Long-term effects of an organophosphate upon the human electroencephalogram

TOXICOLOGY

AND APPLIEDPHARMACOLOGY

47,161-176(1979)

Long-Term Effects of an Organophosphate upon the Human Electroencephalogram FRANK

H. DuFFY,‘,~

JAMES L. BURCHFIEL,’ PETER H. BARTELS,~ MAURICE GAON,~ AND VAN M. SIMS

Seizure Unit Neurophysiology Laboratory, Department of Neurology, Children’s Hospital Medical Center and Harvard Medical School, Boston, Massachusetts 02115; Department of Microbiology and Optical Sciences Center, University of Arizona, Tucson, Arizona; and Biomedical Laboratory, Edgewood Arsenal, U.S. Army Aberdeen Proving Ground, Maryland 21010 Received May 5, 1978; accepted June 15, 1978 Long-term Effects of an Organophosphate upon the Human Electroencephalogram. F.H., BURCHFIEL, J.L., BAR'IZLS, P.H.,GAoN,M., AND SIM, V. M. (1979). Toxicot. Appl. Pharmacol. 47,161-176. The brain electrical activity of workers exposed to the organophosphate compound (OP), sarin, was compared to that of control subjects. Exposed workers had a history of one or more documented accidental exposures to toxic levels of sarin. However, no exposed subject had exposure within 1 year of his examination. The comparison included standard clinical electroencephalograms (EEGs), computer-derived EEG spectral analysis, and standard overnight sleep EEGs. It was not possible to diagnose subjects individually by expert visual inspection of their EEGs. However, statistically significant between-group differences for both the visually inspected and computer-derived data were reported by both univariate and multivariate statistical methods. Different EEG changes revealed by visual inspection and computer-derived spectral analysis appear to reflect the differing sensitivites of these two analytic techniques. Statistically significant group differences included increased beta activity, increased delta and theta slowing, decreased alpha activity, and increased amounts of rapid eye movement sleep in the exposed population. It is suggested that the above findings represent an unexpected persistence of known short-term OP actions. It is also suggested that these results, when taken along with the reported long-term behavioral effects of OP exposure, provide parallel evidence that OP exposure can produce long-term changes in brain function. DUFFY,

Organophosphate (OP) compounds are known to disrupt nervous system function in animals and humans. A biocidal activity, first recognized in 1937, led to their development by Germany during World War II as I Seizure Unit Neurophysiology Laboratory, Department of Neurology, Children’s Hospital Medical Center and Harvard Medical School, Boston, Mass. 02115. * To whom requests for reprints should be sent. 3 Department of Microbiology and Optical Sciences Center, University of Arizona, Tucson, Ariz. 85721. 4 Biomedical Laboratory, Edgewood Arsenal, U.S. Army Aberdeen Proving Ground, Md. 21010. 161

potential chemical warfare agents, e.g., the “nerve gases” soman, sarin, and tabun (Koelle, 1968). More recently OP compounds less toxic to humans have been developed and widely used as pest control agents, e.g., malathion, parathion, mipafox. Casida and Baron (1976) conservatively estimate an annual worldwide production exceeding lo6 kg. Their toxic properties appear related to an ability to inhibit esterases throughout the central and peripheral nervous systems (Davies et al., 1960; Johnson, 1974, 1975). All OP compounds share similar chemical 0041-008X/79/010161-16102.00/0 Copyright 6 1979 by Academic Press, Inc. All rights of reproduction in any form reserved. Printed in Great Britain

162

DUFFY ET AL.

structures and pharmacological actions. They differ mainly in regard to potency and use (Casida and Baron, 1976). Exposure levels that are toxic but too low to threaten life are known to produce a variety of signs and symptoms in humans including miosis, muscular fasciculations, and apprehension (Koelle, 1968; Wadia et al., 1974). These acute manifestations, particularly those involving behavioral alterations, are usually accompanied by marked desynchronization of the electroencephalogram (EEG), a triad of changes consisting of increased high-frequency activity, decreased low-frequency activity, and lowered background voltage (Burchfiel et al., 1976). At higher dose levels, OP compounds can produce convulsions, muscular paralysis, and death (Koelle, 1968). High OP levels cause slowing of the EEG followed by the development of spike wave dischargeswhich accompany clinical convulsions (Burchfiel et al., 1976). Symptomatic recovery from OP exposure is normally complete within 2 to 9 weeks at which time the erythrocyte cholinesterase level usually has returned to normal (Grob and Harvey, 1958; Bowers

OP exposure or a limited period of low-level exposure. Minor disorders of affect, emotion, and memory, lasting more than 6 months, were reported by Tabershaw and Cooper (1966) in 38 % of 114 subjects after acute OP poisoning. Rowntree et al. (1950) suggested that OP exposure might exacerbate preexisting psychiatric problems. Metcalfe and Holmes (1969) suggested that OP exposure may lead to chronic EEG changes. They reported that workers with past histories of both OP and chlorinated hydrocarbon exposure, but with no recent exposures, had EEGs which showed a high incidence of excessiveslowing during drowsinessand after hyperventilation. All-night-sleep EEGs showed a high incidence of “narcoleptic sleep records.” Psychological dysfunctions in this group included disturbed memory and difficulty maintaining alertness and appropriate focusing of attention. Recently, we reported the results of a study designed to examine, under controlled conditions, the long-term effects of OP exposure upon the primate EEG (Burchfiel et al., 1976). We found that a single symptomatic exposure or a series of subclinical exposures to the OP et al., 1964). sarin altered the EEG frequency spectrum Sustained low-level administration of OP for up to 1 year. Fast activity (beta) was compounds can produce symptoms and significantly increased in exposed but not in signs not seen following single exposures to control monkeys. the same dose level. For example, subjects On the basis of the above studies, we given daily injections of diisopropyl fluoroundertook an investigation of a population phosphate (DFP) showed the additional of industrial workers with histories of accisymptomatology of insomnia, excessive dental exposure to sarin. Their waking and dreaming, emotional lability, increasedlibido, sleep EEGs were statistically compared to paresthesias,visual hallucination, and tremor those of a matched but unexposed control (Grob et al., 1947). However, such sympto- population and the results are reported in this matology usually disappears shortly after communication. cessation of exposure. In most instances, EEG abnormalities are reported to disappear METHODS within 2 weeks of an acute exposure or Subjects termination of chronic exposure (Grob, Seventy-seven industrial workers with histories of et al., 1947; Holmes and Gaon, 1956; accidental exposure to sarin were studied. This shall Santolucito and Morrison, 1971). There are, however, a number of studies be referred to as the exposure (E) group. A subgroup of 41 E-group workers had histories of three or more which suggest that some subjects may exposures within the 6 years preceding the study. This experience long-term sequelaefrom a single subgroup shall be referred to as the maximum

