Changes in absolute power values of EEG spectra in the follow-up of Alzheimer’s disease Soininen H, Partanen J , Paakkonen A, Koivisto E, Riekkinen PJ. Changes in absolute power values of EEG spectra in the follow-up of Alzheimer’s disease. Acta Neurol Scand 1991: 83: 133-136. We have studied the absolute and relative power and amplitude of EEG spectra (T6-02) of 24 patients with “probable” Alzheimer’s disease at the early stage of the disease and 1 year later and also compared the values to those of normal elderly controls. A remarkable variability of the absolute values was evident both for the patients and for the controls. The AD patients had significantly higher absolute theta amplitude and power and the absolute beta values tended to decrease compared to controls. Absolute delta and alpha values did not differ from those of the controls. The relative delta, theta and alpha power and amplitude, and beta amplitude showed significant changes in AD patients, whereas the relative beta power was unchanged. In the follow-up of AD patients at 1 year, absolute alpha values decreased and delta values tended to increase. As to relative values, both the alpha and the delta significantly changed but the theta and the beta were unaltered. We conclude that both absolute and relative power and amplitude values should be considered in EEG studies of dementia patients. Absolute values are especially useful in follow-up.
In Alzheimer’s disease (AD) slowing of the EEG, disproportional to normal aging, has been observed in numerous studies using both conventional and quantitative EEG methods (1-7). Parallel to the progression of the disease, further slowing of the EEG has been detected also in a few follow-up studies (8-10). In the quantitative analysis of EEG, relative power values are most commonly used (1 1). The analysis of absolute power is more difficult because of the large variability of the values (1 1). Moreover, there are no follow-up studies of absolute power values in AD. The purpose of this study is to describe changes in absolute amplitude and power of EEG spectra at the time the diagnosis is made and one year later in AD patients, and compare these findings with relative value alterations. Material and methods
We studied 24 patients meeting the criteria of probable AD according to the definition of the NINCDSADRDA work group (12). The patients showed progressive worsening of memory and other cognitive functions without disturbance of consciousness. They had no systemic disorders or other brain diseases accounting for or contributing to the progres-
H. Soininen’, J. Partanen’, A. Paakkonen’, E. Koivisto’, P. J. Riekkinen’
’
Departments of Neurology, Clinical Neurophysiology, University of Kuopio, Finland
Key words: Alzheimer’s disease; dementia; electroencephalogram Hilkka Soininen, University of Kuopio, Department of Neurology, P.O. Box 6, 7021 1 Kuopio, Finland, Tel. 971-1 73004 Accepted for publication September 18, 1990
sive deficits in memory and cognition. They were 16 female and 8 male subjects with the mean age of 68 years, S.D. 6 years, and the mean duration of the disease of 2 years, S.D. 1 year. The patients are participating in the follow-up study of AD patients in the Department of Neurology, Kuopio University Central Hospital, Kuopio, Finland. The study was approved by the local ethics committee. The patients and the controls gave their informed consent for their participation in the study; in case the patient was unable to give the consent, it was given by a near relative. The diagnostic procedure included: clinical neurological examination, psychometric testing, routine blood and CSF tests, chest x-ray, ECG, CT scan of the head, conventional and quantitative EEG, and pattern-reversal visual and somatosensoric evoked potentials. The symptomatic causes of dementia were ruled out using clinical, laboratory and CT data. CT scans of the patients were normal or showed atrophic but no focal changes. All the patients with signs of cerebrovascular disease were excluded; the patients included had 4 or less in the ischemic score modified by Rosen et al. (13). After the diagnostic work-up the patients are carefully followed as long as they live, and ultimately, the diagnosis will be verified at autopsy in as many cases as
133
Soininen et al. possible. During the follow-up, the clinical and psychometric examinations are repeated twice a year and neurophysiological and CSF investigations once a year. The severity of dementia assessed by Clinical Dementia Rating Scale (CDR) (14) at baseline was mild for 14 patients, moderate for 9 patients and severe for one patient. A year later, 10 patients still had mild dementia, 10 had moderate and 4 had severe dementia. Twenty patients were without CNS active medication; one patient received a benzodiazepine, one an antidepressant and three patients were on smalldose neuroleptic regimen at baseline and at 1 year. As controls served 2 1 healthy elderly individuals, matched for age and sex of AD patients. EEG methods
