Healthy volunteers are often used in early phase I studies to explore safety and dosing. This early investigational stage of new drugs presents an opportunity to explore potential therapeutic efficacy, in addition to more traditional pharmacokinetic and adverse event analyses. Given the importance of identifying new treatments for diseases affecting cognition, such as Alzheimer's disease (AD) and schizophrenia, investigation of any potential cognitive changes resulting from novel agents would be of interest in early studies. Unfortunately, many of the cognitive measures [eg, Mini Mental State Examination (MMSE)]1 used to assess change in studies of cognition-enhancing drugs are of limited usefulness in healthy participants because of potential ceiling effects or insensitivity of the instruments. Possible solutions to this problem include manipulation of performance level in healthy controls or selection of cognitive tests with greater sensitivity.
Attempts to induce cognitive improvement in healthy participants through pharmacological intervention have yielded mixed results. Drugs targeted at the cholinergic system have been found to improve cognitive symptoms relative to placebo in AD and more recently in mild cognitive impairment (MCI).2-5 Moreover, the magnitude of the effect relative to placebo appears to be positively correlated with disease severity.6 Conversely, drugs with anticholinergic properties have long been associated with impairment of cognition (ie, memory) in both healthy participants and patients.7-10 Given the link between cognitive performance and cholinergic function, cholinomimetic compounds have been used in healthy participants to examine experimental manipulation of cognitive performance. In a crossover study, Furey et al11 found differential patterns of extrastriatal activation during functional magnetic resonance imaging in healthy participants administered the cholinesterase inhibitor physostigmine relative to the placebo condition during a working memory task. Enhancement of memory, attention, and learning has been demonstrated with cholinomimetics in a handful of studies in healthy participants.12-14 Yet Drachman and Leavitt10 failed to show cognitive improvement in healthy participants after physostigmine administration. Recently, Kitagawa et al15 found improved attention, working memory, and episodic memory in healthy adult men after 5-day treatment sessions with a novel nicotinic cholinergic agonist (GTS-21). However, the limitation of measurements to narrow domains of cognition using experimental instruments is a shortcoming in many of the above studies.
In a recent study,16 we explored the sensitivity of a battery of standard neuropsychological (NP) tests in a sample of healthy adult participants who received donepezil, placebo, or no treatment for 2 weeks. In this study, participants who received donepezil surprisingly showed transient cognitive worsening on tests of memory and attention. Nevertheless, the battery proved to be successful in detecting small cognitive changes. The lack of an improvement with donepezil may have been caused by an already "optimized" cholinergic system in healthy participants who do not benefit from additional cholinergic transmission. Interestingly, Rabey et al7 found that healthy elderly participants with lower basal mental performance were more negatively affected on cognitive tasks by the cholinergic antagonist scopolamine and suggested a connection between the "integrity of the cholinergic system" and responsivity to cholinergic drugs. Thus, selection of subjects in the lower ranges of normal (average to below average) cognitive functioning may provide a better opportunity for improvement with donepezil.
The present investigation was designed to further assess effects of donepezil on healthy participants using several methodological changes. First, elderly adults were recruited rather than young adults. Furthermore, participants with average to low-average initial cognitive performance were recruited specifically (ie, those with above-average cognition were excluded). This population was selected to avoid the ceiling effects observed previously on some cognitive measures in healthy participants and to maximize the ability to detect cognitive improvement. A well-known computerized test battery [Cambridge Neuropsychological Test Automated Battery (CANTAB)]17 with proven sensitivity to cognitive change in several drug trials and in healthy volunteers18-20 was also added to the battery of standard tests. Because the CANTAB had 4 alternate forms available, 4 test sessions were used, rather than the 6 sessions in our previous study. The length of treatment remained the same. Finally, the use of donepezil in the young subjects was chosen because of donepezil's proven effects in patients with AD. The possibility that the negative results in our previous studies was caused by an inadequate dose of donepezil was addressed by increasing the dose from 5 to 10 mg/d, which is the more commonly used therapeutic dose in elderly subjects treated for cognitive dysfunction and in clinical trials.5
In addition to reexamining our previous results showing worsening of performance with donepezil, we also wished to examine learning effects and test performance stability in a small sample of normal subjects, using alternate forms of tests when available. Practice effects were expected to be attenuated by the use of alternate forms and to be confined to the first 2 to 3 sessions.21 Knowledge of the number of practice trials to reach asymptote would allow optimized measurement of drug effects in future clinical trials.
