Advil for Alzheimer’s?

by Jim Schnabel

June 19, 2008

Recent observational studies have kept alive the hypothesis that the long-term use of ibuprofen and related anti-inflammatory drugs could delay or prevent Alzheimer’s disease. But it is unlikely that they will ever be formally sanctioned as Alzheimer’s preventives, and despite two decades of research no one really knows how—or even if—they work against the disease.

For the past two decades, most observational studies of large populations of elderly people have found that those who consistently take ibuprofen and some other non-steroidal anti-inflammatory drugs (NSAIDs) are less likely to have Alzheimer’s disease. The latest such study, published May 6 in the journal Neurology, found that people who had been taking ibuprofen for at least five years were only about half as likely to have Alzheimer’s as those who had not been using NSAIDs at all.

“This is the largest study of its kind by far,” says Steven Vlad, the Boston University clinical epidemiologist who led the project.

Vlad and his colleagues looked at the recent history of NSAID use among nearly 50,000 Alzheimer’s patients and almost 200,000 age-matched controls from a Veteran’s Affairs database. “Consistently with other studies, we found that if you group all NSAIDs together, there is a decreased risk of Alzheimer’s for those using NSAIDs vs. those who haven’t used,” says Vlad.

Those using NSAIDs for more than five years before the study had about 76 percent of the Alzheimer’s incidence of those not using NSAIDs; the effect was statistically significant.

Presumably because it was the most common NSAID used, ibuprofen’s apparent effect stood out more sharply from the statistical noise. It also appeared to be a stronger effect. Five-year ibuprofen users were found to have only about 56 percent of the Alzheimer’s rate of NSAID non-users.

With its statistical power, its ability to begin to distinguish the separate effects of individual NSAIDs and its seeming confirmation of the results of previous observational studies, this study might once have been expected to lead to a major clinical trial of ibuprofen as an Alzheimer’s preventive. But that is now very unlikely to happen, for several reasons.

You would have to start yesterday

The study by Vlad and his colleagues found a clear “usage duration” trend both for ibuprofen’s apparent effect and for that of NSAIDs as a whole: Use for only one year prior to the study was associated with a negligible reduction in Alzheimer’s rates but as the time of usage lengthened, the risk dropped steadily. A similar trend was seen for many of the other NSAIDs, although in most cases their sample sizes at the longer usage durations were relatively small and their results statistically noisy.

Signs of a usage duration effect are in a sense good news: They suggest that the risk-reduction effect is real and not the result of a hidden “confounder” variable that creates the illusion of an effect—for example, something that might make people who have Alzheimer’s less likely to take NSAIDs.

However, another observational (also called “epidemiological”) study, reported last year in Neurology, suggested that the appearance of a usage duration effect in fact showed that NSAIDs were not a useful tool against Alzheimer’s. In that study of elderly residents of Cache County, Utah, a team of researchers from Duke, Johns Hopkins and other centers determined that only if the residents had started using NSAIDs before the age of 65 did their use seem to prevent Alzheimer’s in a significant way.

“The notion [now] is that Alzheimer’s is a chronic disease, with a long period of insidious onset, and if you take the NSAIDs close to when symptoms are clinically manifest, then the drugs may not have a protective effect,” says Johns Hopkins epidemiologist Peter Zandi, one of the co-authors of the Cache County study.

In addition, midlife obesity and diabetes also seem to increase later Alzheimer’s risk, even when these conditions are controlled before age 60, notes Mount Sinai School of Medicine researcher Samuel Gandy. “So there appears to be a middle-aged window in which the process starts, and you may not be able to stop it easily later on,” he says.

The few randomized trials of NSAIDs as Alzheimer’s preventives have produced results that are mostly consistent with this idea. The largest such trial, sponsored by the National Institutes of Health and known as the “ADAPT” study (for the “Alzheimer's Disease Anti-inflammatory Prevention Trial”), gave the NSAIDs celecoxib or naproxen to people 70 years or older who had a family history of Alzheimer’s, and followed them for an average of only two years. (The study was ended early after data from a separate study raised concerns over possible cardiovascular side effects of naproxen.) The truncated study lacked the statistical power to reach strong conclusions, but it made clear at least that the two NSAIDs did not prevent disease when taken so close to clinical onset.

In principle, a longer-term prevention trial including more-promising NSAIDs such as ibuprofen and tracking thousands of patients from their early 60s into their late 70s would provide some clearer answers. But in practice such a study would have many problems: First, it would raise ethical concerns, given NSAIDs’ numerous potential side effects, from gastrointestinal ulcers to kidney and cardiovascular damage. In the time it would take to generate meaningful results, other more powerful therapies would likely have been found. And it would be expensive: “One could design such a prevention study but one could never afford it,” says Gandy, who also chairs the National Medical and Scientific Advisory Council of the Alzheimer’s Association.

