Nonsteroidal Antiinflammatory Drug Use in Alzheimer' s Disease Ugo Lucca, Mauro Tettamanti, Gianluigi Forloni, and Alberto Spagnoli Key W o r d s : Nonsteroidal anti-inflammatory drugs (NSAIDs), Alzheimer's disease, rheumatoid arthritis, elderly, osteoarticular
Introduction An increasing body of findings supports the hypothesis that chronic inflammatory activity sustained by altered immune responses may well be involved in the pathogenesis of Alzheimer's disease (AD). Many immune system proteins have been identified in the AD brain, where they are associated with the characteristic neuropathological features of the disease: senile plaques, neurofibrillary tangles, and dystrophic neurites. The accumulation of cytokines, complement factors, and immunoglobulin receptors does not appear to be merely a consequence of the degenerative processes, but may play a role in the cascade of events inducing neuronal death in AD (McGeer et al 1992). Thus, hope has been aroused for the efficacy of anti-inflammatory drugs, and some nonsteroidal anti-inflammatory drugs (NSAIDs) are currently under clinical investigation in AD patients (Schnabel 1993; Cotton 1994). The prevalence of AD among patients with rheumatoid arthritis (P_A), an autoimmu~ disease often treated chronically with NSAIDs, has been studied in several epidemiological surveys, with discordant results. Jenkinson et al (1989), in a sample of 96 AD patients, found a RA prevalence rate of 2%, while in 92 controls the prevalence was 13%, as expected in a general population comparable for age. McGeer et al (1990) confirmed these data. finding prevalence rates of RA in AD patients ranging from 0.5%--1.2%: however, in a retrospective study whose methodology is not well described. Be~u'det al (1991 ) reported a coincidence rate of RA and AD in 521 RA cases of 4.4%. The association between arthritis and AD also was investigated in a few case-control studies. Broe et al (1990), Li et al (1992), French et al (1985), and Breitner et al (1994) all showed a negative association, but From~ MarioNegriInstituteforPharmacologicalResearch,Milan,Italy. Addressreprintrequeststo Dr.AlbertoSpagnoli,MariaNegriInstituteforPharmacologicalResearch,ViaEritrea62, 20157Milan,Italy;Fax:(02)39001916. ReceivedFebruary18, 1994;revisedJune I0, 1994. © 1994 Societyof BiologicalPsychiatry
only in the first two reports did it reach statistical significance. On the other hand, Graves et al (1990) and Heyman et al (1984) noted a positive but not significant association.
Methods We wished to check for a possible difference between expected (elderly subjects) and observed (AD patients) rates of exposure to NSAIDs. As RA is not the commonest indication for long-term NSAID treatment in the aged, we reviewed the data from three epidemiological surveys conducted in the elderly (n = 1254.~--indicated in Table 1 as surveys A (Spagn~fi et al 1989), B, and C--and from two clinical trials with 195 AD patients, mainly also old (trials A (Spagnoli et al 1991) and B ). For the clinical diagnosis of AD, we followed the DSM-III-R (APA 1987) criteria of primary degenerative dementia in trial A, and the N1NCDS-ADRDA criteria in trial B. Survey A refers to a sequential sample of elderly patients in the general practice, and survey B and C to a randomized sample of very old people drawn from the practice records of the general practitioners (for further details on age, sex, and sample size, see Table 1). Surveys A and B were conducted in an urban area, and survey C in a rural area. ICD 9 and anatomic therapeutic chemical (ATC) were used for h'le classification of diagnoses and drugs, both in the trials and in the surveys.
