[Skip to Content]
Access to paid content on this site is currently suspended due to excessive activity being detected from your IP address 54.205.176.107. Please contact the publisher to request reinstatement.
Sign In
Individual Sign In
Create an Account
Institutional Sign In
OpenAthens Shibboleth
[Skip to Content Landing]
Download PDF
Figure 1.
High vs low log relative beta on electroencephalography (EEG), dichotomized at the medium of the distribution, affecting survival of patients with Alzheimer disease using the Kaplan-Meier method (log rank test, P<.001).

High vs low log relative beta on electroencephalography (EEG), dichotomized at the medium of the distribution, affecting survival of patients with Alzheimer disease using the Kaplan-Meier method (log rank test, P<.001).

Figure 2.
High vs low log relative alpha on electroencephalography (EEG), dichotomized at the medium of the distribution, affecting survival of patients with Alzheimer disease using the Kaplan-Meier method (log rank test, P<.05).

High vs low log relative alpha on electroencephalography (EEG), dichotomized at the medium of the distribution, affecting survival of patients with Alzheimer disease using the Kaplan-Meier method (log rank test, P<.05).

Table 1. 
Characteristics of 101 Patients With Alzheimer Disease*
Characteristics of 101 Patients With Alzheimer Disease*
Table 2. 
Quantitative Spectral Electroencephalogram (EEG) Relative Power Values of 101 Patients With Probable Alzheimer Disease
Quantitative Spectral Electroencephalogram (EEG) Relative Power Values of 101 Patients With Probable Alzheimer Disease
Table 3. 
Proportional Hazards Regression Analysis for Log Relative Percentage Electroencephalogram (EEG) Values at First Time of Diagnosis and Survival in 101 Patients With Probable Alzheimer Disease
Proportional Hazards Regression Analysis for Log Relative Percentage Electroencephalogram (EEG) Values at First Time of Diagnosis and Survival in 101 Patients With Probable Alzheimer Disease
1.
Coben  LADanziger  WStorandt  M A longitudinal EEG study of mild senile dementia of Alzheimer type: changes at 1 year and at 2.5 years. Electroencephalogr Clin Neurophysiol. 1985;61101- 112Article
2.
Penttila  MPartanen  JVSoininen  HRiekkinen  PJ Quantitative analysis of occipital EEG in different stages of Alzheimer's disease. Electroencephalogr Clin Neurophysiol. 1985;601- 6Article
3.
Soininen  HPartanen  JLaulumaa  VPaakkonen  AHelkala  E-LRiekkinen  PJ Serial EEG in Alzheimer's disease: 3 year follow-up and clinical outcome. Electroencephalogr Clin Neurophysiol. 1991;79342- 348Article
4.
Schreiter-Gasser  UGasser  TZiegler  P Quantitative EEG analysis in early onset Alzheimer's disease: correlations with severity, clinical characteristics, visual EEG and CCT. Electroencephalogr Clin Neurophysiol. 1994;90267- 272Article
5.
Prichep  LSJohn  ERFerris  SH  et al.  Quantitative EEG correlates of cognitive deterioration in the elderly. Neurobiol Aging. 1994;1585- 90Article
6.
Claus  JJKwa  VIHTeunisse  S  et al.  Slowing on quantitative spectral EEG is a marker for rate of subsequent cognitive and functional decline in early Alzheimer's disease. Alzheimer Dis Assoc Disord. In press.
7.
Kuskowski  MAMortimer  JAMorley  GKMalone  SMOkaya  AJ Rate of cognitive decline in Alzheimer's disease is associated with EEG alpha power. Biol Psychiatry. 1993;33659- 662Article
8.
Lopez  OLBrenner  RPBecker  JTUlrich  RFBoller  FDeKosky  ST EEG spectral abnormalities and psychosis as predictors of cognitive and functional decline in probable Alzheimer's disease. Neurology. 1997;481521- 1525Article
9.
Rodriguez  GNobili  FArrigo  A  et al.  Prognostic significance of quantitative electroencephalography in Alzheimer patients: preliminary observations. Electroencephalogr Clin Neurophysiol. 1996;99123- 128Article
10.
Kaszniak  AWFox  JGandell  JLGarron  DCHuckman  MSRamsey  RG Predictors of mortality in presenile and senile dementia. Ann Neurol. 1978;3246- 252Article
11.
Helkala  E-LLaulumaa  VSoininen  HPartanen  JRiekkinen  PJ Different patterns of cognitive decline related to normal or deteriorating EEG in a 3-year follow-up study of patients with Alzheimer's disease. Neurology. 1991;41528- 532Article
12.
Saletu  BAnderer  PPaulus  E  et al.  EEG brain mapping in diagnostic and therapeutic assessment of dementia. Alzheimer Dis Assoc Disord. 1991;5(suppl 1)S57- S75Article
13.
Lopez  OLBrenner  RPBecker  JTJungreis  CRezek  DDeKosky  ST Electroencephalographic correlates of periventricular white matter lesions in probable Alzheimer's disease. Dementia. 1995;6343- 347
14.
McKhann  GDrachman  DFolstein  MKatzman  RPrice  DStadlan  EM Clinical diagnosis of Alzheimer's disease: report of the NINCDS-ADRDA work group under the auspices of Department of Health and Human Services task force on Alzheimer's disease. Neurology. 1984;34939- 944Article
15.
Walstra  GJMTeunisse  Svan Gool  WAvan Crevel  H Reversible dementia in elderly patients referred to a memory clinic. J Neurol. 1997;24417- 22Article
16.
Derix  MMHofstede  ABTeunisse  S  et al.  CAMDEX-N: the Dutch version of the Cambridge Examination for Mental Disorders of the Elderly with automatic data processing. Tijdschr Gerontol Geriatr. 1991;22143- 150
17.
Roth  MHuppert  FATym  EMountjoy  CQ CAMDEX, the Cambridge Examination for Mental Disorders of the Elderly.  Cambridge, England Cambridge University Press1988;
