The Role of NSAIDs and Cyclooxygenase on Beta-amyloidosis and Clinical Dementia

Several recent developments in the study of anti-inflammatory drugs as possible therapeutic agents to prevent or treat AD must be reconciled:

1. Epidemiologic evidence continues to suggest that chronic exposure to traditional non-selective NSAIDs reduces subsequent risk of AD. The recent compelling analysis of data from the Rotterdam Study is consistent with prior epidemiologic studies. It suggests that long term use of drugs such as ibuprofen and naproxen by prescription, particularly when used for periods exceeding two years, dramatically reduces risk of AD.

2. Cell culture and transgenic mouse studies indicate that some NSAIDs, but not others, have a favorable impact on amyloid peptide generation and deposition. Ibuprofen is effective in these models, but naproxen and selective COX-2 inhibitors are not. The effective NSAIDs seem to reduce amyloidogenic cleavage of APP by a mechanism independent of COX inhibition.

3. The major trials of anti-inflammatory drugs for the treatment of...  Read more

The Role of NSAIDs and Cyclooxygenase on Beta-amyloidosis and Clinical Dementia

Several recent developments in the study of anti-inflammatory drugs as possible therapeutic agents to prevent or treat AD must be reconciled:

1. Epidemiologic evidence continues to suggest that chronic exposure to traditional non-selective NSAIDs reduces subsequent risk of AD. The recent compelling analysis of data from the Rotterdam Study is consistent with prior epidemiologic studies. It suggests that long term use of drugs such as ibuprofen and naproxen by prescription, particularly when used for periods exceeding two years, dramatically reduces risk of AD.

2. Cell culture and transgenic mouse studies indicate that some NSAIDs, but not others, have a favorable impact on amyloid peptide generation and deposition. Ibuprofen is effective in these models, but naproxen and selective COX-2 inhibitors are not. The effective NSAIDs seem to reduce amyloidogenic cleavage of APP by a mechanism independent of COX inhibition.

3. The major trials of anti-inflammatory drugs for the treatment of AD have been negative. Studies of prednisone, celecoxib and hydroxychloroquine have not demonstrated any benefit in the treatment of patients with mild to moderate AD.

Some unresolved questions:

• Do the epidemiologic studies reflect a causal relationship between NSAID use and reduced risk?
• To what extent does the artificiality of the model systems cloud applicability to AD in humans?
  
• Can the negative results of specific treatment regimens be generalized?
• How feasible is long term treatment with non-selective NSAIDs in the frail population at risk for AD?
• If such treatment is to be effective, how early must it be instituted?
• Is it time to move away from the anti-inflammatory treatment strategy?

View all comments by Paul Aisen

Our published work indicates that (1) COX-2 expression in the AD brain tissues involves blood-derived macrophages, not neurons; (2)macrophages are infiltrating the neuritic plaques and phagocytizing amyloid-beta. COX-2 activation in the brain appears to be associated with inflammatory activation of peripheral monocytes/macrophages. Does the current work in other laboratories support neuronal COX-2 expression

View all comments by Milan Fiala

It appears that expression of Cox-1 and 2 varies with species, dose, route and type of insult. Neurons, microglia, astrocytes and also the peripheral macrophages that migrate into the brain following injury can all express Cox-2. The expression of Cox-1 has also been noted in microglia. I am not sure how much specificity can be expected in Cox-2 expression based on these observations. My hunch is that invading macrophages and microglia may make up a large portion of the Cox-2 activity in inflammatory states but that needs to be certainly challenged and verified with better techniques in localizing COX-2 reactivity in different environments.

References:
1. Minghetti L, Walsh DT, Levi G, Perry VH. "In vivo expression of cyclooxygenase-2 in rat brain following intraparenchymal injection of bacterial endotoxin and inflammatory cytokines." J Neuropathol Exp Neurol 1999 Nov;58(11):1184-91. Abstract

2. Walsh DT, Perry VH, Minghetti L. "Cyclooxygenase-2 is highly expressed in microglial-like cells in a murine model of prion disease." Glia 2000 Feb 15;29(4):392-6. Abstract

3. Matsuoka Y, Picciano M, Malester B, LaFrancois J, Zehr C, Daeschner JM, Olschowka JA, Fonseca MI, O'Banion MK, Tenner AJ, Lemere CA, Duff K. "Inflammatory responses to amyloidosis in a transgenic mouse model of Alzheimer's disease." Am J Pathol 2001 Apr;158(4):1345-54. Abstract

4. Hoozemans JJ, Rozemuller AJ, Janssen I, De Groot CJ, Veerhuis R, Eikelenboom P. "Cyclooxygenase expression in microglia and neurons in Alzheimer's disease and control brain." Acta Neuropathol (Berl) 2001 Jan;101(1):2-8. Abstract

View all comments by Tennore Ramesh

Strong correlations exist between neuroinflammation as observed in Alzheimer's disease brain, antiinflammatory protection as observed in epidemiological studies, and therapeutic benefit as observed in clinical trials of traditional NSAIDs. Nevertheless, there are many comments in the literature expressing doubt about these correlations. They are mostly attributable to some badly targeted clinical trials that have ended in failure.