Twenty-six healthy elderly adults (73% women; 81% white) aged 55 to 75 (mean = 65.4, SD = 5.7) years were recruited from the Indianapolis area, signed informed consents approved by the investigational review board to participate, and were financially compensated for their time (see Table 1 for sample demographics). All participants were screened for health status including laboratory blood work, chest x-ray, and electrocardiogram and completed a drug screen before enrollment in the study. Participants were excluded for significant medical or neuropsychiatric disease, color blindness, use of drugs of abuse or positive findings on urinary drug screening and ethanol testing, known allergies to donepezil and piperidine derivatives, positive human immunodeficiency virus antibodies, or current medication use that may interfere with cognitive performance.
Design and Procedure
This was a 6-week double-blind study of donepezil compared with placebo. Study participants were screened within 30 days before the first test session for medical status as noted above and for neuropsychological status. At screening, participants were given the Wide-Range Achievement Test, Third Edition (WRAT)22 Reading subtest (estimate of premorbid intellect),23 the MMSE, and the Consortium to Establish a Registry for Alzheimer's Disease list learning task of verbal memory for the purposes of excluding people at either extreme of cognitive functioning (eg, above average or with potential early dementia). Exclusion criteria included the following: a WRAT standard score above 100 (average) or below 70 (lower limit of low average), an MMSE total score below 24 (mildly impaired), and a Consortium to Establish a Registry for Alzheimer Disease delayed list recall more than 1.5 SDs below the mean for age and education (mildly impaired). Trained nursing staff under the supervision of a neuropsychologist administered these tests. Subjects completed 2 baseline test sessions to eliminate most practice effects and were randomized into the 2 treatment arms after session 2 (day 14). Subjects received either 5 mg of donepezil or placebo twice a day for 2 weeks. The neuropsychological battery was administered every 2 weeks on days 0, 14 (prerandomization baseline), 28 (end of therapy), and 42 (washout). Subjects, test administrators, and physicians were blinded to the treatment subjects received.
The pharmacodynamic end points for this study were subtests within 2 cognitive batteries: a battery of standard paper-and-pencil NP tests and the CANTAB. The NP battery consisted of tests that are commonly used in clinical neuropsychological assessment24,25 to measure attention, memory, executive functioning, language, and motor ability. Tests were chosen if they were amenable to the creation of alternate forms and if they were commonly used and accepted as valid tests. Alternate forms were created26 for most tests in the NP battery to explore possible attenuation of practice effects with alternate forms. All alternate forms were previously validated, and no form effects were found in a test-retest paradigm.27 The CANTAB is administered on a touch-activated computer screen and consists of tests of memory, attention, executive skills, and psychomotor speed. A psychologist or research assistant under a psychologist's supervision individually administered all tests in a quiet room; day and time of test administration were consistent for each subject throughout the study. The test sessions lasted between 1 and 3 hours, depending on the speed of the examinee and number of requested breaks. Caffeine and tobacco use were monitored for consistency. Form order was randomized across groups. Order of test presentation within the batteries was held constant. The CANTAB battery always preceded the NP battery.
For each selected cognitive measure, an analysis of covariance model was performed as specified in the protocol to assess the effect of donepezil versus placebo on cognition during the treatment phase (session 3). The dependent variable was change from baseline (session 2) to end point (session 3). Model terms included baseline measurements (session 2) as the covariate and treatment effects. An analysis of covariance model with change from baseline (session 3) to washout (session 4) as the dependent variable was used to analyze the washout phase. Model terms included baseline measurements (session 3) as the covariate and treatment effects.
Exploratory analyses were performed to assess the learning effects of the cognitive measures over time. For each measure, a repeated-measures analysis of covariance was performed for the placebo group across the 4 sessions to test the significance of session effect. Model terms include baseline measurement (session 1) as the covariate, session as a fixed effect, and subject as a random effect. Compound symmetry covariance structure was used. Furthermore, for each outcome measure, paired t tests were used to test the trend of improvement across adjacent sessions for the placebo group. For all subjects, paired t tests were calculated on each measure to compare sessions 1 versus 2. Because of the exploratory nature of this study, the P values reported were not adjusted to account for multiple comparisons. Therefore, the maximum probability that 1 or more null hypotheses was rejected incorrectly at the α = 0.05 level if greater than 0.05.