18.
Teunisse  SBollen  AEvan Gool  WAWalstra  GJM Dementia and subnormal levels of vitamin B12: effects of replacement therapy on severity of dementia. J Neurol. 1996;243522- 529Article
19.
John  ERPrichep  LSFridman  JEaston  P Neurometrics: computer-assisted differential diagnosis of brain dysfunctions. Science. 1988;239162- 169Article
20.
John  ERAnn  HPrichep  LSTrepetin  MBrown  PKaye  H Development equations for the electroencephalogram. Science. 1980;2101255- 1258Article
21.
Hijdra  AVerbeeten  BJ Leukoaraiosis and ventricular enlargement in patients with ischemic stroke. Stroke. 1991;22447- 450Article
22.
Hijdra  AVerbeeten  BJVerhulst  JAPM Relation of leukoaraiosis to lesion type in stroke patients. Stroke. 1990;21890- 894Article
23.
Waldemar  GBruhn  PKristensen  MJohnson  APaulson  OBLassen  NA Heterogeneity of neocortical cerebral blood flow deficits in dementia of the Alzheimer type: a [99mTc]-d,l-HMPAO SPECT study. J Neurol Neurosurg Psychiatry. 1994;57285- 295Article
24.
Jacoby  RJLevy  RDawson  JM Computed tomography in the elderly, I: the normal population. Br J Psychiatry. 1980;136249- 255Article
25.
Collet  D Modelling Survival Data in Medical Research.  London, England Chapman & Hall1994;
26.
Folstein  MFFolstein  SEMcHugh  PR "Mini–mental state": a practical method for grading the cognitive state of patients for the clinician. J Psychiatr Res. 1975;12189- 198Article
27.
McKeith  IG Lewy body disease. Curr Opin Psychiatry. 1995;8252- 257Article
28.
McKeith  IGGalasko  DKosaka  K  et al.  Consensus guidelines for the clinical and pathological diagnosis of dementia with Lewy bodies (DLB): report of the consortium on DLB international workshop. Neurology. 1996;471113- 1124Article
29.
Petit  DMontplaisir  JLorrain  DGauthier  S Spectral analysis of the rapid eye movement sleep electroencephalogram in right and left temporal regions: a biological marker of Alzheimer's disease. Ann Neurol. 1992;32172- 176Article
30.
Soininen  HPartanen  VJLaulumaa  VHelkala  E-LLaakso  MRiekkinen  PJ Longitudinal EEG spectral analysis in early stage of Alzheimer's disease. Electroencephalogr Clin Neurophysiol. 1989;72290- 297Article
31.
Steriade  MGloor  PLlinas  RRLopez da Silva  FHMesulam  M-M Report of IFCN Committee on Basic Mechanisms: basic mechanisms of cerebral rhythmic activities. Electroencephalogr Clin Neurophysiol. 1990;76481- 508Article
32.
Steriade  M Cellular substrates of brain rhythms. Niedermeyer  ELopes da Silva  FedsElectroencephalography Basic Principles, Clinical Applications, and Related Fields. Baltimore, Md Williams & Wilkins1993;27- 62
33.
Heyman  APeterson  BFillenbaum  GPieper  C The Consortium to Establish a Registry for Alzheimer's Disease (CERAD), part XIV: demographic and clinical predictors of survival in patients with Alzheimer's disease. Neurology. 1996;46656- 660Article
34.
Burns  ALewis  GJacoby  R  et al.  Factors affecting survival in Alzheimer's disease. Psychol Med. 1991;21363- 370Article
35.
Jagger  CClarke  MStone  A Predictors of survival with Alzheimer's disease: a community-based study. Psychol Med. 1995;25171- 177Article
36.
Walsh  JSWelch  GLarson  EB Survival of outpatients with Alzheimer-type dementia. Ann Intern Med. 1990;113429- 434Article
37.
Leung  LS Fast (beta) rhythms in the hippocampus: a review. Hippocampus. 1992;293- 98Article
38.
Riekkinen  PJRiekkinen  MSirvio  JMiettinen  RKoudstaal  PJ Loss of cholinergic neurons in the nucleus basalis induces neocortical electroencephalographic and passive avoidance deficits. Neuroscience. 1992;47823- 831Article
39.
Boeijinga  PHLopez da Silva  FH Differential distribution of beta and theta EEG activity in the entorhinal cortex of the cat. Brain Res. 1988;448272- 286Article
40.
Gomez-Isla  TPrice  JLMcKeel  DW  JrMorris  JCGrowden  JHHyman  BT Profound loss of layer II entorhinal cortex neurons occurs in very mild Alzheimer's disease. J Neurosci. 1996;164491- 4500
41.
Heinonen  OSoininen  HSorvari  H  et al.  Loss of synaptophysin-like immunoreactivity in the hippocampal formation is an early phenomenon in Alzheimer's disease. Neuroscience. 1995;64375- 384Article
42.
Williamson  PCMerskey  HMorrison  S  et al.  Quantitative electroencephalographic correlates of cognitive decline in normal elderly subjects. Arch Neurol. 1990;471185- 1188Article
43.
Hofman  AOtt  ABreteler  MMB  et al.  Atherosclerosis, apolipoprotein E, and prevalence of dementia and Alzheimer's disease in the Rotterdam Study. Lancet. 1997;349151- 154Article
44.
Jelic  VJulin  PShigeta  M  et al.  Apolipoprotein E ϵ4 allele decreases functional connectivity in Alzheimer's disease as measured by EEG coherence. J Neurol Neurosurg Psychiatry. 1997;6359- 65Article
45.
Tanna  NKKohn  MIHorwich  DN  et al.  Analysis of brain and cerebrospinal fluid volumes with MR imaging: impact on PET data correction for atrophy, part II: aging and Alzheimer dementia. Radiology. 1991;178123- 130Article
46.
Stern  YTang  M-XAlbert  MS  et al.  Predicting time to nursing home care and death in individuals with Alzheimer's disease. JAMA. 1997;277806- 812Article
47.
Stern  YBrandt  JAlbert  M  et al.  The absence of an apolipoprotein epsilon4 allele is associated with a more aggressive form of Alzheimer's disease. Ann Neurol. 1997;41615- 620Article
Original Contribution
August 1998

