Early Event-Related Potential Changes During Working Memory

Journal of Gerontology: MEDICAL SCIENCES 2005, Vol. 60A, No. 5, 660–666

Copyright 2005 by The Gerontological Society of America

Early Event-Related Potential Changes During Working Memory Activation Predict Rapid Decline in Mild Cognitive Impairment Pascal Missonnier,1,2 Gabriel Gold,3 Lara Fazio-Costa,3 Jean-Pierre Michel,3 Reinhild Mulligan,3 Agne`s Michon,2 Vicente Iba´n˜ez,1 and Panteleimon Giannakopoulos2,4 1

Neuroimaging Unit, Service of Geriatric Psychiatry, Department of Psychiatry, and 3 Department of Geriatrics, University Hospitals of Geneva, Switzerland. 4 Service of Old Age Psychiatry, University Hospitals of Lausanne, Switzerland. 2

Background. The conversion of mild cognitive impairment (MCI) to Alzheimer’s disease is associated with substantial compromise of neocortical circuits subserving rapid cognitive functions such as working memory. Event-related potential (ERP) analysis is a powerful tool to identify early impairment of these circuits, yet research for an electrophysiological marker of cognitive deterioration in MCI is scarce. Using a ‘‘2-back’’ activation paradigm, we recently described an electrophysiological correlate of working memory activation (positive-negative working memory [PNwm] component) over parietal electrodes. Methods. Ours was a longitudinal study of 24 MCI patients with ERP analysis at inclusion and neuropsychological follow-up after 1 year. We used ERP waveform subtraction analysis between the n-back and control tasks. Analysis of variance (ANOVA) was used to compare electroencephalograph latencies between progressive MCI (PMCI) and stable MCI (SMCI), and univariate regression was used to assess the relationship between neuropsychological measures at baseline and clinical outcome. Results. Thirteen (54%) MCI patients showed PMCI, and 11 (46%) remained stable (SMCI). In SMCI, a PNwm component with significantly larger density compared to baseline was identified when subtracting the detection task for both the 1- and 2-back tasks. In contrast, in PMCI, the PNwm component was absent in both 1-back and 2-back conditions. Neuropsychological variables and n-back test performance at inclusion did not predict cognitive deterioration 1 year later. Conclusions. In conjunction with recent functional imaging data, the present results support the notion of an early dysfunction of neural generators within the parietal cortex in MCI. They also reveal that the absence of the PNwm component may provide an easily applicable qualitative predictive marker of rapid cognitive deterioration in MCI.

M

ILD cognitive impairment (MCI) has been described as a transitional state between normal aging and dementia. It has attracted great interest from researchers seeking to identify individuals at high risk of developing dementia before the occurrence of significant structural deficits, as such patients are deemed to be the best candidates for therapeutic intervention. MCI patients progress to dementia—in particular, Alzheimer’s disease (AD)—at a greatly accelerated rate compared to normal controls. However, not all MCI patients progress to clinically defined AD or decline at identical rates, and a significant proportion of patients remain stable for several years, or even improve (1). Electroencephalography (EEG) is an easily accessible and low-cost modality that might prove to be a particularly powerful tool for the identification of predictive markers of cognitive deterioration in MCI. Yet, the number of contributions in this field remains very low (2–8). Although prospective studies have proposed that various quantitative EEG parameters may be possible predictors of dementia (5,7,8), most cross-sectional comparisons have failed to identify EEG differences under resting conditions between MCI patients and elderly controls (2,3,6). A recent study (4) 660

confirmed this situation but found significant theta power differences during haptic tasks implying for the first time, to our knowledge, that activation paradigms may improve the prediction of MCI conversion to AD. In fact, because AD symptomatology is mainly due to the progressive damage of neocortical association areas [for review, see (9) and (10)], event-related potential (ERP) analysis, which makes it possible to investigate the functional activation of neocortical circuits with a very high temporal resolution, may represent a particularly sensitive method to identify early alterations of neuronal networks predictive of AD evolution among MCI patients (4,11). Working memory, the capacity to store information and simultaneously manipulate it, is one of the cognitive processes which are mainly subserved by neocortical circuits. In agreement with a previous study suggesting an early activation of neural generators in parietal cortex during the successful performance of spatial and verbal working memory tests (12), we recently identified in healthy individuals an ERP component (a waveform labelled positive-negative working memory [PNwm]) over parietal electrodes, lasting from 140 to 280 ms after the onset of a new visual stimulus

