P300 of auditory event related potentials in occupational chronic solvent encephalopathy

NeuroToxicology 28 (2007) 1230–1236

P300 of auditory event related potentials in occupational chronic solvent encephalopathy Petra Keski-Sa¨ntti a,b,*, Anu Holm b, Ritva Akila b, Katinka Tuisku b, Tero Kovala b, Markku Sainio b a

Helsinki University Central Hospital, Jorvi Hospital, Department of Neurology, Turuntie 150, FIN-02740, Espoo, Finland b Finnish Institute of Occupational Health, Department of Occupational Medicine, Brain and Work Research Centre, Topeliuksenkatu 41 a A, FIN-00250, Helsinki, Finland Received 9 May 2007; accepted 7 August 2007 Available online 10 August 2007

Abstract This retrospective study characterized the P300 component of the auditory event related potential (ERP) and assessed its diagnostic value in occupational chronic solvent encephalopathy (CSE). The P300 was recorded on 86 CSE patients by the classical oddball paradigm. In addition to the laboratory’s reference values, we used an age and education matched control group that consisted of 104 blue-collar workers with no known occupational solvent exposure. The association of P300 values with solvent exposure indices, major depression, alcohol consumption, and neuropsychological parameters was studied. The P300 amplitude was lower in CSE patients (mean 7.5 mV; S.D. 3.6) compared to laboratory controls (mean 11.8 mV; S.D. 4.1; F(1,167) = 24.4; p < 0.001, 95% CI 4.4 to 1.8) and to matched controls (mean 9.0 mV; S.D. 4.0; p = 0.007, 95% CI 2.6 to 0.4). The P300 latency was longer in the CSE patients (mean 358 ms; S.D. 28) compared to laboratory controls (mean 339 ms; S.D. 19, F(1,167) = 7.6, p = 0.006, 95% CI 3.12–18.7) but did not differ from matched controls (mean 358 ms; S.D. 22; p = 0.947, 95% CI 7.4 to 6.9). The solvent exposure indices, major depression, or alcohol consumption did not associate with the P300 values. The P300 amplitude correlated positively with the Digit Symbol test. All the amplitude values in the patient group and in the matched control group were classified as normal (i.e. age corrected mean  2.5S.D.) against the laboratory’s reference values. Thirty percent of the latencies in the CSE patient group and 26% in the matched control group were classified as abnormal. At group level, the decreased P300 amplitudes in CSE patients may reflect solvent-related pathophysiology. However, the P300 measured with the classical oddball paradigm does not seem to be sensitive at individual level or useful in clinical practice. # 2007 Elsevier Inc. All rights reserved. Keywords: Toxic encephalopathy; Organic solvents; Occupational; Event-related potentials; P300

1. Introduction Long-term low-level occupational exposure to organic solvents may lead to permanent central nervous system damage, the chronic solvent encephalopathy (CSE) (Baker, 1994; Edling et al., 1990; Mikkelsen, 1997). The diagnostics of CSE is demanding because the symptoms and the clinical findings are non-specific and the diagnosis relies on the detection of cognitive dysfunction in the neuropsychological

* Corresponding author at: Helsinki University Central Hospital, Jorvi Hospital, Department of Neurology, Turuntie 150, FIN-02740 Espoo, Finland. Tel.: +358 503738938; fax: +358 9 8615918. E-mail address: [email protected] (P. Keski-Sa¨ntti). 0161-813X/$ – see front matter # 2007 Elsevier Inc. All rights reserved. doi:10.1016/j.neuro.2007.08.004

assessment (Lundberg et al., 1995; Ridgway et al., 2003). However, also neuropsychological findings are non-specific to CSE and the interpretation of quantitative and qualitative test results requires an expert in clinical neuropsychology. Event related potentials (ERP) provide a neurophysiological index of subject’s cognitive functions. ERP are generated by a sensory stimulus in tasks where subjects attend and discriminate stimuli that differ from one another on some dimension. It is assumed that a widespread cortical network in areas of the temporoparietal junction, lateral prefrontal, and parietal cortex are involved in the generation of the auditory ERP (Soltani and Knight, 2000). The late components of ERP (e.g. P300) represent aspects of information processing, such as attention allocation, decision making, and activation of immediate memory (Polich and Kok, 1995). It has been proposed that the primary cognitive