Quantitative Spectral Electroencephalography in Predicting Survival in Patients With Early Alzheimer Disease

Author Affiliations

From the Department of Neurology (Drs Claus, Walstra, Hijdra, and van Gool), Division of Clinical Neurophysiology (Drs Claus and Ongerboer de Visser), and Department of Radiology (Dr Verbeeten), Academic Medical Center, Amsterdam, the Netherlands.

Arch Neurol. 1998;55(8):1105-1111. doi:10.1001/archneur.55.8.1105
Abstract

Objective  To determine whether measures of quantitative spectral electroencephalography (EEG) can predict survival in patients with early Alzheimer disease.

Design  Prospective cohort study; median duration of follow-up was 4.4 years in survivors and 2.6 years in nonsurvivors. Cox proportional hazards models, with adjustment for age and sex were used to estimate relationships between EEG measures and survival. Log relative percentage values of EEG bands were used as predictors.

Setting  Outpatient university memory clinic.

Participants  One hundred one consecutively referred patients with early probable Alzheimer disease according to National Institute of Neurological and Communicative Disorders and Stroke–Alzheimer's Disease and Related Disorders Association criteria were studied with EEG at the time of diagnosis. The mean age of the patients was 79.2 years, which was higher than in previous EEG studies.

Main Outcome Measure  Mortality.

Results  Fifty-one patients (50.5%) died during follow-up, with a median survival time in all patients of 4.1 years. The following EEG variables were significantly associated with increased risk of mortality: from parieto-occipital leads, higher theta (hazard ratio, 2.05; 95% confidence interval, 1.15-3.66; P<.05), lower alpha (hazard ratio, 0.43; 95% confidence interval, 0.25-0.76; P<.01), and lower beta (hazard ratio, 0.38; 95% confidence interval, 0.22-0.68; P<.001) activity; and from frontocentral leads, higher theta activity (hazard ratio, 2.07; 95% confidence interval, 1.17-3.66; P<.05). Stepwise Cox regression analysis showed that loss of parieto-occipital beta (P<.01) and alpha (P<.05) power were independent and significant predictors of mortality. Both beta (12.6-35.4 Hz) and alpha (7.5-12.5 Hz) activity remained significantly associated with mortality after adjustment for education, dementia severity, symptom duration, level of cognitive function, presence of extrapyramidal symptoms or hallucinations, presence of vascular risk factors, and presence of leukoaraiosis or local cortical atrophy.

Conclusions  Decreases of beta and alpha activity on quantitative spectral EEG are independent predictors of mortality in patients with early Alzheimer disease. In the clinical context, the use of EEG technology for prediction of survival in individual patients remains to be determined.

ELECTROENCEPHALOGRAM (EEG) abnormalities in Alzheimer disease (AD) include increase of theta and delta activity and reduction of beta power.15 The changes in theta activity may occur early in the disease,5 whereas increase of delta is reported in the late stages.2 Follow-up studies show that slowing on spectral EEG at baseline is associated with more severe cognitive and functional decline in patients with AD.68 Although prediction of survival in patients with AD is important for family and caregivers and for planning of patient care, there are few follow-up studies on the predictive ability of EEG for survival in AD.3,810 The results have been inconsistent, with some studies reporting an association of EEG variables at baseline and survival,3,9,10 while in a more recent report this association was not statistically significant.8 Moreover, study populations included selected hospitalized patients10 and small numbers of patients,3,9 and the mean follow-up time was not more than 3 years.3,810

Clinical features, including extrapyramidal symptoms, and neuroimaging characteristics, such as presence of periventricular white matter lesions or cortical atrophy, are reportedly associated with EEG abnormalities.1113 It is possible that these factors explain the relationship between an abnormal EEG at baseline and the presumed risk of increased mortality. It is therefore necessary to examine the relationship between EEG findings and survival with adjustment for these factors as possible confounders. A similar approach was taken in a recent study by Lopez et al,8 who demonstrated that an abnormal EEG and presence of psychosis were independent predictors of cognitive and functional decline in AD.

To avoid the limitations of previous studies of EEG and survival in AD, we studied quantitative spectral EEG at the time of diagnosis and survival in a prospective long-term follow-up study of consecutively referred patients with probable AD to answer 2 questions. First, can the EEG predict survival in patients with AD? Second, if this is the case, can this relationship be explained by clinical features of AD, including dementia severity and presence of extrapyramidal symptoms, or by presence of white matter lesions or cortical atrophy?