EVENT-RELATED POTENTIAL CHANGES PREDICT RAPID DECLINE IN MCI

and showed that it represents an electrophysiological correlate of working memory load (13). This 1-year longitudinal study reveals that it is possible to predict cognitive deterioration in an MCI group on the basis of the PNwm component. METHODS

Inclusion Procedure and Follow-Up Patients were recruited in a large acute and intermediate care geriatric hospital. All admissions were screened with the Mini-Mental State Examination (MMSE) (14). Patients with MMSE scores between 25 and 28 underwent an additional clinical evaluation which included the Hospital Anxiety and Depression Scale (15) and Lawton’s Instrumental Activities of Daily Living (16). Depressive comorbidity was excluded on the basis of a Hospital Anxiety and Depression Scale score consistently less than 8. Extensive neuropsychological testing of the following was performed: attention (17); orientation (MMSE items); short-term memory [Mattis Dementia Rating Scale (DRS) items; (18), Digit Span Forward (19), Corsi’s Block-Tapping Test (20)]; episodic memory [Buschke’s Double Memory Test (21), Shapes test (22)]; executive functions [verbal fluency test (23), Trail Making Test (24)]; language [Boston Naming Test (25)]; ideomotor (26), reflexive (27), and constructional praxis (28); and visual gnosis (Ghent overlapping figures test) (29). Global cognitive function was assessed by the Clinical Dementia Rating scale [CDR; (30)] and the Mattis DRS. Patients having a test score more than 1.5 standard deviations below the age-appropriate mean in any of the above tests and a CDR score of 0.5 but no dementia, were diagnosed as possible MCI (1). These cases were reviewed independently by two highly experienced clinicians blinded to each other’s findings, and were included in the MCI group only if both clinicians concurred on this diagnosis. All participants had normal or corrected-to-normal visual acuity, and none reported a history of sustained head injury, or neurological or psychiatric disorders. Participants with regular use of psychotropics, stimulants, or b-blockers were also excluded. After formal approval of the local Ethics Committee, informed written consent was obtained from all patients included in this study. The final sample included 24 MCI patients (mean age: 83.0 6 2.5 years, age range: 78–86 years) who underwent a detailed neuropsychological followup evaluation 1 year after inclusion using the same neuropsychological battery. Experimental Design Patients watched a continuous stream of letters (pseudorandom sequences of consonants and vowels) common to the French alphabet on a computer screen and pressed a computer-controlled pushbutton with the index finger of their right hand as soon as a target appeared (response trials). For nontarget trials, no motor response was required (no-response trials). Targets were defined according to the n-back design (Figure 1). Stimuli consisted of white letters, in ‘‘Arial’’ typeface (28 3 2.58 visual angle), with 10% gray noise, embedded in a 50% random-noise gray rectangular background patch

661

(68 3 6.78 visual angle). They were presented for a duration of 0.5 seconds, separated by 5-second intervals (onset to onset), during which a dot helped patients to maintain fixation. Three different tasks were used: in a simple detection task (control), sequential letters or background patches without letters were presented. Patients responded as fast as possible when background patches without letters appeared. In the 1back task, the target was any letter identical to the one immediately preceding it. In the 2-back task, the target was any letter that was identical to the one presented two trials back. Thus, working memory load increased from control (memory-free condition) to 1-back (moderately demanding) and 2-back tasks (highly demanding). Each task was tested in three stimulus sequences of 30 images each, adding up to 90 trials per task. Before each sequence block, patients were informed about the nature of the task, and several warm-up trials were performed. A control task block was followed by one block of the 1back condition, three blocks of the 2-back condition, two blocks of the 1-back condition, and two blocks of the control task. No feedback on performance was provided. Electrophysiological and neuropsychological assessments were performed in the morning.