P. Keski-Sa¨ntti et al. / NeuroToxicology 28 (2007) 1230–1236

deficit in CSE is related to attention and information processing (Haut et al., 2000; Morrow et al., 1992a). Thus, it is reasonable to assume that the brain dysfunction in CSE could cause changes in the P300 component of the auditory ERP. However, the results in previous studies of P300 in CSE patients (Lindgren et al., 1997; Morrow et al., 1992b), or occupationally solvent exposed workers have been inconsistent (Lundberg et al., 1995; Massioui et al., 1990; Steinhauer et al., 1997). The aim of this study was to characterize the P300 component of the auditory ERP in a large, carefully examined CSE patient population excluded for other etiologies of brain dysfunction. In order to reduce confounding, we used an age and education matched group of non-exposed blue-collar workers as a control cohort. We also compared results obtained in CSE patients with our laboratory’s reference auditory ERP values. We studied whether the amplitude or latency of the P300 associates with the duration and intensity of the solvent exposure or with a worker’s performance in neuropsychological tests that measure various aspects of attention and short-term memory functioning. The effect of major depression, alcohol consumption, and medication with central nervous system influence (CNS medication) was also examined. In a subgroup of CSE patients, we studied whether the P300 remains consistent during follow up. Finally, the diagnostic value of P300 in CSE is discussed. 2. Methods 2.1. Assessment of CSE in Finland In the Finnish health care system, all workers suspected to have a work-related nervous system disorder are sent to the outpatient clinic of the Finnish Institute of Occupational Health (FIOH) (Juntunen, 1993). Diagnostic verification includes evaluation of all available documents of exposure and medical records, examination by a specialist in occupational medicine and a specialist in neurology, brain imaging with MRI, and

1231

extensive differential diagnostic laboratory testing. During the diagnostic process, at least one, usually several comprehensive clinical neuropsychological assessments are carried out. If other possible causes for brain dysfunction, such as psychiatric or sleep disorders, are assumed, a comprehensive psychiatric evaluation or polysomnography is carried out. Furthermore, the diagnosis of CSE is made only after a sufficient follow up period of at least 1 year. During the follow up, occupational chemical exposure is ceased or at least minimized and other possible medical disorders are treated where after clinical and neuropsychological re-evaluations are made. 2.2. CSE patients in this study The auditory ERP registration had been done to 104 patients (83%) out of the 125 patients diagnosed with CSE at FIOH during 1993–2002. A second ERP registration on an average 14 months (range from 10 to 18 months) after the first one was done to 19 individuals. All the patients fulfilled the criteria for toxic encephalopathy type 2B according to the classification by the International Solvent Workshop (Cranmer and Golberg, 1987) or class II of the WHO criteria (World Health Organization, 1985). The diagnostic criteria include: verified exposure to neurotoxic solvents, clinical picture of organic nervous system damage with typical subjective symptoms (i.e. fatigue, mood disturbances, difficulty in concentration, impairment of memory, decrease in learning capacity), objective findings in clinical examinations, and exclusion of non-occupational brain disorder and psychiatric disease with appropriate investigations (van der Hoek et al., 2000). Before the diagnosis, the clinical procedure in Finland consists of a 1–3-year follow-up during which treatable etiologies are intervened and progressive disease identified. The clinical information was collected from the patient records. To study the pure solvent effect on P300, we excluded CSE patients with any other brain disorder, previous traumatic

Table 1 Demographics and neuropsychological test results of the CSE patients and the controls CSE patients n = 86

Age (years) Education (years) Alcohol consumption (g/month) Current smokers Hypertension Diabetes Exposure yearsa OELYb Digits Forward (DF)c Digits Backward (DB)c Digit Symbol (DSy) c

Matched controls n = 108

Mean

S.D.

Range

50.1 7.2 158.4

6.3 1.4 198.8

34–61 6–12 0–768

%

Mean

S.D.

Range

49.8 7.1 363.5

6.8 1.9 337.7

40–65 4–15 0–1154

34 17 6 27 10.8 5.4d 4.0d 30.3d

7.5 4.4 1.0 1.0 8.8

11–43 6–26 3–8 2–7 16–60

Laboratory controls n = 84 %

Mean

S.D.

Range

41.1 NA NA

10.2 NA NA

20–63 NA NA

43 13 0 0 0 6.0 4.5 35.5

NA = information not available. a Total years in exposure work. b Occupational Exposure Limit years, a lifetime cumulative exposure estimation. c Test was done to 80–83 out of 86 patients and to 104–105 out of 108 matched controls. d p < 0.001. CSE patients compared to matched controls.