PATIENTS AND METHODS
PATIENTS

Patients met criteria for probable AD according to National Institute of Neurological and Communicative Disorders and Stroke–Alzheimer's Disease and Related Disorders Association criteria14 and were participants in a prospective study of 200 consecutive first referrals to the outpatient memory clinic of the Academic Medical Center, Amsterdam, the Netherlands.15 Inclusion criteria were referral for suspected dementia, age 65 years and older, and presence of an informant involved in the care of the patient; patients who were previously investigated for dementia were excluded. At baseline, all subjects were examined with the Netherlands version of the Cambridge Examination for Mental Disorders of the Elderly (CAMDEX-N),16,17 computed tomography (CT), EEG, and laboratory screening.18 Of the 200 patients initially referred, 105 were diagnosed as having probable AD, but 4 EEGs were missing because of technical reasons, leaving 101 patients in the study (43 men and 58 women). All patients and legally authorized representatives gave informed consent to participate in this study. Survival data were obtained from general practitioners and were available for all patients. Presence of extrapyramidal signs was defined as having at least 1 of the following signs at neurological examination: tremor, rigidity, bradykinesia, abnormal postural reflexes, or a characteristic parkinsonian gait. Clinical indicators used to define presence of 1 or more vascular risk factors included history of diabetes mellitus, hypertension, myocardial infarction, transient ischemic attacks, or stroke. Presence of delusions or hallucinations was determined by questioning the caregiver according to the standardized interview with the CAMDEX-N.16,17 Ten patients used benzodiazepines; no other centrally active medication was used.

EEG METHODS

As described previously,6 EEG recordings were performed at baseline with the use of an on-line 20-channel EEG system, with the international 10/20 system and silver–silver chloride electrodes. The impedance of the electrodes was always lower than 5 kΩ. The high-frequency cutoff was 35 Hz (−3 dB, 12 dB/octave), and the time constant was 0.3 seconds. The EEG was recorded under vigilance control with the patient's eyes closed under resting conditions. With visual inspection to rule out segments with artifacts, such as electromyogram or electrocardiogram activity or eye movement, only artifact-free periods were used for quantitative analysis. Computerized spectral analysis was performed by means of fast Fourier analysis after sampling at a rate of 70 Hz on 6 bipolar frontocentral and parieto-occipital derivations: F3 to C3, T5 to O1, F4 to C4, T6 to O2, P3 to O1, and P4 to O2. After fast Fourier analysis was performed on an epoch of 3.36 seconds with a bin width of 0.298 Hz, the power spectrum of this epoch was calculated and an average power spectrum was computed as the mean of 30 successive epochs. Subsequently, this average power spectrum was then filtered with a 3-point triangular symmetrical filter. This filter did not influence the frequency value of the peak. Leakage of power to other frequencies caused by the triangular window did not significantly influence the power within a spectral band, as the spectral bands consisted of at least 12 points and leakage was restricted to only 1 point. The frequency bands were set as follows: delta, 0.3 to 3.5 Hz; theta, 3.6 to 7.4 Hz; alpha, 7.5 to 12.5; and beta, 12.6 to 35.4 Hz. Since absolute values may suffer from large variability, relative values (percentage) were calculated by dividing the absolute power in each frequency band by the total power of the whole spectrum in all 4 bands.19 Then, a logarithmic transformation was performed to compensate for data skewness with the formula log[x/(1−M)], where x refers to the relative value.20

CT STUDIES

Computed tomography was performed with a scanner (Somatom Plus or HiQ; Bruker AXS, Inc, Erlangen, Germany) in all patients. Consecutive slices were obtained parallel to the canthomeatal line, with a thickness of 5 mm up to and including the third ventricle and of 10 mm above this level. Hard copies of unenhanced CT scans were studied jointly by 2 observers (A.H. and B.V.) who were blinded to the clinical data. The CT ratings included quantitative and qualitative measures. Quantitative measures included the width (in millimeters) of the third ventricle at the level of interventricular foramina, and the frontal horn, occipital horn, and bicaudate indexes.21 Cortical and subcortical infarctions and lacunar infarcts were rated as present or absent in 4 regions, in either left or right anterior or posterior regions. Cerebral white matter lesions were rated as previously described.22Leukoaraiosis was defined as the presence of poorly delineated hypodense areas around the frontal horns and/or around the posterior part of the lateral ventricles. An area was considered hypodense if its density was between that of normal white matter and that of the cerebrospinal fluid. The extent of abnormal white matter was graded as 1 when only the region adjoining the ventricles was abnormal and as 2 when the entire region from the ventricle to the cortex was abnormal. Separate scores were given for the left and right frontal and parietal regions. All scores were added so that leukoaraiosis could be graded from 0 to 8. Absence of leukoaraiosis was defined as a total score of 2 or less (n=70), and presence of leukoaraiosis as a total score of 3 or more (n=31). Cerebral cortical atrophy was rated in left and right frontal and temporoparietal regions. Ratings were performed by visual inspection with the following scale: 0, no atrophy; 1 slight atrophy; 2, moderate atrophy; and 3, severe atrophy.23,24 Scores of 0 and 1 were combined to indicate absence of atrophy, and scores of 2 and 3 were combined to indicate presence of atrophy.