PERSISTENT

ORGANOPHOSPHATE

exposure,(M) group. No subject in this study had a symptomatic exposure, a questioned exposure, or reduced cholinesterase activity during the year immediately preceding the initiation of the study. No member of E group had a history of exposure to chlorinated hydrocarbon or carbamate compounds. For the purpose of this study, an “exposure” is defined as (1) verified history of discrete exposure, (2) resultant clinical signs and symptoms consistent with exposure, and (3) reduction in erythrocyte cholinesterase to a level at least 25 % below the individual’s pm-exposure baseline. Every “exposure,” as defined for the purpose of this study, was associated with equipment failure, operator error, or other industrial accident. Thirty-eight industrial workers from the same plant comprised a control (C) group. No member of this group had access to OP compounds at work; had a history of OP, chlorinated hydrocarbon, or carbamate exposure; had ever complained of symptoms associated with OP exposure; or had ever shown significant fluctuations in semiannual or spotcheck cholinesterase activity. Members of C group worked as janitors, clerks, or guards or worked with chemicals having no known effects on the central or peripheral nervous .System. Stringent safety precautions were taken throughout the entire facility to prevent continual low-level exposure. There was no evidence to suggest that any workers of the C or E group were so exposed. Members of C group were chosen so that their age distribution, sex, and socioeconomic background matched those of the E group. All subjects in both groups were males and received similar salaries. Ages ranged from 26 to 58 with means of 42 for the C group, 46 for the E group, and 44 for the M group. Studies

.

Three separate but related investigations were undertaken: (1) spectral analysis of tape recorded EEG, (2) visual inspection of routine clinical EEG, and (3) visual inspection of an all-night-sleep EEG. Spectral analysis. Subjects were studied in a standard EEG laboratory. EEG activity was recorded on an Ampex FR 1300 tape deck via a Grass Model 78 polygraph with a bandpass of 0.1 to 35 Hz. A simultaneous strip chart was made to monitor for artifact and to assess level of consciousness. Subjects were studied in the following states. (a) Eyes open, EO (2 min), awake with eyes open in a lighted environment and frequently alerted by an auditory stimulus. (b) Eyes closed, EC (2 mitt), same as for (a), but with instructions to close eyes. (c) Drowsy, D (15 min), drowsiness and sleep. Room lights were turned off and extraneous sounds were masked with white noise. One minute of this

EFFECTS

ON

HUMAN

EEG

163

15min period was chosen to represent drowsiness on the basis of the EEG strip chart. Drowsiness was defined as the state beginning after the breakup of alpha activity and ending with the early signs of vertex waves and spindling. (d) Hyperventilation, HV (5 min), 5 min of deep breathing with eyes open and room lights on. The third minute was chosen to represent the HV state. (e) Recovery from hyperventilation pHV (3 miu), the third minute following cessation of HV was used to represent the FHV state (eyes open and room lights on). Off-line data analysis was performed on the EEG activity recorded from the following bipolar linkages (10-20 international electrode placement): 0,-C, (occipital-central), F,-C, (frontal-central), and T,T, (temporal). Spectral analysis was accomplished on a PDP-12 computer (Digital Equipment Corp.) using the comprehensive SIGSYS-12 software system (Agrippa Data Systems, Boston). Signals were played back from FM tape through Khrone-Hite filters set to bandpass 0.5 to 35 Hz (24 db/octave) and analogto-digital converted by the computer at a rate of 128 Hz. Fast-Fourier transforms (FFTs) were performed on 4-set epochs of EEG (computed on 512 points over the frequency range t%64 Hz, with a resolution of a Hz). After FFT computation, the square root was taken of each frequency component to yield a voltage spectrum. Epochs containing excessive eyeblinks, movement artifact, electrode artifact, or muscle activity were not analyzed. Consecutive 4-set epochs were averaged to yield a spectrum representing 1 or 2 min of EEG, i.e., the spectra of 15 or 30 4-set epochs were averaged. Two-minute averages were computed for states*EO and EC. A I-min average was computed for states D, HV, and EHV. For each spectrum, a calculation was made of the energy present in each of 10 3-Hz-wide segments over the total range of 0.5 to 30 Hz. Segmental averaging was done after converting to microvolts to permit additivity. These segments may be related to the standard EEG bands as follows: Delta = 0.5 to 3 Hz, slow theta = 3 to 6 Hz, fast theta = 6 to 9 Hz; alpha = 9 to 12 Hz, slow beta = 12 to 15 Hz, beta = 15 to 30 Hz. The band from 30 to 64 Hz was not analyzed. For each subject frequency spectra were formed for all possible combinations of state and linkage. To facilitate analysis, the energy within each frequency band was averaged across all subjects within a group for each state and linkage. Variance and SD were also calculated. Initial comparisons were made between groups (for each state and linkage) on the basis of two statistical tests: Student’s t test and Mann-Whitney U test. For a test to be considered valid, it had to achieve the 0.05 two-tailed level of significance. The nonparametric U test is sensitive to

164

DUFFY

the relative rank of each member of a group but not to its absolute value. Chical EEG. Routine clinical EEGs were performed using a Grass Model 78 EEG with a complete 24-electrode montage. Both monopolar and bipolar linkage were examined. In addition to the standard declinical “reading,” the electroencephalographer signed as “normal” or “abnormal” 19 separate parameters. The numbers of exposed workers or control workers within the normal or abnormal category for each parameter were placed in a 2 x 2 contingency table. Statistical comparison was made by Fisher’s test (Fisher, 1934). At the time of EEG reading, the electroencephalographer was unaware of the subject’s exposure history. SIeep EEG. An all-night-sleep EEG was recorded from each subject during his usual sleep period. Three bipolar linkages of EEG (Oz-Cz, F3-F4, F4-C4), an electromyogram (EMG) of neck muscle tone, an electrooculogram (EOG) for eye movement potentials, an electrocardiogram (EKG), and a bed movement transducer were recorded on the paper strip chart of a Grass Model 78 polygraph. These records were then scored by the Sleep Laboratory of the University of Maryland, Baltimore, Maryland. Absolute and percentage times in rapid eye movement (REM), stage 1, stage 2, stage 3, and stage 4 sleep were calculated for each subject. Comparisons between groups were made by the I test and U test. Not every subject studied with spectral analysis and daytime clinical EEG had an all-night-sleep study. Furthermore, we excluded from the latter those few subjects below age 30 so as to render the C, E, and M populations more comparable (Williams et al., 1974; Brezinova, 1975). For the sleep analysis, there were 26 members in the C group, 67 in the E group, and 38 in the M group. In addition to the conventional group comparisons of individual parameters as outlined above, we wished to consider the combined effect of the mutual interdependence of all variables upon group differences. Accordingly, we employed the standard multivariate technique of discriminant analysis (Cooley and Lohnes, 1971) to the problem of establishing a single statistically quantifiable measure of group difference. Here, we calculated Wilks’ 1 to assess the significance of group separation. To reduce the dimensionality of the data representation prior to discriminant analysis, we submitted all variables, or “features,” to a merit rating process. The “merit function” is based on (1) the computation of an ambiguity function (Genchi and Mori, 1965) and (2) the computation of a measure of detectability from the receiver operating characteristic curve (Selin, 1965). Furthermore, the merit function takes into account the mutual correlation of all features selected by the first two criteria.