1.1 1.2 10.2

3–9 2–8 13–67

%

NA NA NA 0 0 NA NA NA

NA NA NA

NA NA NA

1232

P. Keski-Sa¨ntti et al. / NeuroToxicology 28 (2007) 1230–1236

brain injury, psychiatric disease apart from depression, or reported alcohol consumption exceeding 1200 g absolute alcohol per month (i.e. 40 g or 3–4 doses absolute alcohol per day) for men and 820 g for women. Also, two patients with encephalopathy due to acute massive solvent intoxication were excluded. The ERP could not be identified for three patients. The final number of patients in the study group was 86 with eight females. Ten out of the 86 patients were using CNS medication, i.e. antidepressants, anxiolytics or analgesic drugs (citalopram, diazepam, clonazepam, moclobemide, doxepin, perfenazin, amitriptyline, zopiclone, carbamazepine, chlordiazepoxide, alprazolam, orphenadrine citrate, fluphenazine). The demographics of the patients are listed in Table 1. The P300 measurements used in this study were a part of the routine clinical investigations during the diagnostic procedure at FIOH. 2.3. The control groups We used two control groups. First, we used the laboratory’s own reference auditory ERP values (laboratory control group) of presumed healthy subjects (n = 84) recruited mainly from the staff of FIOH. This is the normative data we use in clinical practice. Because the laboratory controls were slightly younger and more educated, we secondly formed an age and education matched control group of blue-collar workers without occupational solvent exposure. The subjects of this matched control group had participated in a large study at FIOH during 1991–1995 investigating the predictors of work capacity and disability pension in aging construction workers (Portin et al., 2000). In that study, 200 subjects had been randomly selected from the original sample of 1000 subjects for further medical, neurophysiological and neuropsychological evaluation. The auditory ERP had been recorded on 200 subjects of whom all but three participated in the cognitive testing. From this data set, we excluded 77 subjects (39%) due to a professional title referring to possible occupational solvent exposure. Further 15 subjects were excluded using the same exclusion criteria as for the CSE patients. The ERP could not be identified on two subjects. The total number of matched controls was 108, with eight females. None of the matched controls used CNS medication. The demographics of the control groups are listed in Table 1. 2.4. Neuropsychological parameters The CSE patients had undergone a comprehensive clinical neuropsychological assessment during the diagnostic procedure and the matched controls had been tested with a neuropsychological test battery according to the previous study design (Portin et al., 2000). Three subtests that measure various aspects of attention and short-term memory functioning were chosen from these test batteries. They were the Digit Span and the Digit Symbol tasks from the Wechsler Adult Intelligence Scale or WAIS-revised (Lezak, 1995; Wechsler, 1981). The Digit Span comprises of two subtasks, Digits Forward (DF) and Digits Backward (DB), which are both dependent on the short-term retention capacity of the subject.

The DF performance strongly associates with the subject’s attentional capacity whereas in the DB there is an additional mental operation to perform; i.e. the DB performance requires also working memory capacity. In both tasks the score is the longest sequence correctly repeated. The Digit Symbol substitution task (DSy) is a complex task requiring focused, sustained, and shifting attention with demands on visuomotor coordination and psychomotor speed. The performance score is the number of digits substituted in 90 s. In previous studies, the performance in these three tests, among some other tests, has been consistently reduced in CSE patients (Edling et al., 1990; ¨ sterberg et al., 2000). Morrow et al., 1992a; O 2.5. Psychiatric evaluation To study if major depression associates with the P300 the patients were divided in two groups, patients with and without major depression. During the diagnostic procedure, any objectively detected signs or subjectively reported symptoms referring to possible psychiatric disorders had led to psychiatric consultation (72 out of 86 CSE patients). The psychiatric records were re-evaluated by a specialist in psychiatry. Twentyone patients fulfilled the diagnostic criteria of the fourth edition of the Diagnostic and Statistical Manual of Mental Disorders (DSM-IV) for current major depressive episode (American Psychiatric Association, 1994) at the time of the ERP recording. No depressive symptoms were found on 40 patients. Twenty-five subjects, who had not undergone a psychiatric evaluation (n = 14) or had reported depressive symptom but did not fulfill the DSM-IV diagnostic criteria (n = 11), were excluded from the comparison of the two subgroups (i.e. CSE patients with and without major depression). None of the patients had major depression with psychotic symptoms or any other psychiatric disease. 2.6. Solvent exposure In order to study the dose–response relationship, an occupational hygienist analyzed all the work tasks and categorized the chemical composition of the lifetime exposure (Juntunen, 1993; Kaukiainen et al., 2004). In addition to the duration of occupational solvent exposure in years, the lifetime cumulative exposure estimation was calculated in Occupational Exposure Limit Years (OELY), which means a hypothetical number of years during which the solvent exposure is eight hours daily at the level of the Finnish Occupational Exposure Limit (Kaukiainen et al., 2004). In the clinical practice at FIOH, five or more OELY is considered sufficient to cause CSE when the other diagnostic criteria for CSE are met. All the patients had been exposed to mixtures of industrial solvents. The most common chemicals were toluene, xylene, white spirit and industrial alcohols. The chemical composition of the solvent mixtures and the working conditions of CSE patients had been heterogeneous. Thus, in order to study the dose–response relationship in the CSE patient group, we also formed a subgroup of car painters (n = 16), the largest occupational group with similar working conditions and