Twenty-five subjects completed the study (13 taking donepezil and 12 taking placebo) with a requirement of 28 oral doses of donepezil (5 mg twice a day) or placebo (twice a day). One subject randomized to donepezil dropped out because of drug-associated discomfort (nausea and headache) after 2 doses. All adverse symptoms were rated as mild to moderate by the donepezil group, and the only adverse symptom attributed to donepezil was 1 case of nausea. The most frequently reported symptom was nausea (n = 4). Reports of headache were equal (n = 2) in the donepezil and placebo groups. There were no statistically significant differences at the α = 0.05 level across therapy assignments in demographic measures: sex, age, education, ethnic origin, handedness, WRAT, MMSE, and Consortium to Establish a Registry for Alzheimer's Disease. The average participant was high school-educated (12.1 ± 2.4 years) and was of average intellectual endowment, as estimated with the WRAT Reading subtest. Mean performances of this sample were within the average range on all cognitive screening tests relative to published norms (Table 1). The lowest Consortium to Establish a Registry for Alzheimer's Disease delayed list recall in the sample was −1.1 SDs below the norm (low average). Therefore, this sample appears representative of the population.
Donepezil Effects on Cognition
Table 2 lists the least square means change from baseline from the comparison between the donepezil and placebo groups during the treatment phase (session 3), with session 2 as a baseline covariate. There was no evidence of improved performance with donepezil at the 2-sided α level of 0.05 compared with the placebo group. For both Trails B (P = 0.017) and CANTAB Delayed Match to Sample (DMS) (P = 0.028), the placebo group was superior to donepezil after 2 weeks of treatment. Subjects in the donepezil group spent 29 seconds longer completing Trails B and were 16% less accurate on the short delay of DMS than the placebo group (Fig. 1). Calculation of effect sizes (Table 2) suggests that the 2 significant cognitive tests produced a medium (>0.5) and a large (>0.8) effect size, according to Cohen's cutoffs.28 In addition, 11 measures met the criterion for a small effect size. No other significant changes were found.
With regard to the washout period, subjects randomized to donepezil performed significantly worse than the placebo group when donepezil was withdrawn on the Stroop Interference (P = 0.021; raw score = 4 fewer words), with session 3 performance as a covariate. However, subjects receiving donepezil were faster on the CANTAB Reaction Time test (P = 0.028; raw score = −31 milliseconds) compared with placebo at washout. No other significant changes were found.
Learning Effects of the Neuropsychological Measures
Using paired t tests, evidence of a learning effect (ie, significant improvement in session 2 compared with session 1) was observed in all subjects on Digit Symbol (P = 0.001) and Paced Auditory Serial Addition Test 2-second interval (P = 0.001), despite the use of alternate forms. Improvements in CANTAB Match to Sample (P = 0.020) and CANTAB Intra/Extradimensional Shift (P = 0.034) were also significant for all subjects from sessions 1 to 2.
Examination of the placebo group alone allows for conclusions about learning over all 4 sessions. Repeated-measures analysis of covariance for all 4 sessions in the placebo group revealed significant session effects on the following measures: Trails B (P = 0.025), Stroop Interference (P < 0.001), Finger Tapping: Dominant Hand (P = 0.014), and CANTAB Reaction Time (P = 0.028). With paired t tests across 4 sessions, the placebo group showed a significant improvement on Stroop Interference at sessions 2 (P = 0.035) and 4 (P = 0.006), Trails B at session 3 (P = 0.020), and CANTAB Rapid Visual Information Processing at session 3 (P = 0.030).