ET AL.

RESULTS For all analyses, control subjects were compared to all the exposed subjects (C x E analysis) and again to the subgroup of subjects with multiple (three or more) exposures (C x M analysis). We considered doing a regression analysis but sufficient medical information to permit quantification of an individual’s exposure history was not available. Spectral Analysis Results of the spectral analysis are shown in Tables 1, 2, and 3 for the C, E, and M population, respectively. Results of the univariate C x E and C x M comparisons are summarized in Tables 4 and 5. Univariate analysis. The C x E analysis (Table 4) showed only eight significant comparisons. This raises the important question of how many individual comparisons must be statistically significant to have a “physiological significance.” For example, there are 50 comparisons-5 states and 10 bands-for the C x E analysis of F3-C3. At the 0.05 level, 0.05 x 50 or 2.5 comparisons might be expected to show significance by chance. As only eight comparisons in the entire C x E comparison showed significance, these could have arisen by chance alone and have no definite physiological significance. However, since five comparisons within the Oz-Cz linkage were statistically significant and since two of all the Cx E comparisons reached the 0.02 level, it is unlikely that chance was a major factor. Furthermore, six of the eight significant comparisons represented increases in beta activity (1230 Hz) making the likelihood of a physiologically significant difference between the C and E groups more secure. On the other hand, the C x M analysis (Table 5) shows a more definitive picture. There are 14 significant comparisons for T3-T5, 5 for F3-C3, and 14 for Oz-Cz. Clearly, we have exceeded the “chance level” for all linkages (especially T3-T5 and

PERSISTENT

ORGANOPHOSPHATE

Oz-Cz) and the entire Cx M analysis (33>7.5). Note that 13 of the 24 Cx M comparisons were significant at the 0.02 level, 6 at the 0.01 level, and 1 at the 0.001 level. The physiological significance is increased by the clear tendency for the results to cluster in the beta range and to represent an increase of beta in every case. Thus, the results of the univariate spectral analysis show (a) probable differences between the C and E populations, (b) highly significant differences between the C and M populations, and (c) a tendency of the differences to cluster in the beta range and represent an increase in beta within the exposed group. Multivariate analysis. The merit function values for the selection of best features for discrimination were computed for all 150 features available per subject (10 spectral bands x 5 states x 3 electrode pairs). The percentage time spent in REM sleep was added as the 15lst feature. The 10 best features were chosen for inclusion in the discriminant analysis. All had highly significant differences for the comparison between controls and subjects who suffered maximum exposure. The data show that the spectral values in the beta frequency range contribute most to the discriminating function. All of the selected features showed approximately the same high correlation with the discriminant function. The overall group separation is measured by the value of Wilks’ 1, which was found to be 0.644 at 10 and 93 df, respectively (C x M), for the data without the REM feature, and 0.620 at 9 and 94 df for the data set including the REM sleep feature. This statistic is also known as “determinant ratio” and since the determinant of a multivariate distribution expresses the scatter, Wilks’ I may be seen as a multivariate analog to the F ratio in univariate statistics, which is a variance ratio. The asymptotic distribution of Wilks’ I can be approximated by the F distribution. The value for Wilks’ 1 for the C x M comparison, with and without the sleep feature, rendered a

EFFECTS

ON

HUMAN

EEG

165

highly significant F ratio, p = 0.00001. Very similar data, with only a slightly decreased level, were obtained for the C x E discrimination, p = 0.001. Thus, the results of the multivariate spectral analysis showed that there were significant differences (a) between the C and E populations and (b) significant differences between the C and M populations and (c) that beta frequencies contribute most heavily to the differentiation. Visual Inspection of Clinical EEG Table 6 shows the results of the C x E and C x M analyses for the parameters designated normal or abnormal by the electroencephalographer. Note the significant results for the Cx E and Cx M comparisons in the Background Rhythms column. This indicates that EEG background activity was more likely to deviate from normal in the E group, and especially in the M group, than in the C group. As the High and Low columns failed to show significant levels, the background EEG abnormalities in the E and M groups were not consistently toward either more slow or more fact activity. However, the significant values in the Delta and Theta columns indicate that more E- and M-group EEGs showed excessive amounts of slow activity at some time during the record than C-group EEGs. Furthermore, the Alpha column revealed more EEGs with diminished or absent alpha activity in the E and M populations. No other columns showed significant values. There were no more significant numbers of abnormal EEGs in the E and M groups than in the C group (Entire Record). Also, the electroencephalographer was not able to place EEGs into their correct group (Doctor’s Guess). He was able to correctly categorize 47 % of the C-, 51% of the E- and 56% of the M-group EEGs. The overall classification success for the electroencephalographer was 49.6x, clearly at the chance level. The large number of EEG records read as “abnormal” deserve comment. Note that

166

DUFFY

ET AL.

TABLE MEAN

1

VALUES OF SPECTRAL ENERGY:

CONTROL

POPULATION”

Spectral

0.5-3

3-6

1 9.902

2 6.550

13 8.872 ~__

14 6.885

-T3-T5

EO -EC

D

HV

25 8.874 ~~

26 6.756

37 20.551

38 13.274

49 9.984

50 7.544

$HV -F3-C3

, ---

6-9 -------3 5.918 ~~ 15 9.023 --

band

(Hz)

