A new observational study has drawn wide media coverage for its suggestion that the taking of cholesterol-lowering statins may help to prevent Alzheimer’s and other dementia-related conditions. Previous observational studies also have associated statin use with an apparent reduced risk of dementia and Alzheimer’s.
However, this accumulating evidence may serve only to highlight the hazards of drawing conclusions from observational studies, which are considered much less rigorous than clinical trials. In the past few months, researchers have announced that in two large clinical trials, the leading statins failed to show any effect against established Alzheimer’s disease. “For the moment, further clinical studies on statins [against dementia-related conditions] seem unlikely,” says Benjamin Wolozin, an Alzheimer’s researcher at Boston University.
The Latest Observational Study
The recent report on statin use and cognition, published in the July 29 issue of Neurology, comes from a team of researchers at the University of Michigan led by epidemiologist Mary Haan. In a five-year follow-up of 1,674 older Mexican-Americans initially free of cognitive impairment according to standard tests, the researchers found that those taking statins at some point during the study were about half as likely to develop cognitive impairment or dementia, according to those same tests.
The association between statin use and less cognitive decline was statistically significant, and the apparent “reduced risk” associated with statin use was in the same range as that seen in other studies showing such a link. Haan says that “analyses we recently completed based on more follow-up time, and restricted to dementia cases, suggest that the ‘benefit’ of statin use over time is even greater than in the published paper.”
How statins affect the aging brain isn’t well understood, and that they protect from dementia-related conditions has never been proved. But researchers have proposed that they might reduce the deposition of amyloid, reduce inflammation and boost growth factors that generally help to protect brain cells. It is also somewhat plausible that statins could prevent dementia-related conditions by lowering the production of cholesterol. A cholesterol-transport protein known as apolipoprotein E has been strongly linked to Alzheimer’s. People with a common variant of this protein, known as apoE4, have a much higher risk of the disease.
Other observational studies have linked cognitive impairment and/or dementia to obesity, diabetes, stroke and metabolic syndrome. The Mexican-Americans in Haan’s statins and cognition study, she says, have “a high level of type 2 diabetes, and at the study [outset], 51 percent met criteria for metabolic syndrome.”
How Observational Studies Work
In the past decade or so, researchers have published close to a dozen observational studies on statins and dementia-related conditions. Most have suggested a link between statin use and reduced cases of these conditions. In July 2007, for example, Benjamin Wolozin and colleagues reported that in a Veterans Affairs database with more than 700,000 patient records, the group of people who had used simvastatin (Zocor) within the study window developed, proportionately, about half the number of Alzheimer’s cases as the group who didn’t use statins.
Such studies, also known as epidemiological studies and population studies, are designed to find a statistical association between an exposure—to a medicine, behavior, or environmental toxin—and a disease-related outcome. When the researchers conducting such a study find the association they had suspected, they usually also suspect that the association is a causative one—that the exposure caused the outcome. But in principle there are an incalculable number of alternative connections that might partly or fully explain the link between exposure and outcome.
Properly designed and conducted, a formal clinical trial essentially “controls for” these alternative routes of causation by directly assigning the exposure, usually a drug, to subjects. But clinical trials are much more expensive than observational studies, which often rely only upon existing hospital records.
In an observational study, researchers try to simulate the experimental method of a clinical trial by isolating the major alternative routes of causation, which they term confounding biases. They then use standard statistical techniques to reduce the influence of these “biases” on the data. The result is an estimate of the “relative risk” of a disease associated with a particular exposure, and the assumption is that the data approximate what would have been found in a formal clinical trial.
When it is biologically plausible that the drug in question can affect the disease in question and when multiple observational studies link the taking of that drug to an apparent reduced risk of disease, it can be tempting to treat these studies as conclusive. “The bottom line is that if a person takes statins over a course of about 5 to 7 years, it reduces the risk of dementia by half, and that’s a really big change,” Haan was quoted saying in a University of Michigan press release announcing her team’s results.
The Problem with Observational Studies
In reality, observational studies on their own can never support such a conclusion. These studies come in a variety of designs with greater or lesser susceptibility to so-called confounding biases. But the claim that these biases can be identified and excluded sufficiently in any of these studies is controversial, in part because many potential biases are simply unknown. In the case of statins and dementia, it is also suspected that dementia reduces the likelihood of being prescribed statins.
Some observational studies have tried to exclude this possibility. For example, Wolozin and his colleagues, in their VA database study, looked at the association between other, non-statin cardiovascular drugs and a reduced incidence of dementia, yet found that such associations were minor in comparison to simvastatin’s. Even so, the elimination of that alternative route of causation still leaves an unknown number of others. Wolozin acknowledges that the links between statins and reduced dementia in these studies may simply “reflect a bias in the data sets.”
It is actually common for the results of observational studies to be contradicted by subsequent clinical trials. In the past few decades, such trials have indicated that a variety of drugs, from estrogen to vitamin E, fail to work against diseases as observational studies had suggested—and in some cases may even be harmful.
Effects of Statins on Cognition in Clinical Trials
There have been several large clinical trials that bear on the question of statins and dementia, and none has found suitable evidence for an effect.
In May, researchers funded by the drug company Pfizer announced that, as measured by pre-specified endpoints, atorvastatin (Lipitor) had failed to slow the progress of Alzheimer’s disease in a clinical trial of 640 patients. In July, at the International Conference on Alzheimer’s Disease in Chicago, Mary Sano of the Mount Sinai School of Medicine announced that simvastatin also had shown no effect in a clinical trial of 400 Alzheimer’s patients.
Patients in such trials are, as Haan notes, “much more advanced in disease” than they would be in a trial designed to determine whether statins could prevent dementia. Conceivably, statins have little or no effect in halting established disease, but are somewhat effective if used for years in a prevention mode.
However, statins arguably already have been used in a prevention mode. Two large clinical trials of statins as preventives of heart disease, conducted earlier this decade, also looked for any effect of the drugs in preventing cognitive impairment and dementia—and failed to find any.
Haan argues that the measures used to evaluate cognitive outcomes in these studies were less sophisticated than in her study. She and others have called for a large clinical trial designed to determine once and for all whether statins prevent dementia.
Such a “primary prevention trial” would require the enrollment and medication of thousands of elderly subjects over five to ten years. To Wolozin, in the face of all the other, negative clinical trial data on statins and dementia, the cost of such a huge undertaking would be “too high.”
Sano, a senior Alzheimer’s clinical trial specialist, doesn’t think another statins trial is even necessary, given the data from the heart disease prevention studies. “To my mind, the primary prevention studies have been done and are quite strikingly negative,” she says.