Aisen PS, Schmeidler J, Pasinetti GM.
Randomized pilot study of nimesulide treatment in Alzheimer's disease.
Neurology. 2002 Apr 9;58(7):1050-4.
Please login to recommend the paper.
To make a comment you must login or register.
The main purpose of our Neurology study was to determine tolerability in AD of nimesulide, a NSAID preferentially inhibiting cyclooxygenase (COX)-2 but also, to a lesser degree, COX-1. The study was designed to test if nimesulide is suitable in chronic treatment. This is a fundamental question for the use of anti-inflammatory drugs in AD because the poor tolerability of all the currently tested non-steroidal anti-inflammatory drugs in elderly, frail AD patients does not allow a long-term, efficacious treatment plan in these patients. The study was supported by Helsinn Healthcare SA.
The primary analysis of the influence of nimesulide on cognition was covariance analysis, with baseline ADAScog score as a covariate, to assess change in ADAScog during treatment in the nimesulide group versus the placebo group. There was no difference in cognitive decline between groups. We note, however, that at 12-week double-blind phase, only a symptomatic benefit (as seen with cholinesterase inhibitors such as donepezil) would be likely to be detected. Analysis of the 12-week time point found no adverse effect of nimesulide on cognition, an important consideration with a CNS-penetrating drug. Further, there was no significant adverse effect on clinical stage (CDR), activities of daily living, mood, or behavior.
A secondary analysis, which is ongoing, concerns the impact of nimesulide on specific items of the assessment scales among those who completed the 12 week phase on full-dose therapy without interruption. As with the intent-to-treat analysis, there was no significant difference between groups on any of the total outcome measures. However, analysis of individual items revealed differences between nimesulide and placebo. These pos-hoc analyses of item scores suggest that nimesulide treatment had a favorable effect on judgment, orientation, and sleep. But while the raw p values suggest statistical significance, it must be noted that these are not corrected for the large number of comparisons.
Because of the high tolerability, a compassionate use phase program was offered at the end of the scheduled double-blind and open-label phases. The timing of visits during the compassionate use phase was somewhat irregular, so we estimated rate of decline at 12, 24, 52, 78 and 104 weeks using a 4-week margin around these target intervals. For comparison, we generated estimates of expected decline over these intervals using the annual rate of decline of historical controls in studies performed in our institution, adjusted for baseline ADAScog score. We found that nimesulide treatment over the course of 104 weeks compared favorably to the course of disease progression in historical controls. We point out that this comparison to historical controls has questionable validity, which may be particularly true in this instance, as the control study was conducted prior to the availability of donepezil, which may reduce the rate of cognitive decline (88 percent of subjects in this study were concomitantly treated with donepezil).
In conclusion, our study suggests that nimesulide therapy is well-tolerated in the AD population, in contrast to other, non-selective NSAIDs for periods exceeding 2 years. Most importantly, the rate of cognitive decline of subjects on nimesulide compares favorably to published data on historical, untreated controls followed in our department at Mount Sinai School of Medicine. Interestingly, we also found that post-hoc analysis of individual items suggests possible benefit on clinical assessment of orientation and judgment, and possible mood elevation in response to short-term therapy with nimesulide, compared to placebo. Helsinn Healthcare SA and I are considering a further efficacy study to fully assess the potential of nimesulide in treatment of Alzheimer's.
The epidemiology supporting an association of chronic and early NSAID use with reduced risk for AD is consistent enough to take seriously. Results from the recent Rotterdam and Cache County studies, as well as many others, suggest that risk for AD can be reduced to about one fifth compared to control populations. This is without optimization of NSAID choice, dose, or time of intervention, suggesting that even more risk reduction would result from optimization of NSAID use.