It is now incontrovertible that neuroinflammation is a prominent characteristic of AD pathogenesis. More than a thousand reports on the subject have so far been published. The most recent papers can be found in a special issue of the Neurobiology of Aging devoted to the subject (Vol. 22:6, Nov-Dec, 2001).

It is also incontrovertible that antiinflammatory agents protect against AD. More than 20 epidemiological studies involving diverse subjects from 4 continents have confirmed this phenomenon. The most recent report was that of Veld et al. (see ARF news story) who analyzed the Rotterdam cohort of 6,989 subjects. They found that...  Read more

Strong correlations exist between neuroinflammation as observed in Alzheimer's disease brain, antiinflammatory protection as observed in epidemiological studies, and therapeutic benefit as observed in clinical trials of traditional NSAIDs. Nevertheless, there are many comments in the literature expressing doubt about these correlations. They are mostly attributable to some badly targeted clinical trials that have ended in failure.

It is now incontrovertible that neuroinflammation is a prominent characteristic of AD pathogenesis. More than a thousand reports on the subject have so far been published. The most recent papers can be found in a special issue of the Neurobiology of Aging devoted to the subject (Vol. 22:6, Nov-Dec, 2001).

It is also incontrovertible that antiinflammatory agents protect against AD. More than 20 epidemiological studies involving diverse subjects from 4 continents have confirmed this phenomenon. The most recent report was that of Veld et al. (see ARF news story) who analyzed the Rotterdam cohort of 6,989 subjects. They found that use of traditional NSAIDs for 2 years or more provided an 80 percent protection against AD. In other studies, some protection, but of a considerably smaller degree, was found for steroids (Breitner et al., 1995; Canadian Study of Health and Aging, 1994; McGeer et al, 1996.) A correlation of sorts is the identification of fewer reactive microglia in postmortem brains of NSAIDs users compared with controls (Mackenzie & Munoz, 1998), although no reductions were found for steroid users (Mackenzie, 2000).

Translation of these findings into practice has been inexplicably weak. The only clinical trial within the boundaries of these prominent signposts was conducted nine years ago (Rogers et al., 1993). In that small, double-blind, placebo-controlled study of patients with probable AD, therapeutic doses of the COX-1 inhibiting NSAID indomethacin appeared to arrest progress of the disease. Five of 24 treated patients and 1 of 20 placebo patients withdrew because of gastrointestinal side effects. Two test patients and 5 controls withdrew for other reasons. Overall, 58 percent of test patients and 70 percent of controls patients completed the trial.

An equally small trial of the mixed COX inhibitor diclofenac was not straightforward because the prostaglandin analog misoprostol was added as a gastroprotective agent (Scharf et al., 1999.) That strategy was not successful since 12 of 24 test patients dopped out because of GI problems. Even so, there were non-significant trends towards less deterioration in the test group.

Other, poorly targeted clinical trials have been outright failures. Hydroxychloroquine proved to be of no benefit (Van Gool et al., 2001) but there were no data, either epidemiological or pathological, to indicate that it would work. Hydroxychloroquine is not an antiinflammatory agent. Rather it is an antimalarial drug which, for reasons yet unknown, is helpful in treating rheumatoid arthritis.

Similarly, a trial of the COX-2 inhibitor celecoxib failed (1). Once again there was no persuasive evidence to suggest that it would be effective. Selective COX-2 inhibitors have not been in use long enough for epidemiological data to accummulate, but immunohistochemical evidence was available which indicated they should fail. Unlike COX-1, COX-2 is highly expressed in normal pyramidal neurons, as well as in pyramidal neurons of AD cases (Yasojima et al., 1999). Therefore, selective COX-2 inhibitors will primarily target pyramidal neurons rather than microglia.

The prednisone trial also failed, despite the mild epidemiological support. The problem here was probably the use of a subthreshold dose of 10 mg/day (2). Based on the doses of steroids required to control acute relapses of multiple sclerosis, levels approximately three times higher would probably have been required. Whether such levels could be tolerated for any length of time is questionable, given the many physiological functions of glucocorticoids in the body. An additional problem is the known toxicity of steroids to hippocampal neurons.

The preferred target for NSAID treatment of AD is clearly COX-1. GI side effects are an inevitable problem with such treatment. The question to be considered is whether to put brains ahead of guts. Currently approved treatments for AD do not alter neuronal loss. They only mask the disease by boosting cholinergic function. Since 100 percent of AD cases will inevitably deteriorate and only some will experience GI side effects, the balance would appear to favor NSAID treatment. Nevertheless, better strategies are required if the full benefits of COX-1 inhibition in brain are to be obtained. Perhaps the NO-NSAIDs, which have greatly reduced GI toxicity (Wallace JL et al., 1994) while retaining the beneficial effects of their parent NSAIDs, will provide the answer." (See also ARF news story.)

References:
(1) Sainati SM et al. Results of a double-blind, placebo-controlled study of Celebrex for the progression of Alzheimer's disease. 6th International Stockholm-Springfield Symposium of Advances in Alzheimer Therapy, page 180..

(2) Aisen P et al. Treatment of Alzheimer's disease with prednisone: results of pilot study and design of multicenter trial. J Am Geriatric Soc 43, SA27 (1995)

View all comments by P.L. McGeer