NSAIDS, ALZHEIMER'S DISEASE AND PUBLIC CITIZEN

With regard to this news item please see my Open letter to Public Citizen's Health Research Group on Alzheimer's disease research. Science SAGE KE (21 Feb., 2003) [ Full Text ] ; BMJ (27 Feb., 2003) [ Full Text ] .


View all comments by Alexei R. Koudinov

This is the fifth published trial of antiinflammatory treatments for disease modification or symptomatic improvement in Alzheimer's dementia. Two early small studies showed an encouraging suggestion of disease modification (slowing or progression) in patients randomized to indomethacin and diclofenac. These trials encountered severe difficulties, however, with high rates of dropouts, and they were too small, in any case, to produce definitive results.

More recently published trials have tested low-dose prednisone, a broad-spectrum corticosteroidal antiinflammatory agent, and hydroxychloroquine, an antimalarial drug that is widely acknowledged to have potent antiinflammatory activity. The prednisone trial showed a predictable range of safety concerns, but the treated group showed worse performance, if anything, on their measures of disease progression. The hydroxycholoroquine trial results were null, despite that trial's having especially abundant statistical power to demonstrate a relatively modest level of benefit. This latest trial with rofecoxib and naproxen is again...  Read more

This is the fifth published trial of antiinflammatory treatments for disease modification or symptomatic improvement in Alzheimer's dementia. Two early small studies showed an encouraging suggestion of disease modification (slowing or progression) in patients randomized to indomethacin and diclofenac. These trials encountered severe difficulties, however, with high rates of dropouts, and they were too small, in any case, to produce definitive results.

More recently published trials have tested low-dose prednisone, a broad-spectrum corticosteroidal antiinflammatory agent, and hydroxychloroquine, an antimalarial drug that is widely acknowledged to have potent antiinflammatory activity. The prednisone trial showed a predictable range of safety concerns, but the treated group showed worse performance, if anything, on their measures of disease progression. The hydroxycholoroquine trial results were null, despite that trial's having especially abundant statistical power to demonstrate a relatively modest level of benefit. This latest trial with rofecoxib and naproxen is again null.

With the benefit of more recent laboratory data, there are reasons to wonder whether the alleged beneficial effects of NSAIDs on AD dementia could relate to activities of these compounds other than their inhibition of cyclooxygenases. In this connection, it is unfortunate that there has been no trial yet published of ibuprofen for AD treatment. In a broad range of observational studies, ibuprofen is the agent most widely associated with reduced risk of incident AD, and the laboratory data suggest that, among widely used NSAID agents, ibuprofen may have particular advantages in its influence on Aβ formation and possibly on glutamate uptake through its influence on PPARγ. Particularly in view of the findings produced by ibuprofen in Tg2597 transgenic mice, I believe that a treatment trial with ibuprofen is needed before we should sound the "death knell" for treatment trials with antiinflammatory agents.

More importantly, however, I would note that all of the above trials have tested various agents for the treatment of Alzheimer's dementia, whereas all of the observational data suggest that NSAIDs may reduce the risk of incident dementia, i.e., may protect those at risk from the progression of the AD pathogenetic pathway, from its latent stages (believed to occur over a period of decades) through the recognized AD prodrome of mild cognitive impairment (MCI), and eventually to onset of AD dementia. While laboratory evidence might offer some hope that NSAIDs could mitigate disease pathogenesis after the onset of dementia, the epidemiological data offer no support for this perspective. Instead, they suggest that NSAIDs may be effective for the prevention (not treatment) of AD dementia. Therefore, the most promising avenue for future research with NSAIDs in AD would appear to lie with their being tested for prevention. Such prevention could either operate from the point when individuals have developed MCI (i.e., from the standpoint of AD dementia, secondary prevention) or in yet earlier pathogenetic stages that are still symptomatically silent (primary prevention). The Alzheimer's Disease Antiinflammatory Prevention Trial, or ADAPT, is testing two NSAIDs, naproxen and celecoxib, for their safety and efficacy in preventing AD dementia in asymptomatic individuals. Given the recent laboratory data, it may be that a similar trial using ibuprofen would be justified.

View all comments by John Breitner

The study by Aisen et al. shows no benefit to rofecoxib or naproxen in patients with mild to moderate AD. Nevertheless, this negative study is an important contribution to the collective effort to evaluate therapeutic strategies to treat or prevent AD. The excellent discussion highlights the salient points and limitations of this study:

• The epidemiologic data supports a role for NSAIDs in prevention of AD and not in treatment. Thus, it is not necessarily surprising that no benefit was observed in this patient trial.
• The most consistent reduction of risk in the epidemiologic studies of NSAIDs and AD was in the group that used NSAIDs for greater than two years; thus, it is possible that a one-year study with any NSAID will not show a beneficial effect.
• Unlike some other NSAIDs (ibuprofen), rofecoxib or naproxen do not lower Ab42. They also have not been reported to have any beneficial effect in AbPP mouse models.
• Treatment trials with other NSAIDs that reduce Ab42 and are proven...  Read more
  

The study by Aisen et al. shows no benefit to rofecoxib or naproxen in patients with mild to moderate AD. Nevertheless, this negative study is an important contribution to the collective effort to evaluate therapeutic strategies to treat or prevent AD. The excellent discussion highlights the salient points and limitations of this study:

• The epidemiologic data supports a role for NSAIDs in prevention of AD and not in treatment. Thus, it is not necessarily surprising that no benefit was observed in this patient trial.
• The most consistent reduction of risk in the epidemiologic studies of NSAIDs and AD was in the group that used NSAIDs for greater than two years; thus, it is possible that a one-year study with any NSAID will not show a beneficial effect.
• Unlike some other NSAIDs (ibuprofen), rofecoxib or naproxen do not lower Ab42. They also have not been reported to have any beneficial effect in AbPP mouse models.
• Treatment trials with other NSAIDs that reduce Ab42 and are proven to have efficacy in animal models may still be warranted.
• Prevention studies of NSAIDs, especially those that lower Ab42, may be warranted.
• NSAID use in an elderly population that is associated with risk of morbidity.

Of note, although not published, both Merck and Pharmacia have conducted or are still conducting therapeutic trials with rofecoxib and celecoxib, respectively. Preliminary data presented at some meetings is consistent with a lack of efficacy of these agents.

From our perspective, we believe that it is important to consider testing ibuprofen in both therapeutic and prevention trials. Although there are no rigorous statistical data from the epidemiology studies that prove that ibuprofen use was associated with the reduction in risk, it is generally believed that the majority of long-term NSAID users will, in fact, be taking ibuprofen. Furthermore, ibuprofen is the only approved NSAID that has been tested in AD animal models and shown to be of benefit. The only current prevention trial of NSAIDs (ADAPT evaluating naproxen and celecoxib) is not administering a compound that alters Ab42 production. Moreover, neither of the agents in the ADAPT trial have been shown to have a beneficial effect in AD animal models.

View all comments by Todd E. Golde
View all comments by Eddie Koo