Electrophysiological Recordings Continuous electroencephalography (Brain Quick system 98, BQ 3200; Micromed, Treviso, Italy) was recorded, using 20 surface electrodes placed over the scalp according to the International 10–20 System of Electrode Placement (31). Linked earlobes were used as a reference. Skin impedance was kept below 5 k . Physiological signals were sampled at 1024 Hz, the lower cutoff was 0.33 Hz, and the upper cutoff was 120 Hz (DC amplifiers; Micromed). Right, left, supra-, and infra-orbital electrodes monitored horizontal and vertical electro-oculograms. Synchronized with stimulus onset, Transistor-Transistor Logic (TTL)-pulses were used off-line to segment the continuous EEG data into epochs. Reaction times and patients’ performances were recorded. Waveform Analysis Data were analyzed with NeuroScan software (NeuroScan, Herndon, VA). After artefact removal and off-line correction of ocular artefacts (threshold reduction algorithm), data from trials with correct answers were averaged according to the task condition (control, 1-back, 2-back). All correct answers were combined (response to a target and no response in the absence of a target), because working memory is required in both situations and because prior research has shown that there is no difference between response and no response conditions during the time window used to study the ERPs (13). Combined ERPs were then averaged over a time window of 700 ms with a 200-ms prestimulus onset and were band-pass filtered between 0.33 Hz and 30 Hz, 24 dB/octave for low-pass filter. Latencies of the ERP components were measured at the time of the peak maximum. Data Analysis The visual P100 component latency was measured at the O1, Oz, and O2 electrode sites. The maximal peak latencies of the N160, P200, and N200 components occurred at fron-

662

MISSONNIER ET AL.

Figure 1. (A) Schematic representation of the control task (first row) and the two working memory tasks (second and third rows). In the control task, the patient’s task was to press a pushbutton as soon as a nonletter patch appeared. During 1-back and 2-back tasks, the patient was asked to press the button only when the presented letter was identical to the one that had been shown one trial before (1-back task) or two trials before (2-back task) in the sequence (response trials). Note that both response and no-response trials require working memory. (B) Sequence timing: each stimulus was presented for 0.5 seconds, followed by a delay interval (4.5 seconds) during which the patient fixed a small point.

tal, midline, and posterior (Fz, Cz, and Pz) electrode sites in the two groups, and these sites were thus selected for their measurement (Figure 2). To isolate the PNwm component, we performed an ERP waveform-subtraction analysis between the n-back tasks and the control task. Electrodes F3, Fz, F4, C3, Cz, C4, P3, Pz, and P4 were selected for analysis in keeping with the anteroposterior distribution of the PNwm component. The intersection between the detection waveform and either the 1-back or 2-back waveforms in the 175- to 317-ms timerange window was used to define the temporal limits for the measurement of density (area in lV2). These temporal limits were not identical in both tasks. To correct for this difference, the density of the PNwm component was divided by the measurement time range defined by the temporal limits for each task in all patients; it was then normalized through logarithmic transformation.

Statistical Comparisons Between Progressive and Stable MCI Neuropsychological values at baseline and upon follow-up were compared in progressive MCI (PMCI) and stable MCI (SMCI) groups using the Wilcoxon nonparametric test. Differences in reaction times and performances between SMCI and PMCI groups were assessed using the Mann– Whitney nonparametric U test. Univariate regression analysis was also performed to examine whether neuropsychological measures at baseline predict further cognitive decline after 1 year in the present MCI cohort. Analyses of variance (ANOVAs) were used to compare EEG latency and densities among the different task conditions (control, 1-back, and 2back). The significance values were determined using the Greenhouse–Geisser correction. The Scheffe´ test was used

Figure 2. Grand average waveforms, from a midline electrode site, for the control (dotted lines), 1-back (dashed lines), and 2-back (solid lines) tasks, averaged over response and no-response trials for stable mild cognitive impairment (SMCI) (A) and progressive mild cognitive impairment (PMCI) (B) patients.

for post hoc analyses. Two-tailed differences of p , .05 were considered significant. RESULTS Upon follow-up examination which took place 1 year after baseline (range 12.0 6 0.4 months), 13 (54%) of the original MCI patients demonstrated significant cognitive decline and constituted the PMCI group. The remaining 11 patients (46%) showed no change in cognitive function and were included in the SMCI group. PMCI patients showed deterioration in one or more of the following domains: executive function, attention, and memory. No patients in the SMCI group developed AD 1 year later (Table 1). There was no statistical difference in baseline scores for any of the neuropsychological tests between SMCI and PMCI patients. This finding was confirmed by the univariate regression analysis, which showed that neuropsychological parameters at baseline did not predict cognitive deterioration 1 year later. Age, sex distribution, educational level, func-

EVENT-RELATED POTENTIAL CHANGES PREDICT RAPID DECLINE IN MCI

663

Table 1. Neuropsychological Variables of SMCI and PMCI Patients at Inclusion and After 1 Year SMCI Patients (N ¼ 11)