9-12

12-15

15-18

18-21

4 6.971

5 5.349

6 5.165

7 5.161

16 12.118

17 5.425

18 5.099

21-24

24-27

27-30

8 5.004

9 4.479

10 4.238

19 4.800

20 3.706

21 3.124

22 2.921

30 3.846

31 3.504

32 2.924

33 2.521

34 2.284

--

27 7.045 ~___

28 7.954

29 4.560

39 113.919 -- ~-

40 15.666

41 7.922

42 7.088

43 6.320

44 5.573

45 5.101

46 4.753 -__

51 9.432 -- -------62 63 6.977 5.603 -- -74 75 6.389 6.566 __~

52 13.566

53 6.325

54 5.230

55 4.775

56 3.999

57 3.481

58 3.182

64 6.466

65 4.250

66 4.064

67 4.269

68 3.798

69 3.213

70 2.891

76 8.269

77 4.618

78 4.276

79 3.876

80 3.201

81 2.695

82 2.367

88 6.416

89 4.336

90 3.647

91 3.261

92 2.807

93 2.344

94 1.977

100 10.469

I01 6.079

102 5.490

103 4.602

104 3.458

106 2.345

112 9.557

113 5.725

114 4.606

115 4.365

116 3.813

105 2.745 -___ 117 3.240

124 11.451

125 7.076

126 5.874

127 5.141

128 4.069

138 7.361

139 6.650

140 4.909

--

~-

EO

61 10.917

EC

73 9.323

D

85 9.544

86 6.752

97 22.945

98 13.157 110 6.982

EO

109 10.776 -~ 121 15.037

122 12.068

EC

133 13.866

134 13.180

135 16.039 __~

136 23.183

137 8.768

146 14.726

149 7.871

150 6.063

151 5.337

152 4.704

158 21.392

160 28.432

161 12.717

162 10.110

163 8.451

164 5.778

I>HV

169 15.078

170 13.309

147 1 4.033 ~__ 159 2 4.514 ___~ 171 I16.815

148 14.430

HV

145 14.508 -~ 157 27.630

172 24.305

173 10.444

174 8.088

175 7.099

176 5.601

HV

@HV

oz-cz

--

EEG

D

87 6.232 -~____ 99 1 1.653 __~ 111 7.501 -~-----123 10.818 __~

-~

---

129 3.335 ___~ 141 3.857 -___ 153 3.885 165 4.313 ~___ 177 4.410

118 2.737 130 2.860 142 3.214 154 2.980 166 3.681 178 3.638

’ This table gives the mean values for the integrated spectral energy content for various spectral EEG bands for differing electrode montages and subject states for the control (C) subject group. The columns refer to spectral bands from 0.5 to 30 Hz. The rows refer to particular electrode montages and subject states. T3-T5 = temporal electrode montage; F3-C3 = frontakentral montage; OZ-CZ = occipital-central montage; EO = eyes-open, alerted subject state; EC = eyes-closed state; D = drowsy state; HV = hyperventilation state; FHV = recovery from HV state. The feature number from 1 through 177 is given. Missing entries refer

PERSISTENT

ORGANOPHOSPHATE

EFFECTS

TABLE

I

MEAN

T3-T5

EO

VALUES OF SPECTRAL ENERGY:

EXPOSED POPULATION’

Spectral 3-6

6.9

9-12

1 8.730 ___-

2 6.329

3 6.448

4 7.021

13 8.988

14 6.927

15 9.764

16 11.803

25 8.698

26 7.374

27 7.745

37 19.550

38 13.260

EEG

18-21

6 6.121

7 6.417

18 5.815

28 8.625

29 5.158

40 13.767

24-27

27-30

8 6.263

9 5.813

10 5.324

19 5.417

20 4.232

21 3.655

22 3.362

30 4.601

31 4.229

32 3.701

33 3.240

34 2.878

41 8.317 -~---___

42 7.466

43 6.691

44 5.952

45 5.427

46 5.263

49 9.898

50 7.515

51 9.858

52 12.310

53 6.559 ------

54 5.745

55 5.336

56 4.537

57 4.137

58 3.650

EO

61 9.426

62 6.438

63 6.033

64 6.520

66 4.074

67 4.212

68 3.911

73 9.887

74 6.590

75 7.212

76 8.278

78 4.177

80 3.492

82 2.344

D D

85 9.530

86 7.172

87 6.533

88 6.873

90 3.845

79 4.102 -__ 91 3.544

69 3.411 ___81 2.830

70 3.088

EC

92 3.171

93 2.626

94 2.145

97 21.098

98 13.423

102 5.366

103 4.767

104 3.952

105 3.135

106 2.582

109 10.449 --

110 7.039

111 8.061

112 9.254

65 4.124 -~77 4.734 -____ 89 4.732 --101 6.170 -113 5.534 ------

114 4.729

115 4.563

116 3.943

117 3.222

118 2.695

121 14.623 -133 14.434 --

122 12.196

123 11.852

124 11.532

125 7.354 ---______-

126 6.390

127 5.390

128 4.343

129 3.637

130 2.855

134 13.382

135 18.294

136 23.918

138 8.246

139 7.433

140 5.548

141 4.329

142 3.340

145 15.296 ___157 26.902 -169 15.294

146 16.228

147 16.095

148 16.497

137 10.273 -______ 149 9.225

151 6.172

152 5.364

153 4.409

154 3.363

158 22.429

159 25.202

160 27.805

161 14.540 ---~

150 6.918 -___ 162 11.071

163 8.847

164 6.570

165 5.090 __-

166 3.986

170 13.848

171 18.966

172 24.403

173 11.776

174 9.002

175 8.054

176 6.454

177 5.137

178 3.954

:HV --

-HV

SHV

oz-cz

21-24

-39 13.44

--

F3-C3

(Hz)

15-18

-HV

band

12-15 -----5 5.811 --___--17 6.046 -___--

-D

167

EEG

2

0.5-3

EC

ON HUMAN

EO

EC

D

HV

CHV

99 12.20

D This table gives the mean values for the integrated (E). The conventions are the same as for Table 1.

100 9.649

spectral

energy

--

--

content

---

for the entire

,,

exposed

population

to computational entries not analyzed or reported. Below each feature number is shown the summed spectral energy value for the particular spectral band, electrode montage, and subject state. These represent the mean values for the C group.

168

DUFFY

ET AL.

TABLE MEAN

VALUES

OF SPECTRAL

ENERGY:

MAXIMALLY

Spectral

0.5-3

T3-T5

(

3-6

(

6-9

EEG

Band

POPULATION’

(Hz)

12-15

15-18

18-21

21-24

24-27

27-30

-- -------

2 6.459

3 6.396

4 7.558 --~

5 6.251

6 6.789

7 7.093

8 6.847

9 6.225

10 5.729

13 8.936

14 6.919

15 9.800

16 12.294 --

17 6.433

18 6.506 -___-__-__-

19 6.206

20 4.728

21 4.131

22 3.936

25 8.880 ____

26 7.677

27 7.862

28 9.036 -~

29 5.577

30 4.960

31 4.412

32 3.786

33 3.280 __-

34 3.077

37 19.910 -__~ 49 10.099 ---

38 13.101

39 14.075

40 13.935 --51 52 13.265 10.035 -- ___------

41 8.876

42 8.409

43 7.706 -__

44 6.800

45 6.276 -__-

46 6.194

53 7.118

54 6.407

55 6.064

56 5.136

57 4.804

58 4.267

EO

61 10.033 _____

62 6.602

63 5.878

64 6.808 _-_____

65 4.036

66 4.032

67 4.209

68 3.755

69 3.151

70 2.774

EC

73 10.245 ___-

74 6.591

75 7.026

76 8.668 _-__~

77 4.656

78 4.326

79 4.370

80 3.629

81 2.887

82 2.393

D

85 9.862 --c_

86 7.358

87 6.570

88 7.141

89 5.070

90 4.082

91 3.706

92 3.286

93 2.753 -___

94 2.329

102 5.748

103 5.785 -_____

104 4.311

105 3.362

106 2.776

113 5.448

114 4.686

115 4.641

116 3.931

117 3.123

118 2.523

124 125 12.547 7.464 ___--_________ 137 136 25.129 10.321 ______ 148 149 17.397 10.018 --__I 160 161 28.222 14.559 ______

