Of course mom was right; she always is. But for those who like evidence before they believe, consider this: At the Philadelphia conference, Francine Grodstein at Harvard Medical School presented data demonstrating that taking β-carotene starting in late mid-life protects later on against the kind of cognitive decline that heralds the beginning of dementia. This is the result of a cognitive function sub-study of the Physician’s Health Study, a large randomized prevention trial.

The rub lies in the length of exposure. The group who showed protection had been taking 50 mg every other day of β-carotene supplements for a whopping 18 years. A separate cohort who took it for approximately six years showed no benefit. This trial is one of the few to date that targeted the treatment at hand to a period in life when the underlying disease process had not yet begun or was truly at an early stage. (Researchers now believe this happens a decade or more before symptoms set in.) The trial also represents a vindication of sorts for antioxidant vitamins, which are generally thought to have some effect in staving off disease but have been disappointing in dementia-related trials.

The Physician’s Health Study (PHS) enrolls physicians, randomizes them to various treatments or placebo, and then follows them. The original PHS, of aspirin and β-carotene, began in 1982 with about 7,600 participants. It ended in 1998 and its result established low-dose aspirin as a preventive treatment against cardiovascular disease. At that time, β-carotene showed no protective effect, Grodstein said, and the trial continued. In 1997, the scientists recruited another cohort of physicians and randomized them to either β-carotene, vitamin E, vitamin C, or a multivitamin to ensure adequate folate levels. (Why physicians? To control cost, this trial relies on phone interviews to gather relevant information; physicians tend to provide that sort of information accurately and are considered highly motivated to participate long-term. Self-reporting is often cited as a technical weakness of other prospective studies, including the Cache County study.)

When the β-carotene arm ended in March 2003, the researchers assessed cognitive status in 6,012 people aged 67 and older: 4,118 from the original PHS and 1,894 new recruits. Ninety-two percent in each group participated in this assessment, Grodstein said; the others either refused or the scientists could not get them on the phone. The phone interview measuring cognitive ability comprised five tests including the East Boston memory test and a 10-word list. The primary outcome was a global composite score averaging performance across the cognitive tests. Risk of cognitive impairment was defined as being in the worst 10 percent of the performance distribution. This was pre-specified—something to keep in mind when reading epidemiological results. Frequently, when a study produces a disheartening null result in its pre-specified outcome, the researchers analyze it again for a second, more modest outcome, or they adjust their analysis to correct unexpected problems, such as a difference in the drug/placebo groups. Such results can be statistically weaker. Here, the two populations were identical in all measured confounding variables, as was participation in the cognitive sub-study.

The results showed that doctors who had taken β-carotene for about 18 years performed significantly better than did those on placebo (p= 0.2), and their risk of cognitive impairment was 20 percent lower (relative risk=0.81, 95 percent confidence interval). By contrast, the second cohort on short-term β-carotene performed just like those on placebo (p=0.5), and had the same risk (rr=1.09, 95 percent confidence interval). Performance differences were greatest in the verbal memory score, Grodstein said.

As far as epidemiological trial design goes, these data can be considered as showing strongly that long-term β-carotene helps maintain cognitive abilities and protects somewhat against dementia. The risk reduction was small, but given the prevalence of dementia in an aging population, delaying its onset by one year could eliminate several hundred thousand cases per year, Grodstein said.

The cognitive assessment measures researchers use in add-on studies to large trials fall far short of the clinical and neuropsychologic test batteries established for AD trials. Even so, Grodstein said, this phone interview is validated. “Poor performance on it correlates with later diagnosis of dementia,” Grodstein said. As a first-pass test of the interview’s relevance, the researchers determined the physicians’ ApoE phenotype, and it correlated with cognitive performance, Grodstein reported.

The length of exposure needed, and the long period of time between the exposure and the outcome in this study illustrate a problem that besets scientists investigating other experimental therapies, such as vitamin E, estrogen, or statins (see ARF related Philadelphia story). While the details of each case vary, generally good cell-, molecular-, and neurobiological data suggest these substances can help, but prospective observational epidemiology and treatment trials can come up empty-handed.

Consider vitamin E. Abundant biological data support the role of oxidative stress in neurodegenerative diseases, and at this meeting many more such data are pouring in. A prior two-year treatment trial reported an effect in patients with moderate AD upon adjusting the analysis (Sano et al., 1997), which prompted hope that perhaps this antioxidant would help earlier-stage patients even more. For this reason, vitamin E was included in the first treatment trial of MCI cases, led by Ron Petersen and Leon Thal. In Philadelphia, results of this trial (which will appear in this space in a separate story) were presented, and the vitamin E arm was negative. It did not slow the progression of people with MCI to full-blown AD at all. This looks at first blush like vitamin E does not work, and many people may spare themselves the expense now. The trouble is that vitamin E may still prove to be somewhat useful eventually, but this MCI trial exposed people to the vitamin only for three years, and perhaps too late in their lives.

Similarly, with estrogen the basic underlying science supports a protective role, but the Women’s Health Initiative Memory Study was stopped when both estrogen and estrogen/progesterone appeared to increase dementia (Espeland et al., 2004). One of the questions researchers are trying to sort out now is whether estrogen might help at earlier stages around peri-menopause, but hastens a disease process once it is underway. With statins, the current situation is puzzling in that prospective observational studies do not support the basic biology, either, though these studies may not have measured statin exposure nearly long enough (see ARF related Philadelphia story).

For β-carotene, surely cautionary remarks, qualifiers, and questions will surface. For example, might hearing loss have interfered with the assessment? Were the men in the first cohort younger than in the second when they started taking the vitamin? How could changes in dietary habits between the earlier and the later cohorts play into the result? This writer invites all readers to round off this conference summary with their comments and arguments.—Gabrielle Strobel.


  1. There was a 15,000-person study in Finland of carrot consumption and heart disease. No improvement was noticed. This study was about five years ago (seen in a newspaper). Have you seen any thing about using liquorice to increase memory? A concentrate was used... some blood pressure problem.
    This research is done at University of Edinburgh.

    La Salute... an Italian weekly health news notices..Republica news paper. number 414

Make a Comment

To make a comment you must login or register.


News Citations

  1. Philadelphia: All Is Not Well with the Statin Story

Paper Citations

  1. . A controlled trial of selegiline, alpha-tocopherol, or both as treatment for Alzheimer's disease. The Alzheimer's Disease Cooperative Study. N Engl J Med. 1997 Apr 24;336(17):1216-22. PubMed.
  2. . Conjugated equine estrogens and global cognitive function in postmenopausal women: Women's Health Initiative Memory Study. JAMA. 2004 Jun 23;291(24):2959-68. PubMed.

Further Reading

No Available Further Reading