P. Keski-Sa¨ntti et al. / NeuroToxicology 28 (2007) 1230–1236

1233

solvent exposure. The mean age of the car painters was 49.3 (range 43–57, S.D. 4.4), the mean exposure time was 28.2 years (range 21–40, S.D. 6.5) and the mean OELY was 9.6 (range 6– 16, S.D. 3.5). They were all male and the mean alcohol consumption was 140.3 g/month (range 0–624, S.D. 192.6). None of the car painters used CNS medication. The solvent exposure of 59 out of 86 patients had ended before the first ERP registration (mean 23 months prior to the registration, range from 0.5 to 136 months, S.D. 25.5) and 27 patients were still occupationally exposed. Four out of the 19 follow-up patients were still at exposure work at the time of the second registration.

averaged on-line using Cadwell Spectrum 32 system, which also controlled the quality of the stimulus and rejected artifacts: trials exceeding 80 mV on any channel were automatically rejected on-line. The analysis time was 650 ms including 50 ms prestimulus baseline. The mean responses were digitally filtered (6 Hz lowpass) and both responses were compared visually to ensure repeatability. Only waveforms present in both stimulus blocks were considered adequate responses for the analyses. The amplitude of P300 was measured relative to the 50 ms prestimulus baseline and the latency from the onset of the stimuli.

2.7. Event related potentials

The P300 latencies and amplitudes of CSE patients and laboratory controls were compared using an analysis of covariance with adjustment for age. Degrees of freedom were adjusted (Greenhouse-Geisser) when appropriate. The comparisons of P300 values and neuropsychological performance between CSE patients and matched controls were made using the independent measures t-test. The homogeneity of variances was evaluated with Levene’s test. The associations of alcohol consumption, neuropsychological test results and exposure indices with the P300 values were studied by the Spearman rank order correlation analysis. The subgroups of CSE patients with or without major depression or CNS medication were compared using the independent samples t-test. Also, the subgroup of patients still at exposure work at the time of the first ERP registration and the subgroup of patients with ceased solvent exposure were compared using the independent samples t-test. The follow up comparison of P300 in the CSE patient group was made with the paired samples t-test. Along with summary statistics and p-values, 95% confidence intervals (CI) were also reported. All statistical tests were two-tailed, and the level for significance was set at p < 0.05. The analyses were performed using the SPSS for Windows, Version 12.0 (SPSS Inc., Chicago).

An auditory oddball paradigm was completed in a counting paradigm using a Cadwell Spectrum 32 system. Tones were presented with varying interstimulus interval (1.72–1.96 s) binaurally at 70 dB sound pressure level having a 10 ms rise time, a 30 ms plateau time and a 10 ms fall time. The standard tone frequency was 1000 Hz, and the target tone frequency was 2000 Hz. Two 256-trial blocks were presented, each containing 64 (25%) target stimuli in a random order. Before recording, an example of the task was presented to ascertain that all the subjects could easily discriminate the target from the standard tones. The subjects were asked to count silently the number of target stimuli and report the final number of targets at the end of the session. The repeatability of the recording was controlled by carrying out two consecutive 256 trial blocks. The subjects were examined while awake with their eyes closed and sitting in a chair. The experimenter observed each subject’s alertness and performance and in case of artifacts the recording was momentarily interrupted. The electroencephalogram (bandpass 0.5–80 Hz) was recorded at the Pz electrode site (International 10–20 system). Linked ears (A1A2) were used as a reference. Waveforms were

2.8. Statistical methods

Fig. 1. The median, quartiles, and extreme values for the P300 amplitudes (A) and latencies (B) of the CSE patient (n = 86), the matched control (n = 108) and the laboratory control groups (n = 84). The P300 amplitude was significantly smaller in the patient group compared to the laboratory control group (***p < 0.001) and to the matched control group (**p = 0.007) (A). The P300 latency was longer in the patient group compared to the laboratory control group (**p = 0.006) but did not differ from the matched control group ( p = 0.947) (B).

