Visual search and eye movements in patients with chronic solvent-induced toxic encephalopathy

NeuroToxicology 27 (2006) 1013–1023

Visual search and eye movements in patients with chronic solvent-induced toxic encephalopathy Helena Ojanpa¨a¨ a,b,*, Risto Na¨sa¨nen b, Juha Pa¨a¨llysaho b, Ritva Akila b, Kiti Mu¨ller b, Ari Kaukiainen b, Markku Sainio b a

Department of Psychology, P.O. Box 9 (Siltavuorenpenger 20 D), FIN-00014 University of Helsinki, Finland b Finnish Institute of Occupational Health, Topeliuksenkatu 41 a A, FIN-00250 Helsinki, Finland Received 17 November 2005; accepted 28 April 2006 Available online 7 May 2006

Abstract Various aspects of visual perception have been found to be impaired in patients with occupational chronic solvent-induced toxic encephalopathy (CSE). The purpose of the study was to characterise the changes in eye movements and visual search performance in CSE patients. We measured eye movements of 13 CSE patients and 22 healthy controls during dynamic visual search task by using a fast video eye tracker. The task was to search for and identify a target letter among numerals presented in a rectangular stimulus matrix (3  3–10  10 items). Threshold search time, i.e. the duration of stimulus presentation required for identifying the target with a given probability was determined by using a psychophysical staircase method. The visual search times of the CSE patients were clearly longer, and they needed considerably more eye fixations than healthy controls to find the target. Thus, their reduced performance in this task was mainly related to the reduction in the number of items which could be processed during a single eye fixation ( perceptual span). This reduction probably reflects a limited capacity of visual attention, since visual acuity, contrast sensitivity, and the oculomotor saccade velocity were found to be normal. The results suggest that motor slowness or low-level visual factors do not explain the poor performance of CSE patients in visual search tasks. The results are also discussed with respect to the effects of education, and compared to the performance in the widely used neuropsychological Trail Making Test, which uses similar stimuli and requires visual search. # 2006 Elsevier Inc. All rights reserved. Keywords: Attention; Eye movements; Main sequence; Perceptual span; Solvent exposure; Trail Making Test; Visual search

1. Introduction Long-term occupational exposure to organic solvents can cause a psycho-organic syndrome, chronic solvent encephalopathy (CSE), which includes a combination of neurological complaints, perceptual and cognitive impairments, as well as an increased incidence of anxiety and mood symptoms (Condray et al., 2000; Juntunen, 1993; van der Hoek et al., 2000; White and Proctor, 1997). Most frequently documented deficits related to chronic solvent exposure are cognitive impairments, such as deficits in attention, short-term memory, associative learning, and psychomotor speed (see, e.g. Anger, 1990; Baker, 1994). The further characterisation of each separate cognitive and perceptual process that can be impaired in chronic solvent exposure is important in order to better understand the nature of * Corresponding author. Tel.: +358 44 3241376; fax: +358 9 588 4759. E-mail address: [email protected] (H. Ojanpa¨a¨). 0161-813X/$ – see front matter # 2006 Elsevier Inc. All rights reserved. doi:10.1016/j.neuro.2006.04.009

the toxic effects, and to develop a set of sensitive screening tests for early detection of nervous system dysfunction. 1.1. Long-term solvent exposure and visual system In vision, chronic solvent exposure can be associated with partially sub-clinical defects in colour vision (Mergler et al., 1987; Muttray et al., 1997; for reviews, see Gobba and Cavalleri, 2003; Iregren et al., 2002). Contrast sensitivity can also be affected, especially in medium spatial frequencies (Donoghue et al., 1995; Mergler et al., 1991). However, defects in different visual functions are not concurrent, that is, some of the CSE patients may have defective colour vision but normal contrast sensitivity, whereas others may have normal contrast sensitivity and colour vision but be slow in visual search (Na¨sa¨nen et al., 2005). Therefore, several aspects of visual perception must be investigated in order to characterise the neurotoxic effects.

