Parminder Raina, PhD; Pasqualina Santaguida, PhD; Afisi Ismaila, MSc; Christopher Patterson, MD; David Cowan, MD; Mitchell Levine, MD; Lynda Booker, BSc; Mark Oremus, PhD
Disclaimer: The authors are solely responsible for the content of this review. The opinions expressed herein do not necessarily reflect the opinions of the Agency for Healthcare Research and Quality, the Ontario Ministry of Health and Long-Term Care, or the McMaster Evidence-based Practice Center.
Grant Support: Parminder Raina holds a Canadian Institute of Health Research Investigator award and an Ontario Premier's Research Excellence award. The original systematic review was funded by the Agency for Healthcare Research and Quality, U.S. Department of Health and Human Services (contract no. 290-02-0020). The update to this review was funded by the Ontario Ministry of Health and Long-Term Care and the McMaster University Evidence-based Practice Center.
Potential Financial Conflicts of Interest:Honoraria: P. Santaguida (American College of Physicians).
Requests for Single Reprints: Parminder Raina, PhD, McMaster University Evidence-based Practice Center, Department of Clinical Epidemiology and Biostatistics, McMaster University, 1280 Main Street West, DTC Room 306, Hamilton, Ontario L8S 4L8, Canada; e-mail, firstname.lastname@example.org.
Current Author Addresses: Drs. Raina, Santaguida, Ismaila, Patterson, Cowan, Levine, Booker, and Oremus: McMaster University Evidence-based Practice Center, Department of Clinical Epidemiology and Biostatistics, McMaster University, 1280 Main Street West, DTC Room 306, Hamilton, Ontario L8S 4L8, Canada.
Raina P., Santaguida P., Ismaila A., Patterson C., Cowan D., Levine M., Booker L., Oremus M.; Effectiveness of Cholinesterase Inhibitors and Memantine for Treating Dementia: Evidence Review for a Clinical Practice Guideline. Ann Intern Med. 2008;148:379-397. doi: 10.7326/0003-4819-148-5-200803040-00009
Download citation file:
Published: Ann Intern Med. 2008;148(5):379-397.
Appendix: Studies of Cholinesterase Inhibitors or Memantine That Were Excluded
The effectiveness of the 5 U.S. Food and Drug Administrationâ€“approved pharmacologic therapies for dementias in achieving clinically relevant improvements is unclear.
To review the evidence for the effectiveness of cholinesterase inhibitors (donepezil, galantamine, rivastigmine, and tacrine) and the neuropeptide-modifying agent memantine in achieving clinically relevant improvements, primarily in cognition, global function, behavior, and quality of life, for patients with dementia.
Cochrane Central Register of Controlled Trials, MEDLINE, PREMEDLINE, EMBASE, Allied and Complementary Medicine Database, CINAHL, AgeLine, and PsycINFO from January 1986 through November 2006.
English-language randomized, controlled trials were included in the review if they evaluated pharmacologic agents for adults with a diagnosis of dementia, did not use a crossover design, and had a quality score of at least 3 on the Jadad scale.
Data were extracted on study characteristics and outcomes, including adverse events. Effect sizes were calculated and data were combined when appropriate.
96 publications representing 59 unique studies were eligible for this review. Both cholinesterase inhibitors and memantine had consistent effects in the domains of cognition and global assessment, but summary estimates showed small effect sizes. Outcomes in the domains of behavior and quality of life were evaluated less frequently and showed less consistent effects. Most studies were of short duration (6 months), which limited their ability to detect delay in onset or progression of dementia. Three studies directly compared different cholinesterase inhibitors and found no differences in cognition and behavior.
Limitations of available studies included short duration, inclusion of only patients with mild to moderate Alzheimer disease, poor reporting of adverse events, lack of clear definitions for statistical significance, limited evaluation of behavior and quality-of-life outcomes, and limited direct comparison of different treatments.
Treatment of dementia with cholinesterase inhibitors and memantine can result in statistically significant but clinically marginal improvement in measures of cognition and global assessment of dementia.
Dementias have become a major public health concern because of their increasing prevalence, chronicity, caregiver burden, and high personal and financial costs of care. Currently, there are no cures for most dementias. For the most common types (Alzheimer disease, vascular dementia, and mixed dementias), clinicians often prescribe pharmacotherapy to alleviate symptoms and delay disease progression. The pharmacotherapeutic agents available to treat problems associated with dementias (for example, psychosis) have varying levels of evidence to support their efficacy and have been reviewed elsewhere (1). Some drugs, although not approved, are being used in populations with mild cognitive impairment; in such patients, the rate of conversion to dementias is 0.3 to 2.3 per 100 person-years (2). Currently, 5 drugs have U.S. Food and Drug Administration (FDA) approval for managing dementias. The cholinesterase inhibitors (donepezil, galantamine, rivastigmine, and tacrine) degrade acetylcholinesterase, allowing levels of acetylcholine (a neurotransmitter critical to the neurons involved in cognition) to increase. Memantine partially blocks the N-methyl-d-aspartic acid receptor and prevents excess stimulation of the glutamate system, which influences memory and learning. Although FDA approval specifies use of these 5 drugs for Alzheimer disease, in clinical practice the drugs are also prescribed for other dementias.
This review systematically evaluates the evidence for the effectiveness of these 5 drugs in improving outcomes in cognition, global function, behavior, and quality of life among patients with dementia.
We searched the Cochrane Central Register of Controlled Trials, MEDLINE, PREMEDLINE, EMBASE, Allied and Complementary Medicine Database, CINAHL, AgeLine, and PsycINFO for relevant evidence published in English from January 1986 through November 2006. We also reviewed the bibliographies of retrieved papers.
All populations with major dementias (including Alzheimer disease, vascular dementia, and Parkinson dementia) and mild cognitive impairment were included. Only parallel randomized, controlled trials that compared a cholinesterase inhibitor or memantine with placebo or another drug were eligible. We excluded crossover trials because of potential bias due to period effects or period-by-treatment interaction. Our content-expert panel reached consensus and established that eligible studies also had to have a minimum modified Jadad score of 3 of 5 (original scale), indicating moderate study quality. Study outcomes primarily encompassed 4 broad domains: cognition, global function, behavior, and quality of life (including activities of daily living [ADLs] and caregiver burden). We classified most clinical outcomes within these 4 domains; other outcomes were rate of institutionalization, mortality, or adverse events. Two independent reviewers evaluated each study for eligibility. Appendix Table 1 describes the eligibility criteria in detail.
This systematic review was done in the context of an Agency for Healthcare Research and Quality–funded review that evaluated 92 pharmacologic agents for dementias (1).
Two independent reviewers abstracted data from and assessed the quality of all studies that met the eligibility criteria. The modified Jadad scale (which includes additional domains that concern collection of adverse events, description of statistical analysis, and reporting of eligibility criteria) (3) and a checklist for the quality of reporting of adverse events were used to evaluate methodological quality; the latter measures included questions on frequency of reporting harms, withdrawals, and method of collection (1).
Evaluation of benefit was based on reported changes in the principal outcome within the domains of interest. Although we did not restrict studies by the type of outcome, we did anticipate that some outcomes would be more commonly used in these drug studies. We searched the literature to establish the magnitude of change considered to be clinically important in key outcomes.
Specifically, within the domain of cognition, we considered the Alzheimer's Disease Assessment Scale (ADAS)—consisting of the cognitive subscale (ADAS-cog), noncognitive subscale (ADAS-noncog), and total ADAS score (ADAS-tot)—the Mini-Mental State Examination (MMSE) (or the standardized MMSE version), and the Severe Impairment Battery (SIB) to be commonly used measures that have established properties and are scored by a trained evaluator or clinician. The ADAS-cog is a validated psychometric assessment scale for the domains of attention, memory, orientation, language ability, and praxis in Alzheimer disease (4). Scores range from 0 to 70, with higher scores indicating greater impairment. A change of 4 points is considered clinically significant for patients with mild to moderate dementia, but the ADAS-cog is not uniformly sensitive to change over the course of the disease (5). The ADAS-noncog evaluates behavioral changes. The MMSE is a widely used measure of cognitive function validated in dementia populations (6). Scores range from 0 to 30, with lower scores indicating greater impairment. The MMSE measures orientation, attention, recall, and language, but it does not evaluate mood or disordered forms of thinking. The SIB is a validated measure of cognitive function for moderate to severe dementias in the areas of orientation, attention, language, and praxis (7). Scores on the SIB range from 0 to 100, with lower scores indicating greater deficits. There are no established clinically important differences for the MMSE or SIB.