PMCI Patients (N ¼ 13)

Inclusion

Follow-Up

p Value

Inclusion

Follow-Up

p Value

26.64 6 1.63 21.00 6 0.46

25.45 6 0.39 20.55 6 0.46

NS NS

25.85 6 1.91 20.85 6 0.43

23.46 6 0.56 19.38 6 0.52

NS NS

16.55 6 0.75 241.82 6 3.31

16.91 6 0.87 181.27 6 2.82

NS NS

16.00 6 0.65 242.31 6 3.07

10.92 6 0.65 296.31 6 3.28

.01 NS

19.36 6 0.31

18.45 6 0.32

NS

18.85 6 0.31

17.38 6 0.41

NS

4.09 6 0.29 3.36 6 0.21 10.00 10.09 6 0.29

4.36 6 0.25 3.18 6 0.23 10.00 9.91 6 0.39

NS NS NS NS

4.69 3.38 9.92 9.62

6 6 6 6

0.19 0.22 0.14 0.32

4.54 6 0.20 2.92 6 0.22 10.00 9.08 6 0.38

NS NS NS NS

3.82 6 0.19

4.00 6 0.00

NS

3.61 6 0.20

3.85 6 0.21

NS

9.55 6 0.45 77.00 6 1.60

9.64 6 0.41 81.64 6 1.59

NS NS

7.85 6 0.43 102.38 6 1.79

7.38 6 0.41 135.38 6 2.34

NS .01

5.00 6 0.33 3.73 6 0.42

4.82 6 0.35 3.73 6 0.36

NS NS

4.54 6 0.34 3.69 6 0.30

4.54 6 0.33 3.38 6 0.32

NS NS

FR tot CR tot Rec LTR LTCR

18.09 6 0.60 24.27 6 0.72 16.00 6.82 6 0.48 7.82 6 0.49

19.64 6 0.62 23.64 6 0.48 16.00 6.27 6 0.56 8.36 6 0.53

NS NS NS NS NS

17.00 25.85 15.69 5.92 8.61

6 6 6 6 6

0.52 0.68 0.24 0.43 0.38

.02 NS NS NS NS

Shapes test

26.55 6 0.82

27.91 6 0.78

NS

18.23 6 0.57

17.38 6 0.77

NS

Variables Global assessment MMSE DRS Executive functions Verbal fluency Trail Making Test B, s Language abilities Boston Naming Test Standardized praxis Ideomotor praxis Reflexive praxis Constructional praxis CERAD Gnosis Ghent Overlapping Figures Test Attention WAIS-R; Code Trail Making Test A, s Memory tests Direct span forward Corsi Blocks Buschke Double Memory Test (16 items)

6 6 6 6 6

0.51 0.46 0.22 0.42 0.40

13.31 23.69 15.69 4.15 7.77

Notes: Data are presented as mean 6 standard error of the mean. Note the progressive decline in the PMCI group as compared to the preserved cognitive abilities in the SMCI group. MMSE ¼ Mini-Mental State Examination; DRS ¼ Mattis Dementia Rating Scale; CERAD ¼ Consortium to Establish a Registry for Alzheimer’s disease; WAIS-R ¼ Wechsler Adult Intelligence Scale, revised; FR tot ¼ free recall total; CR tot ¼ cued recall total; Rec ¼ recognition; LTR ¼ long-term recall; LTCR ¼ long term cued recall. SMCI ¼ stable mild cognitive impairment; PMCI ¼ progressive mild cognitive impairment; NS ¼ not significant.

tional status, and assessments of anxiety and depression were also similar in the SMCI and PMCI groups (Table 2). During the n-back testing, reaction times increased as the task became more difficult ( p , .01). However, this increase was similar in both PMCI and SMCI patients (Table 3). All MCI patients performed well in both 1-back and 2-back tasks and responded correctly in more than 80% of the trials. Performance was related to task difficulty ( p , .001); this effect was restricted to the 2-back condition (Scheffe´ p , .001). More importantly, performance of PMCI and SMCI patients was not significantly different (Table 3).