STATISTICAL METHODS

Data analysis was performed with SPSS, version 7.0 for Windows, a commercially available statistical package (SPSS Inc, Chicago, Ill). Cox proportional hazards multiple regression analysis was used to estimate the effects of baseline variables on survival. Hazard ratios (HRs) with 95% confidence intervals (CIs) are presented as estimates of these relationships. Statistical analysis was performed in several segments. Predictors of survival, the log relative percentage values of the 4 EEG bands, were entered in the regression model as independent variables, with adjustment for age at entry and sex. Then, using statistically significant predictors from these analyses, we used stepwise Cox proportional hazards regression analysis to identify the best predicting variable or a combination, adjusted for age and sex. The next step was to investigate whether the relationship between a significant predictive EEG variable from this stepwise analysis could be explained by confounding with other factors. This would be the case if the association were to disappear when a putative confounder was added as explanatory factor to the model. We examined the possible confounding effect of education, and clinical and neuroimaging features, by adding these variables to the regression model as independent variables, for each variable in a separate analysis. Clinical variables included dementia severity, symptom duration, level of general cognitive function as measured with the overall score on the cognitive test from the CAMDEX-N, presence of extrapyramidal symptoms, presence of delusions or hallucinations, and presence of vascular risk factors. Neuroimaging variables included anterior and posterior horn index, bicaudate index, width of the third ventricle, presence of cortical or subcortical infarctions, ratings of local cortical atrophy, and presence of leukoaraiosis. With the EEG variable that was most strongly associated with survival after these adjustments, we constructed survival curves by the Kaplan-Meier method with dichotomized groups according to the median of the distribution. Log rank tests were performed to assess differences between survival curves. Proportionality of the hazard was visually inspected from log minus log survival curves,25 stratified by the variable of interest and adjusted for the other variables, from plots of weighted Schoenfeld residuals vs time,25 and from the interaction between variables and log(time). Linearity of the log(hazard) with prognostic variables was tested from plots of deviance residuals against the variables25 and from improvement of the fit by adding quadratic terms.

RESULTS

Some subject characteristics are presented in Table 1. Relative power EEG values in percentage of delta, theta, alpha, and beta band, respectively, are shown in Table 2. Means of combined right and left EEG variables were calculated for frontocentral and parieto-occipital regions, since no differences between homologous left and right measurements were observed (Table 2). Fifty-one patients (50.5%) died during the follow-up period, with a median (±SD) survival time of 4.1±0.2 years. The following variables were statistically significantly associated with increased risk of mortality: higher age (per 5-year increase, hazard ratio ([HR]), 1.26; 95% CI, 1.01-1.57; P<.05), male sex (HR, 1.92; 95% CI, 1.12-3.23; P<.05), lower score on the cognitive test from the CAMDEX-N (HR per 10-point decrease, 1.43; 95% CI, 1.19-1.69; P<.001), lower score on the Mini–Mental State Examination26 (HR per 5-point decrease, 1.47; 95% CI, 1.19-1.82; P<.001), and more severe dementia (HR, 2.18; 95% CI, 1.19-3.99; P<.05), but not education (HR, 1.07; 95% CI, 0.89-1.28; P =.58) and symptom duration (HR, 1.0; 95% CI, 0.99-1.01; P =.54).

Higher parieto-occipital (P<.05) and frontocentral (P<.05) theta activity, and lower parieto-occipital alpha (P<.01) and parieto-occipital beta (P<.001) activity were statistically significantly associated with increased risk of mortality, adjusted for age and sex (Table 3). Stepwise Cox regression analysis showed that parieto-occipital beta activity was the strongest predictor of survival in AD (P<.01), while parieto-occipital alpha power was retained in the model with additional predictive ability (P<.05). Thus, both parieto-occipital beta and alpha power were used for further analyses. The relationships between parieto-occipital beta and alpha activity and survival remained statistically significant after adjustment for all possible confounders examined in this study and after exclusion of subjects with a history of transient ischemic attacks or stroke. Since our patients had relatively high delta power in parieto-occipital leads and delta power may increase with advancing age,1 the relatively high age of our study population might suggest that delta is a marker of dementia severity, independent of the CAMDEX-N dementia severity marker used in our study. However, delta power was statistically significantly associated with general level of cognitive function, both the cognitive test of the CAMDEX-N (M<.01) and the Mini–Mental State Examination (P<.01), as well as dementia severity (P<.05). Further, the association between beta and alpha power and survival was not changed with the addition of parieto-occipital delta power to the Cox regression model as correction factor, with the use of multiple linear regression analysis with adjustment for age and sex. The possibility that we included patients with Lewy body dementia was considered, since this subgroup of patients appears to have more abnormal EEGs than patients with only AD.27 However, no patient in our study satisfied clinical criteria for Lewy body dementia.28 Further, there were no significant differences in alpha (P>.50) or beta (P>.80) power between groups with or without extrapyramidal symptoms, or in alpha (P>.30) or beta (P>.20) power between those with and without delusions or hallucinations, by means of multiple linear regression analysis, with adjustment for age and sex. After a mean of 5.4 years of follow-up, Kaplan-Meier survival curves for groups dichotomized according to the median (−1.14; range,−1.85 to−0.24) log relative percentage beta activity (Figure 1) showed that 59% of the patients with high beta power were alive (median survival time, at least 5.4 years), while 14% of the patients with low beta power were alive (median survival time, 3.2 years). Log rank test for comparison between patients with high vs low beta activity was statistically significant (P<.001). Kaplan-Meier survival curves are also shown for log relative percentage alpha activity in Figure 2, dichotomized according to the median of the distribution. Similar results were obtained when patients using benzodiazepines were excluded (n=10). Proportionality assumptions of the hazard were met for all EEG variables.