For the domain of global function, a commonly used outcome is the clinician-based impression of change (CIBIC), with caregiver input (CIBIC-plus) and other modified versions (New York University–CIBIC-plus, clinician's global impression of change [CGIC], Alzheimer's Disease Cooperative Study CGIC, and clinician interview–based impression). The CIBIC-plus is a validated measure of change that requires a clinician to judge global patient function in 4 areas: general, cognitive, behavioral, and ADLs (8). This measure is scored on a 7-point scale, with 1 reflecting marked improvement, 4 indicating no change, and 7 denoting marked worsening. Because the CIBIC-plus is a global rating by clinicians, any change in score is considered clinically significant. Most other measures commonly used in clinical settings do not have established effect sizes that reflect clinically important differences.
To evaluate adverse effects, we used a standardized instrument that assessed rates of withdrawals due to adverse effects, the method (active vs. passive and standardized vs. nonstandardized approaches) and frequency of collection of harms, and the definition and collection of serious and severe harms. A priori, we selected specific events (nausea, diarrhea, dizziness, accidental injury, agitation, urinary disorder, serious adverse events) and expressed these as a percentage for each study. Where 2 or more studies provided sufficient information, we calculated the summary estimate for the specific adverse event evaluated.
We used standard meta-analytic techniques to estimate effect sizes for each drug in studies with the same outcomes. The effect measure selected varied according to the manner in which the outcome was reported and included change scores or, for dichotomous data, relative risks (RRs). Reasonableness of pooling was assessed on clinical and biological grounds in terms of clinical heterogeneity (drugs, similarity of populations, and outcomes); therefore, meta-analysis was not appropriate for all outcomes. We did not include summary estimates when studies provided only end point scores. Similarly, we excluded studies that did not provide a measure of variance for outcomes when computing summary estimates.
When meta-analyses were undertaken, the weighted mean difference (WMD) was selected as the pooled estimate instead of the standardized mean difference. When only the proportions of patients whose condition improved or worsened were reported, the RR was used as a measure of the summary effect size. In all meta-analyses, a random-effects model was used; tests for statistical heterogeneity were based on the chi-square statistic and the I2 statistic. In some cases (9–12), estimates of mean changes in the study outcomes used for the meta-analyses were based on best estimates derived from figures in the citations.
Figure 1 shows the process of study selection. Of the papers in the larger review, 127 evaluated donepezil, galantamine, rivastigmine, tacrine, and memantine. We excluded 22 of these that scored less than 3 on the Jadad scale, 8 that were crossover trials, and 1 that administered tacrine to both study groups. The Appendix lists all excluded studies. The remaining 96 reports included 59 unique study cohorts. Seventy-five different outcomes were measured across the domains of interest. Cognition and global function were the domains from which efficacy was most frequently determined.
The term companion refers to multiple reports from a single study. The authors considered the first published study as the main paper and referred to all associated reports as “companion papers.” DSM = Diagnostic and Statistical Manual of Mental Disorders; ICD = International Classification of Diseases; NINCDS = National Institute of Neurological and Communicative Disorders and Stroke.
Twenty-four unique studies (9, 10, 12–33) from 34 different reports evaluating donepezil versus placebo (or vitamin E) were eligible for this systematic review. Three additional studies (4 reports) directly compared donepezil with galantamine (34, 35) and rivastigmine (36, 37) and are discussed in the section on comparative effectiveness. Appendix Tables 2, 3, and 4 describe study characteristics and outcome effect sizes, the frequencies of a priori–selected harms, and all reported adverse events. A total of 7556 participants (sample size, 12 to 818 participants) were randomly assigned in these placebo-controlled trials. Most studies addressed Alzheimer disease, with fewer focusing on vascular dementia (22, 23), Parkinson dementia (28), dementia in patients with the Down syndrome (12), or patients with mild cognitive impairment (21, 32). Dementia severity was described as “probable” or mild to moderate in most studies, moderate to severe in 2 studies (14, 16), and severe in 1 study (33). Many studies inaccurately used the term probable to describe severity rather than a measure of diagnostic certainty. Most studies evaluated daily doses of 10 mg (10, 12–16, 20, 21, 25–30, 32, 33), whereas 2 studies used 5 mg or less daily (19, 31). Five studies compared 5-mg and 10-mg doses (9, 17, 18, 22, 23), and 1 study (24) presented combined data. The duration of the drug intervention (including titration) was 12 to 16 weeks (18, 19, 26), 18 weeks (28) to 23 or 24 weeks (9, 10, 12, 14, 16, 17, 20–23, 25, 27, 29–31, 33), 52 to 54 weeks (13, 15), or 156 weeks (32). One long-term study (33) reported 2-year follow-up, but participants did not receive donepezil continuously.
All studies that compared donepezil with placebo evaluated some form of cognitive outcome, and all but 3 of these studies (12, 16, 29) showed a positive effect in at least 1 measure used in this domain. Four trials (20, 21, 28, 31) evaluated more than 1 outcome in the cognition domain, and results varied. Figure 2 shows summary estimates of effect sizes for the ADAS-cog at the highest dosage (10 mg/d) and across all levels of disease severity. The summary effect sizes were largest in patients with Alzheimer disease, next largest in those with vascular dementia, and smallest in those with mild cognitive impairment; no group achieved a change of 4 points (the change considered clinically significant) (5). The meta-analysis (Figure 2) shows a consistent and statistically significant treatment effect for improvement in the ADAS-cog; the exception is patients with mild cognitive impairment, in whom the effect was nonsignificant (P = 0.31). However, tests for heterogeneity were also significant for this group (I2 = 75.5%; P = 0.043). The length of these two trials differed appreciably (from 6 months to 3 years), as did their criteria for mild cognitive impairment.
For donepezil (10 mg/d) versus placebo (Alzheimer disease [AD], all severity levels), the estimate was statistically significant (P < 0.001) and tests for heterogeneity were not significant (I2 = 0.0%; P = 0.94). For donepezil (10 mg/d) versus placebo (mild cognitive impairment), the estimate was not significant (P = 0.31) and tests for heterogeneity were significant (I2 = 75.5%; P = 0.043). For donepezil (10 mg/d) versus placebo (mild to moderate vascular dementia), the estimate was significant (P < 0.001) and tests for heterogeneity were not significant (I2 = 0.0%; P = 0.84). For galantamine (24 mg) versus placebo (mild to moderate AD), the estimate was significant (P < 0.001) and tests for heterogeneity were significant (I2 = 75.5%; P = 0.001). For galantamine (24 mg) versus placebo (mild to moderate AD and vascular dementia), the estimate was significant (P < 0.001). For rivastigmine (6 mg and 12 mg) versus placebo (AD, all severity levels), the estimate was significant (P < 0.001) and tests for heterogeneity were significant (I2 = 90.8%; P < 0.001). For memantine (20 mg) versus placebo (mild to moderate AD), the estimate was not significant (P = 0.25). For memantine (20 mg) versus placebo (mild to moderate vascular dementia), the estimate was significant (P < 0.001) and tests for heterogeneity were not significant (I2 = 11.4%; P = 0.29).
Table 1 shows summary effect sizes for the MMSE and the SIB among patients with Alzheimer disease and vascular dementia. Consistent statistically but not clinically significant effects were observed.
All but 8 studies (15, 20, 24–27, 29, 30) used some measure of global function assessment. All but 4 trials (10, 12, 21, 28) showed a statistically significant difference in this domain. One study (32) evaluating a population with mild cognitive impairment showed significant differences at 18 but not 36 months. On the basis of 3 studies that provided sufficient information, the summary RR for improvement (CIBIC score, 1 to 3) relative to baseline for the CIBIC-plus (Figure 3) in patients with Alzheimer disease indicates a significant improvement (RR, 2.01 [95% CI, 1.58 to 2.57]). The magnitude of the effect decreased in 1 study that dichotomized the CIBIC-plus score as improved or stabilized (CIBIC score, 1 to 4) and deteriorated (CIBIC score, 5 to 7). The summary RR estimate for improvement or stabilization of vascular dementia was not statistically significant and showed moderate heterogeneity; however, this estimate was based on only 2 trials. Table 1 shows the summary estimate (WMD) for 4 studies that provided the mean change scores for the CIBIC-plus; this estimate also showed statistical significance. In addition, Table 1 shows improvement in the WMD for the Clinical Dementia Rating scale, another measure of global function; however, the tests for heterogeneity were significant.
For donepezil versus placebo (Alzheimer disease [AD], all severity levels), the relative risk (RR) for improvement was statistically significant (P < 0.001) and tests for heterogeneity were not significant (I2 = 0.0%; P = 0.762). For donepezil versus placebo (AD, all severity levels), the RR for improvement or stabilization was significant (P < 0.001). For donepezil versus placebo (mild to moderate vascular dementia), the RR for improvement or stabilization was not significant (P = 0.633) and tests for heterogeneity were not significant (I2 = 55.1%; P = 0.136). For galantamine versus placebo (mild to moderate AD), the RR for improvement or stabilization was significant (P < 0.001) and tests for heterogeneity were not significant (I2 = 19.9%; P > 0.20). For galantamine versus placebo (mild to moderate AD and vascular dementia), the RR for improvement or stabilization was significant (P = 0.002). For memantine versus placebo (AD, all severity levels), the RR for improvement was significant (P < 0.001) and tests for heterogeneity were not significant (I2 = 0.0%; P > 0.20). For memantine versus placebo (AD, all severity levels), the RR for improvement or stabilization was significant (P < 0.001) and tests for heterogeneity were not significant (I2 = 13.8%; P > 0.20). For rivastigmine versus placebo (AD, all severity levels), the RR for improvement or stabilization was not significant (P = 0.114).