Sensory ERPs Latencies of sensory P100 and N160 ERPs were no different in the PMCI and SMCI groups for any of the tasks (116.4 ms vs 115.4 ms and 162.1 ms vs 167.5 ms, respectively) (Figure 2). The PNwm Component Figure 3 shows the PNwm component, obtained by

subtracting ERP waveforms for the detection task from the 2-back (Figure 3A) and 1-back (Figure 3B) tasks. In the SMCI group, a PNwm component is clearly visible. Its density was significantly larger than baseline for both the 1- and 2-back tasks ( p , .01 and p , .001, respectively). In the PMCI group, no significant PNwm component could be identified. No electrode site effect or interactions between group and electrode sites were observed (Figure 3).

DISCUSSION Strengths of the present study include the large series of longitudinally studied MCI patients, confirmation of the clinical diagnosis of MCI by two independent clinicians blind to the study purposes, as well as control for medications which influence ERP. The present data indicate that the absence of an early ERP component (i.e., PNwm component) during working memory activation is associated with rapid cognitive deterioration in MCI. Neither neuropsychological measures nor n-back performances at baseline were related to the clinical outcome, suggesting that the

MISSONNIER ET AL.

664

Table 2. Demographic and Clinical Characteristics of SMCI and PMCI Patients at Inclusion Variable

SMCI Patients (N ¼ 11)

PMCI Patients (N ¼ 13)

Age, y Sex, f/m Educational level* IADLy HADz (anxiety) HADz (depression)

84.91 6 8/3 1.64 6 12.91 6 5.00 6 4.45 6

83.54 6 4.31 11/2 1.31 6 0.48 12.77 6 4.28 5.46 6 2.60 3.54 6 2.37

4.78 0.81 4.46 2.53 2.16

Notes: Data are presented as mean 6 standard deviation. Educational levels from 1 to 3. y Lawton’s Instrumental Activities of Daily Living (IADL) score. z Hospital Anxiety and Depression (HAD) scales, cutoff is 8 for both anxiety and depression. The two diagnostic groups were comparable with respect to the above mentioned variables. SMCI ¼ stable mild cognitive impairment; PMCI ¼ progressive mild cognitive impairment. *

observed ERP change may be an independent predictor of MCI decline. Importantly, all PMCI patients displayed high performances in n-back task at baseline and only few additional specific cognitive deficits on follow-up (with preserved global scores such as MMSE and Mattis DRS), indicating that the inability to visualize a PNwm component in this group represents an early EEG phenomenon which may predict even subtle decrease in the cognitive abilities of MCI patients. In this context, the successful completion of the task in PMCI patients may be due to the compensatory activation of additional cortical areas. This possibility is supported by a recent functional imaging study (32) showing that high- but not low-performing elderly individuals may compensate for the neural decline of parietal areas during working memory tasks, through the activation of distinct neurocognitive networks within prefrontal and limbic areas. The only other longitudinal ERP study focusing on memory in MCI reported that a significant reduction of late potential repetition (P600) in a word repetition paradigm may help to distinguish PMCI from SMCI in a 2-year follow-up (11). This study included only 14 MCI patients and used post hoc determined cutoff values of the P600 to separate converters from nonconverters. More importantly, it focused on verbal episodic memory processes thought to reflect mesial temporal lobe function. However, several lines of evidence suggest that metabolic or electrophysiological changes within the mesial temporal lobe may be not the best candidates for the prediction of MCI conversion to AD. In fact, this brain region is a site of very early neurofibrillary tangle (NFT) formation which was quite comparable in patients with no cognitive impairment and in patients with questionable dementia [CDR score of 0.5, (33,34)]. In addition, MCI patients display extensive (up to 50%) neuron loss in layer II of entorhinal cortex, and NFT counts in this area are elevated in these patients even in comparison to those in AD patients with several years of evolution (33,35). In contrast, the progression of NFT within neocortical association areas is clearly associated with clinical signs of dementia (for review, see 9,10,34,35). These neuropathological observations are also confirmed by a recent prospective single photon emission tomography (SPECT) study

Table 3. Reaction Times (ms) and Performances (%) With Standard Errors (6SE ) for the Control, 1-Back, and 2-Back Tasks in SMCI (N ¼ 11) and PMCI (N ¼ 13) Patients Reaction Times (Mean 6 SE) Tasks

SMCI

PMCI

p Value

Performances (Mean 6 SE) SMCI

PMCI

p Value

Control 977.67 6 4.13 1081.22 6 4.06 NS 98.89 6 0.44 97.78 6 0.44 NS 1-Back 1069.03 6 4.56 1140.21 6 4.33 NS 97.78 6 0.44 95.98 6 0.60 NS 2-Back 1294.79 6 5.38 1431.29 6 5.77 NS 86.87 6 0.51 83.08 6 0.73 NS