1234

P. Keski-Sa¨ntti et al. / NeuroToxicology 28 (2007) 1230–1236

3. Results The P300 amplitude was smaller in the patient group (mean 7.5 mV; S.D. 3.6) compared to the laboratory control group (mean 11.8 mV; S.D. 4.1; F(1,167) = 24.4; p < 0.001, 95% CI 4.4 to 1.8) and to the matched control group (9.0 mV; S.D. 4.0; p = 0.007, 95% CI 2.6 to 0.4) (Fig. 1A). The P300 latency was longer in the patient group (mean 358 ms; S.D. 28) compared to the laboratory control group (mean 339 ms; S.D. 19, F(1,167) = 7.6, p = 0.006, 95% CI 3.12–18.7) but did not differ from the matched control group (mean 358 ms; S.D. 22; p = 0.947, 95% CI 7.4 to 6.9) (Fig. 1B). Age correlated negatively ( 0.394, p < 0.001) with the amplitude and positively (0.417, p < 0.001) with the latency of P300 and was thus taken as a covariate in the group comparisons of the CSE patients and the laboratory controls. The duration of exposure in years or OELY did not correlate with the P300 amplitude ( 0.06, p = 0.585) or latency (0.009, p = 0.932). Thus, there was no dose–response relationship in the patient group or in the subgroup of car painters. There was no difference in the P300 values of the patients still at exposure work at the time of the first ERP registration and the patients with ceased solvent exposure. Major depression or use of CNS medication did not associate with the P300 amplitude or latency within the patient group, nor did the amount of alcohol consumed in the patient or in the matched control groups. The CSE patients performed worse in all the three neuropsychological tests when compared to the matched controls (Table 1). The DSy correlated positively with the P300 amplitude both in the patient group (0.279, p = 0.011) and in the matched control group (0.220, p = 0.026). The DF had a tendency to correlate positively with the P300 amplitude in CSE patient group (0.20, p = 0.074) but not in the matched control group (0.017, p = 0.864). The DB did not correlate with the amplitude in either group. The P300 latency did not correlate with any of the three neuropsychological tests in either group. The neuropsychological test results did not associate

with major depression, the amount of alcohol consumed, the duration of exposure in years, or OELY. In the follow up group of 19 patients, there were no significant changes in the P300 latency or amplitude values between the first and the second ERP registration. The P300 amplitude improved in four patients, deteriorated in four patients and remained unchanged in 11 patients when at least 25% change in the amplitude was taken as the criterion for improvement or deterioration (dashed line in Fig. 2). There seemed to be no significant difference in age, duration of exposure in years or OELY between the patients with improved or deteriorated amplitude. However, it seemed that the exposure-free time had been longer in the four patients with improved P300 amplitude compared to the four patients with deteriorated amplitude (Fig. 2A). All the changes in the P300 latency values were less than 25% and were thus regarded nonsignificant (Fig. 2B). All the amplitudes in the CSE patient group and in the matched control group were classified as normal (i.e. age corrected mean  2.5S.D.) against the laboratory’s reference values. Thirty percent of the latencies in the CSE patient group and 26% in the matched control group were classified as abnormal. 4. Discussion We studied P300 in an exceptionally large CSE patient group (n = 86) compared to previous P300 studies with CSE patients (n = 12–20) (Lindgren et al., 1997; Morrow et al., 1992b) or solvent exposed workers (n = 13–53) (Lundberg et al., 1995; Massioui et al., 1990; Steinhauer et al., 1997). Also, our clinical differential diagnostic procedure and the evaluation of the solvent exposure have been extensive and strict. The main result of this study is that CSE patients have lower P300 amplitudes than non-exposed age and education matched controls but the latencies do not differ. The result is congruent with the previous findings in a small CSE patient group very

Fig. 2. The ratio of the P300 amplitude (A) and latency (B) values in the two ERP registrations in the follow-up group (n = 19). Ratio > 1 = amplitude improved (A), ratio < 1 = latency improved (B). Dashed line = change more than 25%.