126 6.651

127 5.692

128 4.535

130 3.031

138 8.599

139 8.022

140 5.845

150 7.371

151 6.643

152 5.770

129 3.784 -___ 141 4.576 ____153 4.775

154 3.624

162 11.554

163 9.498

164 6.974

165 5.416

166 4.312

174 9.415

175 8.546

176 6.858

177 5.523

178 4.176

-D

HV

EHV

50 7.642

_---

97 22.385

98 13.557

99 12.109

100 10.014 _-______

110 7.065

111 7.723

112 9.839 -- -------

EO

109 10.653 --121 15.072

122 12.445

123 11.93

EC

133 14.378

134 13.237 --__-146 16.629

135 18.08 147 16.32

HV

EHV

oz-cz

9-12

EXPOSED

1 9.056 ___-

EO

EC

F3-C3

3

D

145 15.468 -__-I_

HV

157 27.508

158 22.304

159 25.89

pHV

169 14.582

170 13.448

171 18.67

172 15.842

101 6.274

173 12.245

‘This table gives the mean values for the integrated spectral population (M). The conventions are the same as for Table 1.

energy

content

--

for

the maximally

142 3.584

exposed

PERSISTENT

ORGANOPHOSPHATE

EFFECTS

TABLE

ON

HUMAN

169

EEG

4

Cx E ANALYSISOF SPECTRA' Spectral EEG band (Hz) 0.5-3 1 3-6

T3-TS

EO

----

1 6-9

) 9-12

1 12- 15 1 15- 18 1 18- 21 1 21-24 1 24-27 1 27-30

-----___-

--

___-

EC D HV FHV F3-C3

EO

--$ T* --

-----_____

EC D HV

oz-cz

-------EO

---

EC D

--

-_

---

~~

TV

HV jiHV

--

tu

fT

-tT*V*

-t TV

@This table summarizes the univariate statistical analysis of the spectral energy content for the comparison of the control group (C) with the entire exposed population (E). Labeling of the columns and rows is the same as for Table 1. Each entry in the table signifies a statistically significant difference in the mean spectral energy content for that particular spectral band, electrode montage, and subject state between the C and E groups. The absence of an entry signifies the lack of a significant difference. T = a significant parametric Students t test; V = a significant nonparametric Mann-Whitney U test (two-tailed). The absence of an asterisk signifies statistical significance at the 0.05 level. The presence of an asterisk(s) signifies statistical significance at the *0.02, **O.Ol, and ***O.OOl levels.

29% of the E records (22 of 77) and M records (12 of 41) were read as abnormal. However, as 24% of the C records (9 or 38) were also “abnormal,” the C x E and C x M comparisons were not statistically significant. It is not clear why so many records in all groups were “abnormal” in comparison to the electroencephalographer’s hypothetically

normal model. Possibly the electroencephalographer overread or possibly some unknown environmental factor was present, e.g., effect of living at a 1500- to 2000-m elevation. Whatever the explanation, the effect was equally distributed across all groups. The essenceof the experimental method used in this study involved the comparison of groups

170

DUFFY

chosen to be as similar as possible with the exception of a single factor, in this case, OP exposure. Thus any statistically significant differences between the EEG of the groups could be argued to be attributable to this single factor. The effect of other factors (reader bias, etc.) should be randomly distributed across all groups. Thus the visual inspection of the clinical EEG revealed more EEGs in the E and M groups containing (1) diminished alpha, (2) TABLE C

x M

ANALYSIS

ET AL.

increased epochs of slow deviation of background faster or slower rhythms. were not clearly diagnostic electroencephalographer’s cess was at chance level.

activity, and (3) activity to either However, these features since the classification suc-

Visual Inspection of Sleep EEG Table 7 shows the results of the C x E and C x M analyses of the all-night-sleep EEG records. Note the highly significant increase 5 OF SPECTRA

Spectral EEG band (Hz.) 10.5-3 1 3-6 1 6-9 1 9-12 1 12-15 1 15-18 1 18-21

------~

T3-T5

EO

-___

--

EC

f TV

T TV

---

24-27

TT**u

f T

TT**V*

f T*V

t TV

f T*U

tT*lJ

f T*V

t T*U --

t T*

___---

D

21-24

___--

27-30

1 f T*V

tT

HV FHV F3-C3

EO EC D HV

---~ ___-

-~

~~__~ 1‘T

~~-~

t T**U*’

--__-

SU --t T**U**

t V

@HV ---OZ-CZ

EO

-~-__

EC

t TV

__~

D --

HV ijHV

--

tu

f TV*

f T*V

f T*V* t T**U*

______-~ t TV

t T*V*

f T**V*

’ Results of the spectral analysis between the control subjects (C) and the maximally exposed subjects (M) with histories of three or more exposures. Convention is the same as in Table 1 and Table 4. Note the clustering of significant results in the beta range. Note also that all changes represent an increase in beta for the maximum exposure group.

PERSISTENT

ORGANOPHOSPHATE

EFFECTS

TABLE HUMAN

C

E CxE M CxM

C E CxE M CxM

EEG

READINGS:

ON

HUMAN

171

EEG

6 EXACT

Posthyperventilation

PROBABIL~Y

TEST’

Waking

Drowsy

Sleep

Hyperventilation

Strobe

Background voltage

Background rhythms

33/5b 61/16 0.13 30/11 0.07

3315 57120 0.06 30/l 1 0.07

3612 67110 0.13 3714 0.26

28110 58119 0.18 3219 0.19

3414 7017 0.25 3813 0.27

38/O 7314 0.20 3912 0.27

3117 66/l 1 0.18 33/8 0.22

Parox. Activity

Focal Activity

Asymmetries

Delta

Theta

Alpha

23115 44133 0.15 25116 0.18

3711 7512 0.45 3912 0.39

3612 76/l 0.22 W 0.36

3513 58119 0.02* 3219 0.06

27/l 1 39138 0.02* 21120 0.04*

2919 44133 0.02* 22119 0.02*

Background rhythms High

Low

3216 54123 0.05* 26115 0.02*

3214 54113 0.13 2619 0.07

3212 54110 0.11 2616 0.09

Beta

Doctor’s guess

Entire record

3216 58119 0.11 28113 0.06

18/20 39139 0.15 18123 0.17

2919 55122 0.16 29112 0.17

’ Results of analysis, by visual inspection of standard clinical EEG, on all subjects. C, E, M, C x E, and C x M have the same meaning as in previous tables. Results of the exact probability test are shown. Statistically significant levels are marked with an asterisk. * Normal/abnormal.