In this study, a mock phase I clinical trial with donepezil was conducted to explore the sensitivity of a battery of NP tests designed to detect a cognitive change in elderly participants. After a 2-week treatment period with donepezil, there was evidence of slight cognitive worsening, or a lack of improvement from practice, on 2 tests of short-term memory, attention, and psychomotor speed. This is consistent with our previous study.16 Importantly, this worsening was not accompanied by subjective reports of cognitive change or adverse effects such as sedation reported by the subjects treated with donepezil and was thus a relatively subtle effect that was detectable with this particular cognitive battery. Furthermore, worsening was observed in the same tests or cognitive domains as in our previous study. Specifically, after 14 days of treatment with donepezil, both young subjects in our previous study and the older subjects in the present study took longer to complete Trails B, an executive skills measure of attention and set shifting. Worsened performance was also evident on a measure of immediate recall (Delayed Matching to Sample, short delay) after 14 days of donepezil in the present study.
The above results are also in the same domain in which younger subjects demonstrated worsened performance after 7 days of donepezil on a verbal learning and memory task in our previous study. In addition, an investigation performed in collaboration with FORENAP (Rouffach, France) showed a slight worsening of immediate recall of a word list, Trails B, and attentional measures in young healthy participants treated with donepezil for 14 days (Eli Lilly and Company, unpublished data on file). This study design is limited by low power because of a small sample and multiple statistical comparisons, so results should be considered preliminary. However, these converging results, including the consistency of domains, suggest that a true mild cognitive depressing effect of donepezil in healthy participants 2 weeks after initiation of treatment may occur. The fact that the worsening in immediate recall was no longer present after 14 days of donepezil treatment in younger participants suggests that it may have been a transient effect during the acute adjustment to the drug. One recent publication5 has reported the efficacy of donepezil in older adult subjects with mild cognitive impairment. In this study, no effect of donepezil was demonstrated for the primary outcome variables, the New York University Paragraph Recall, and Alzheimer's Disease Cooperative Study Clinical Global Impression of Change-MCI. Statistically significant improvements were seen in some secondary outcome measures; however, the first assessment was performed after 6 weeks of treatment. Thus, the acute cognitive effects of this drug in patients with MCI remain unknown.
There are several possible explanations for the lack of improvement in cognitive functioning for older adults taking donepezil in this study. First, most NP tests were designed to detect abnormality or deviation from the normal population. Although we attempted to circumvent this problem by excluding participants with above-average performance on the cognitive screening tests, healthy participants did make up the sample. Therefore, the range for cognitive improvement could continue to be narrow and difficult to detect. Second, donepezil has a known plasma half-life of approximately 70 hours, yet we do not know when the maximal cognitive effect of the drug, if any, is in normal subjects. In previous studies of donepezil in AD patients, the duration of treatment was usually much longer than 2 weeks, and the first assessments often did not occur within the first 2 weeks of treatment, making comparison to the current results impossible. Based on a half-life of 70 hours, steady-state equilibrium would be expected at about 14 days; however, future studies could explore the time course of the effect of donepezil by using a longer treatment phase.
A cognitive enhancing effect of cholinesterase inhibitors in normal subjects has not been consistently demonstrated. Although acetylcholinesterase inhibitors may benefit patients with deficient cholinergic neurotransmission, such as those with AD, cognitive improvement might not be seen in healthy subjects who have presumably intact cholinergic innervation. This is particularly relevant for clinicians who must distinguish between MCI with impending conversion to AD and the normal memory decline that occurs with aging. As noted above, in a recently published study of donepezil in patients with MCI,4,5 cognitive improvement was shown on 4 secondary measures in the drug-treated group relative to a placebo-treated group after 24 weeks. Importantly, the subjects in this study were characterized as the "amnestic"-type MCI with memory impairments a necessary inclusion criterion. The authors noted that therapeutic treatment should be considered for "those individuals who are at the highest risk for developing AD." Further research is needed to identify the optimal time to administer these types of drugs to patients with mild cognitive impairment and to determine how patients with other defined subtypes of MCI react to cholinergic compounds. This diagnostic subtyping may prove to be a worthwhile predictor variable for response to drug interventions. Multisite trials of donepezil in MCI patients, including the Donepezil 401 Study Group and the Alzheimer's Disease Cooperative Study,29 will provide additional information on effects of donepezil in patients with MCI.
