EEG as a neurotoxicological indicator

Toxicology, 49 (1988) 237-246 Elsevier Scientific Publishers Ireland Ltd.

EEG AS A NEUROTOXICOLOGICAL INDICATOR

JAN KUBAT, JAROSLAV FORMANEK, ALEXANDR FUCHS, PETR ttEHAK, PETR ZAJICEK, JAN DVOttAK and MILENA VANi~KOVA

Institute of Hygiene and Epidemio~gy, Centre of Industrial Hygiene and Occupational Diseases, Department of Ergonomics, Srobdrova 48, 100 42 Prague 10 (Czechoslovaki~ SUMMARY

In living organisms both chemical agents and physical factors [16] may produce neurophysiological change that affects EEG activity. EEG signals are very suitable for non-invasive measurement of CNS reactions, but quite complicated equipment is necessary for measurement and analysis. We have implemented a system that permits the study of EEG changes both in time and frequency domains using broad-band analysis or fast Fourier transformation (FFT). Experimental animals were influenced by high doses of toxic agents (CO, CS 2, barbiturates, pesticides), drugs and both nonionizing and ionizing radiation. The EEG changes reflecting the influencing factor, respectively its quantity may be divided into several classes: (1) appearance of new activities; (2) disappearance of some activities; (3) increase of amplitudes, respectively spectral power densities (SPD) in certain frequencies; and (4) decrease of amplitudes, respectively SPD in certain frequencies. All changes are related to the controls, i.e. to the relatively normal state of CNS. Furthermore it is possible to investigate the temporal dynamics of these changes. Physiological concordance of these findings is sometimes possible from clinical analogues, but in other cases is unknown and considerable effort will be necessary to elucidate these correlates. Anyhow in some toxic substances, the EEG may be quite insensitive as an indicator of neurotoxicity. The best way to solve these problems is to collect sufficient experimental data for complex analysis. Although few relevant data are currently available, temporal and frequency domain measures of EEG activity appear to have promise as neurotoxicity indicators.

Key words: Neurotoxicology; EEG; Spectral power density; Broad-band analysis; Indicator of neurotoxicity INTRODUCTION

Adverse environmental factors may produce various health disturbances. Several electrophysiological methods are especially suitable for detecting 0300-483X/88/$03.50 © Elsevier Scientific Publishers Ireland Ltd. Printed and Published in Ireland

237

possible effects of chemicals on the central nervous system [7]. These methods are non-invasive and can be used both in humans and in experimental animals. Papers claiming EEG changes detected with visual evaluation arouse suspicion because they are not quantitative enough. Even with computer-aided EEG analysis, questions arise concerning what approaches are best for detection of early toxic changes, how to validate the different quantitative methods and how to extrapolate from animal experimentation. Several examples of possible EEG changes associated with various chemical and physical factors will be presented from research in progress. MATERIALSAND METHODS Since 1963 we have implemented various methods suitable for EEG evaluation. Figure 1 reviews the on-line and off-line system of EEG processing used in our laboratory. This system contains: a control unit for data collection of amplitude or broad-band frequency measures; a minicomputer (Czech analog of PDP 11/05) serves also for additional data processing; a 2-channel signal analyser Briiel-Kjaer 2034 and a HP 2116C minicomputer, with software specialized to fast Fourier transformation (FFT) and spectral power densities (SPD) plotting in form of compressed spectral array (CSA) [2,3,5,10], connected to the HP 2100S for additional statistical processing of the digitized data (on a computer tape). Some of the digitized data are used for further testing of EEG analysis methods in other computing systems (EC 1040, HP 9831A, ADT 4500). The described hardware was used for the following methods:

(1) Amplitude analysis (generally restricted to integration) serves as a coarse evaluation that we use to assess individual frequency bands. (~) FFT spectra plotted as SPD in the form of CSA give a clear survey of changes both in time and frequency of the measured signal. The results are subjected to specialized statistical evaluation [1,3,12]: (a) the significance of changes based on estimates of confidence intervals for mean power densities; (b) "optimal projection" of multi~limensional data matrix so as to get a "trajectory" of the intoxication development. (3) Non-FFT spectral analyses (e.g., autoregressive models, maximum entropy method, Pisarenko harmonic decomposition, Prony method) [6,8,12] give smoother spectra but memory and time demands necessary for processing are much higher. (4) Broad-band analysis (BBA) using analog filters may be executed on-line in real time including data collection in a computer memory medium. The subsequent procedures of presentation and statistical evaluation need naturally more time. We prefer this simple method now because it yields both frequency and time composition of EEG signals [4,5,11]. (5) Special spectrum processor makes efficient analysis and evaluation of electrophysiological signals possible. Additional analysis/modelling requires

238

SIGNALS, electrophyslologlcal contra 1 tlme callbratlng

gram8

_

.L, .