Of the 9 studies that evaluated behavior, all but 1 used the Neuropsychiatric Inventory (10). Summary estimates for this outcome were not significant in patients with Alzheimer disease (Table 1).
Eight (9, 13–15, 18, 22, 24, 33) of 12 studies showed statistically significant improvement in the various outcomes assessing ADLs. However, only 2 studies used the same outcome to allow computation of a summary estimate. Table 1 shows the summary estimate for the Alzheimer's Disease Functional Assessment and Change Scale. The effect size is small and of borderline statistical significance (P = 0.053) for patients with vascular dementia. With the exception of 3 studies (9, 24, 33), ADLs were evaluated as a secondary quality-of-life outcome. Courtney and colleagues (24) found statistically significant changes in the Bristol Activities of Daily Living score, but this difference was not clinically significant (a threshold had been set a priori as an absolute change of 3 points).
One large trial (24) measured rate of institutionalization as the primary outcome but did not show statistically significant differences. This study had the longest duration (2 years) of any trial, but there were anomalies in the design. After initial randomization, patients in the donepezil group had treatment interruptions (described as washout periods) and were randomly assigned twice during the trial. The purpose of discontinuation of drug therapy was not clearly specified. A study evaluating patients with mild cognitive impairment showed statistically significant differences in the rate of conversion to Alzheimer disease at 12 months but not at 36 months. Another study (30) showed some statistical differences in certain sleep variables.
Five (9, 15, 17, 18, 32) of 7 studies showed statistically significant differences between groups for diarrhea, nausea, and vomiting, which are consistent with expected effects of cholinesterase inhibitors. Six studies (9, 17, 18, 22, 23, 38) reported a dose–response effect, with increasing frequency of adverse events as dose increased. The summary effect size could be computed for 29 different harms in studies of Alzheimer disease, 11 for vascular dementia, and 6 for mild cognitive impairment; these effect sizes were based on 2 studies for these latter 2 patient populations but a variable number of studies for the patients with Alzheimer disease. Many of the effect sizes were not statistically significant (Appendix Table 4). Of the 29 different adverse effects examined in patients with Alzheimer dementia, 9 had statistically significant effect sizes. Diarrhea (RR, 2.57 [CI, 1.93 to 3.41]) and nausea (RR, 2.54 [CI, 1.97 to 3.29]) were reported most frequently. Anorexia had the largest effect size (RR, 3.21 [CI, 1.94 to 5.33]) and dizziness the smallest (RR, 1.47 [CI, 1.06 to 2.03]). The pooled estimate for vomiting was moderately heterogeneous. For patients with vascular dementia, abnormal dreams, diarrhea, nausea, and muscle and leg cramps were statistically more frequent with donepezil; muscle cramps had the highest RR (9.62 [CI, 3.48 to 26.58]). The effect sizes for the group with mild cognitive impairment were similar to those for patients with vascular dementia, with the addition of insomnia.
Rates of withdrawal due to adverse events ranged from 0% to 57% in treatment groups and 0% to 20% in placebo groups (Appendix Table 2). In general, the quality of reporting harms was low to moderate in all but 2 trials (22, 23) evaluating vascular dementia. The methods of recording harms varied; 10 trials did not specify the mode (13, 20, 21, 23, 26, 29–33), and a minority used standardized instruments (12, 16, 17, 19, 27, 28). Six trials specified an operational definition of serious adverse events (14, 16, 17, 19, 23, 33); however, no serious harms were attributed to donepezil in any study.
We included 10 studies (12 reports) of galantamine (39–48) that evaluated 3997 patients total (sample size range, 182 to 978 patients). All but 3 of the studies (43, 45, 47) included only patients with Alzheimer disease. Two studies enrolled patients with Alzheimer disease and cerebrovascular disease, and 1 included only patients with mild cognitive impairment (47). All studies aimed for a final treatment dose of 24 mg or 32 or 36 mg/d (1 study used 32 mg/d and 1 used 36 mg/d). One study (46) compared extended-release galantamine with the usual formulation. Trials lasted 12 to 16 weeks (40, 44, 47), 20 weeks (39), and 24 to 26 weeks (41–43, 45, 46) (Appendix Table 2). One trial (48) evaluated the difference between a 3-day and a 7-day washout period when patients were switched from donepezil to galantamine; the subsequent follow-up lasted 48 months.
Eight trials (39–43, 45–47) showed significant improvement in cognitive function. One trial (44) reported mixed effects—improvement on the ADAS-cog with 24 mg but not with 32 mg. Another study (48) of the length of 2 washout periods showed no difference between groups, suggesting that the washout period had no effect on cognition. Figure 2 shows the summary estimate for improvement on the ADAS-cog with 24 mg (currently the maximum dose recommended by the manufacturers); this finding was statistically significant, but so were tests for heterogeneity. One small study (47) evaluated patients with mild cognitive impairment and found significant changes after 4 months.
Of the 6 studies that evaluated global function with the CIBIC-plus, all but 1 (46) showed significant changes relative to placebo (Table 2). Figure 3 shows the summary estimates for the CIBIC-plus with 24 mg, and the summary RR for improvement or stabilization from baseline was 1.22 (CI, 1.12 to 1.33), a statistically significant finding with minimal heterogeneity.
Five studies (39, 40, 43, 45, 46) measured behavior using the Neuropsychiatric Inventory. The summary estimate of improvement based on the 2 studies that reported sufficient data was significant (Table 2) and shows no important inconsistency.
Five studies (40–43, 45) used the Disability Assessment for Dementia to measure ADLs. Only 2 were included in the summary estimate; this improvement was statistically significant. Two studies used the Alzheimer's Disease Cooperative Study–ADL subscale, and the summary effect was also statistically significant (Table 2).
The most common harms reported in galantamine studies were gastrointestinal symptoms (nausea and vomiting, diarrhea), eating disorders or weight loss, and dizziness (Appendix Table 3). Table 2 shows that the effect sizes for all these harms were significant, with anorexia having the largest effect (RR, 3.41 [CI, 2.36 to 4.93]) and dizziness the smallest (RR, 1.90 [CI, 1.43 to 2.51]). One study (49) evaluated a subgroup of patients who reported nausea or vomiting and found that women and patients with lower body weight at baseline were more likely to report these adverse events. Most trials did not report statistical testing of adverse events for differences between groups, but 2 trials (41, 42) reported a statistically significantly greater weight loss in the treatment groups. Some studies (40, 42, 43, 46) showed a dose–response relationship for adverse events during titration.
Rates of withdrawal due to adverse events ranged from 4% to 17% for placebo and 8% to 54% for active treatment. Although not consistently reported for half the studies, no important differences emerged in the rates of serious adverse events between the placebo and galantamine groups. Most studies did not report using a standardized instrument to collect harms, but some (39, 41, 42, 45) used standardized coding to classify adverse events.
Nine eligible studies (11, 50–57) (11 reports) compared rivastigmine with placebo. These studies evaluated 2164 patients, with sample sizes ranging from 27 to 725. One trial (52) evaluated dementia associated with Parkinson disease, 1 study (54) evaluated Lewy body dementia, and the remainder evaluated Alzheimer disease. These studies included all levels of severity. Daily rivastigmine doses ranged from 1 mg (51) to 12 mg (11, 52–54, 56), and treatment lasted 14 to 52 weeks.
Eight studies evaluated general cognitive function. Those using the ADAS-cog (11, 51–53, 56) showed statistically significant improvement, whereas those using other measures (SIB, specific neuropsychological tests) did not. Two studies (51, 56) (1 North American and 1 European site) of the same protocol had different findings: one (51) found improvements with both higher (6 to 12 mg) and lower (1 to 4 mg) doses, but the other (56) failed to show significance for the lower doses. Figure 2 shows the summary estimate for trials that provided sufficient data on the ADAS-cog for all levels of severity and mixed doses. The effect was statistically significant and larger, but with significant heterogeneity. Table 3 shows the summary estimate for the MMSE, which was not significant and showed a high level of heterogeneity.
For global changes, 7 (11, 50–53, 55, 56) of 8 studies showed significant improvements, but 3 studies used the higher dose only. One of these 3 (50) defined the higher dosage as 6 mg/d, which was the minimum dosage for the other 2 studies. Figure 3 shows the summary estimate of the RR for improving from baseline (RR, 1.76 [CI, 1.35 to 2.29]) for CIBIC-plus in 3 studies that provided sufficient information. Table 3 shows that the CIBIC-plus summary estimate (WMD, −0.36), based on 5 studies, was statistically significant and consistent across studies.