A major obstacle to more widespread use of NSAIDs for AD prevention is toxicity, particularly in the elderly. Both COX-1 and COX-2 targets are arguably important in AD pathogenesis. The positive epidemiology relies entirely on mixed COX inhibitors, suggesting they should be tested. Toxicity, particularly of mixed COX inhibitors, has been a significant limiting factor in previous NSAID trials in AD. To date, the option of using less toxic COX-2 specific inhibitors has failed to produce good results. Therein lies the significance of this Neurology report from Aisen et al., which demonstrates a relatively favorable toxicity profile with nimesulide, a widely used mixed COX-1 and COX-2 inhibitor.
These new clinical trial results show no short-term effect on cognition, but the authors suggest that long-term nimesulide use may slow disease progression. This is a testable hypothesis, and based on nimesulide's demonstrated neuroprotective and anti-inflammatory effects in animal models. While it is hard to imagine a downside to neuroprotective activity, the relative importance of a range of potentially beneficial NSAID effects on AD pathogenesis remains uncertain. In addition to control of reactive microglia and direct neuroprotection, a number of other relevant potentially important NSAID activities have been observed, including effects on amyloid production, on pro-amyloidogenic factors (ApoE, ACT) and on amyloid phagocytosis/clearance. Whether NSAIDs proposed for use in AD have these or other AD-relevant activities at acceptable doses should be an active area of investigation. More work needs to be done at the preclinical level.
This is the third small report of a trial of a non-steroidal anti-inflammatory drug (NSAID) for the treatment of AD. As in each prior study, the intent was to test the efficacy of an NSAID (in this instance, the COX-2 preferential agent nimesulide) for improvement in symptoms of AD over a relatively brief time span. In the present instance, somewhat more emphasis was placed also on assessing the tolerability of the treatment in often-frail elderly with AD.
Participants were observed in the masked portion of this trial for 12 weeks, with a follow-on period of open label observation lasting another 12 weeks. Neither the masked, placebo-controlled portion or the trial, nor the 24-week assessment of symptoms before and after treatment showed any hint of therapeutic benefit. This may not be surprising, given the size and the relatively short duration of treatment in this trial, but it is disappointing nonetheless. A more hopeful outcome is that this agent was better tolerated than the drugs tested previously, diclofenac (Scharf et al., 1999) and (especially) indomethacin (Rogers et al, 1993).
Although suggesting a therapeutic benefit, the results of the indomethacin trial were of questionable interpretation because of the experiment's very high drop-out rate and the method of statistical analysis, which combined four non-informative results post hoc into an aggregate outcome that was marginally "statistically significant." The diclofenac results were ambiguous for efficacy. The present results are less hopeful but, as an offset, suggest the best tolerability yet observed in a published NSAID trial. All three trials were under-powered, but one might have hoped here for a weak signal suggesting efficacy. If the results are interpreted literally, they seem to suggest greatest efficacy with the COX-1 selective agent indomethacin, weaker effect with the relatively balanced agent diclofenac, no apparent benefit with COX-2 selective nimesulide, and the inverse order for tolerability. The present observations fit with the null results (but relatively good tolerability) reported from an unpublished large one-year trial of the specific COX-2 inhibitor celecoxib.
Clearly, more data are needed from adequately powered trials on this important topic. Given the above, these trials should include agents targeting COX-1. Presumably, commercial considerations have precluded the initiation of a trial of ibuprofen, the agent with the greatest body of epidemiological evidence to suggest its benefit. Such a trial is clearly needed. Also important will be trials for efficacy of NSAIDs' effects at earlier stages of AD pathogenesis, including the "conversion" to dementia from milder cognitive syndromes and the primary prevention of symptoms altogether.
The nimesulide trial represents another failure in applying the epidemiological evidence of NSAIDs protecting against AD toward successful treatment. There are some simple explanations for this failure. First of all, a mismatch exists between the NSAIDs used in the failed clinical trials and those consumed by the cohorts studied epidemiologically. The latter all involved classical NSAIDs, which are strong COX-1 or mixed inhibitors, while the failed clinical trials, including that of nimesulide, have involved preferential COX-2 inhibitors. The COX-2 inhibitors have lower gastrointestinal side effects, but this is hardly important if the appropriate target in brain is COX-1. COX-2 in human brain is concentrated in pyramidal neurons while COX-1 is preferentially expressed in microglia, so COX-1 is the principal target. To be clinically effective, the NSAID chosen must be effective against COX-1, able to cross the blood-brain barrier, and be administered in a dose sufficient to inhibit neuroinflammation.