in absolute and percentage REM sleep for both Cx E and Cx M comparisons. There were no such changes for any single slow wave sleep state for combined stages 3 and 4 or for total sleep time. The decreased stage 2 sleep for the C x M percentage comparison suggests that the additional time spent in REM came at the expense of stage 2 sleep. Percentage REM for all three of our groups, including the controls, fell below published values for middle-aged men (Agnew et al., 1967; Williams et al., 1974). However, there are no absolute “normal” values for REM sleep. REM amount is variable and very sensitive to differing laboratory conditions. Therefore, as we have done, each investigator must establish his own control values for his own particular laboratory environment. Our control population values fell below values obtained in other laboratories for two probable reasons. First, REM is usually decreased during the first night of sleep in a laboratory situation, i.e., the first-night effect (Mendels and

Hawkins, 1968). Second, most REM tends to occur in the final third of a night’s sleep (Williams et al., 1974). All our subjects were awakened at 0530 hr, approximately 30 to 45 min before their usually waking time. Thus a significant portion of this lost sleep might have been REM. The results of the all-night-sleep study show that the entire exposed group showed an increase in REM sleep (compared to the controls) and that this was of slightly greater magnitude for the maximum exposure group. DISCUSSION The results of this investigation show that the workers in our study with histories of exposure to sarin have waking and sleeping EEGs that differ significantly from those of workers with no exposure history by both univariate and multivariate statistical analysis. The multivariate discriminant analysis shows that the control (C) group, when represented as a multidimensional vector,

NS 183.161

NY 46.996

NS

T** u** NS

196.790 193.913

39.078 43.305

Stage 2

34.865 46.318 T* II* 47.736

Stage 1

NS

NS 20.701

16.639 16.483

Stage 3

NS

NS 6.663

6.332 4.141

Stage 4

NS

NS 27.364

22.970 20.595

Stages 3+4

EEG”

NS

NS 305.261

289.857 304.194

Total Sleep

SLEEP

7

NS T** u**

NS

14.974

T** u*

15.645

15.732 13.967

Stage 1

11.414 15.117

REM

T* lJ**

NS 60.710

67.718 64.319

Stage 2

NS

NS 6.818

5.575 5.419

Stage 3

Percentage time

NS

NS 1.841

1.967 1.168

Stage 4

NS

NS 8.669

7.548 6.583

Stages 3+4

a Results of analysis of all-night-steep EEG on human subjects. C, E, M, C x E, and C x M have the same meaning as in previous tables. Time in each phase of sleep is shown in minutes and again as percentage of total sleep time. Statistical comparison was made by t test and U test. Asterisks show level of significance (two-tail test). NS, Nonsignificant. * pGO.05; ** pCO.02.

M CxM

C E CxE

REM

Absolute time (min)

HUMAN

TABLE

PERSISTENT

ORGANOPHOSPHATE

can be statistically separated from the entire exposure (E) group at the p = 0.001 level and from the maximum exposure (M) group at the p = 0.00001 level. Similarly, the univariate Cx E and C x M comparisons indicated statistically significant group differences, the natures of which lend themselves, directly, to physiological and pharmacological discussion. Results of the univariate analysis showed the E and M groups to differ from the C group in the following manner. (1) Spectral analysis : increased high-frequency activity (12-30 Hz, beta); (2) visual reading of standard EEG: decreased amounts of alpha (9-12 Hz), increased amounts of slow activity (O-8 Hz, delta and theta), and nonspecific abnormalities in the EEG background; (3) sleep EEG: increased amounts of REM sleep. The increase in delta and theta slowing seen by visual inspection were not noted by spectral analysis. It is likely that the electroencephalographer was influenced by transient periods of slow activity in the paper EEG record which, however, were not of sufficient duration to significantly influence the mean slow activity as measured by spectral analysis. This contention is supported by (1) the absence of generalized slowing of background rhythms by visual inspection (Table 6, Low column) and (2) reports by others of periods of EEG theta slowing in subjects with recent histories of OP exposure (Grob et al., 1947; Rowntree et al., 1950; Grob and Harvey, 1953, 1958; Metcalf and Holmes, 1969). A discrepancy for alpha activity between the visual and spectral results is similarly explained. On the other hand, the marked increase in beta (12-30 Hz) revealed only by spectral analysis probably reflects the opposite phenomenon. Low levels of beta activity are very difficult to perceive by visual inspection. However, beta is easily quantitated by spectral analysis. Thus spectral analysis and visual EEG reading have provided complementary evidence of the nature of the brain wave

EFFECTS

ON

HUMAN

EEG

173

differences between subjects exposed and unexposed to OP compounds. Despite the real statistical group differences, expert visual EEG inspection did not permit diagnosis of an individual subject. Furthermore, no single computer-derived parameter was sufficient to permit individual diagnosis by itself. However, preliminary studies in our laboratory suggest that the combinations of computerappropriate derived EEG parameters, when analyzed using modern automated classification techniques (Wied et al., 1968; Bartels and Wied, 1977), will permit individual subject classification as control or exposed. This is to be the subject of a future report. The increased beta activity noted in this study is of particular interest for similar increases in temporal lobe beta were also found in monkeys 1 year after sarin exposure (Burchfiel et al., 1976). The significance of this finding is not yet established. Beta activity may be due to any number of factors. It is commonly increased during drowsiness and sleep (Kiloh et al., 1972), by many drugs (Fink, 1969), and by aging (MundyCastle, 1953). Rarer causes are mental retardation (Kooi, 1971) organic brain lesions (Joffee and Jacobs, 1972; Nealis and Duffy, 1978), epilepsy (Nealis and Duffy, 1978), and emotional tension (Kamp et al., 1972). Many pharmacological agents are especially prone to produce beta as part of generalized desynchronization. These include some of the most potent psychoactive agents, DMT, mescaline, and e.g., amphetamine, LSD (Fink, 1969). OP compounds are also among those known to cause desynchronization and increased beta at subconvulsive levels (Grob et al., 1947; Grob and Harvey, 1953; Burchfiel et al., 1976). The significance of increased amounts of REM sleep is unclear. Of 20 or more pathological states, other than narcolepsy, associated with clinical derangements of sleep (Williams et al., 1974), almost all involve no change or a decrease in REM. In a review of REM sleep pharmacology, Hartman (1968)

174

DUFFY

ET AL.