S I GNAL TREATtlENT

I,

Fig. 1. Block scheme of possible ways of EEG evaluation.

**I,=

Z

-

I

S ch 10 bit

32 ch 12 bll

-

HP 2116C Oflo 16 k 16

tlINICOnPUTER 0tl1 32 k 16 word8

f

bit-_,

blt

I

result8

-

-

SPD in CSA form Slgnlflcant renges comparlaon Optimal proJectlon

Under preparationl - SPECTRA - CORRELATIONS - COHERENCES - TIIIE SERIES - STATISTICS - IIODELLING - AUTOflAT I C ADRPT IUE SEGtlENJATION

Done~ - BBR

data storage in computer memory media. The preliminary results prove this way as very suitable in applications in experimental neurotoxicology and other branches of occupational health. (6) Automatic adaptive segmentation (AAS) cuts signal to "quasi stationary" segments on base of adaptive autoregressive filtering or autocorrelation function [13,14]. We have used modified AAS only for a few EEG samples [9]. Segmentation is a very exacting method for contemporary practise, but we hope that the technical development will enable us to use it in future. A new insight into the EEG signal evaluation and interpretation related to environmental factors is expected.

Subject preparation We have used various subjects like rats, guinea pigs, cats, rabbits and dogs (beagle). In all animals the epidural electrodes were implanted under anaesthesia. The experiments were carried out in the majority on freely moving subjects. RESULTS

The main purpose of this study is to show the quality and quantity of EEG changes which serve as a neurotoxicological indicator. Considerable study will be necessary to achieve this goal. Due to the complexity of the methods, the present data must be considered preliminary. We have chosen examples of EEG evaluation during spreading depression, under the influence of a neuroleptic drug, carbon monoxide and ionizing irradiation.

(1) Spreading depression Spreading cortical depression (SD) is a well known reversible phenomenon that may be induced by electric, chemical, mechanical or other stimuli. We have examined the course of EEG changes in the time and the frequency domains during chemically induced SD in 6 rats. The typical results show (see Fig. 2) simultaneous depression at all frequencies and also simultaneous but slower restoration of EEG activity [11]. These well-balanced dynamic EEG changes may be regarded as one of the reference patterns for neurotoxicological studies.

(2) Carbon monoxide (CO) Knowledge of CO intoxication is relatively extensive [15] and therefore, we have studied CO-induced EEG changes. The following sample shows interesting changes. We have managed to make 2 consecutive CO exposures (inhaled air contained 10 000 ppm of CO) in 1 rat in a single experiment interrupted by a recovery period (see Fig. 3). In the first part of CO intoxication from the normal state to the total disappearance of EEG activity, lasting only about 12 min, some remarkable patterns of spectra development were observed. In the subsequent recovery period (about 15 min) a nearly normal EEG state was achieved; the SPD recovered sooner at

240

KCI

rapt If'it Im

bBpollr

el octPo¢le

PofePence • I ectrocle

O" ~ ' d u r

I¢:I~.

Y31¢~1¢0 ~.

• ~n

|

8OC

ZS

lnE |

4

•

•

•

IlLS

IS

•

•

•

•

~

TI~

~

•

•

•

•

I@

w ~

=ll

FREQUENCY 1 0 . 2 - ~0 H-z]

Fig. 2. Spreading depression in an anaesthetized rat: T o p : - electrode a r r a n g e m e n t and site of KC1 administration; M/dd/e -- samples of EEG activity before, 36 s after KC1 administration and 9 min later; B o t t o m : - - time course of EEG spectra (FFT) during 15 rain experiment.

241

L~

bO

?ig. i.74 ;ion

.

.

.

.

°-

3. Time course of the EEG spectral power densities of a freely-moving rat intoxicated by twice repeated CO inhalation (10 000 ppm): Axis: frequency from 0.2 to 30 Hz; y - spectral power density in W/Hz; z - real time of the experiment in min (columns to be continued). 0-min, control period; 5.75--18.3 min, the first intoxication (the beginning and the end marked by arrows); 18.3--33.25 min, period of restoraof the EEG activity; 33.25--57.75 min, the second intoxication (between arrows); from 57.75 min, prelethal stadium.

iN RAT

SPECTRAL PC

EFFECT OF CC

slower frequencies than at higher ones. The latter intoxication lasting about 24 min had a course completely opposite to the recovery (the S P D at higher frequencies disappeared sooner than at lower frequencies). The significance of changes demonstrated with the use of estimates of confidence intervals of S P D revealed 9.4 and 11 Hz as the frequencies most sensitive to CO. The additional statistical method "optimal projection" gave new insight to the whole intoxication development [5,10,11]. These findings are consistent with other results. The temporal pattern of compressed E E G spectra arrays could be used as a neurotoxicological reference pattern for comparison with the effects of other chemical agents.