Two studies (11, 50) evaluated behavior using the Nurses Observation Scale for Geriatric Assessment–Mood subscale. Although there were no consistently significant changes in individual trials, the summary effect size estimate (Table 3) was significant but showed moderate heterogeneity across studies. The Neuropsychiatric Inventory was used to evaluate patients with different dementias (Parkinson dementia  and dementia with Lewy bodies ); statistically significant differences between the treatment and placebo groups were found. Three studies (51, 53, 56) that evaluated ADL in patients with Alzheimer disease using the Progressive Deterioration Scale failed to show a significant summary effect size and found moderate heterogeneity (Table 3). One study (52) of patients with Parkinson dementia found significant improvement as measured with the Alzheimer's Disease Clinical Scale.
All the summary adverse effect sizes presented in Appendix Table 4 are statistically significant, except diarrhea. Table 3 demonstrates that vomiting had the greatest effect size (RR, 6.06 [CI, 3.88 to 9.45]) and that dizziness had the smallest (RR, 2.24 [CI, 1.45 to 3.46]). This finding is consistent with data for the other cholinesterase inhibitors evaluated. Two trials demonstrated a dose response; 1 (56) showed significant differences for rates of nausea and vomiting only, whereas the other (51) found significant differences for all reported adverse events.
Rates of withdrawal due to adverse events ranged from 0% to 11% in the placebo groups and from 12% to 29% in the treatment groups. One trial (55) did not report the withdrawal rates or the types of adverse events observed. One study (11) prescribed antiemetics to increase the tolerance of patients taking rivastigmine. Only 3 trials (50, 51, 56) provided an operational definition of serious adverse events, but overall the frequency of severe adverse events did not differ between the treatment and placebo groups. The quality of reporting for harms varied widely.
Seven studies (17 reports) evaluating tacrine were eligible for this review. Tacrine was compared with placebo in 6 trials (58–63) and with idebenone in 1 (64). The placebo-controlled studies evaluated a total of 1203 randomly assigned patients (range, 13 to 663 patients) (Appendix Table 2). All but 1 study (64) included patients with Alzheimer disease; the exception included patients with “primary degenerative dementia” and those with Alzheimer disease. All studies enrolled participants with dementias of mild to moderate severity or specified as “probable” disease. Daily doses ranged from 80 mg/d (60) to 160 mg/d (58). Treatment duration varied from 12 to 13 weeks (60, 61, 63) and 30 to 36 weeks (58, 59, 62) for all placebo-controlled studies, whereas the trial with idebenone (64) treatment lasted 60 weeks.
No summary effect size could be computed for tacrine trials because insufficient information was provided or different outcomes were evaluated.
Of the 6 placebo-controlled studies, only 1 (58) that used the ADAS-cog showed statistically significant improvement. Three doses (80, 120, and 160 mg) were compared in 1 trial (58), and only the 120- and 160-mg doses were shown to be statistically significantly better than placebo (mean change in ADAS-cog score, approximately 2 points). The remaining studies used other measures of cognition and showed no difference or inconsistent results; these trials had small sample sizes (12 to 32 participants). One trial (63) that found no statistical difference for general cognitive function was short (12 weeks) and used 80 mg/d, a dosage that another trial showed to be ineffective (58). Three placebo-controlled studies (58, 62, 63) evaluated global function; 2 (58, 63) showed statistically significant improvement. The trial that found no benefit (62) also had inconclusive findings for general cognitive function. Five trials evaluated behavior, and 4 (58–60, 63) showed no difference between groups. Two trials (58, 61) that used different outcomes for quality of life failed to show significant improvement (Appendix Table 2). Two studies (59, 61) evaluated caregiver burden and showed no benefit from tacrine.
Appendix Table 3 shows the frequencies of various harms. No studies provided sufficient information on any of these harms to allow us to compute summary effect sizes. Elevated alanine aminotransferase levels or other hepatic abnormality (4% to 13% in the placebo groups; 7% to 67% in the treatment groups [all doses]) was reported in 6 studies, suggesting the potential for serious liver damage. No trial tested for differences in adverse events between treatment and placebo groups. Five of the studies reported nausea and vomiting, gastrointestinal problems, and dizziness; these findings are consistent with other effects of cholinesterase inhibitors.
The proportion of patients withdrawing because of adverse events ranged from 0% to 12% in the placebo groups and 0% to 55% in the treatment groups. Rates of withdrawal were greater with higher doses. In general, the quality of procedures used to collect harms was moderate to low across studies (Appendix Table 3). Only 1 study (58) reported methods for reporting adverse events, only half reported the frequency of collection, and no study provided an operational definition of a serious event.
Five eligible studies (65–69) (6 reports) compared memantine with placebo (Appendix Table 2). In 1 study (70) (3 reports), all participants also received donepezil for at least 6 months before randomization to memantine or placebo. Sample sizes ranged from 166 to 579 (1944 patients total). The study populations included vascular dementia (65, 66), mixed groups (67), and Alzheimer disease (68–70). Half of the studies evaluated mild to moderate disease, and the rest evaluated moderate to severe disease. All but 1 study (68) used a final dosage of 20 mg/d; this study had the shortest duration (12 weeks) compared with the other trials (range, 24 to 28 weeks).
Two studies (65, 66) in patients with mild to moderate vascular dementia showed significant improvement on the ADAS-cog. The summary estimate was also significant (Table 4). Two studies (68, 70) found changes on the SIB to be significantly different in patients with moderate to severe Alzheimer disease. In 1 of these trials, patients also received donepezil; the summary effect size was significant but also positive for heterogeneity. Two studies (65, 66), both evaluating vascular dementia, showed varied results for the MMSE. A single study (68) in patients with Alzheimer disease found no difference in MMSE scores. Another (69) showed significant differences on the ADAS-cog. The summary estimate was not significant (Table 4) but was positive for heterogeneity. Figure 2 shows that in studies that used the ADAS-cog to evaluate patients with vascular dementia, the summary estimate was significant.
Two trials (65, 66) that evaluated patients with vascular dementia did not show significant differences on the CIBIC-plus, and a summary effect size could not be computed. In studies that enrolled patients with Alzheimer disease, the summary effect size was statistically significant (Table 4). In 1 of these trials, all patients received donepezil concurrently. Figure 3 shows that the summary effect size for the CIBIC-plus was statistically significant. This meta-analysis combines patients with Alzheimer disease and those with vascular dementia. Sensitivity analyses showed no change in the summary effect size when the studies were stratified according to the types of dementia.
Of the 5 studies that evaluated behavior, only 2 showed statistically significant differences. However, the summary estimate (Table 4) showed a significant effect size for both the Nurses Observation Scale for Geriatric Patients in patients with vascular dementia and the Neuropsychiatric Inventory in patients with Alzheimer disease. One study (70), in which all participants were also taking donepezil, showed improvements in the Neuropsychiatric Inventory caregiver distress subscale at 12 weeks (P = 0.006) but not 24 weeks (P = 0.059).
Three (67, 68, 70) of the 4 studies that evaluated ADLs showed statistically significant differences; the summary effect size for the Alzheimer's Disease Clinical Scale–ADL subscale was also statistically significant (Table 4). Two trials (68, 70) evaluated caregiver burden and resource utilization and found statistically significant improvements in patients with moderate to severe Alzheimer disease.
Appendix Table 3 reports the frequencies of various harms. Reported adverse events included nausea, dizziness, diarrhea, and agitation (8% to 32% in the placebo groups; 4% to 18% in the treatment groups [all doses]), and none reported eating disorders. In all but 1 study (69) that evaluated agitation as a potential adverse event, the treatment group experienced less agitation; the pooled estimate was not significant, but memantine showed a protective effect for agitation (Appendix Table 4). One trial (65) tested and found no significant differences between the treatment and placebo groups. None of the summary effect size estimates for different harms reported for memantine was statistically significant (Appendix Table 4).
All but 1 study (67) reported withdrawal rates; the proportion of patients withdrawing because of any adverse events varied from 5% to 17% in the placebo groups and 8% to 13% in the treatment groups. The quality of methods used to collect and report harms was moderate to high (Appendix Table 3); however, no study provided an operational definition of serious events or indicated that a standardized instrument for collection was used.
Two studies (34, 35) compared donepezil (10 mg/d) with galantamine in 251 patients. One study (35) was a pilot undertaken primarily to evaluate the potential of rivastigmine to affect sleep in patients with mild to moderate Alzheimer disease and lasted only 8 weeks. It was insufficiently powered because of the small sample size. The second study (34) showed no statistical differences in the primary outcome of function (measured with the Bristol Activities of Daily Living Scale) in patients with Alzheimer disease over 52 weeks. Changes in secondary outcomes of cognition (measured with the ADAS-cog and MMSE) showed statistical differences favoring galantamine only in a subgroup of patients with MMSE scores between 12 and 18. One study (34) showed differences favoring galantamine over donepezil in scores on the Screen for Caregiver Burden. However, many caregivers were missing from the analysis, and the results were presented in a limited manner. In this trial, the adverse events most frequently reported were nausea, agitation, vomiting, headache, and falls. Although not statistically evaluated, the rates for all these harms were marginally higher with galantamine. Galantamine and donepezil did not differ with respect to serious adverse events.