Notes: There was no statistically significant difference in reaction times and performances between the two diagnostic groups. See text for details. SMCI ¼ Stable mild cognitive impairment; PMCI ¼ progressive mild cognitive impairment; NS ¼ not significant.

which showed that reduction of regional cerebral blood flow in the medial temporal lobe has no value as predictor because it also occurs in SMCI (36). In contrast, two recent functional imaging studies proposed that decreased metabolism in parietal areas is a sensitive early marker of MCI progression toward AD (37,38). In line with these observations and our previous work (13), the prediction of MCI decline based on the early impairment of neural generators within the parietal

Figure 3. Frontal (Fz), central (Cz), and parietal (Pz) grand average waveforms of the positive-negative working memory (PNwm) component, representing event-related potential (ERP) differences between 2-back task and control task (A) and between 1-back task and control task (B), respectively, for stable mild cognitive impairment (SMCI) (dashed lines) and progressive mild cognitive impairment (PMCI) (solid lines) patients. Response and no-response conditions were averaged. No PNwm component was observed in PMCI patients.

EVENT-RELATED POTENTIAL CHANGES PREDICT RAPID DECLINE IN MCI

cortex reported here further supports a key role for this area in MCI transition to AD. Numerous structural and functional neuroimaging reports addressed the issue of a predictive marker of further cognitive deterioration in MCI with conflicting data. Early longitudinal and cross-sectional MRI studies supported the value of hippocampal (39–44) and entorhinal cortex (40,45) atrophy in MCI with regard to subsequent cognitive decline. However, volumetric alterations in hippocampus and entorhinal cortex reflect the presence of an already substantial loss of pyramidal neurons (46) and cannot thus be considered as early predictors of the degenerative process in brain aging. Because functional alterations are thought to occur prior to structural deficits in the course of MCI, recent studies have attempted to identify an early functional marker of MCI deterioration. Decreased parietal and increased prefrontal regional cerebral blood flow (37), posterior cingulate (38), prefrontal (36), and right temporoparietal cortex hypometabolism (47,48) have all been associated with MCI conversion to AD. These variable results have led to increasing scepticism about the validity of positron emission tomography and SPECT in the prediction of AD. Furthermore, poor time resolution, low practicability, and cost-effectiveness issues limit the routine use of positron emission tomography and SPECT in clinical settings. In contrast, the PNwm component may provide an easily applicable qualitative predictive EEG marker of rapid cognitive deterioration in MCI. Two main limitations should, however, be considered when interpreting these data. First, no distinction was made between different MCI subgroups (i.e., vascular vs AD MCI) in the present series. Second, as is frequently the case in imaging studies (49), the definition of an individual risk is still not possible on the basis of the present activation paradigm due to the relatively low signal-to-noise ratio. Additional ERP studies in large, prospectively studied MCI subgroups with a longer follow-up period are needed to determine how future paradigm changes including an increased number of n-back trials and improved noise-filtering methodology could lead to predictive models of MCI conversion to PMCI and ultimately dementia that could be applied to a single individual.

ACKNOWLEDGMENTS This project was funded by the Swiss National Foundation for Scientific Research (grant 3100-59110.99 to PG), the Research and Development University Hospitals of Geneva (grant 03-I-15 to VI), and the Je´roˆme Tissie`res Foundation (to PG). Address correspondence to Dr. Gabriel Gold, Department of Geriatrics, University of Geneva Hospitals, ch. Pont-Bochet 3, 1226 Thoˆnex–Geneva, Switzerland. E-mail: [email protected]

REFERENCES 1. Petersen RC, Doody R, Kurz A, et al. Current concepts in mild cognitive impairment. Arch Neurol. 2001;58:1985–1992. 2. Jelic V, Shigeta M, Julin P, Almkvist O, Winblad B, Lo W. Quantitative electroencephalography power and coherence in Alzheimer’s disease and mild cognitive impairment. Dementia. 1996;7:314–323. 3. Grunwald M, Busse F, Hensel A, et al. Correlation between cortical theta activity and hippocampal volumes in health, mild cognitive impairment, and mild dementia. J Clin Neurophysiol. 2001;18:178–184.