Results Results are summarized in Table 1. Even though "younger-old" subjects are more represented in the trials, the prevalence of NSAID use among subjects aged 60 or more does not correlate with age; thus, age differences in these studies cannot be considered a bias in the interpretation of these results. Data on drugs, NSAIDs, and osteoarticular disorders show good consistency both among surveys and trials, despite differences in sampling proceC:J06-3223]94/$07.00
Table 1. Main Findings on NSAIDs in Epidemiological and Clinical Studies Considered No. and sex of subjects (M = men, W = women)
Epidemiological studies in the e/dedy Survey A, 1986 802 (M = 3 ! 7; W---485)
Mean age (range)
Mean No. of drugs daily (~ SD)
261 80.8 (M = 135; (75-95) W= 126) Survey C, 1992 191 80.3 (M = 99; 75-94) W=92) Clinical trials in patients with AIzheimer "sdisease Trial A, 1989 130 75.2 (M = 38; (56-.89) W=92) TrialB, 1992 65 67.5 (M = 22; (50-80) W=43)
Subjects on NSAIDs
Prevalence of osteaanicol~ luehologies (ICD9 7 1 0 - 7 3 9 )
Subjects ea NSAIDs fo¢ p
3.6 4- 2.3
20.3% (I 5.4-25.2)
3.0 -+ 2.2
15.0% (9.9-20. I)
!.2 --- 1.2
22.3% (15.5-30.4 )
0.8 4- 1.1
Survey B, i 99 !
Age,numberofdrugs,andprevalencerateswereadjustedaccordingto actualsexwevalencein the generalpopulations NSAIDs= nonsteroidalanti-inflammatorydrugs.CI= confidenceinmrvals;HA= notavailable.
dures. The lower total drag use in AD patients (about ! vs. 3 in the eldedy) is due primarily to the effect of the exclusion criteria operating at the level of disorders and drug therapies. The use of NSAIDs shows a striking and consistent difference between AD patients and unselected elderly subjects: NSAIDs are virtually absent in AD patients (1/195), whereas they are taken by about 20% of the elderly, 40% of whom are on long-term treatment. This finding, like that of the lower drug use in AD patients, might be the product of a selection bias related to the exclusion criteria of the trials. In fact, if we take cardiovascular drugs and cardiovascular disorders (the latter was an exclusion criterion when of recent onset or with a chronic but unstable course), prevalence rates are 60% for both drugs and disorders in the surveys, and of 25% and 30%, respectively, in the trials. However, neither NSAID treatment nor osteoarticular disorder was an exclusion criterion. Furthermore, in the prevalence of osteoarticular disorders, which accounted for nearly 85% of NSAID use in the surveys, no relevant differences in the five studies were seen (see column 5 of Table 1). Thus, even if selection bias cannot be fully excluded, it seems highly unlikely to be the primary cause of the difference found in NSAID use.
Discussion Semla et al (1993), studying drug use patterns of persons with AD, multi-infarct dementia (MID), and mixed dementia, found that despite salicylate use being much lower in AD patients, 12.2% of AD males were on salicylates. Previous case-control studies mainly investigated the association of AD with the use of"analgesics", much too broad a class oftreatmants. No statl,,:ticallysignifi-
cant differences between AD patients and elderly comzois were reported by Heyman et al (1984) (odds ratio [OR], 1.23); Amaducci et al (1986) (OR, 1.00); Li et al (! 992) (OR, 1.00; confidence interval [CI]: 0.09-11.03); and Broe et al (1990) (OR, 0.71; CI: 0.46-1.17). Graves et al (1990) noted a trend toward an inverse association with steroid use (OR, 0.73; CI: 0.38-1.38); this finding was reproduced by Breitner et al (1994), who showed, in a co-twin control study, that the onset of AD was inversely associated with prior use of anti-inflammatory drugs (OR, 0.24; CI: 0.07-0.74). If Breitner et al's results were confirmed, they could suggest that the low prevalence of AID among RA patients is not due to genetic factors, as hypothesized by Jenkinson et al (1989), but to a delayed onset related to the anti-inflammatory drug use. A possible explanation of the unexpectedly low NSAI~ use that we observed in AD patients could be a more conservative attitude towards the drug treatment of osteoarticular problems, either because they are considered marginal in relation to the dementia syndrome or because of less clearcut complaints by ,he patients. This could partially explain the differences in Table 1 oetween the prevalence of osteoarticular disorders (column 5) and of NSAID use (column 6) in AD patients. However, it is reasonable to assume that our findings, like those of McGeer et al (1990) and Breitner et al (1994), point to a possible retarding action of NSAIDs on the onset or progression of AD. There are obvious and unavoidable limitations when we try to use epidemiological findings to understand biological processes. However, it has to be noted that research on AD lacks reliable animal models. Thus epidemiological data with sound biological plausibility could serve as a valuable source of suggestions for the understanding of the disease process and for the search of new
therapeutic strategies. Several experimental data have supported the hypothesis of the involvement of immune system proteins in the pathogenesis of AD. in particular, the considerable accumulation of IL- 1 in AD has been associated with this cytokine's ability to induce the neuronal expression of amyloid precursor deposited in senile plaques (Forloni et al 1992), while in turn complement is
activated by ~-amyloid protein (Rogers et al ! 992). The inflammatory hypothesis of AD needs further biological investigation, but there are already enough data to justify the passage from the encouraging results of a small study (Rogers et al 1993) to largescale drug trials to test the long-term risk-benefit ratio of NSAID treatment in AD patients.