One large trial (36, 37) compared donepezil (up to 10 mg/d) with rivastigmine (up to 12 mg/d) in patients with moderately severe Alzheimer disease over 2 years. Measures of cognition (SIB and MMSE) and behavior (Neuropsychiatric Inventory) did not significantly differ. However, statistically significant differences in global function (Global Deterioration Scale) and function (Alzheimer's Disease Cooperative Study–ADL subscale) favored rivastigmine. A subgroup analysis of patients age 75 years or older showed statistical differences favoring rivastigmine in some measures of behavior and function compared with younger patients. The collection of harms was well reported (maximum quality score) and showed higher frequency of nausea for rivastigmine than donepezil during the titration phases. This finding was attributed to the faster escalation rate with rivastigmine. In general, patients receiving rivastigmine reported more adverse events than those receiving donepezil, but serious events did not differ.
Data on clinically significant changes for the ADAS-cog, MMSE, and CIBIC-plus were noted during data extraction. For donepezil, only 10 of 27 studies reported on the percentage of patients with a clinically significant change in at least 1 of these outcomes; 1 study (34) compared donepezil with galantamine. Six of 10 studies on galantamine, 5 of 9 studies on rivastigmine, and 3 of 6 studies on memantine reported this information. For tacrine, 2 studies reported the proportion of change for the treatment group or placebo group but not both. When data were presented on patients with clinically significant changes on the ADAS-cog, MMSE, or CIBIC-plus, they suggest that a small proportion of patients may respond to the drug therapy. However, the characteristics of these responders were not adequately detailed within the studies.
For each drug except tacrine, the evidence consisted of 1 or more well-designed and well-executed randomized, controlled trials yielding consistent, directly applicable results for most of the outcomes. However, most trials were of less than 1 year's duration. For tacrine, the evidence reviewed was of moderate quality because it was obtained from randomized, controlled trials with important limitations. These included great variation in the tacrine doses used in the studies and the fact that few studies selected the ADAS-cog or CIBIC-plus as outcomes; this limited comparability across studies and with other drugs used to manage dementias.
Methodological caveats that may affect the interpretation of systematic reviews in this areas of research centered on 3 main areas: 1) classification of dementias and severity levels, 2) capture of adverse events, and 3) definition of clinically meaningful changes in outcome measures.
Classification systems used for diagnosing the various types of dementias and other forms of cognitive impairment are not interchangeable. Moreover, concerns about the accuracy of these criteria remain (71). Defining severity in patients with dementia raises another concern, and a variety of methods were used across studies. Although popular, the MMSE may not best capture severity levels, and the categories (mild, moderate, and severe) may not always reflect cognitive and functional differences in a clinically meaningful manner. These factors may contribute to heterogeneity and limit the inferences that can be drawn across studies. Trials were inconsistent in classifying serious events or the severity of typical events. Capturing information on the basis of self-report from individuals with cognitive decline can be problematic even if done by a caregiver. Most of the trials in this systematic review were of relatively short duration and included relatively healthy individuals with mild to moderate dementias. Patients with dementia seen in practice often have more complex medical illnesses and are at greater risk for side effects and pharmacologic interactions. Published rates of adverse events in controlled trials may underestimate the rates seen in clinical practice.
Consensus is lacking about which outcomes best reflect clinical importance in the domains we evaluated. For most studies in this review, cognition and global assessment (measured with >40 different instruments) were the domains from which efficacy was determined; the emphasis on these 2 domains reflects the “dual efficacy” recommended by the FDA for dementia drugs. European guidelines emphasize the importance of functional and behavioral outcomes to evaluate efficacy of drugs (72). A clinically relevant treatment can be defined as one in which the change is both relevant and important to the patient or caregiver and to clinicians. In contrast, a statistically significant difference, which is associated with probabilities of events, does not always reflect clinically meaningful changes. The magnitude of a clinically relevant change may vary depending on whether importance is defined by the patient, caregiver, or clinician. Moreover, the goals for treatment vary with disease stage. In early stages, the aim is to improve cognition and slow progression of disease. In the mid-stages of the disease, the emphasis is on preserving function (that is, ADLs), maintaining safety, and delaying institutionalization; support in the home becomes increasingly important. In the late stages, the emphasis moves toward management of difficult behaviors, which can be addressed with both pharmacotherapy and manipulation of the physical and social environment. Ultimately, clinical significance is a complex issue; its definition can vary among individuals and clinicians as well as with the stage of disease.
Despite these methodological caveats, the literature evaluated in this systematic review can be used to guide the development of practice guidelines. The strength of the evidence considers the methodological quality, consistency, and directness of the findings and the relevance to populations likely to be prescribed the drug.
On the basis of 34 reports (from 24 distinct studies), there is consistent evidence that donepezil, at both 5 and 10 mg, improves cognition and global function assessment for patients with Alzheimer disease and vascular dementia. The summary estimates for the ADAS-cog and the CIBIC-plus suggest that these effects are small; the exception is for patients with mild cognitive impairment, in whom no benefit occurred. Clinically significant changes are demonstrated with the CIBIC-plus but not the ADAS-cog. Improvement in behavioral symptoms and quality of life were not evaluated as extensively or with consistent outcomes. Adverse events are primarily associated with gastrointestinal problems consistent with this class of drugs and are dose related. Most studies evaluated patients with mild to moderate Alzheimer disease for relatively short periods. Although differences in cognition and global assessment were maintained between treatment and placebo groups during the studies, short study durations prevent us from drawing conclusions about the potential of donepezil to delay the progression of the disease or about longer-term use (>6 months) in those already given a diagnosis. For patients with mild cognitive impairment, donepezil reduced rates of conversion to Alzheimer disease in the short term, but differences relative to placebo disappeared by 36 months.
In a previous review, Passmore and colleagues (73) pooled individual-patient data from 4 trials and concluded that donepezil affected cognition and global function in both Alzheimer disease and vascular dementia. Similarly, Whitehead and colleagues (74) pooled individual-patient data across 11 trials for Alzheimer disease and showed differences between the 5-mg and the 10-mg doses in terms of cognition and global assessment (the only 2 outcomes considered). Two Cochrane reviews (75, 76) evaluating donepezil in Alzheimer disease and vascular dementia also highlight the limited number of studies that evaluated other important outcomes (such as behavior and caregiver burden). The major findings from these previous reviews are consistent with those in this systematic review.
In the 10 studies (12 reports) evaluated, consistent evidence indicates that galantamine positively affects cognition and global assessment, as measured by the ADAS-cog and CIBIC-plus; only the latter achieved clinical significance. The evidence is inconsistent with respect to change in ADLs. The single study evaluating caregiver burden demonstrated positive results. Adverse events were primarily gastrointestinal problems. All of these studies evaluated patients with mild to moderate Alzheimer disease and vascular dementia for up to 6 months. The short duration of these studies and the open-label design of the single longer study (1 year) limit interpretation of the findings for use of the drug beyond 6 months.
Wilkinson and colleagues (77) pooled individual-patient data from 4 phase III trials (24 and 32 mg) in patients with moderate Alzheimer disease. Changes in cognition (ADAS-cog) and global assessment (CIBIC-plus) in trials of up to 6 months' duration were evaluated. The findings show statistical and clinical (change >4 points) significance for the ADAS-cog in a subgroup with a baseline ADAS-cog score greater than 30 and an MMSE score greater than 12. Their data also suggest that galantamine may delay deterioration on the Neuropsychiatric Inventory scores compared with placebo. A Cochrane review (78) concluded that galantamine improved cognition (ADAS-cog) and global assessment (CIBIC-plus) in the trials evaluated, with less consistent evidence for functional and behavior changes.
The 9 studies (11 reports) showed consistent results for the outcome of cognition and global assessment. However, the summary effect size for the ADAS-cog was not statistically significant. Although this effect was smaller relative to other cholinesterase inhibitors, fewer studies were combined; in addition, heterogeneity was present. In a Cochrane review, Birks and colleagues (79) reported an effect size of −2.09, which is similar to that estimated in our review. Efficacy in other domains has not been evaluated as extensively, but findings suggest some benefit in ADLs. Adverse events are primarily associated with gastrointestinal problems consistent with this class of drugs. Half of the studies evaluated patients with disease of all severity levels (1 study included patients with moderate to severe disease) for a duration of up to 6 months. One study with a small sample size maintained blinding for up to 12 months, suggesting that statistical differences between groups were maintained.