P. Keski-Sa¨ntti et al. / NeuroToxicology 28 (2007) 1230–1236

similar in age, exposure, and education (Lindgren et al., 1997). The previous finding that CSE patients have prolonged P300 latency compared to healthy controls is partly corroborated in our study where the CSE patients had longer P300 latencies only when compared to the laboratory control group (Morrow et al., 1992b). However, compared to the CSE patients, the laboratory controls may represent healthier subjects with better motivation, higher education and socio-economic status. On the other hand, the amount of alcohol consumed in the matched control group was notably higher compared to the CSE patient group that may contribute the lack of difference in the P300 latency values between the CSE patients and the matched controls. In the few studies with solvent exposed workers without evidence of CSE, no changes in the P300 amplitude or latency have been detected (Lundberg et al., 1995; Massioui et al., 1990). However, acute exposure within the last 1–66 h prior to the ERP registration has shown to be associated with prolonged P300 latency in solvent exposed workers (Steinhauer et al., 1997). In our study, the continuation of occupational solvent exposure at the time of ERP registration did not have an effect on the P300 values. However, we could not study in detail the possible acute solvent effects due to missing information on the time interval between last exposure and testing in patients still at exposure work (27 out of 86 patients). None of the patients had been at exposure work on the day of testing and usually exposure had also been reduced during the follow-up. We studied if major depression in CSE patients associated with P300 because solvent exposed workers have higher rate of mood disorders (Condray et al., 2000; Morrow et al., 2000) and major depression may affect P300 (Sara et al., 1994). In CSE patients, the neuropsychiatric symptomatology measured by the Global Severity Index from the Symptom Checklist-90— Revised (SCL-90) has shown to be associated with reduced P300 amplitude (Morrow et al., 1996) but depressive symptoms measured by the Beck Depression Inventory (BDI) scores have not shown to be associated with the P300 values (Morrow et al., 1992b). In our study, comparison to SCL-90 or BDI could not be done because of retrospective psychiatric re-evaluation of the psychiatric medical records. We used the DSM-IV diagnostic criterion for the classification of CSE patients and did not find an association with the P300 amplitude or latency and major depression at the time of the ERP registration. Serotonin reuptake inhibitors, benzodiazepine derivates or medication with anticholinergic effects have shown to associate with reduced P300 amplitude (Martin and Siddle, 2003; Mu¨ller et al., 2001; d’Ardhuy et al., 1999). In our study, there was no association with the CNS medication and the P300 amplitude or latency values. In alcoholic subjects, the P300 latency may be delayed and amplitude decreased, but social alcohol consumption has not shown to associate with changes in P300 (Boutros et al., 2000; Polich and Kok, 1995). As expected, there was no association between the P300 values and the amount of reported alcohol consumption in our study groups because heavy drinkers were excluded from the study. However, the amount of alcohol consumed in the matched control group was notably higher

1235

compared to the CSE patient group that may explain why the P300 latencies were delayed in the matched controls compared to the laboratory controls. Actually, this strengthens our finding of reduced P300 amplitude in CSE. However, this may also mean that the P300 latency is prolonged in CSE, but it is not specific for solvent effects. Although, in a previous study with 12 CSE patients, a correlation was found between the length of exposure and the P300 latency (Morrow et al., 1992b), we did not find dose– response relationships between the P300 variables and the solvent exposure, i.e. duration in years or the OELY. To study this in more detail we formed a subgroup of car painters but no dose–response relationship could be found. The poor association with exposure parameters and P300 may suggest many other influencing factors including individual differences in toxicokinetics and metabolism which are very difficult to control in clinical studies (Lo¨f and Johanson, 1998). Also, even in car painters the solvent exposure had been heterogeneous. In CSE patients, longer P300 latency has shown to associate with poorer cognitive tests scores especially in tasks requiring attention (Morrow et al., 1996). In our study, we did not find an association between the P300 latency and the three neuropsychological tests (i.e. DF, DB and DSy) but the DSy correlated positively with the P300 amplitude both in the patient and in the matched control group. This is not surprising, as the P300 amplitude reflects the efficacy of information processing and DSy is a highly sensitive time limited task, where the score is affected by various performance components, e.g. response speed, selective attention and incidental learning (Kok, 2001; Lezak et al., 2004). During the follow-up period, no significant changes in the P300 values were found. However, longer exposure-free time prior to the ERP registration was seen in the four patients with improved P300 amplitude compared to the four patients with deteriorated amplitude, which is congruent with the previous findings (Morrow et al., 1998). The individual changes in the P300 amplitude are difficult to interpret due to a large intraindividual variability. The follow up results suggest that removal from solvent exposure work may prevent further deterioration of the brain dysfunction in CSE patients. Against the laboratory’s reference values, all the individual amplitude values in the CSE patients and in the matched controls were classified as normal (i.e. age corrected mean  2.5S.D.). Even using the age corrected mean  2S.D. as the criterion for normality, the amplitude values were classified as abnormal only in 1.2% and 4.5%, respectively. Thirty percent of the latencies in the CSE patient group and 26% in the matched control group were classified as abnormal by the 2.5S.D. criterion, and 43% and 41% by the 2S.D. criterion, respectively. The difference between the sensitivity of the amplitude and the latency of P300 may be due to a larger interindividual variability of the amplitude values. Altogether, the P300 measured with the classical oddball paradigm does not seem to be a sensitive marker of the brain dysfunction in CSE patients and not useful as such in the clinical practice. However, sensitivity of the ERP may be improved by the use of ERP-adapted neuropsychological tests (Lefebvre et al., 2005; Lindgren et al., 1997).