References Amaducci LA, Fratiglioni L, Rocca WA, et al (1986): Risk factors for clinically diagnosed Alzheirner's disease: a case control study of ~,~Italian population. Neurology 36:922-93 I. American Psychiatric Association (I 987): Diagnostic and Statistical Manual of Mental Disorders, 3rd ed rev. Washington DC: American Psychiatric Press. Beard CM, Kokman E, Kurland LT (1991): Rheumatoid arthritis and susceptibility to Alzheimer's disease. Lancet 337:1426. Brae GA, Henderson AS, Creasey H, et al ( 1990): A case-control study of Alzheimef s disease in Australia. Neurology 40:16981707. Breituer JCS, Gau BA, Welsh KA, et al (1994): Inverse association of anti-inflammatory treatments and Alzheirner's disease: Initial results of a co-twin control study. Neurology 44:227232. Cotton P (1994): Constellation of risks and processes seen in search for Alzheimer's clues. JAMA 271:89--9I. French LR. Schuman LM, Mortimer JA, Hutton JT, Boatman RA, Christians B (1985): A case-control study of dementia of the Alzheirner type. Am J Epidemiol 121:414--421, Forloni G, Demicheli F, Giorgi S, Bendotti C, Angeretti N (1992): Expression of amyloid precursor protein mRNAs in endothelial, neuronal and glial cells: modulation by interleukin-1. Mol Brain Res 16:128-134. Graves AB, Wlfite E, Koepsell TD, et al (1990): A case-control study of Alzheimer's disease. Ann Neural 28:766--774. Heyman A, Wilkinson WE, Stafford JA, Helms MJ, Sigmon Aid, Weinberg T (1984): Alzheimer's disease: a study of epidemiological aspects. Ann Neural 15:335-341,
Jenkinson ML, Bliss MR, Brain AT, Scott DL (! 989): Rheumatoid ~tis and senile dementia of the Aizheirner's Type. Br l Rheumato128:86--88, Li G, Shen YC, Li YT, Chen CH, Zhau YW, Silvemtan JM (! 992): A case-control study of Alzheimer's disease in China. Neurology 42:1481-1488. McGeer PL, McGeer E, Rogers J, Sibley J ( 1990): Anti-inflammatory drugs and Alzheimer's disease. Lancet 335:1037. McGeer PL, Rogers J (I 992): Anti-inflammatory agents as a therapeutic approach to Alzheimer's disease. Neurology 42:447449. Rogers J, Cooper NR, Webster S, et al (1992): Complement activation by [~-amyloid in Alzheimer disease. Proc NatiAcadSci USA 89:10016-10020. Rogers J, Kirby LC, Hempelman SR, et al (1993): Clinical trial of indomethacin in Alzheimer's disease. Neurology 43:16091611. Schnabei J (1993): New Alzheimer's therapy suggested. Science 260:1719-1720. Semla TP, Cohen D, Paveza G, et al (1993): Drug use patterns of persons with Alzheimer's disease and related disorders living in the community. J Am Geriatr Sac 41:408--413. Spagnoli A, Osfino G, Dana Borga A, et al (1989): Drug compliance and unreported drugs in the elderly. J Am Geriatr Sac 37:619--624. Spagnoli A, Lucca U, Menasce G, et al (1991): Long-term acetylL-carnifine treatment in Alzheinaer's disease. Neurology 41:1726-1732.