Interestingly, 1 of the few other published results of worsened cognitive performance in healthy adults after administration of a cholinomimetic was in a study of the herbal remedy Melissa officinalis (lemon balm), which has cholinesterase inhibiting effects. Kennedy et al30,31 found differential effects on cognitive tests depending on dose and extract/leaf preparation and suggested a U-shaped dose-response curve. An inverted U-shaped curve is suggested in the Kitagawa et al15 study for a measure of sustained attention using doses of 25, 75, and 150 mg of the cholinergic agonist GTS-21. In a study by Davis et al,32 higher doses of physostigmine (above 1.5 mg) facilitated a "physostigmine syndrome" resulting in mild sedation, bradyphrenia, and decreased speech in healthy adults. Consistent with our results, they found worsening on a test of short-term memory and concentration (digit span), but not long-term memory. Of course, direct side effects of the medication may also explain worsened performance. In the present study, only 1 report of nausea was attributed to donepezil, and the overall adverse effects reported were roughly equivalent between the treatment and control groups (Table 1); therefore, adverse effects do not appear to explain the cognitive differences between the 2 groups. However, cognitive performance could be assessed in a follow-up study with a titration phase (eg, 5 mg for 1 to 3 weeks, then increase to 10 mg) to answer this question.
A long-term goal of this type of methodological study is to develop a measure of cognitive function that can be integrated into phase I clinical trials. Ideally, this tool should detect some signal of the drug effect (cognitive improvement or deterioration) after a relatively short interval of treatment (eg, 2 to 4 weeks), as is typical in phase I studies. Our findings suggest that false-negative results could be obtained when studying cognitive enhancing drugs in healthy volunteers; conversely, the sensitivity of these instruments suggests that adverse effects related to cognition, such as those present for several centrally acting drugs including anticonvulsants, could be detected in early stages of drug development.
Despite the use of alternate forms, a learning effect was suggested from sessions 1 to 2 on 2 measures (Digit Symbol and Paced Auditory Serial Addition Test 2-second presentation rate). The Stroop Color/Word test suggested a learning effect in the placebo group across 3 sessions (1, 2, and 4). These measures may require practice trials before randomization in future studies. Alternate forms appear to have reduced practice effects in healthy older adults for the majority of other measures used in this study.
In summary, this study in elderly participants is consistent with our previous results in younger healthy participants. Taken together, these studies provide preliminary evidence that donepezil 5 to 10 mg/d may cause a slight but statistically significant worsening, or lack of learning, on tests of attention and immediate recall over a brief treatment period. Although these findings could be caused by assessment at early time points after treatment initiation and reflect only transient worsening, they also are consistent with the view that healthy subjects have intact cholinergic systems and do not benefit from additional cholinergic neurotransmission. Future studies with longer treatment periods and multiple dose levels can address these questions and provide important information about the appropriateness of using healthy adults as surrogates for patients in cognitive research.
The authors thank Jane Crawford, MS, for assistance with NP testing; the Lilly Clinic CRU staff; Marianna Schneider, MEd, study coordinator; William Z. Potter, MD, for support of these studies; and Brenda Gaydos, PhD, for statistical guidance.
1. 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
2. Freo U, Pizzolato G, Dam M, et al. A short review of cognitive and functional neuroimaging studies of cholinergic drugs: implications for therapeutic potentials. J Neural Transm
3. Mohs RC, Doody RS, Morris JC, et al. A 1-year, placebo-controlled preservation of function survival study of donepezil in AD patients. Neurology
4. Salloway SP, Kumar D, Ieni J, et al. Benefits of donepezil treatment in patients with mild cognitive impairment [abstract]. Neurology
. 2003;60(suppl 1):A411.