One explanation may be that the low tissue listed over 20 drugs causing a depression in cholinesterase activity and resultant acetylREM (including barbiturates, common trancholine excess at time of exposure induce quilizers, mood elevators, alcohol, and amlong-term changes in synaptic morphology phetamine), but only four caused an increase in REM (reserpine, L-tryptophane, 5-OHor biochemical organization which do not tryptophane, and LSD). Thus, it is unusual completely reverse when tissue cholinesterase to find classes of drugs that increase REM. activity returns to normal. For example, Fenichel et al. (1972, 1974) found that OP compounds, however, form one such excessive acetylcholine concentrations could class. Hernandez-Peon and Chavez-Ibarra produce skeletal muscle necrosis in rats even (1963) and Hernandez-Peon (1965) were where cholinesterase systems were intact. A among the first to suggest that the sleep-wake cycle was partially under chohnergic control. central analog of this effect could be postulated where healing is not as rapid as reported More recently, several reports have demonstrated a cholinergic link in REM sleep in muscle (Marsden and Reid, 1961; Ariens mechanisms (Magherini et al., 1972; Thoden et al., 1969) when an excess of aceylcholine is et al., 1972; Matsuzaki et al., 1968; Henno longer present. We suggest that high acetylcholine concenricksen et al., 1972; Sitaram et al., 1976). It has been postulated that cholinergic trations at the time of exposure might produce long-term changes in the postsynaptic recepmechanisms are involved in REM initiation, tor which render it more sensitive to endobased upon the demonstration that cholingenous acetylcholine. We have recently esterase inhibitors, such as OP compounds, can elicit REM sleep in both jntact animals demonstrated the presence of such a and brain-stem preparations. mechanism in the hippocampus of the rat (J. L. Burchfiel, M. Duchowny, and F. H. It is possible, therefore, that the long-term After an EEG changes in this study represent the Duffy, unpublished observation). electrically induced seizure discharge we unexpected persistence of the well-known found that hippocampal pyramidal cells effects of acute OP exposure. What might the pharmacological mechanshowed prolonged hypersensitivity to iontoisms underlying this persistent effect be? phoretically applied acetyicholine. Symptoms and signs of acute OP exposure Another explanation may be that sarin has are believed to occur secondarily to reductions actions beside that of a potent anticholinin cholinesterase activity. However, as tissue esterase. Similar compounds, triorthocresylcholinesterase returns to normal at an phosphate (TOCP) and diisopropyl fluoroapproximate rate of 1% per day (Grob and phosphate (DFP), are known to produce Harvey, 1958; Milby, 1971), recovery should peripheral neuropathies of delayed onset. be complete 3 to 4 months following exposure. The primary lesion is axonal degeneration On the other hand, monkeys exposed to sarin with secondary demyelination. When the showed EEG abnormalities 1 year after peripheral nerves are involved, lesions are exposure (Burchfiel et al., 1976). Furthermore, always found in the spinal cord, often in the the abnormalities reported in the current brain stem, and occasionally in the hemistudy were found in subjects with normal spheres (Johnson, 1970, 1974, 1975). HowRBC cholinesterase values at the time they ever, sarin is not reported to produce delayed were studied and who were exposure free for neurotoxic responses. There is no evidence more than 1 year. Thus, evidence of CNS to suggest that the persistent CNS effects of abnormality attributable to OP compounds sarin in primates are related to a direct can be found in monkeys and humans at a action on central axons or myelin. time when tissue cholinesterase activity is A further explanation may relate to the probably within normal limits. recent finding that cardiovascular regulatory

PERSISTENT

ORGANOPHOSPHATE

mechanisms are under partial cholinergic control (Aoyagi et al., 1975). OP exposure may result in alterations in cerebral flow secondarily resulting in hypoxia unrelated to respiratory compromise. However, this would not explain why monkeys with multiple subclinical (low dose) exposures show as much long-term EEG change as monkeys with single massive exposures (Burchfiel et al., 1976). In any event, the long-term findings are also seen as part of the acute exposure syndrome. Thus, the pathogenesis of the long-term effects may well represent an unexpected extension of the primary drug effect the exact details of which do not appear to be completely established. Regardless of the pathogenic mechanisms, results of the current study confirm the ability of OP compounds to induce persistent abnormalities in the electrical activity of the human brain. Whether these long-term neurophysiological changes can be directly related to reported behavioral abnormalities is not known. On the other hand, the longterm behavioral changes and long-term electrophysiological abnormalities provide parallel evidence that OP exposure can produce long-term change in the brain function of monkeys and humans. ’ REFERENCES H. W., WEBB, W. W., and WILLIAMS, R. L. (1967). Sleep patterns in late middleage male: An EEG study. Electroencephalogr. Clin. Neurophsiol. 23, 168-l 7 1. AOYAGI, M., MEYER, J. S., AND DESHMUKH, V. D. (1975). Central cholinergic control of cerebral blood flow in the baboon. Effect of cholinesterase inhibition with neostigmine on autoregulation and CO, responsiveness. J. Neurosurgery43(6),689-705. ARIBNS, A. TH., MEETER, E., WOLTHUS, 0. L., AND VON BENTHEM, R. M. T. (1969). Reversible necrosis of the end plate region in striated muscles of the rat poisoned with cholinesterase inhibitors. Experientia AGNEW,

25, 57-59. BARTELS, P.

H., AND WIED, G. L. (1977). Computer analysis and interpretation of microscopic images: Current problems and future directions. Proc. IEEE 65, 252-261.

EFFECTS

ON

HUMAN

EEG

175

M. B., GOODMAN, E., AND SIM, V. M. (1964). Some behavioral changes in man following anticholinesterase administration. J. Nero. Merit. Dis.

BOWERS,

138, 383-389. BREZINOVA, V.

(1975). The number and duration of the episodes of the various EEG stages of sleep in young and older people. Elecfroeneephalogr. C/in. Neurophysioi.

39, 273-278.

BURCHFIEL, J. L., DUFFY, F. H., AND SIM, V. M. (1976). Persistent effect of sarin and dieldrin upon the primate electroencephalogram. Toxicol. Appl. Pharmacol. CASIDA, J.

35, 365-379.

E., and BARON, R. L. (1976). Recognition and overview of the organophosphate induced delayed neurotoxicity problem. In Pesticide Induced Delayed Neurotoxicity (R. L. Baron, ed.), pp. 7-23. U.S. Environmental Protection Agency EPA600/l-26-025, Research Triangle Park, N.C. COOLEY, W. W., and LOHNES, P. R. (1971). Multivariate Data Analysis. Wiley, New York. DAVIES, D. R., HOLLAND, P., AND RURENS, H. J. (1960). The relationship between the chemical structure and neurotoxicity of alkyl organophosphate compounds. Brit. J. Pharmacol. 15, 271-278.

FEN~CHEL,G. M., DETTBARN, W. D., AND NEWMAN, T. M. (1974). An experimental myopathy secondary to excessive acetylcholine release. Neurology 24, 41-45.