On the basis of the 6 trials that compared tacrine with placebo, the evidence is less consistent for a significant difference in cognition. In part, this may be due to the choice of outcome instruments, small sample sizes, low doses, and insufficient study duration. Two of 3 trials found benefit in global assessment. Benefit in other domains has not been evaluated as extensively. Adverse events were primarily associated with gastrointestinal problems. The high rates of withdrawal due to adverse events and the potential for liver damage suggest that this drug is less well tolerated by patients.
Many previous reviews have considered the class effects of cholinesterase inhibitors in patients with dementia. Trinh and colleagues (80) selected studies that used the Neuropsychiatric Inventory and instrumental ADLs. Their conclusions were based on various cholinesterase inhibitors (including velnacrine, physostigmine, eptastigmine, metrifonate, and those included in the present review) and noted only small changes in function and neuropsychiatric symptoms. A recent review by the Canadian Centre for Health Technology (81) on cholinesterase inhibitors (donepezil, rivastigmine, and galantamine) in mild to moderate Alzheimer disease also concluded that the long-term benefit of using these drugs was difficult to evaluate given the short duration of the trials. Comparisons between the cholinesterase inhibitors could not be made because of the paucity of head-to-head trials. Lanctôt and colleagues (82), in a review of 16 cholinesterase trials, also concluded that the mean proportion of global responders in excess of placebo was 9% (CI, 6% to 12%).
On the basis of 6 distinct studies (from 9 reports), consistent evidence indicates that memantine improves cognition and global assessment, but the magnitude of the effect size for the ADAS-cog does not approximate those considered clinically significant. Outcomes of benefit in other domains are limited but suggest improvement in quality of life in patients with moderate to severe Alzheimer disease. Adverse events included gastrointestinal symptoms, dizziness, and headache; in most studies, agitation was less frequently reported in the treatment group than in the placebo group. Memantine was well tolerated as monotherapy and in conjunction with donepezil. The memantine trials evaluated populations with mild to moderate vascular dementia and moderate to severe Alzheimer disease. Although differences in cognition, global assessment, and quality of life were maintained throughout the study, the short study duration cannot inform use for more than 28 weeks. Other reviews on memantine (83, 84) concur with the findings of this systematic review.
For the treatment of dementias, cholinesterase inhibitors and memantine can improve symptoms, primarily in the domains of cognition and global function. Clinically important differences were not consistently evaluated or demonstrated in these 2 domains for all drugs. Direct comparisons among these drugs are limited and do not suggest important differences.
Cummings JL, Schneider L, Tariot PN, Kershaw PR, Yuan W. Reduction of behavioral disturbances and caregiver distress by galantamine in patients with Alzheimer's disease. Am J Psychiatry. 2004;161:532-8. [PMID: 14992980]
Davis KL, Thal LJ, Gamzu ER, Davis CS, Woolson RF, Gracon SI, et al. A double-blind, placebo-controlled multicenter study of tacrine for Alzheimer's disease. The tacrine Collaborative Study Group. N Engl J Med. 1992;327:1253-9. [PMID: 1406817]
Davis KL, Yang RK, Davidson M, Mohs RC, Ryan TM, Schmeidler J et al. Alzheimer's disease: tacrine and tacrine metabolite concentrations in plasma and cognitive change. Drug Dev Res. 1995;34:55-65.
Farlow M, Gracon SI, Hershey LA, Lewis KW, Sadowsky CH, Dolan-Ureno J. A controlled trial of tacrine in Alzheimer's disease. The Tacrine Study Group. JAMA. 1992;268:2523-9. [PMID: 1404819]
Farlow MR, Lahiri DK, Poirier J, Davignon J, Hui S. Apolipoprotein E genotype and gender influence response to tacrine therapy. Ann N Y Acad Sci. 1996;802:101-10. [PMID: 8993489]
Feldman H, Gauthier S, Hecker J, Vellas B, Subbiah P, Whalen E, et al. Donepezil MSAD Study Investigators Group. A 24-week, randomized, double-blind study of donepezil in moderate to severe Alzheimer's disease. Neurology. 2001;57:613-20. [PMID: 11524468]
Forette F, Hoover T, Gracon S, de Rotrou J, Hervy MP, Lechevalier B, et al. A double-blind, placebo-controlled, enriched population study of tacrine in patients with Alzheimer's disease. Eur J Neurol. 1995;2:229-38.
Fuschillo C, La Pia S, Campana F, Pinto A, De Simone L. Cognitive deficits in Alzheimer's disease: treatment with acetylcholinesterase inhibitor agents. Arch Gerontol Geriatr Suppl. 2001;7:151-8. [PMID: 11431059]
Galasko D, Kershaw PR, Schneider L, Zhu Y, Tariot PN. Galantamine maintains ability to perform activities of daily living in patients with Alzheimer's disease. J Am Geriatr Soc. 2004;52:1070-6. [PMID: 15209643]
Gauthier S, Feldman H, Hecker J, Vellas B, Ames D, Subbiah P, et al. Donepezil MSAD Study Investigators Group. Efficacy of donepezil on behavioral symptoms in patients with moderate to severe Alzheimer's disease. Int Psychogeriatr. 2002;14:389-404. [PMID: 12670060]
Homma A, Takeda M, Imai Y, Udaka F, Hasegawa K, Kameyama M, et al. Clinical efficacy and safety of donepezil on cognitive and global function in patients with Alzheimer's disease. A 24-week, multicenter, double-blind, placebo-controlled study in Japan. E2020 Study Group. Dement Geriatr Cogn Disord. 2000;11:299-313. [PMID: 11044775]
Kim JM, Shin IS, Yoon JS. Correlates of dropout, efficacy, and adverse events in treatment with acetylcholinesterase inhibitors in Korean patients with Alzheimer's disease. Int Psychogeriatr. 2002;14:187-95. [PMID: 12243209]
Markowitz JS, Gutterman EM, Lilienfeld S, Papadopoulos G. Sleep-related outcomes in persons with mild to moderate Alzheimer disease in a placebo-controlled trial of galantamine. Sleep. 2003;26:602-6. [PMID: 12938815]
Möbius HJ, Stöffler A. Memantine in vascular dementia. Int Psychogeriatr. 2003;15 Suppl 1:207-13. [PMID: 16191242]
Moretti R, Torre P, Antonello RM, Cazzato G, Bava A. Depression and Alzheimer's disease: symptom or comorbidity? Am J Alzheimers Dis Other Demen. 2002;17:338-44. [PMID: 12501480]
Nakano S, Asada T, Matsuda H, Uno M, Takasaki M. Donepezil hydrochloride preserves regional cerebral blood flow in patients with Alzheimer's disease. J Nucl Med. 2001;42:1441-5. [PMID: 11585854]
Olin JT, Schneider LS. Assessing response to tacrine using the factor analytic structure of the Alzheimer's Disease Assessment Scale (ADAS)-cognitive subscale. Int J Geriatr Psychiatry. 1995;10:753-56.
Onofrj M, Thomas A, Luciano AL, Iacono D, Di Rollo A, D'Andreamatteo G, et al. Donepezil versus vitamin E in Alzheimer's disease: Part 2: mild versus moderate-severe Alzheimer's disease. Clin Neuropharmacol. 2002;25:207-15. [PMID: 12151908]
Orgogozo JM, Small GW, Hammond G, Van Baelen B, Schwalen S. Effects of galantamine in patients with mild Alzheimer's disease. Curr Med Res Opin. 2004;20:1815-20. [PMID: 15537482]
Raskind MA, Peskind ER, Wessel T, Yuan W. Galantamine in AD: A 6-month randomized, placebo-controlled trial with a 6-month extension. The Galantamine USA-1 Study Group. Neurology. 2000;54:2261-8. [PMID: 10881250]
Rockwood K, Mintzer J, Truyen L, Wessel T, Wilkinson D. Effects of a flexible galantamine dose in Alzheimer's disease: a randomised, controlled trial. J Neurol Neurosurg Psychiatry. 2001;71:589-95. [PMID: 11606667]
Smith F. Mixed-model analysis of incomplete longitudinal data from a high-dose trial of tacrine (Cognex) in Alzheimer's patients. J Biopharm Stat. 1996;6:59-67. [PMID: 8838779].
Tariot PN, Solomon PR, Morris JC, Kershaw P, Lilienfeld S, Ding C. A 5-month, randomized, placebo-controlled trial of galantamine in AD. The Galantamine USA-10 Study Group. Neurology. 2000;54:2269-76. [PMID: 10881251]
Wilcock GK, Lilienfeld S, Gaens E. Efficacy and safety of galantamine in patients with mild to moderate Alzheimer's disease: multicentre randomised controlled trial. Galantamine International-1 Study Group. BMJ. 2000;321:1445-9. [PMID: 11110737]
Wilkinson DG, Passmore AP, Bullock R, Hopker SW, Smith R, Potocnik FC, et al. A multinational, randomised, 12-week, comparative study of donepezil and rivastigmine in patients with mild to moderate Alzheimer's disease. Int J Clin Pract. 2002;56:441-6. [PMID: 12166542]
Wimo A, Winblad B, Engedal K, Soininen H, Verhey F, Waldemar G et al. Donepezil Nordic Study Group. An economic evaluation of donepezil in mild to moderate Alzheimer's disease: results of a 1-year, double-blind, randomized trial. Dement Geriatr Cogn Disord. 2003;15:44-54. [PMID: 12457078].