(3) Neuroleptic drug A new potential neuroleptic was prepared for production in a Czechoslovak research institute and we participated during preliminary tests by measuring EEG changes (experiments of Dr. J. Mety§). Figure 4 summarizes the principal results (1 example, EEG measured only in the *JbJl

~1 ! ~u--

J |

BEFORE

4 HRS RFTER flPPLICRTI~

rREOu¢.c' [0.2

-

30 I'tzl

Fig. 4. Influence of the neuroleptic "isofloxythepin" on the EEG SPD in a rat: Control patterns -- before p.o. application (left side); all other controls including placebo effect are consistent with these. Patterns after (right side) small dose 2 mg/kg (top), respectively larger dose 10 mg/kg (bottom).

243

range 0.2--15 Hz): - a small therapeutic dose, 2 mg/kg, induces changes of EEG spectra -- increased activities at nearly all frequencies (4 h after administration); -- a larger therapeutic dose, 10 mg/kg, multiplies the changes of spectra patterns (high activity practically in all frequencies), residual drug effect is apparent 24 h after administration. All controls (before drug application and 48 h after and during placebo administration) showed consistent normal EEG spectra [11]. This example shows the possible sensitivity of the EEG method in differentiating reliably 2 doses (2 levels of a specific effect).

f4j y-Irradiation With respect to possible radiation accidents in nuclear power plants and in order to predict their effect on behaviour and the CNS of the staff during the earliest stages of the accident, we studied changes of EEG activity in 10 dogs (beagle) after y-irradiation with high doses 50, 100 and 200 Gy. Some radioprotective drugs were tested. The BBA made on several occasions after irradiation revealed the following results: (a) the time dynamics of the EEG activity increased in the early stage was in the later stage altered by restoration or depression (mainly in a and ~ bands). (b) the duration of the stages was related both to the dose and to the administered radioprotective drug. Figure 5 shows 1 example of simplified results - only significant changes (both positive and negative for P ~ 0.001) in individual frequency bands in the time course [11]. This study again represents a certain validation of the EEG signal for defining the states (or changes of dynamics) of the CNS. DISCUSSION

Some results of our preliminary studies are consistent with existing neurotoxicological knowledge, but other findings are still difficult to understand. Distinct changes in EEG spectra, at least to certain environmental factors were observed and presented. A discussion of drawbacks and merits of these methods will be more fruitful after more data Qn the EEG changes due to chemical agents are collected. A properly planned interlaboratory validation study would be best for this purpose. CONCLUSION

These studies show that EEG spectra may in some cases serve as a useful indicator in neurotoxicology. The principal EEG changes that may be observed are: (1) appearance of new activities; (2) disappearance of some activities; (3) increase of amplitudes, respectively SPD in certain frequencies; (4) decrease of amplitudes, respectively SPD in certain frequencies, which all may be reflected in: (5) a change of the factorial structure of the EEG signal, as with the described effect of the lethal CO intoxication in the rat.

244

TmI

lm.61

0 1 2 6X? 2’520360.~19.n.14

3 4 Km. 6.7.t263 6

6

7 0

3

to

DOSE

II

12

100

13

14

Gy

16

16

17

16

Id

26

21

Fig. 6. EEG and y-irradiation: Simplified time course of significant changes (positive and negative with limit P 4 0.001) in individual frequency bands (d, 0, u 1, (I 2, fl 1 and fl 2) obtained from broad-band analysis results for left (top) and right (bottom) hemisphere of a dog (beagle) irradiated by high dose 100 Gy from the right side.

.?U36@?- *lo

IRRADIATION

GfWlFl IRRfUlI~TION & SIGNIFICNYT EEG CHF)NGESIN TII’IE~; Fim

C o m p u t e r - a i d e d e v a l u a t i o n of t h e E E G s i g n a l m a y y i e l d f u r t h e r p a r a m e t e r s of t h e s t a t e o r d y n a m i c c h a n g e s . A c o m p u t e r m a y p r o v i d e t h e s u b s e q u e n t modelling or o t h e r m a t h e m a t i c a l / s t a t i s t i c a l evaluation. EEG activity may sometimes differentiate qualities, sometimes even q u a n t i t i e s a n d i n o t h e r c a s e s m a y b e q u i t e i n s e n s i t i v e t o c h e m i c a l or o t h e r e n v i r o n m e n t a l f a c t o r s . T o e l u c i d a t e all t h e s e p r o b l e m s it is n e c e s s a r y to o b t a i n m o r e d a t a for c o m p l e x a n a l y s i s . A p r o p e r l y p l a n n e d i n t e r l a b o r a t o r y v a l i d a t i o n s t u d y w o u l d b e b e s t for t h i s p u r p o s e . REFERENCES 1