For some researchers, there is no basis for further clinical study of NSAIDs as Alzheimer’s preventives, even if ethical and practical hurdles were overcome. “The fact is that the epidemiologic studies are not randomized, and so it is impossible, I believe, to remove all the biases in those studies” says Paul Aisen, an Alzheimer’s clinician and researcher at Georgetown University Medical School. “I’m convinced that the [positive epidemiological] results are related to systemic confounds,” he says.

Steven Vlad disagrees. “The fact that we saw such obvious differences between drugs makes me more confident that what we were seeing, for ibuprofen at least, was true. If this was all-confounding, then I think we’d be seeing [an association with Alzheimer’s risk reduction] for most of these drugs, and we just didn’t.”

If they do work, no one really knows how

Despite years of laboratory, clinical and epidemiological research, no one really knows how NSAIDs might work against Alzheimer’s.

From the first epidemiological observations in the late 1980s it had been assumed that NSAIDs affect the disease process by damping harmful inflammation in the brain—inflammation that is provoked by the initial causes of Alzheimer’s (perhaps the deposition of harmful amyloid) and that ends up burning chronically, somewhat like arthritis.

But a report in Nature in 2001 by researchers at the University of California–San Diego suggested that a subset of NSAIDs, including ibuprofen, appeared to affect the disease process more directly, by reducing the production of the specific form of amyloid—amyloid-beta-42—that constitutes the characteristic Alzheimer’s plaques. By 2003 researchers working at the Mayo Clinic in Jacksonville, Fla., and Myriad Pharmaceuticals in Salt Lake City had isolated an NSAID-like compound, R-flurbiprofen, that had a strong amyloid-lowering effect in transgenic mice and lab-dish assays, but—because its molecular structure made it only an enantiomer, or “mirror image” of an NSAID—lacked the usual NSAID side effects.

As has been the case for many strategies against Alzheimer’s, this one so far has failed to produce a dramatic disease-modifying therapy. Renamed tarenflurbil (brand name Flurizan), R-flurbiprofen showed significant but very modest effects in delaying disease progression in a phase II clinical trial among people with mild, early-stage Alzheimer’s. Results from a phase III trial of about 1,700 patients are expected this summer. [The results, announced June 30, were disappointing.]

Recent epidemiological studies also have suggested that NSAIDs—assuming they work at all in preventing Alzheimer’s—don’t do so simply by reducing the production of amyloid-beta-42. In a study reported online May 28 in Neurology, Peter Zandi and colleagues at Johns Hopkins and other institutions pooled the data from six previous NSAID epidemiological studies and concluded that there was no obvious difference, in terms of apparent impact on Alzheimer’s risk, between NSAIDs that reduce the production of amyloid-beta-42 in lab experiments and those that don’t.

Steven Vlad’s even larger study also failed to find signs of a strong difference between the amyloid-lowering NSAIDs and the rest. “It looked like the amyloid-beta-42–lowering NSAIDs might have had a bigger effect,” says Vlad. “But the ones that didn’t lower amyloid-beta-42 also looked like they had an effect.”

Vlad suspects that ibuprofen’s apparent effect in his study is “just a clearly visible example of an effect that’s probably true of a bunch of drugs. And whether that’s an amyloid-beta-42–lowering effect or not, I’m unsure at this point.”

It might be that NSAIDs, to the extent that they work against Alzheimer’s, do so largely through anti-inflammatory pathways and merely differ in their pharmacologic ability to reach relevant brain areas. It is also conceivable that they work through a process that is not as closely related to inflammation, or even to amyloid levels. For example, researchers at the University of Minnesota, reporting in the March issue of the journal Brain, found that amyloid-beta-42 in its soluble form appears to assist in blocking a basic memory-related function at the synapses of brain cells, whereas a subset of NSAIDs stops this from happening, measurably improving cognition in mice genetically engineered to overexpress amyloid.

Neurologist Karen Ashe, who heads the laboratory where this work was conducted, notes that the cognition-preserving effects of NSAIDs in the study were relatively insensitive to decreases in inflammation, and also to decreases in amyloid-beta-42 levels, implying that NSAIDs act downstream from these factors in the disease process.  The synaptic-protection effect of NSAIDs, she says, is therefore “a third possible mechanism” by which these drugs impact the Alzheimer’s process.