EEG for quantitative analysis was recorded from the derivation T6-02. The records were amplified and filtered by DISA 15C01 amplifier with bandpass set at 0.5-50.0 Hz. The EEG was digitized at 125 samples/s with a 12-bit multichannel analyzer (Trator Northern 1710). The EEG was visually inspected on the oscilloscope screen and the samples with artifacts were manually rejected. Four epochs of 8.192 s were recorded. This EEG sample (4 x 1024 points) was then transferred to a Tektronix 4052 computer and stored on disc for offline analysis. The Fast Fourier Transform (FFT) was computed on a series of half-overlapping sections for each of the 4 epochs; The first section comprised the first half, the second section the second and the third quarters, and the third section the second half of the epoch. To reduce leakage and time truncation errors, the time-domain wave form of each section was multiplied by a cosine (Hanning) window before the FFT operation (1 1). FFTs from a total of 12 sections were averaged to obtain the frequency spectrum of the whole EEG sample of 32.772 s. Absolute and relative (percentage of the total) amplitude and power of delta (1.46-3.91 Hz), theta
(4.15-7.32 Hz), alpha (7.57-13.92) and beta (14.16-20.02) bands were calculated. The total range of the analysis was 1.46-20.02 Hz. During the EEG recordings the patients were told to relax with their eyes closed. Verbal communication was used if necessary to maintain the alertness of the patient. According to the EEG findings, the 24AD patients could be divided into two subgroups, 12 patients with stable EEG in 1 year follow-up and another group of 12 patients with deteriorating EEG (10). The EEG was considered to deteriorate if at least 2 of following 4 criteria were fulfilled; 1) decrease of the alpha power more that 10% ; 2) increase of the theta power more that 10%; 3) increase of the delta power more that 10% ; 4)decrease of the mean frequency (range 1.5-20.0 Hz) by more than 1 Hz. Statistics
The significance of the differences between AD patients and controls was analyzed using Students's t-test for independent samples, and baseline and 1-year data of AD patients were compared using Student's t-test for paired samples. To change the data into Gaussian distribution natural logarithm transformation log(x/( 1 - x)) was applied for relative values, where x is the relative power or amplitude of a frequency band, and log(x) for absolute values, where x is the corresponding absolute value (15). Significances of differences were calculated using both raw data and logarithm transformed data. As multiple tests were performed, differences were considered statistically significant at the level of 0.0 1. Results
In comparison between AD patients and controls the logarithm transformation improved significances of the differences in relative values but the effect was smaller in absolute values; the p levels of logarithm
Table 1. Absolute and relative amplitude and power of EEG spectra of normal elderly controls (NI (n = 21) and Alzheimer patients (A) (n = 24) Delta Abs. amp.
Theta
Alpha
Beta
(UVI
4.46 (1.551 5.95 (3.23)
4.79 (1.921 7.63 (4.12)"
18.51 (8.27) 14.68 (6.17)
7.56 (2.51) 5.78 (2.02)
Abs. pow. (UV?
2.69 (1.87) 5.43 (7.751
2.73 (2.55) 7.33 (7.72)*
30.47 (28.61) 15.63 (14.44)
3.27 (2.12) 1.97 (1.28)
Rel. amp.
13.01 (2.50) 17.42 (5.08)**
13.72 (2.66) 21.93 (7.79)***
5 1.06 (6.44) 42.78 (7.62)***
22.21 (4.57) 17.88 (6.27)*
9.09 (5.03) 17.50 (10.19)***
8.00 (3.40) 22.70 (15.50)***
70.91 (13.65) 50.97 (18.55)***
12.00 (7.93) 8.83 (6.88)
(%I Rel. pow.
(%I
Values are means (SDI. The log (1/ 1 -XI) for relative values and log(x) for absolute values was used for statistical calculations. Student's t-test for independent samples; *p
In Alzheimer’s disease (AD) slowing of the EEG, disproportional to normal aging, has been observed in numerous studies using both conventional and quantitative EEG methods (1-7). Parallel to the progression of the disease, further slowing of the EEG has been detected also in a few follow-up studies (8-10). In the quantitative analysis of EEG, relative power values are most commonly used (1 1). The analysis of absolute power is more difficult because of the large variability of the values (1 1). Moreover, there are no follow-up studies of absolute power values in AD. The purpose of this study is to describe changes in absolute amplitude and power of EEG spectra at the time the diagnosis is made and one year later in AD patients, and compare these findings with relative value alterations. Material and methods
We studied 24 patients meeting the criteria of probable AD according to the definition of the NINCDSADRDA work group (12). The patients showed progressive worsening of memory and other cognitive functions without disturbance of consciousness. They had no systemic disorders or other brain diseases accounting for or contributing to the progres-
H. Soininen’, J. Partanen’, A. Paakkonen’, E. Koivisto’, P. J. Riekkinen’
’
Departments of Neurology, Clinical Neurophysiology, University of Kuopio, Finland
Key words: Alzheimer’s disease; dementia; electroencephalogram Hilkka Soininen, University of Kuopio, Department of Neurology, P.O. Box 6, 7021 1 Kuopio, Finland, Tel. 971-1 73004 Accepted for publication September 18, 1990
