Psychometric and EEG changes after carotid endarterectomy

Abstract

The influence of carotid stenosis and its surgical treatment on brain function is still poorly defined. We therefore performed a study to assess psychometric and quantified EEG findings after carotid endarterectomy (CEA). Sixty-nine non-demented patients (aged 72 ± 7 years) with severe carotid stenosis (≥70 %) eligible for CEA were studied. Forty patients (group A) had unilateral stenosis, and 29 patients (group B) had bilateral stenosis. Before and 5 months after CEA all the patients were evaluated by the Trail Making Test A, the Symbol Digit Test, and spectral EEG analysis. At baseline, compared to group A, group B patients performed slowly the Trail Making Test A (Z: 1.45 ± 1.4 vs. 0.76 ± 1.3; p  <  0.05), but not the Symbol Digit Test (Z: 0.83 ± 1.38 vs. 0.64 ± 1.26; p = 0.59). Altogether, the patients with at least one abnormal psychometric test were 29 % (group A: 26 %; group B: 33 %, p = 0.56). The EEG did not differ significantly between patients of group A compared to group B. After CEA, psychometric tests improved (mean Z score from 0.73 ± 1.12 to 0.45 ± 1.15, p  <  0.05). The improvement was similar in group A and B. The EEG mean dominant frequency improved only in group B patients and it was related to the improvement in psychometric tests (r = 0.43, p = 0.05). Low psychometric performance was detectable in about 1/ 3 of non-demented patients with severe carotid stenosis. CEA improved mental performance and, in patients with severe bilateral stenosis, accelerated the EEG frequency.

Appendix

Methods

Details about spectral analysis

The interval of the EEG analysed had length of 60 s, subdivided in 30 epochs of 2 s, Each epoch, below denoted by {x(n)}n = 0,1, …M-1, results in a number of samples equal to M = 512 when the sampling frequency Fs is equal to 256 Hz. Spectral estimation was performed by using: the periodogram method, numerically implemented by FFT. For each subject, the final spectral estimate was obtained by averaging 30 spectra, each estimated from a 2 s epoch. Inorder to estimate the spectrum by the periodogram method the DFT {X(k)}k = 0, …M-1 of each epoch {x(n)}n = 1, …M-1

$$ \mathrm{X}\left(\mathrm{k}\right)={\displaystyle \sum_{\mathrm{n}=0}^{\mathrm{M}\hbox{-} 1}\mathrm{x}\left(\mathrm{n}\right)}\cdot {\mathrm{e}}^{\hbox{-} \mathrm{j}\frac{2\uppi}{\mathrm{M}}\mathrm{nk}}\kern3em \mathrm{k} = 0,\dots,\ \mathrm{M}\hbox{-} 1 $$

(1)

is first computed by employing a FFT algorithm. Then, the spectrum is obtained as

$$ {\mathrm{S}}_{\mathrm{x}}\left(\mathrm{k}\right)=\frac{{\left|\mathrm{X}\left(\mathrm{k}\right)\right|}^2}{\mathrm{M}}\kern3em \mathrm{k} = 0, \dots,\ \mathrm{M}\hbox{-} 1 $$

(2)

the first M/ 2 samples of which represent (under certain mild assumptions concerning the bandwidth of the original analog signal and the sampling frequency) samples of the spectrum at frequency points equally-spaced between 0 and Fs/2.

The Mean Dominant Frequency MDF, representing the barycentre of the spectrum, results from the following

$$ MDF=\frac{{\displaystyle \underset{0}{\overset{f_{\max }}{\int }}{S}_x(f) df}}{{\displaystyle \underset{0}{\overset{f_{\max }}{\int }} f\cdot {S}_x(f) df}} $$

(3)

Relative Power on four bands, representing the energy of the signal in the frequency range defined by the band :

$$ RP=\frac{{\displaystyle \underset{f_l}{\overset{f_h}{\int }}{S}_x(f) df}}{{\displaystyle \underset{0}{\overset{f_{\max }}{\int }}{S}_x(f) df}}\cdot 100 $$

(4)

where fmax = 25.5 Hz, fl = 1 Hz fh = 4 Hz for the delta band; fl = 4 Hz fh = 8 Hz for the theta band; fl = 8 Hz fh = 13 Hz for the alpha band and fl = 13 Hz fh = 17 for the beta 1 band, and fl = 17 Hz fh = 25.5 Hz for the beta 2 band.