COMMENT

We studied the relationship between EEG at the time of diagnosis and survival in consecutively referred patients with early AD. Low parieto-occipital beta power was the strongest predictor of mortality, while decrease in alpha power also independently contributed to this mortality risk, although to a lesser extent. The predictive power of the EEG for survival was not explained by differences in demographic, clinical, or CT features. Also, the higher percentage delta power in our patients as compared with that in some previous studies1,29,30 is probably related to the greater age of our study population than in these previous studies, since in healthy older subjects delta activity may increase with age.1 Indeed, delta activity appeared not to be a marker of severity, independent of the CAMDEX-N dementia severity marker used. In this study, we extended our earlier observations that, from all EEG bands, parieto-occipital beta values were the strongest predictor of cognitive and functional decline in patients with probable AD.6 This prognostic value was further specified by the observation that low values of beta on EEG were related to rapidly progressive decline.6

From a methodological viewpoint, our EEG study on survival in AD differs from previous studies in several aspects, including choice of study population, length of follow-up, inclusion of possible confounders, and approach for statistical analysis. In contrast to some reports,3,8,10 we studied consecutively referred patients for evaluation of memory complaints or (suspected) dementia, and these patients were part of a prospective follow-up study.15 None of our patients satisfied any of currently used clinical criteria for dementia with Lewy bodies,28 and alpha and beta power in our study were not different when compared in patients with and without extrapyramidal symptoms, or in those with and without delusions or hallucinations. Electroencephalographic studies in selected groups of patients with AD8 may limit the external validity to clinical practice. Further, in our study 51.5% of the patients died during follow-up, while in some previous studies these percentages varied from 12.5%3 to 27.4%.8 The shorter follow-up in these previous studies may limit the statistical power and may there fore explain differences in study results. In the study by Kaszniak et al,10 45.7% of the patients died, but these were hospitalized patients with dementia. Not all previous reports use analysis of separate EEG bands,8,10 while the notion that these bands have different regional correlates in the human brain31,32 suggests the necessity of separate EEG band analysis. Our approach to use a wide selection of clinical and CT variables to identify factors, rather than the EEG alone, that also may account for the observed relationship between EEG and survival was not used in other studies.3,810 Furthermore, we extended our analyses by identifying a subset of independent predictive EEG bands.

The predictive ability of the EEG for survival in patients with AD was noted by some authors,3,9,10 but not by others.8 Our results are in agreement with an earlier study by Kaszniak et al.10 They examined hospitalized patients, probably more severely demented than patients in our population, with a follow-up period of 1 year and reported that 79% of survivors had either normal EEGs or mild diffuse slowing, contrasting with 13% of the deceased patients with similar EEG findings.10 Interestingly, they found no differences between survivors and nonsurvivors in baseline demographic and severity of disease characteristics, or in degree of cerebral atrophy on CT. Our results are further supported by 2 other studies reporting that increase of slow activity on EEG is associated with increased risk of mortality.3,9 In contrast to our observations, Lopez et al8 found that mean frequency as pooled score from parasagittal derivations was not associated with survival in patients with AD. In this study, however, the HR for increased risk of mortality was in the expected direction. Since 27.4% of their patients died, compared with 50.5% in our study, with similar total patient numbers in both studies, their limited statistical power may render their study sensitive to a type II error. Indeed, age and sex were not significant predictors in the study by Lopez et al, whereas these are well-established predictors of survival.3336

Beta and alpha activity were recognized as independent predictors of survival in our study. These results suggest that beta and alpha activity have pathophysiological correlates in the brains of patients with AD and that these correlates are different for both EEG bands. Preclinical and clinical studies show that several brain structures intimately involved in the pathophysiology of AD are linked to beta activity, including the hippocampus,37 the cholinergic nucleus basalis,38 and layer II of the entorhinal cortex.39 Even in patients with very mild AD, there is a marked loss of layer II entorhinal cortex neurons,40 and synaptic cell loss in the hippocampal formation is an early phenomenon in AD.41 The observation by Williamson et al42 that normal elderly subjects with cognitive decline had a marked reduction in beta activity, compared with those without cognitive decline, is intriguing in this context. This observation might have resulted from the incorporation of subjects with unrecognized AD in the sample. Reduction of beta activity in patients with early AD may thus reflect very early neuropathological changes. Further, the predictive ability of decreased beta activity for survival in AD does not simply reflect dementia severity, cortical atrophy, or leukoaraiosis, as suggested by our findings. In analogy with Parkinson disease, where only substantial neuronal loss in the substantia nigra results in clinical impairment, a certain amount of neurons in hippocampus and entorhinal cortex may need to degenerate for demonstrable dementia symptoms to be present. The additional predictive ability for survival of decreased alpha activity may be explained by the finding that alpha is generated in corticocortical pathways.31

There may be certain methodological limitations to our study. Although we made strenuous efforts to examine the effect of factors that could explain the relationship between EEG values and survival, the possible confounding effect of both clinical and neuroimaging features cannot be excluded. Vascular risk factors and indicators of atherosclerosis, recently implicated in the pathogenesis of AD,43 may especially be associated with EEG abnormalities and they may also play a role in the prognosis. We examined leukoaraiosis on CT, but more sensitive assessment of white matter abnormalities and hippocampal atrophy is possible with magnetic resonance imaging. Further, we used presence of diseases associated with vascular risk, but no quantified measures of atherosclerosis were used. A recent study suggested that patients with AD homozygous for apolipopotein E4 allele had selective decreases in functional connectivity, assessed with EEG coherence in the alpha frequency band, compared with those without or with only 1 apolipopotein E4 allele.44 Although several studies did not find an association between apolipoprotein E genotype and survival in AD,45,46 another study reported that the presence of at least 1 apolipoprotein E4 allele reduced the risk of mortality in AD.47 These results suggest that the role of apolipoprotein E genotype, subclinical vascular disease, and AD progression in relation to EEG findings need further examination. Medication in our patients probably did not play a role, since exclusion of those with centrally active medication yielded similar results. Recently, efforts were made to construct prediction algorithms based on clinical features, including age at onset, presence of extrapyramidal or psychotic symptoms, and level of cognitive function.46 Since we used these features as correction factors, our results suggest that the EEG should be used as an additional variable in future endeavors to construct a predictor index.