Indomethacin and ibuprofen are two classical NSAIDs that meet the first two criteria, but to be effective they must be administered in a sufficiently high dose. Post-mortem studies show that the degree of inflammation in AD brain is higher than in joints removed surgically for chronic arthritis. Therefore it may be that treatment doses will need to be higher than preventive doses. Side effects may be significant, but managing such side effects is preferable to not managing the deterioration in AD.
The involvement of an inflammatory component in AD has been amply
documented over the last decade and has led to a number of clinical trials
of anti-inflammatory agents. The epidemiological data clearly demonstrate
that a subset of NSAIDs are effective in reducing AD risk. However, clinical
trials of two COX-2 specific inhibitors, as well as the recent nimesulide
study, do not yet provide compelling evidence that NSAID therapy will be
therapeutically effective in AD. Moreover, and as pointed out by others, the
trials have been limited by being underpowered, of short duration, and
perhaps not done at the appropriate clinical stage.
One aspect of this issue that has not been discussed in this forum is the
existence of other potential targets of NSAIDs that could account for their
therapeutic effects (Heneka et al. 2001, Landreth and Heneka, 2001).
Indomethacin and ibuprofen, the NSAIDs that have been shown to exert the
most significant therapeutic effect in reducing the risk and delaying the
onset of AD, also bind to and activate the Peroxisome Proliferator Activated
Receptor gamma (PPARγ) a member of the nuclear hormone receptor family
(Lehman et al. 1997). PPARγ activation exerts strong anti-inflammatory
actions in many cell types, including microglia, astrocytes, and neurons
(Bernardo et al, 2000; Combs et al. 2000; Klegeris et al, 1999; Heneka et al
2000; Heneka et al. 1999). These anti-inflammatory effects of PPARγ agonists
are mediated through transcriptional inhibition of proinflammatory gene
expression. The mechanisms for these events have been shown to be due, in
part, to scavenging of transcriptional cofactors (i.e. CBP, P300) by
activated PPARγ, which therefore prevents them from interacting with
inflammatory transcription factors such as NFkB and results in decreased
inflammatory gene expression. Also, PPARγ activation induces expression of
inhibitory IkB proteins, which would be expected to further reduce NFkB
The possibility that NSAIDs exert protective actions through PPARγ
activation, and not by inhibition of COXs, is supported by several studies
showing that 1) non-selective NSAIDs were effective at high doses (mM) where
they bind to PPARγ, but not at lower doses at which they would be expected
to inhibit COX-dependent prostanoid generation; 2) direct application of
COX2 selective inhibitors were without effects; and 3) direct application of
selective PPARγ agonists were protective. In our studies we found that the
inflammatory reaction in vitro (Combs et al. 2000) or in vivo (Heneka et al
2000) in brain was prevented by drugs that bind to and activate PPARγ
(15-deoxy D12, 14-prostaglandin J2; troglitazone, and ibuprofen) but not by
the COX2 inhibitor NS398.
The possibility that PPARγ activation could be therapeutically useful for
the treatment of AD is particularly relevant given that two PPARγ agonists
(pioglitazone and rosiglitazone) are FDA-approved for treatment of type II
diabetes. This should facilitate the design and implementation of clinical
trials for AD. An NIH-sponsored pilot clinical trial of pioglitazone in a
small cohort of patients has recently been initiated at University Hospitals
Cleveland/Case Western Reserve University.
We should also mention that oral PPARγ agonists provided protection in other
neuropathological conditions that contain an inflammatory component,
including EAE , the model for MS, (Niino et al, 2001, and other work in
press), suggesting that at least some of these drugs can sufficiently pass
through the blood-brain- barrier and exert intraparenchymal