results from Cache County suggested no benefit—indeed, possibly an increased risk of AD incidence—in women who had used HRT recently but not years earlier.

    Saturated and partially hydrogenated fats are risk factors for cardiovascular and cerebrovascular disease, and these are increasingly recognized as risk factors for AD. Until we can control for this sort of confounding, we must be cautious in interpreting the new findings in terms of direct implications for AD risk. Even if one could do a randomized trial with dietary fat, there would be no easy way to differentiate between effects mediated by fat intake directly versus those mediated by resulting vascular risks.

    Studies such as these test the limits of inference possible in the customary kinds of epidemiological analyses.

  5. The paper by Den Heijer et al. referred to in the above article reported that, in a normal ageing cohort, women with higher estradiol levels had smaller hippocampal volumes and poorer memory performance than did women with lower estradiol levels. This finding is consistent with work by the Oxford Project to Investigate Memory and Ageing (OPTIMA) that has shown, contrary to expectation, that women with AD have higher blood levels of estradiol than age-matched controls (Hogervorst and Smith, 2002). We have suggested that one reason why others have not obtained this result is related to the sensitivity of the assay method used for estradiol (Hogervorst et al., 2003). It is, of course, well-known that patients with AD have smaller hippocampi than do controls. There is a clear need for reappraisal of the role of female sex hormones in dementia (see Cochrane review by Hogervorst et al., 2002).


    . The interaction of serum folate and estradiol levels in Alzheimer's disease. Neuro Endocrinol Lett. 2002 Apr;23(2):155-60. PubMed.

    . Hormone replacement therapy to maintain cognitive function in women with dementia. Cochrane Database Syst Rev. 2002;(3):CD003799. PubMed.

    . Measuring serum oestradiol in women with Alzheimer's disease: the importance of the sensitivity of the assay method. Eur J Endocrinol. 2003 Jan;148(1):67-72. PubMed.

    View all comments by A. David Smith

Make a Comment

To make a comment you must login or register.


News Citations

  1. Rotterdam Study Questions Links between Fat and Dementia Risk
  2. Fat and Calories Mean Higher AD Risk
  3. Epidemiology Strengthens Link of Vitamins E and C to Alzheimer’s
  4. Hormone Replacement Therapy Still up on Balance Beam

Paper Citations

  1. . Dietary fat intake and the risk of coronary heart disease in women. N Engl J Med. 1997 Nov 20;337(21):1491-9. PubMed.
  2. . Dietary fat and risk of coronary heart disease in men: cohort follow up study in the United States. BMJ. 1996 Jul 13;313(7049):84-90. PubMed.
  3. . Dietary fat intake and the risk of incident dementia in the Rotterdam Study. Ann Neurol. 1997 Nov;42(5):776-82. PubMed.
  4. . Cognitive and neuroendocrine response to transdermal estrogen in postmenopausal women with Alzheimer's disease: results of a placebo-controlled, double-blind, pilot study. Psychoneuroendocrinology. 1999 Aug;24(6):657-77. PubMed.
  5. . High-dose estradiol improves cognition for women with AD: results of a randomized study. Neurology. 2001 Aug 28;57(4):605-12. PubMed.

Further Reading

Primary Papers

  1. . Dietary fats and the risk of incident Alzheimer disease. Arch Neurol. 2003 Feb;60(2):194-200. PubMed.
  2. . Antioxidant vitamin intake and risk of Alzheimer disease. Arch Neurol. 2003 Feb;60(2):203-8. PubMed.
  3. . Higher estrogen levels are not associated with larger hippocampi and better memory performance. Arch Neurol. 2003 Feb;60(2):213-20. PubMed.
  4. . Estrogen levels do not correlate with improvement in cognition. Arch Neurol. 2003 Feb;60(2):209-12. PubMed.