2 3

4

5

6 7

8 9 10

11

12 13 14 15 16

246

J. AndS1, T. Cipra, J. Form~nek, J. Kub~tt and L. Tom~ek, Spectral analysis of EEG signals, in R. Sinz (Ed.), Psychophysiology 1980, Proc. of the 22nd Int. Congr. of Psychology, Gustav Fischer Verlag, Leipzig, Elsevier North Holland, Amsterdam, 1982, pp. 157-158. R.G. Bickford et al., The compressed spectral array (CSA) - A pictorial EEG. Proc. San Diego Biomed. Confer., 11, 1972, pp. 365-370. J. Dvoi'~k, J. Form~nek, J. Kub~t, J. Plevov~, M. Vani~kov~, M. Fire~, J. And~l, T. Cipra, L. Tom~§ek, Z. Pr~§kov~, E. Holoubkov~ and Z. Fabi~in, A principal component model of the time series of the broad-band EEG frequency spectra and of the EEG spectral power densities, in L. Fedina, B. Kany~r, B. Kocsis and M. Kollai (Eds.), Advanc. Physiol. Sci. Vol. 34. Mathematical and Computational Methods in Physiology, Akaddmiai Kiad5, Budapest, 1980, pp. 301--310. J. Form~inek and J. Dvorak, Broad-band EEG frequency analysis, in M. Horv~th (Ed.), Adverse Effects of Environmental Chemicals and Psychotropic Drugs, Vol. 1. Elsevier, North Holland, Amsterdam, London, N.Y., 1973, pp. 177-184. J. Form~nek, E. Frantlk, J. Plevov~i, M. Vani~kov~ and E. Holoubkov~, The effect of carbon disulphide, carbon monoxide and of barbiturates on the EEG frequency spectra in rats, in E. Klimkov~-Deutschov~ and E. Luk~t§ (Eds.), Proc. 2nd Int. Industrial and Environmental Neurology Congress, Charles University, Prague, 1976, pp. 83-92. A. Isaksson, A. Wennberg and L.H. Zetterberg, Computer analysis of EEG signals with parametric models. Proc. IEEE, 69 (1981) 451. B.L. Johnson, Electrophysiological methods in neurotoxicity testing, in P.S. Spencer and H.H. Schaumburg (Eds.), Experimental and Clinical Neurotoxicology, Williams and Wilkins, Baltimore, 1980, pp. 726--742. S.M. Kay and S.L. Marple, Spectrum analysis - a modern perspective. Proc. IEEE, 69 (1981) 1380. V. Kraj~a, Automatic classification of EEG segments and extraction of representative ones by dynamic clusters method. Act. Nerv. Super., 26 (1984) 118. J. Kub~tt, J. Dvorak, M. Vani~kov~i,M. Fire§, J. And~l, T. Cipra, L. Tom~§ek, Z. Fabian and J. Form~nek, Effect of carbon monoxide on the EEG spectral power densities in an animal model, in Abstract of papers, 28th Int. Congress of Physiological Sciences, Budapest, 1980, p. 531. J. Kub~t, The influence of supralethal 6°Co irradiation on some biophysical phenomena of the brain functions (EEG) in dogs. Thesis for the PhD degree, Prague, 1986, 235 pp. (in Czech). P. Matthis, D. Scheffner and C. Benninger, Spectral analysis of the EEG: Comparison of various spectral parameters. Electroencephalogr. Clin. Neurophysiol., 52 (1981) 218. D. Michael and J. Houchin, Automatic EEG Analysis: A segmentation procedure based on the autocorrelation function. Electroencephalogr. Clin. Neurophysiol., 46 (1979) 232. H.M. Praetorius, G. Bodenstein and O.D. Creutzfeldt, Adaptive segmentation of EEG records: A new approach to automatic EEG analysis. Electroencephalogr. Clin. Neurophysiol., 42 (1977) 84. P.F. Prior, The EEG in acute cerebral anoxia, Excerpta Med., Amsterdam, 1973, 314 pp. WHO -- IPCS, Principles and methods for the assessment of neurotoxicity associated with exposure to chemicals. Environmental Health Criteria 60, WHO Geneva 1986, 180 pp.