665

4. Grunwald M, Busse F, Hensel A, et al. Theta-power differences in patients with mild cognitive impairment under rest condition and during haptic tasks. Alzheimer Dis Assoc Disord. 2002;16:40–48. 5. Huang C, Wahlund L, Dierks T, et al. Discrimination of Alzheimer’s disease and mild cognitive impairment by equivalent EEG sources: a cross-sectional and longitudinal study. Clin Neurophysiol. 2000; 111:1971–1967. 6. Frodl T, Hampel H, Juckel G, et al. Value of event-related P300 subcomponents in the clinical diagnosis of mild cognitive impairment and Alzheimer’s disease. Psychophysiology. 2002;39:175–181. 7. Stam C, van der Made Y, Pijnenburg Y, et al. EEG synchronization in mild cognitive impairment and Alzheimer’s disease. Acta Neurol Scand. 2003;108:90–96. 8. Jelic V JS, Almkvist O, Shigeta M, et al. Quantitative electroencephalography in mild cognitive impairment: longitudinal changes and possible prediction of Alzheimer’s disease. Neurobiol Aging. 2000; 21:533–540. 9. Giannakopoulos P, Hof PR, Michel JP, et al. Cerebral cortex pathology in aging and Alzheimer’s disease: a quantitative survey of large hospital-based geriatric and psychiatric cohorts. Brain Res Brain Res Rev. 1997;25:217–245. 10. Morrison JH, Hof PR. Life and death of neurons in the aging brain. Science. 1997;278:212–219. 11. Olichney JM, Morris MS, Ochoa C, et al. Abnormal verbal event related potentials in mild cognitive impairment and incipient Alzheimer’s disease. J Neurol Neurosurg Psychiatry. 2002;73:377–384. 12. McEvoy LK, Smith ME, Gevins A. Dynamic cortical networks of verbal and spatial working memory: effects of memory load and task practice. Cereb Cortex. 1998;8:563–574. 13. Missonnier P, Leonards U, Gold G, et al. A new electrophysiological index of working memory load in humans. NeuroReport. 2003;14: 1451–1455. 14. 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;12:189–198. 15. Zigmond AS, Snaith RP. The hospital anxiety and depression scale. Acta Psychiatr Scand. 1983;67:361–370. 16. Berberger-Gateau P, Chaslerie A, Darigues JF, et al. Health measures correlates in a French elderly community population: the PAQUID study. J Gerontol. 1992;47:S88–S95. 17. Wechsler D. Adult Intelligence Scale, Revised (WAIS-R). 1981. 18. Gardner R Jr, Oliver-Munoz S, Fisher L, et al. Mattis Dementia Rating Scale: internal reliability study using a diffusely impaired population. J Clin Neuropsychol. 1981;3:271–275. 19. Wechsler D. Manual for the Wechsler Adult Intelligence Scale. New York, NY: Psychological Corporation; 1955. 20. Milner B. Interhemispheric differences in the localization of psychological processes in man. Br Med Bull. 1971;27:272–277. 21. Buschke H, Sliwinski MJ, Kuslansky G, Lipton RB. Diagnosis of early dementia by the Double Memory Test: encoding specificity improves diagnostic sensitivity and specificity. Neurology. 1997;48:989–997. 22. Baddeley A, Emslie H, Nimmo-Smith L. The Doors and People Test: A test of visual and verbal recall and recognition. Bury St Edmonds, Thames Valley Test; 1994;20. 23. Butters N, Granholm E, Salmon DP, et al. Episodic and semantic memory: a comparison of amnesic and demented patients. J Clin Exp Neuropsychol. 1987;9:479–497. 24. Reitan RM. Validity of the Trail Making Test as an indicator of organic brain damage. Percept Mot Skills. 1958;8:271–276. 25. Kaplan EF, Goodglass H, Weintraub S. The Boston Naming Test. Philadelphia, PA: Lea and Febiger; 1983. 26. Schnider A, Hanlon RE, Alexander DN, et al. Ideomotor apraxia: behavioral dimensions and neuroanatomical basis. Brain Lang. 1997; 58:125–136. 27. Poeck K. Clues to the nature of disruption to limb praxis. In: Roy EA, ed. Neuropsychological Studies of Apraxia and related Disorders. Amsterdam, The Netherlands: Elsevier; 1985:99–109. 28. Welsh KA, Butters N, Mohs RC, et al. The Consortium to Establish a Registry for Alzheimer’s Disease (CERAD). Part V. A normative study of the neuropsychological battery. Neurology. 1994;44:609–614. 29. Ghent L. Perception of overlapping and embedded figures by children of different ages. Am J Psychol. 1956;69:575–581.