Winblad B, Engedal K, Soininen H, Verhey F, Waldemar G, Wimo A, et al. Donepezil Nordic Study Group. A 1-year, randomized, placebo-controlled study of donepezil in patients with mild to moderate AD. Neurology. 2001;57:489-95. [PMID: 11502918]
The In the Clinic® slide sets are owned and copyrighted by the American College of Physicians (ACP). All text, graphics, trademarks, and other intellectual property incorporated into the slide sets remain the sole and exclusive property of the ACP. The slide sets may be used only by the person who downloads or purchases them and only for the purpose of presenting them during not-for-profit educational activities. Users may incorporate the entire slide set or selected individual slides into their own teaching presentations but may not alter the content of the slides in any way or remove the ACP copyright notice. Users may make print copies for use as hand-outs for the audience the user is personally addressing but may not otherwise reproduce or distribute the slides by any means or media, including but not limited to sending them as e-mail attachments, posting them on Internet or Intranet sites, publishing them in meeting proceedings, or making them available for sale or distribution in any unauthorized form, without the express written permission of the ACP. Unauthorized use of the In the Clinic slide sets will constitute copyright infringement.
Norman L. Foster
University of Utah
March 13, 2008
Appropriate Drug Treatment Depends on Cause of Dementia
To the Editors:
The evidence review by Raina et al. (1) and the associated ACP-AAFP Guidelines (2) published in the March 4, 2008 issue are disappointing and do not reflect current diagnostic practice or FDA indications for treatment. Both fail to recognize that the appropriate choice of drug treatment depends upon the cause of the dementia. They also leave the incorrect impression that the FDA has specifically approved drugs for the treatment of dementia. Instead, FDA indications are much more explicit (3). Tacrine, galantamine, and donepezil are approved only for treatment of dementia of the Alzheimer's type. Rivastigmine is only approved for treatment of the Alzheimer's type and Parkinson's disease with dementia. There are many other causes of dementia that warrant different treatment. For example, there is little disagreement that dementia due to B12 deficiency should be treated with B12 supplementation and that drugs known to reduce the risk of recurrent stroke should be used in patients with vascular dementia.
Although the Raina review was intended to evaluate the treatment of dementia, it included studies of mild cognitive impairment, even though patients with this condition by definition do not have dementia. There are no FDA-approved drug treatments for mild cognitive impairment. It is unlikely that a single drug will effectively treat all causes of dementia and all symptoms that patients with dementia can develop. The drugs evaluated in these articles were developed primarily to address cognition. A discussion of the relative effectiveness of these drugs for specific symptoms would have been helpful. While more studies are needed, antidepressants and psychotropic agents are more likely to prove effective in treating depression and behavior than the drugs considered in these papers. Future drug treatments will have even more targeted mechanisms of action making it even more important that the cause of dementia is determined early and accurately.
Norman L. Foster, M.D.
Center for Alzheimer's Care, Imaging and Research, University of Utah
1. Raina P, Santaguida P, Ismaila A, Patterson C, Cowan D, Levine M, et al. Effectiveness of cholinesterase inhibitors and memantine for treating dementia: evidence review for a clinical practice guideline. Ann Intern Med. 2008;148:379-97. [PMID: 18316756].
2. Qaseem A, Snow V, Cross JT Jr, Forciea MA, Hopkins R Jr, Shekelle P, et al. Current pharmacologic treatment of dementia: a clinical practice guideline from the American College of Physicians and the American Academy of Family Physicians. Ann Intern Med. 2008;148:370-8. [PMID: 18316755].
3. FDA. Labels and patient package inserts.
Dr. Foster received personal compensation for speaking and serving on the Scientific Advisory Board for Myriad Pharmaceuticals and GE Healthcare. He received research support to conduct clinical trials from Merck, Myriad, Eisai, and Wyeth.
March 18, 2008
Can we learn more from dementia trials?
In their comprehensive systematic review, Raina et al. non- controversially conclude that the effect sizes for cholinesterase inhibition in Alzheimer's disease typically are small, and that translation of these effects into patients' quality of life remains problematic . There is perhaps even more to be concluded from the existing literature. If we accept that the average small effect size represents a distribution of effects, it is reasonable to ask why some people seem to do well in response to treatment and others do not. Three types of answers seem possible.
At the group level, a dose-response can be demonstrated, such that patients who are able to tolerate the highest doses commonly demonstrate effect sizes in the moderate range . This should not be gainsaid; there is also demonstrable convergence of the measures within and across trials, which suggests, a priori, that the effects are real . Even so, current evidence is only for dose, and not for type of cholinesterase inhibitor, or use in combination with other drugs. A second conclusion, also consistent with Raina et al., is that the existing measures do not adequately allow us to identify people who respond to treatment from those who do not. One reason for this might be that we have discounted improved executive function in the standard trials [4-6].
A third conclusion is that we need to pay attention to the experience of patients and caregivers to better understand not just their satisfaction with treatment (something that has been caught up in what, as argued elsewhere,  is an often-dubious "quality of life" evaluation effort). Regulatory agencies endorse patient-centred measurement, and the accompanying ACP/AAFP consensus statement equally sees this as a priority for additional dementia research. Even so, most clinical trials have not taken patient/caregiver views into account, although one that did seems to show that important aspects of the treatment response are missed by current measures [9,10]
There are reasons other than efficiency to listen to what patients and caregivers tell us about the effects of treatment. Most compellingly, what they say in trials can address clinical meaningfulness directly, and shorten the translation between the trials and what we can look for in daily practice. In addition, just as patients and caregivers stressed to us the importance of executive function long before it became fashionable to measure it in clinical trials, their experience can help us better understand how our treatments work. Especially in matters of higher cortical function, these are effects which are unlikely to emerge from pre -clinical models.
To the extent that it discounts listening to patients as "mere anecdote", the evidence-based movement needs to get to grips with the gap between what trials say and what people experience. Systematic listening to patients in quantifiable ways in properly designed trials allows anecdote to become evidence, but the trials reviewed by Raina et al. show what distance we are from attaining that goal.
1. Raina P, Santaguida P, Ismaila A, et al., Effectiveness of cholinesterase inhibitors and memantine for treating dementia: evidence review for a clinical practice guideline. Ann Intern Med. 2008;148:379-97.
2. Rockwood K. Size of the treatment effect on cognition of cholinesterase inhibition in Alzheimer's disease. J Neurol Neurosurg Psychiatry. 2004;75:677-85.
3. Rockwood K, MacKnight C. Assessing the clinical importance of statistically significant improvement in anti-dementia drug trials. Neuroepidemiology. 2001;20:51-6.
4. Royall DR, Lauterbach EC, Cummings JL, et al., Executive control function: a review of its promise and challenges for clinical research. A report from the Committee on Research of the American Neuropsychiatric Association. J Neuropsychiatry Clin Neurosci. 2002;14:377-405.
5.Voss SE, Bullock RA. Executive function: the core feature of dementia? Dement Geriatr Cogn Disord. 2004;18:207-16.
6. Rockwood K, Fay S, Gorman M, Carver D, Graham JE. The clinical meaningfulness of ADAS-Cog changes in Alzheimer's disease patients treated with donepezil in an open-label trial. BMC Neurol. 2007 Aug 30;7:26.
7. Rockwood K. Quality of life outcomes. In Rockwood K, Gauthier S (eds.) Trial Designs and Outcomes in dementia Therapeutic Research. London: Taylor & Francis, 2006, 131-140.
8.Qaseem A, Snow V, Cross JT Jr, et al. Current pharmacologic treatment of dementia: a clinical practice guideline from the American College of Physicians and the American Academy of Family Physicians. Ann Intern Med. 2008;148:370-8.
9. Rockwood K, Fay S, Song X, MacKnight C, et al., Attainment of treatment goals by people with Alzheimer's disease receiving galantamine: a randomized controlled trial. CMAJ. 2006;174:1099-105.
10. Rockwood K, Fay S, Jarrett P, Asp E. Effect of galantamine on verbal repetition in AD: a secondary analysis of the VISTA trial. Neurology. 2007;68:1116-21.