1014

H. Ojanpa¨a¨ et al. / NeuroToxicology 27 (2006) 1013–1023

1.2. Visual search task in investigating attentional and visual impairments in CSE It has been proposed that the deficient cognitive performance of CSE patients may be mainly related to a difficulty of allocating attentional resources, especially with increasing processing load, which in turn affects the ability to acquire information at the same rate as healthy controls (Akila et al., in press; Morrow et al., 1992; Stollery, 1996; see also Reinvang et al., 1994). The performance in some neuropsychological paper-andpencil tests can be affected by low-level factors such as uncontrolled visual acuity of the patient (see Skeel et al., 2003) or motor slowness. Therefore, the use of psychophysical experimental setting designed to take into account some of these confounding variables could be useful in addition to traditional tests. More generally, these experimental methods may provide new possibilities for the characterisation of early neural effects of solvents by providing means for more specific hypothesis testing about the cognitive and perceptual processes that are thought to be affected. Recently, Na¨sa¨nen et al. (2005) used a psychophysical computerised visual search task in investigating visual perception of CSE patients. They found that patients performed poorly in both pre-attentive pop-out search task as well as in a serial letter search task requiring eye movements and quick processing of visual information. Since the letter contrast sensitivity of the patients (i.e. contrast transfer and letter recognition per se) was only slightly impaired in comparison to age-matched control subjects, Na¨sa¨nen et al. suggested that the poor performance of CSE patients in visual search may be mainly related to the impaired speed of visual information processing, and/or limitations in attentional capacity. However, since eye movements were not measured, further characterisation of the dysfunction could not be made.

sequences of saccades and fixations form the scan path, which is related to search strategy. The number of eye fixations needed to find the target depends on the number of items that can be processed during a single eye fixation (i.e. the size of the perceptual span or visual span) (Legge et al., 1997; in reading, see O’Regan et al., 1983; Rayner, 1998). The ultimate limits of perceptual span are anatomically determined and depend on the factors affecting stimulus visibility, such as contrast of the stimuli and visual acuity of the subject (Legge et al., 1997; Na¨sa¨nen et al., 2001). However, if the stimulus visibility is good, the higher-level factors, such as task demands and attentional capacity begin to affect the perceptual span more strongly. 1.4. Purpose of the study In this paper, we investigate the visual perception of CSE patients as measured by a computerised visual search task and simultaneous eye movement recordings. The experimental setting allows us to exclude the effects of motor reaction speed from the speed of perceptual information processing during visual search. If the CSE patients need markedly more eye fixations to find a target than healthy controls, it would suggest that the area around the point of eye fixation from which information can be acquired (i.e. perceptual span) is reduced. If, on the other hand, their fixation durations are substantially increased in comparison to controls, the speed of information processing during a fixation could be reduced. In addition, we analysed the eye movement data with respect to a global visual search strategy, and oculomotor properties of saccadic eye movements. Finally, we investigated whether the size of the perceptual span of the subjects is related to their performance in the Trail Making Test, a paper-and-pencil test, which is widely used in behavioural neurotoxicology, and which is considered to measure visual search and effectiveness of attention switching (Lezak et al., 2004).

1.3. Eye movements and visual search 2. Materials and methods Eye movement parameters measured during an intensive visual information processing task can give further information about various factors which affect performance, such as the speed of information processing, global search strategy, capacity of spatial visual attention, as well as integrity of low-level oculomotor processes. Information for visual analysis is mostly gathered during eye fixations (see, e.g. Ross et al., 2001), in which the fovea, the retinal area of the highest visual resolution, is fixated momentarily to a given spatial location. Thus, during visual information processing task, eye fixations take most of the time. In visual search task, the duration of single eye fixation in healthy people is about 200 ms (Na¨sa¨nen et al., 2001), although there is considerable variation depending on the task and stimuli. Since visual resolution strongly decreases with increasing eccentricity (Curcio and Allen, 1990; Curcio et al., 1990; Wertheim, 1894), fast saccadic eye movements are needed to shift the foveal area to successive locations in the stimulus matrix while searching for targets. Together the