1236

P. Keski-Sa¨ntti et al. / NeuroToxicology 28 (2007) 1230–1236

5. Conclusions The P300 amplitude is decreased in CSE suggesting a solventrelated pathophysiology of CSE and reflecting the deficit in attention and information processing detected in the neuropsychological examination of CSE patients. Also, the latency may be prolonged in CSE. However, the classical oddball paradigm does not seem to be sensitive at individual level and not useful in the clinical practice. Combining neuropsychological tests with neurophysiological indices may improve the clinical usefulness of ERP in the early diagnosis of CSE and also reveal new information on the neural basis of CSE. Acknowledgements The authors want to thank MD Ari Kaukiainen and industrial hygienist Kaarina Rantala for the exposure analyses. We also want to thank Mrs. Outi Fischer, Mrs. Seija Karas, Mrs. Hannele Kataja, and Mrs. Riitta Velin for their valuable help in collecting the study material. The authors wish to declare no competing interests. References Baker EL. A review of recent research on health effects of human occupational exposure to organic solvents. A critical review. J Occup Med 1994;36:1079– 92. Boutros NN, Reid MC, Petrakis I, Campbell D, Torello M, Krystal J. Similarities in the disturbances in cortical information processing in alcoholism and aging: a pilot evoked potential study. Int Psychogeriatr 2000;12:513–25. Condray R, Morrow LA, Steinhauer SR, Hodgson M, Kelley M. Mood and behavioral symptoms in individuals with chronic solvent exposure. Psychiatry Res 2000;97:191–206. Cranmer JM, Golberg L. Human aspects of solvent neurobehavioral effects. Report of the workshop session on clinical and epidemiological topics.In: Proceedings of the work-shop on neuro-behavioral effect of solvents. Neurotoxicology 1987;7:45–56. d’Ardhuy XL, Boeijinga PH, Renault B, Luthringer R, Rinaudo G, Soufflet L, et al. Effects of serotonin-selective and classical antidepressants on the auditory P300 cognitive potential. Neuropsychobiology 1999;40:207–13. DSM-IV. Diagnostic and statistical manual of mental disorders. Washington DC: American Psychiatric Association; 1994. Edling C, Ekberg K, Ahlborg G Jr, Alexandersson R, Barregard L, Ekenvall L, et al. Long-term follow up of workers exposed to solvents. Br J Ind Med 1990;47:75–82. Haut MW, Leach S, Kuwabara H, Whyte S, Callahan T, Ducatman A, et al. Verbal working memory and solvent exposure: a positron emission tomography study. Neuropsychology 2000;14:551–8. Juntunen J. Neurotoxic syndromes and occupational exposure to solvents. Environ Res 1993;60:98–111. Kaukiainen A, Vehmas T, Rantala K, Nurminen M, Martikainen R, Taskinen H. Results of common laboratory tests in solvent-exposed workers. Int Arch Occup Environ Health 2004;77:39–46. Kok A. On the utility of P3 amplitude as a measure of processing capacity. Psychophysiology 2001;38:557–77. Lefebvre CD, Marchand Y, Eskes GA, Connolly JF. Assessment of working memory abilities using an event-related brain potential (ERP)-compatible digit span backward task. Clin Neurophysiol 2005;116:1665–80.