In addition to the usual academic conflict of interest (evident from the references) I have worked with each of the drug companies that makes cholinesterase inhibitors, and the VISTA trial (references 9,10) was an investigator initiated, peer-reviewed double blind clinical trial, funded in part by the Canadian Institutes of Health Research, and in the majority by Janssen-Ortho Canada. I also am founder and majority shareholder of DementiaGuide Inc., a web-based company that provides information on dementia and its treatment, and that has a contract with Pfizer Canada to improve patient-centred measurement in a clinical trial.
Nancy S. Foldi
Queens College, City University of New York
April 7, 2008
Choose better neuropsychological measures
To the editors:
The review of Raina et al.,(1) and commentary of Quaseem et al.,(2) continue our debate of how to evaluate the pharmacological treatments of Alzheimer's disease (AD). However, until new drugs emerge that produce more obvious "˜clinically significant' symptom improvement, the research community needs to use objective measures that better demonstrate an individual's response to treatment. Indeed, a great frustration for physicians and families alike is the lack of any sensitive, valid, reliable, objective outcome metric to determine efficacy. Families report vague changes (e.g., the patient is more "˜alert'), but we cannot even corroborate their impression. Our current outcome measures and methodologies need correction. The industry standard ADAS-Cog(3) assesses placebo versus treatment groups, but was never designed to assess individual response. Moreover, most significant group effects appear only after 4-6 months of treatment, confounding treatment efficacy and disease progression. We need to use sensitive outcome measures that detect efficacy after short-term treatment before additional natural disease progression occurs.
Not all neuropsychological measures are created equal. Global measures, such as ADAS-Cog (3)or the MiniMental Status Examination (4)collapse multiple cognitive domains, and while they have served important roles in efficacy studies, global measures can diffuse a true effect(5). In contrast, targeted cognitive measures exist that are sensitive to AD are delayed recall(6), learning strategy of the recency effect(7), selective(8;9) or divided attention(10). Moreover, the choice of outcome measure should directly relate to known associations with the neurotransmitter system in question: we already know that acetylcholine mediates the effect of attentional demands and memory consolidation (see 11;12). Of note is that the 70-point ADAS-Cog lacks any measures of attention, learning strategy, or consolidation.
Any neuropsychological outcome measure of drug efficacy should fulfill three mandates. First, the outcome measure has to be sensitive and valid, and has to assess a targeted cognitive function. Estimating height using a scale could tell us that heavier people tend to be taller, but using a yardstick would be better. Second we have to detect efficacy after a short-term treatment (e.g., 4-8 weeks), not only to obviate the confound of treatment and progression, but also to aid in a timely decision whether or not to continue treatment. Lastly, a best outcome measure is a response in an individual patient. AD presents as a heterogeneous disease resulting in large variance of performance, necessitating greater group differences between placebo and treatment and resulting in smaller effect sizes. The goal should be an individual's change as a function of treatment on the sensitive measure. Better measures already exist and should be applied.
(1) Raina P, Santaguida P, Ismaila A, Patterson C, Cowan D, Levine M et al. Effectiveness of cholinesterase inhibitors and memantine for treating dementia: evidence review for a clinical practice guideline. Ann Intern Med. 2008;148:379-97.
(2) Qaseem A, Snow V, Cross JT, Jr., Forciea MA, Hopkins R, Jr., Shekelle P et al. Current pharmacologic treatment of dementia: a clinical practice guideline from the American College of Physicians and the American Academy of Family Physicians. Ann Intern Med. 2008;148:370-378.
(3) Rosen WG, Mohs RC, Davis KL. A new rating scale for Alzheimer's disease. Am J Psychiatr. 1984;141:1356-64.
(4) Folstein MF, Folstein SE, McHugh PR. Mini-Mental State: A practical method for grading the cognitive state of patients for the clinician. J Psychiatr Res. 1975;12:189-98.
(5) Rockwood K. Size of the treatment effect on cognition of cholinesterase inhibition in Alzheimer's disease. J Neurol Neurosurg Psychiatry. 2004;75:677-85.
(6) Delis DC, Kaplan E, Kramer JH, Ober BA. California Verbal Learning Test - II (second edition). San Antonio, TX: The Psychological Corporation; 2000.
(7) Foldi NS, Brickman AM, Schaefer LA, Knutelska ME. Distinct serial position profiles and neuropsychological measures differentiate late life depression from normal aging and Alzheimer's disease. Psychiatry Res. 2003;120:71-84.
(8) Foldi NS, Schaefer LA, White REC, Johnson R.Jr, Berger JT, Carney MT et al. Effects of graded levels of physical similarity and density on visual selective attention in patients with Alzheimer's disease. Neuropsychology. 2005;19:5-17.
(9) Foldi NS, White RE, Schaefer LA. Detecting effects of donepezil on visual selective attention using signal detection parameters in Alzheimer's disease. Int J Geriatr Psychiatry. 2005;20:485-88.
(10) Gron G, Brandenburg I, Wunderlich AP, Riepe MW. Inhibition of hippocampal function in mild cognitive impairment: targeting the cholinergic hypothesis. Neurobiol Aging. 2006;27:78-87.
(11) Arnold HM, Nelson CL, Sarter M, Bruno JP. Sensitization of cortical acetylcholine release by repeated administration of nicotine in rats. Psychopharmacology. 2003;165:346-58.
(12) Sarter M, Bruno JP, Givens B. Attentional functions of cortical cholinergic inputs: what does it mean for learning and memory? Neurobiol Learn Mem. 2003;80:245-56.
Vifor AG, consultant
May 14, 2008
Authors: P. Raina, P.L. Santaguida, C. Patterson
Our recent publication(1) presents the results of a systematic review on the efficacy of cholinesterase inhibitors and memantine in the treatment of dementias. We thank Dr Rockwood for his thoughtful and perceptive response to this review and also thank Ms. Carins for bringing to our attention new information with regard to the publication by Raskind.(2)
We completely agree with Dr. Rockwood that aggregate scores on instruments which do not measure executive function may fail to capture useful improvements in individuals receiving cholinesterase inhibitors or other cognitive enhancing agents. Of the many attempts to isolate meaningful individual responses to treatment, Goal Attainment Scaling is among the most sensitive and robust. While not always a means of determining outcome, and rarely quantified in as rigorous a manner as advocated by Dr Rockwood, we share his enthusiasm for "listening to patients," the most basic and satisfying element of clinical care.
Dr. Rockwood also cites two related publications that demonstrate this point. One of these studies was eligible for our review. (3) Inclusion of this study does not alter the effect size for the mean difference ADAS-cog score (-2.39 (95% CI, -3.32"”1.46)) in the meta analysis of treatment effects of galantamine for Alzheimer's disease; nor is the heterogeneity improved (Chi Squared p=0.003, I2 =70.3%). The other study was published after completion of our review.(4)
It was also brought to our attention that there was an error within the Raskind study of galantamine. (2) The mean change for the ADAS-cog score was -1.9 for the treatment group rather than +1.9 which was stated in the original study publication. Corrections or corrigenda for previously published articles are uniformly considered by NLM to be errata. In this case searching in MEDLINE by the author's name does not return the correction; neither does the citation listed within PUBMED or OVID indicate that a correction was issued. In short, detection of such corrections is problematic due to the manner in which they are indexed; uniformity amongst journals and consistency in indexing is required. None the less, the effect size of the ADAS-cog summary estimate is still not clinically significant and does not change our interpretation of the evidence report.(1) Figure 1 shows both the addition of the study by Rockwood and the effect of this correction on the meta-analysis for galantamine; it shows a slight increase in the effect size (-3.06 (95% CI -3.53"”2.58)) and the tests for heterogeneity are no longer significant (Chi Squared p = 0.715, I2 = 0.0%). The effect size for galantamine is statisitically, but not clinically significant.
(1) Raina P, Santaguida P, Ismaila A, Patterson C, Cowan D, Levine M, et al. Effectiveness of cholinesterase inhibitors and memantine for treating dementia: evidence review for a clinical practice guideline. Ann Intern Med 2008 Mar 4;148(5):379-9.
(2) Raskind MA, Peskind ER, Wessel T, Yuan W. Galantamine in AD: A 6 -month randomized, placebo-controlled trial with a 6-month extension. The Galantamine USA-1 Study Group. Neurology 2000 Jun 27;54(12):2261-8.
(3) Rockwood K, Fay S, Song X, MacKnight C, Gorman M, Video-Imaging Synthesis of Treating Alzheimer's Disease (VISTA) Investigators. Attainment of treatment goals by people with Alzheimer's disease receiving galantamine: a randomized controlled trial.[see comment]. CMAJ Canadian Medical Association Journal 2006 Apr 11;174(8):1099-105.
(4) Rockwood K, Fay S, Jarrett P, Asp E. Effect of galantamine on verbal repetition in AD: a secondary analysis of the VISTA trial. Neurology 2007 Apr 3;68(14):1116-21.
Results provided by:
Copyright © 2016 American College of Physicians. All Rights Reserved.
Print ISSN: 0003-4819 | Online ISSN: 1539-3704
Conditions of Use
This PDF is available to Subscribers Only