sive deficits in memory and cognition. They were 16 female and 8 male subjects with the mean age of 68 years, S.D. 6 years, and the mean duration of the disease of 2 years, S.D. 1 year. The patients are participating in the follow-up study of AD patients in the Department of Neurology, Kuopio University Central Hospital, Kuopio, Finland. The study was approved by the local ethics committee. The patients and the controls gave their informed consent for their participation in the study; in case the patient was unable to give the consent, it was given by a near relative. The diagnostic procedure included: clinical neurological examination, psychometric testing, routine blood and CSF tests, chest x-ray, ECG, CT scan of the head, conventional and quantitative EEG, and pattern-reversal visual and somatosensoric evoked potentials. The symptomatic causes of dementia were ruled out using clinical, laboratory and CT data. CT scans of the patients were normal or showed atrophic but no focal changes. All the patients with signs of cerebrovascular disease were excluded; the patients included had 4 or less in the ischemic score modified by Rosen et al. (13). After the diagnostic work-up the patients are carefully followed as long as they live, and ultimately, the diagnosis will be verified at autopsy in as many cases as
133
Soininen et al. possible. During the follow-up, the clinical and psychometric examinations are repeated twice a year and neurophysiological and CSF investigations once a year. The severity of dementia assessed by Clinical Dementia Rating Scale (CDR) (14) at baseline was mild for 14 patients, moderate for 9 patients and severe for one patient. A year later, 10 patients still had mild dementia, 10 had moderate and 4 had severe dementia. Twenty patients were without CNS active medication; one patient received a benzodiazepine, one an antidepressant and three patients were on smalldose neuroleptic regimen at baseline and at 1 year. As controls served 2 1 healthy elderly individuals, matched for age and sex of AD patients. EEG methods
EEG for quantitative analysis was recorded from the derivation T6-02. The records were amplified and filtered by DISA 15C01 amplifier with bandpass set at 0.5-50.0 Hz. The EEG was digitized at 125 samples/s with a 12-bit multichannel analyzer (Trator Northern 1710). The EEG was visually inspected on the oscilloscope screen and the samples with artifacts were manually rejected. Four epochs of 8.192 s were recorded. This EEG sample (4 x 1024 points) was then transferred to a Tektronix 4052 computer and stored on disc for offline analysis. The Fast Fourier Transform (FFT) was computed on a series of half-overlapping sections for each of the 4 epochs; The first section comprised the first half, the second section the second and the third quarters, and the third section the second half of the epoch. To reduce leakage and time truncation errors, the time-domain wave form of each section was multiplied by a cosine (Hanning) window before the FFT operation (1 1). FFTs from a total of 12 sections were averaged to obtain the frequency spectrum of the whole EEG sample of 32.772 s. Absolute and relative (percentage of the total) amplitude and power of delta (1.46-3.91 Hz), theta
(4.15-7.32 Hz), alpha (7.57-13.92) and beta (14.16-20.02) bands were calculated. The total range of the analysis was 1.46-20.02 Hz. During the EEG recordings the patients were told to relax with their eyes closed. Verbal communication was used if necessary to maintain the alertness of the patient. According to the EEG findings, the 24AD patients could be divided into two subgroups, 12 patients with stable EEG in 1 year follow-up and another group of 12 patients with deteriorating EEG (10). The EEG was considered to deteriorate if at least 2 of following 4 criteria were fulfilled; 1) decrease of the alpha power more that 10% ; 2) increase of the theta power more that 10%; 3) increase of the delta power more that 10% ; 4)decrease of the mean frequency (range 1.5-20.0 Hz) by more than 1 Hz. Statistics
The significance of the differences between AD patients and controls was analyzed using Students's t-test for independent samples, and baseline and 1-year data of AD patients were compared using Student's t-test for paired samples. To change the data into Gaussian distribution natural logarithm transformation log(x/( 1 - x)) was applied for relative values, where x is the relative power or amplitude of a frequency band, and log(x) for absolute values, where x is the corresponding absolute value (15). Significances of differences were calculated using both raw data and logarithm transformed data. As multiple tests were performed, differences were considered statistically significant at the level of 0.0 1. Results
In comparison between AD patients and controls the logarithm transformation improved significances of the differences in relative values but the effect was smaller in absolute values; the p levels of logarithm
Table 1. Absolute and relative amplitude and power of EEG spectra of normal elderly controls (NI (n = 21) and Alzheimer patients (A) (n = 24) Delta Abs. amp.
Theta
Alpha
Beta
(UVI
4.46 (1.551 5.95 (3.23)
4.79 (1.921 7.63 (4.12)"
18.51 (8.27) 14.68 (6.17)
7.56 (2.51) 5.78 (2.02)
Abs. pow. (UV?
2.69 (1.87) 5.43 (7.751
2.73 (2.55) 7.33 (7.72)*
30.47 (28.61) 15.63 (14.44)
3.27 (2.12) 1.97 (1.28)
Rel. amp.
13.01 (2.50) 17.42 (5.08)**
13.72 (2.66) 21.93 (7.79)***
5 1.06 (6.44) 42.78 (7.62)***
22.21 (4.57) 17.88 (6.27)*
9.09 (5.03) 17.50 (10.19)***
8.00 (3.40) 22.70 (15.50)***
70.91 (13.65) 50.97 (18.55)***
12.00 (7.93) 8.83 (6.88)
(%I Rel. pow.
(%I
Values are means (SDI. The log (1/ 1 -XI) for relative values and log(x) for absolute values was used for statistical calculations. Student's t-test for independent samples; *p