FENICHEL, G. M., KILBLER, W. B., OLSON, W. H., AND DE~BARN, W. D. (1972). Chronic inhibitioq of cholinesterase as a cause of myopathy. Neurology 22, 1026-1033. FINK, M. (1969). EEG and human psychopharmacology. Ann. U. Rev. Pharmacol. 9, 241-258. FISHER, R. A. (1934). Statistical Methods for Research Workers. Oliver & Boyd, Edinburgh. GENCHI, H., AND MORI, K. (1965). Evaluation and feature extraction on automated pattern recognition system. Dendi-Tsushin Gakki, Part 1. [Japanese]. GROB, D., AND HARVEY, A. M. (1953). The effects and treatment of nerve gas poisoning. Amer. J. Med. 14, 52-63.

GROB, D., AND HARVEY, J. C. (1958). Effects in man of the anticholinesterase compound sarin (isopropyl methyl phosphonofluoridate). J. Clin. invest. 37, 350-368.

GROB, D., HARVEY, A. M., LANGWORTHY, 0. R., AND LILIENTHAL, J. L., JR., (1947). The administration of diisopropyl fluorophosphate (DFP) to man: 1III, Effect of the central nervous system with special reference to the electrical activity of the brain. Bull. Johns Hopkins Hosp. 81, 257-266. HARTMANN, E. (1968). On the pharmacology of dreaming sleep (the D. state). J. New. Merit. Dis. 146, 165-173.

176

DUFFY

HENRICKSEN, S. J., JACOBS, B. L., AND DEMENT, W. C. (1972). Dependence of REM sleep PC0 waves on cholinergic mechanisms. Brain Res. 48, 412-416. HERNANDEZ-PEON, R. (1965). Central neuro-humoral transmission in sleep and wakefulness. In Sleep Mechanism, Progress in Brain Research (C. Bally, and J. P. Schade, eds.), p. 18. Elsevier, Amsterdam. HERNANDEZ-PEON, R., AND CHAVEZ-IBARRA, G. (1963). Sleep induced by electrical or chemical stimulation of the forebrain. Ekctroenceph. C/in. Neurophysiol. Suppl. 24, 188-198. HOLMES, J. H., AND GAON, M. D. (1956). Observations on acute and multiple exposure to anticholinesterase agents. Trans. Amer. Clin. Climarol. Ass. 68, 86-101. JOFFEE, R.*AND JACOBS, L. (1972). The beta focus: Its nature and significance. Acta Neural. Stand. 48, 191-203. JOHNSON, M. K. (1970). Organophosphorus and other inhibitors of “neurotoxic esterase” and the development of delayed neurotoxicity in hens. Biochem. J. 120, 523-531. JOHNSON, M. K. (1974). The primary biochemical lesion leading to the delayed neurotoxic effects of some organophosphorus compounds. J. Neurothem. 23, 785-789. JOHNSON, M. K. (I 975). Organophosphorous esters causing delayed neurotoxic effects: Mechanism of action and structure/activity studies. Arch. Toxicol. 34, 259-288. KAMP, A., SCHRIJER, C. F. M., AND STORM VAN LEEUWEN, W. (1972). Occurrence of “beta bursts” in human frontal cortex related to psychological parameters. Electroenceph. Clin. Neurophysiol. 33, 257-268. KILOH, L. G., MCCORNAS, A. J., and OSSELTON, J. W. (1972). Clinical Electroencephalography. Appleton-Century-Crofts, New York. KOELLE, G. B. (1968). Anticholinesterase agents. In The Pharmacological Basis of Therapeutics (L. S. Goodman and A. Gilman, eds.). Macmillan Co., New York. Koor, K. A. (1971). Fundamentals qf Electraencephalography. Harper & Row, New York. MAGHERINI, P. C., POMPEIANO, 0.. AND THODEN, U. (1972). Cholinergic mechanisms related to REM sleep: I, Rhythmic activity of the vestibulooculomotor system induced by an anticholinesterase male decerebrate cat. Arch. Ital. Biol. 110, 234-259. MARSDEN, A. T. H., AND REID, H. A. (1961). Pathology of seasnake poisoning. Brit. Med. J. 1, 1290-1293.

ET AL. MATSUZAKI, M., OBADA, Y., AND SHOTO, S. (1968). Cholinergic agents related to parasleepstateinacute brain stem preparations. Bruin Res. 9. 253-267. MENDELS, J., AND HAWKINS, D. R. (1968). Sleep laboratory adaptation in normal subjects and depressed patients (“first night effects”). Electroencephalogr. Clin. Neurophysiol. 22, 556-558. METCALF, D. R., AND HOLMES, J. H. (1969). EEG, psychological, and neurological alterations in humans with organophosphate exposure. Ann. N. Y. Acad. Sri. 160, 357-365. MILBY, T. (1971). Prevention and management of organophosphate poisoning. J. Amer. Med. Ass. 216, 2131-2133. MUNDY-CASTLE, A. C. (1953). Theta and beta rhythm in the elctroencephalograms of normal adults. Electroencephalogr. Clin. Neurophysiol. 4, 165-179. NEALIS, J. G. T., AND DUFFY, F. H. (1978). Paroxysmal beta activity in the pediatric electroencephalogram. Ann. Neural. 4, 112-l 16. ROWNTREE, D. W., NEVIN, S., AND WILSON, A. (1950). The effects of diisopropyl fluorophosphate in schizophrenic and manic depressive psychosis. J. Neurol. Neurosurg. Psychiat. 13, 47-62. SANTOLUCITO, J. A., AND MORRISON, G. (1971). EEG of rhesus monkeys following prolonged lowlevel feeding of pesticides. Toxicol. Appl. Pharmaeel. 19, 147-154. SELIN, I. (1965). Detection Theory. Princeton Univ. Press, Princeton. SITARAM, N., WYATT, R. J., DAWSON, S., AND GILLIN, J. C. (1976). REM sleep induction by physostigmine infusion during sleep. Science 191, 1281-1282. TABERSHAW, I. R., AND COOPER, W. C. (1966). Sequelae of acute organic phosphate poisoning. J. Occup. Med. 8, S-10. THYODEN, J., MAGHERINI, P. C., AND POMPEIANO, 0. (1972). Cholinergic mechanisms related to REM sleep: II, Effects of an anticholinesterase on the discharge of central vestibular neurons in the decerebrate cat. Arch. Stat. Biol. 110, 260-283. WAD~A, R. S., SADAGOPAU, C., AMIN, R. B., AND SARDESAIR, H. V. (I 974). Neurological manifestations of organophosphorous insecticide poisoning. J. Neurol. Neurosurg. Psychiat. 37, 841-847. WIED, G. L., BARTELS, P. H., BAHR, G. F., AND OLDFIELD, D. G. (1968). Taxonomic intracellular analytic system (TICAS) for cell identification. Acta Cytol. 12, 177-210. WILLIAMS, R. L., KARACAN, 1. K., and HURSCH, C. J. (1974). EEG ofhuman sleep. Wiley, New York.