In conclusion, decreased beta and decreased alpha activity on EEG are predictors of survival in patients with early AD, independent of demographic, clinical, and CT features, including leukoaraiosis and cortical atrophy. These results suggest that the EEG should be included as an independent variable in survival models of AD. However, the use of the EEG for prediction of survival in individual patients in clinical practice remains to be determined.

Back to top
Article Information

Accepted for publication January 16, 1998.

Reprints: Jules J. Claus, MD, PhD, Department of Neurology, Academic Medical Center, University of Amsterdam, Meibergdreef 9, 1105 AZ Amsterdam, the Netherlands (e-mail: j.j.claus@amc.uva.nl).

References
1.
Coben  LADanziger  WStorandt  M A longitudinal EEG study of mild senile dementia of Alzheimer type: changes at 1 year and at 2.5 years. Electroencephalogr Clin Neurophysiol. 1985;61101- 112Article
2.
Penttila  MPartanen  JVSoininen  HRiekkinen  PJ Quantitative analysis of occipital EEG in different stages of Alzheimer's disease. Electroencephalogr Clin Neurophysiol. 1985;601- 6Article
3.
Soininen  HPartanen  JLaulumaa  VPaakkonen  AHelkala  E-LRiekkinen  PJ Serial EEG in Alzheimer's disease: 3 year follow-up and clinical outcome. Electroencephalogr Clin Neurophysiol. 1991;79342- 348Article
4.
Schreiter-Gasser  UGasser  TZiegler  P Quantitative EEG analysis in early onset Alzheimer's disease: correlations with severity, clinical characteristics, visual EEG and CCT. Electroencephalogr Clin Neurophysiol. 1994;90267- 272Article
5.
Prichep  LSJohn  ERFerris  SH  et al.  Quantitative EEG correlates of cognitive deterioration in the elderly. Neurobiol Aging. 1994;1585- 90Article
6.
Claus  JJKwa  VIHTeunisse  S  et al.  Slowing on quantitative spectral EEG is a marker for rate of subsequent cognitive and functional decline in early Alzheimer's disease. Alzheimer Dis Assoc Disord. In press.
7.
Kuskowski  MAMortimer  JAMorley  GKMalone  SMOkaya  AJ Rate of cognitive decline in Alzheimer's disease is associated with EEG alpha power. Biol Psychiatry. 1993;33659- 662Article
8.
Lopez  OLBrenner  RPBecker  JTUlrich  RFBoller  FDeKosky  ST EEG spectral abnormalities and psychosis as predictors of cognitive and functional decline in probable Alzheimer's disease. Neurology. 1997;481521- 1525Article
9.
Rodriguez  GNobili  FArrigo  A  et al.  Prognostic significance of quantitative electroencephalography in Alzheimer patients: preliminary observations. Electroencephalogr Clin Neurophysiol. 1996;99123- 128Article
10.
Kaszniak  AWFox  JGandell  JLGarron  DCHuckman  MSRamsey  RG Predictors of mortality in presenile and senile dementia. Ann Neurol. 1978;3246- 252Article
11.
Helkala  E-LLaulumaa  VSoininen  HPartanen  JRiekkinen  PJ Different patterns of cognitive decline related to normal or deteriorating EEG in a 3-year follow-up study of patients with Alzheimer's disease. Neurology. 1991;41528- 532Article
12.
Saletu  BAnderer  PPaulus  E  et al.  EEG brain mapping in diagnostic and therapeutic assessment of dementia. Alzheimer Dis Assoc Disord. 1991;5(suppl 1)S57- S75Article
13.
Lopez  OLBrenner  RPBecker  JTJungreis  CRezek  DDeKosky  ST Electroencephalographic correlates of periventricular white matter lesions in probable Alzheimer's disease. Dementia. 1995;6343- 347
14.
McKhann  GDrachman  DFolstein  MKatzman  RPrice  DStadlan  EM Clinical diagnosis of Alzheimer's disease: report of the NINCDS-ADRDA work group under the auspices of Department of Health and Human Services task force on Alzheimer's disease. Neurology. 1984;34939- 944Article
15.
Walstra  GJMTeunisse  Svan Gool  WAvan Crevel  H Reversible dementia in elderly patients referred to a memory clinic. J Neurol. 1997;24417- 22Article
16.
Derix  MMHofstede  ABTeunisse  S  et al.  CAMDEX-N: the Dutch version of the Cambridge Examination for Mental Disorders of the Elderly with automatic data processing. Tijdschr Gerontol Geriatr. 1991;22143- 150
17.
Roth  MHuppert  FATym  EMountjoy  CQ CAMDEX, the Cambridge Examination for Mental Disorders of the Elderly.  Cambridge, England Cambridge University Press1988;
18.
Teunisse  SBollen  AEvan Gool  WAWalstra  GJM Dementia and subnormal levels of vitamin B12: effects of replacement therapy on severity of dementia. J Neurol. 1996;243522- 529Article
19.
John  ERPrichep  LSFridman  JEaston  P Neurometrics: computer-assisted differential diagnosis of brain dysfunctions. Science. 1988;239162- 169Article
20.
John  ERAnn  HPrichep  LSTrepetin  MBrown  PKaye  H Development equations for the electroencephalogram. Science. 1980;2101255- 1258Article
21.
Hijdra  AVerbeeten  BJ Leukoaraiosis and ventricular enlargement in patients with ischemic stroke. Stroke. 1991;22447- 450Article