5. Salloway S, Ferris S, Kluger A, et al. Efficacy of donepezil in mild cognitive impairment. Neurology
6. Rogers SL, Farlow MR, Doody RS, et al. A 24-week, double-blind, placebo-controlled trial of donepezil in patients with Alzheimer's disease. Donepezil Study Group. Neurology
7. Rabey JM, Neufeld MY, Treves TA, et al. Cognitive effects of scopolamine in dementia. J Neural Transm Gen Sect
8. Gelenberg AJ, Van Putten T, Lavori PW, et al. Anticholinergic effects on memory: benztropine versus amantadine. J Clin Psychopharmacol
9. Beatty WW, Butters N, Janowsky DS. Patterns of memory failure after scopolamine treatment: implications for cholinergic hypotheses of dementia. Behav Neural Biol
10. Drachman DA, Leavitt J. Human memory and the cholinergic system. A relationship to aging? Arch Neurol
11. Furey ML, Pietrini P, Haxby J. Cholinergic enhancement and increased selectivity of perceptual processing during working memory. Science
12. Davis KL, Mohs RC, Tinklenberg J, et al. Physostigmine: improvement of long-term memory processes in normal humans. Science
13. Sitaram N, Weingartner H, Gillen JC. Human serial learning: enhancement with arecholine and choline impairment with scopolamine. Science
14. Yesavage JA, Mumenthaler MS, Taylor JL, et al. Donepezil and flight simulator performance: effects on retention of complex skills. Neurology
15. Kitagawa H, Takenouchi T, Azuma R, et al. Safety, pharmacokinetics, and effects on cognitive function of multiple doses of GTS-21 in healthy, male volunteers. Neuropsychopharmacology
16. Beglinger LJ, Gaydos B, Kareken DA, et al. Neuropsychological test performance in healthy volunteers before and after donepezil administration. J Psychopharmacol
17. Morris RG, Evenden JL, Sahakian BJ, et al. Computer-aided assessment of dementia: comparative studies of neuropsychological deficits in Alzheimer-type dementia and Parkinson's disease. In: Stahl SM, Iversen SD, Goodwin EC, eds. Cognitive Neurochemistry
. New York: Oxford University Press; 1987:21-36.
18. Harmer CJ, McTavish SF, Clark L, et al. Tyrosine depletion attenuates dopamine function in healthy volunteers. Psychopharmacology (Berlin)
19. Young AH, Sahakian BJ, Robbins TW. The effects of chronic administration of hydrocortisone on cognitive function in normal male volunteers. Psychopharmacology (Berlin)
20. Coull JT, Middleton HC, Robbins TW. Contrasting effects of clonidine and diazepam on tests of working memory and planning. Psychopharmacology (Berlin)
21. Van Gorp WG, Lamb DG, Schmitt FA. Methodologic issues in neuropsychological research with HIV-spectrum disease. Arch Clin Neuropsychol
22. Wilkinson GS. WRAT-3: Wide-Range Achievement Test (Third Edition)
. Delaware: Wide Range, Inc; 1993.
23. Kareken DA, Gur RC, Saykin AJ. Reading on the Wide Range Achievement Test-Revised and parental education as predictors of IQ: comparison with the Barona formula. Arch Clin Neuropsychol
24. Lezak M. Neuropsychological Assessment, Third Edition
. New York: Oxford University Press; 1995.
25. Wechsler D. Manual for the Wechsler Adult Intelligence Scale (WAIS-III)
. Texas: Psychological Corporation; 1997.
26. Fastenau PS, Hankins WT, McGinnis CM. Content validity of five alternate forms for six established neuropsychological tests [abstract]. Arch Clin Neuropsychol
27. Fastenau PS, Hankins WT, McGinnis CM, et al. Effects of alternate forms on retest effects in clinical testing [abstract]. J Int Neuropsychol Soc
28. Cohen J. Statistical Power Analysis for the Behavioral Sciences
. Hillsdale, NJ: Lawrence Erlbaum Associates; 1988.
29. Petersen RC, Grundman M, Thomas R, et al. Donepezil and vitamin E as treatments for mild cognitive impairment [abstract]. Neurol Aging
. 2004;25(S2):S20. (O1-05-05).
30. Kennedy DO, Scholey AB, Tildesley ND, et al. Modulation of mood and cognitive performance following acute administration of Melissa officinalis
(lemon balm). Pharmacol Biochem Behav
31. Kennedy DO, Wake G, Savelev S, et al. Modulation of mood and cognitive performance following acute administration of single doses of Melissa officinalis
(lemon balm) with human CNS nicotinic and muscarinic receptor-binding properties. Neuropsychopharmacology
© 2005 Lippincott Williams & Wilkins, Inc.
32. Davis KL, Hollister LE, Overall J, et al. Physostigmine: effects on cognition and affect in normal subjects. Psychopharmacology (Berlin)