666

MISSONNIER ET AL.

30. Hughes CP, Berg L, Danziger WL, et al. A new clinical scale for the staging of dementia. Br J Psychiatry. 1982;140:566–572. 31. Homan R, Herman J, Purdy P. Cerebral location of international 10–20 system electrode placement. Electroencephalogr Clin Neurophysiol. 1987;66:376–382. 32. Cabeza R, Anderson ND, Locantore JK, et al. Aging gracefully: compensatory brain activity in high-performing older adults. Neuroimage. 2002;17:1394–1402. 33. Gold G, Ko¨vari E, Corte G, et al. Clinical validity of Aß-protein deposition staging in brain aging and Alzheimer’s disease. J Neuropathol Exp Neurol. 2001;60:946–952. 34. Gold G, Bouras C, Ko¨vari E, et al. Clinical validity of Braak neuropathological staging in the oldest-old. Acta Neuropathologica. 2000; 99:579–582. 35. Gomez-Isla T, Price J, McKeel DJ, et al. Profound loss of layer II entorhinal cortex neurons occurs in very mild Alzheimer’s disease. J Neurosci. 1996;16:4491–4500. 36. Encinas M, De Juan R, Marcos A, et al. Regional cerebral blood flow assessed with (99m)Tc-ECD SPET as a marker of progression of mild cognitive impairment to Alzheimer’s disease. Eur J Nucl Med Mol Imaging. 2003;30:1473–1480. 37. Huang C, Wahlund L, Almkvist O, et al. Voxel- and VOI-based analysis of SPECT CBF in relation to clinical and psychological heterogeneity of mild cognitive impairment. Neuroimage. 2003;19: 1137–1144. 38. Drzezga A, Lautenschlager N, Siebner H, et al. Cerebral metabolic changes accompanying conversion of mild cognitive impairment into Alzheimer’s disease: a PET follow-up study. Eur J Nucl Med Mol Imaging. 2003;30:1104–1113. 39. Wolf H, Grunwald M, Ecke G, et al. The prognosis of mild cognitive impairment in the elderly. J Neural Transm Suppl. 1998;54:31–50.

40. Visser P, Scheltens P, Verhey F, et al. Medial temporal lobe atrophy and memory dysfunction as predictors for dementia in subjects with mild cognitive impairment. J Neurol. 1999;246:477–485. 41. Marquis S, Moore M, Howieson D, et al. Independent predictors of cognitive decline in healthy elderly persons. Arch Neurol. 2002;59:601–606. 42. Kaye J, Swihart T, Howieson D, et al. Volume loss of the hippocampus and temporal lobe in healthy elderly persons destined to develop dementia. Neurology. 1997;48:1297–1304. 43. Jack CR Jr, Petersen RC, Xu YC, et al. Prediction of AD with MRIbased hippocampal volume in mild cognitive impairment. Neurology. 1999;52:1397–1403. 44. de Leon M, Golomb J, George A, et al. The radiologic prediction of Alzheimer disease: the atrophic hippocampal formation. Am J Neuroradiol. 1993;14:897–906. 45. Dickerson B, Goncharova I, Sullivan M, et al. MRI-derived entorhinal and hippocampal atrophy in incipient and very mild Alzheimer’s disease. Neurobiol Aging. 2001;22:747–754. 46. Kordower J, Chu Y, Stebbins G, et al. Loss and atrophy of layer II entorhinal cortex neurons in elderly people with mild cognitive impairment. Ann Neurol. 2001;49:202–213. 47. Chetelat G, Desgranges B, de la Sayette V, et al. Dissociating atrophy and hypometabolism impact on episodic memory in mild cognitive impairment. Brain. 2003;126:1955–1967. 48. Chetelat G, Desgranges B, De La Sayette V, et al. Mild cognitive impairment: can FDG-PET predict who is to rapidly convert to Alzheimer’s disease? Neurology. 2003;60:1374–1377. 49. Wolf H, Jelic V, Gertz H, et al. A critical discussion of the role of neuroimaging in mild cognitive impairment. Acta Neurol Scand Supp. 2003;179:52–76. Received January 29, 2004 Accepted February 16, 2004 Decision Editor: John E. Morley, MB, BCh