22.
Hijdra  AVerbeeten  BJVerhulst  JAPM Relation of leukoaraiosis to lesion type in stroke patients. Stroke. 1990;21890- 894Article
23.
Waldemar  GBruhn  PKristensen  MJohnson  APaulson  OBLassen  NA Heterogeneity of neocortical cerebral blood flow deficits in dementia of the Alzheimer type: a [99mTc]-d,l-HMPAO SPECT study. J Neurol Neurosurg Psychiatry. 1994;57285- 295Article
24.
Jacoby  RJLevy  RDawson  JM Computed tomography in the elderly, I: the normal population. Br J Psychiatry. 1980;136249- 255Article
25.
Collet  D Modelling Survival Data in Medical Research.  London, England Chapman & Hall1994;
26.
Folstein  MFFolstein  SEMcHugh  PR "Mini–mental state": a practical method for grading the cognitive state of patients for the clinician. J Psychiatr Res. 1975;12189- 198Article
27.
McKeith  IG Lewy body disease. Curr Opin Psychiatry. 1995;8252- 257Article
28.
McKeith  IGGalasko  DKosaka  K  et al.  Consensus guidelines for the clinical and pathological diagnosis of dementia with Lewy bodies (DLB): report of the consortium on DLB international workshop. Neurology. 1996;471113- 1124Article
29.
Petit  DMontplaisir  JLorrain  DGauthier  S Spectral analysis of the rapid eye movement sleep electroencephalogram in right and left temporal regions: a biological marker of Alzheimer's disease. Ann Neurol. 1992;32172- 176Article
30.
Soininen  HPartanen  VJLaulumaa  VHelkala  E-LLaakso  MRiekkinen  PJ Longitudinal EEG spectral analysis in early stage of Alzheimer's disease. Electroencephalogr Clin Neurophysiol. 1989;72290- 297Article
31.
Steriade  MGloor  PLlinas  RRLopez da Silva  FHMesulam  M-M Report of IFCN Committee on Basic Mechanisms: basic mechanisms of cerebral rhythmic activities. Electroencephalogr Clin Neurophysiol. 1990;76481- 508Article
32.
Steriade  M Cellular substrates of brain rhythms. Niedermeyer  ELopes da Silva  FedsElectroencephalography Basic Principles, Clinical Applications, and Related Fields. Baltimore, Md Williams & Wilkins1993;27- 62
33.
Heyman  APeterson  BFillenbaum  GPieper  C The Consortium to Establish a Registry for Alzheimer's Disease (CERAD), part XIV: demographic and clinical predictors of survival in patients with Alzheimer's disease. Neurology. 1996;46656- 660Article
34.
Burns  ALewis  GJacoby  R  et al.  Factors affecting survival in Alzheimer's disease. Psychol Med. 1991;21363- 370Article
35.
Jagger  CClarke  MStone  A Predictors of survival with Alzheimer's disease: a community-based study. Psychol Med. 1995;25171- 177Article
36.
Walsh  JSWelch  GLarson  EB Survival of outpatients with Alzheimer-type dementia. Ann Intern Med. 1990;113429- 434Article
37.
Leung  LS Fast (beta) rhythms in the hippocampus: a review. Hippocampus. 1992;293- 98Article
38.
Riekkinen  PJRiekkinen  MSirvio  JMiettinen  RKoudstaal  PJ Loss of cholinergic neurons in the nucleus basalis induces neocortical electroencephalographic and passive avoidance deficits. Neuroscience. 1992;47823- 831Article
39.
Boeijinga  PHLopez da Silva  FH Differential distribution of beta and theta EEG activity in the entorhinal cortex of the cat. Brain Res. 1988;448272- 286Article
40.
Gomez-Isla  TPrice  JLMcKeel  DW  JrMorris  JCGrowden  JHHyman  BT Profound loss of layer II entorhinal cortex neurons occurs in very mild Alzheimer's disease. J Neurosci. 1996;164491- 4500
41.
Heinonen  OSoininen  HSorvari  H  et al.  Loss of synaptophysin-like immunoreactivity in the hippocampal formation is an early phenomenon in Alzheimer's disease. Neuroscience. 1995;64375- 384Article
42.
Williamson  PCMerskey  HMorrison  S  et al.  Quantitative electroencephalographic correlates of cognitive decline in normal elderly subjects. Arch Neurol. 1990;471185- 1188Article
43.
Hofman  AOtt  ABreteler  MMB  et al.  Atherosclerosis, apolipoprotein E, and prevalence of dementia and Alzheimer's disease in the Rotterdam Study. Lancet. 1997;349151- 154Article
44.
Jelic  VJulin  PShigeta  M  et al.  Apolipoprotein E ϵ4 allele decreases functional connectivity in Alzheimer's disease as measured by EEG coherence. J Neurol Neurosurg Psychiatry. 1997;6359- 65Article
45.
Tanna  NKKohn  MIHorwich  DN  et al.  Analysis of brain and cerebrospinal fluid volumes with MR imaging: impact on PET data correction for atrophy, part II: aging and Alzheimer dementia. Radiology. 1991;178123- 130Article
46.
Stern  YTang  M-XAlbert  MS  et al.  Predicting time to nursing home care and death in individuals with Alzheimer's disease. JAMA. 1997;277806- 812Article
47.
Stern  YBrandt  JAlbert  M  et al.  The absence of an apolipoprotein epsilon4 allele is associated with a more aggressive form of Alzheimer's disease. Ann Neurol. 1997;41615- 620Article
×