Lezak MD, Howieson DB, Loring DW, Hannay HJ, Fischer JS. Neuropsychological assessment. New York: Oxford University Press; 2004 p. 368– 370. Lezak MD. Neuropsychological assessment. New York: Oxford University Press; 1995 p. 356–368. ¨ sterberg K, Ørbæk P, Rosen I. Solvent-induced toxic encephaloLindgren M, O pathy: electrophysiological data in relation to neuropsychological findings. J Clin Exp Neuropsychol 1997;19:772–83. Lundberg I, Michelsen H, Nise G, Hogstedt C, Hogberg M, Alfredsson L, et al. Neuropsychiatric function of housepainters with previous long-term heavy exposure to organic solvents. Scand J Work Environ Health 1995;21(Suppl. 1):1–44. Lo¨f A, Johanson G. Toxicokinetics of organic solvents: a review of modifying factors. Crit Rev Toxicol 1998;28:571–650. Martin FH, Siddle DA. The interactive effects of alcohol and temazepam on P300 and reaction time. Brain Cogn 2003;53:58–65. Massioui FE, Lille F, Lesevre N, Hazemann P, Garnier R, Dally S. Sensory and cognitive event related potentials in workers chronically exposed to solvents. J Toxicol Clin Toxicol 1990;28:203–19. Mikkelsen S. Epidemiological update on solvent neurotoxicity. Environ Res 1997;73:101–12. Morrow LA, Gibson C, Bagovich GR, Stein L, Condray R, Scott A. Increased incidence of anxiety and depressive disorders in persons with organic solvent exposure. Psychosom Med 2000;62:746–50. Morrow LA, Steinhauer SR, Condray R. Predictors of improvement in P300 latency in solvent-exposed adults. Neuropsychiatry Neuropsychol Behav Neurol 1998;11:146–50. Morrow LA, Steinhauer SR, Condray R. Differential associations of P300 amplitude and latency with cognitive and psychiatric function in solventexposed adults. J Neuropsychiatry Clin Neurosci 1996;8:446–9. Morrow LA, Robin N, Hodgson MJ, Kamis H. Assessment of attention and memory efficiency in persons with solvent neurotoxicity. Neuropsychologia 1992a;30:911–22. Morrow LA, Steinhauer SR, Hodgson MJ. Delay in P300 latency in patients with organic solvent exposure. Arch Neurol 1992b;49:315–20. Mu¨ller TJ, Kalus P, Strik WK. The neurophysiological meaning of auditory P300 in subtypes of schizophrenia. World J Biol Psychiatry 2001;2:9–17. ¨ sterberg K, Ørbæk P, Karlson B, Bergendorf U, Seger L. A comparison of O neuropsychological tests for the assessment of chronic toxic encephalopathy. Am J Ind Med 2000;38:666–80. Polich J, Kok A. Cognitive and biological determinants of P300: an integrative review. Biol Psychol 1995;41:103–46. Portin R, Kovala T, Polo-Kantola P, Revonsuo A, Muller K, Matikainen E. Does P3 reflect attentional or memory performances, or cognition more generally? Scand J Psychol 2000;41:31–40. Ridgway P, Nixon TE, Leach JP. Occupational exposure to organic solvents and long-term nervous system damage detectable by brain imaging, neurophysiology or histopathology. Food Chem Toxicol 2003;41:153–87. Sara G, Gordon E, Kraiuhin C, Coyle S, Howson A, Meares R. The P300 ERP component: an index of cognitive dysfunction in depression? J Affect Disord 1994;31:29–38. Soltani M, Knight RT. Neural origins of the P300. Crit Rev Neurobiol 2000;14:199–224. Steinhauer SR, Morrow LA, Condray R, Dougherty GG. Event-related potentials in workers with ongoing occupational exposure. Biol Psychiatry 1997;42:854–8. van der Hoek JA, Verberk MM, Hageman G. Criteria for solvent-induced chronic toxic encephalopathy: a systematic review. Int Arch Occup Environ Health 2000;73:362–8. Wechsler D. Wechsler adult intelligence scale—revised manual. New York: Psychological Corporation; 1981. WHO. Chronic effects of organic solvents on the central nervous system and diagnostic criteria. Copenhagen: World Health Organization; 1985.