The enrolment of the subjects in this study adhered to the tenets of the declaration of Helsinki. All subjects gave their informed consent to participate in the study. The experimental protocol was accepted by the Ethics Committee for Research on Occupational Health and Safety in Work of the Helsinki and Uusimaa Hospital District. 2.1. Patient group There were 13 voluntary patients, 12 of them were male (see Table 1). Their mean age was 56.2 (standard deviation; S.D. 4.8) years. Education years in the patient group were on average 8.7 (S.D. 1.4). All patients had previously received a diagnosis of chronic solvent-induced toxic encephalopathy caused by a long-term occupational exposure to mixtures of several organic solvents. Consecutive CSE patients were taken to the study when they came to their regular yearly follow-up visit at the Finnish Institute of Occupational Health during the year 2002.

H. Ojanpa¨a¨ et al. / NeuroToxicology 27 (2006) 1013–1023

1015

Table 1 CSE patients by occupation, age, education years, and clinical vision test results Background information

Vision test results

Occupation

Age

Education years

Visual acuitya

CSb

Colour visiony

Visual fieldz

CSE patient group 1. Printing trade worker (flexography) 2. Factory worker (plastics) 3. Painter 4. Painter 5. Painter 6. Laboratory worker 7. Painter (cars) 8. Painter 9. Painter (cars) 10. Painter (railroad cars) 11. Factory worker (paints/dyes) 12. Painter 13. Painter

59 56 60 61 52 54 54 45 55 56 57 56 65

8 8 8 8 8 11 11 8 8.75 7.1 7.5 11 9

1.20 1.00 0.95 1.20 1.20 0.95 1.20 1.20 1.50 1.20 0.95 1.50 0.75

Ok Ok Ok Ok Ok Ok Ok Ok Ok Ok Ok Ok Ok

Cong (mild deut.) Ok Ok Ok Cong (deutan) Ok Ok Acq (mild tritan) Ok Ok Acq (mild tritan) Ok Cong (deutan)

Left mild defect Ok Red (tubular field) Ok Ok Ok Ok Red (superior def.) Red (tubular field) Ok Ok Ok Red (superior def.)

For vision tests, ‘Red’ refers to reduced sensitivity, and ‘Ok’ refers to normal performance. ‘Cong’ refers to congenital, ‘Acq’ to acuired defect. a Visual acuity test for near distances (logMAR chart by Precision Vision) was recalibrated for viewing distance of 57 cm, and acuity was measured with similar refractive corrections as the visual search test. b Contrast sensitivity was measured by using Vistech VCTS-6500 grating test. Normal values provided by the manufacturer of the test were used. y Farnsworth-Munsell 100 Hue test was performed in artificial 2000 lux daylight (6500 K). The colour vision sensitivity was regarded impaired if the square root of the total error score exceed the 97.5th percentile of the age matched control population (9.78). z Octopus 101 automated static perimetry, G2 Program. Deviation of 6 dB or more of the mean defect level of age-matched controls was considered reduced (normal data provided by the manufacturer of the test).

For this study, the exclusion criteria were the following: nonoccupational causes for dysfunctioning of the central nervous system such as major depression or other psychiatric diagnosis, major sleep disorder, previous heavy alcohol consumption, current strong medication with central nervous system effects, and a disease or condition affecting eyes or visual system (unrelated to CSE, such as glaucoma or cataract). The CSE can been classified as WHO level II toxic encephalopathy (WHO, 1985; see also van der Hoek et al., 2000). The diagnosis had been based on detailed medical investigations including an interview by an occupational physician, symptom and health questionnaires (Chouanie´re et al., 1997; Kaukiainen et al., 2004a), an extensive neuropsychological assessment and neurological examination, psychiatric consultation when needed, clinical ocular examination, consultation of occupational hygienist for assessing the exposure level (Kaukiainen et al., 2004b), and at least a 1 year follow-up period and a re-evaluation after the cessation of the solvent exposure. The history of solvent exposure was assessed based on the work description and the exposures in each work assignment were evaluated separately (Kaukiainen et al., 2004b). The average lifetime duration of solvent exposure in the patient group was 31 (S.D. 5.7, range 21–41) years, and 10.9 (S.D. 4.0, range 6–19) years as expressed in occupational exposure limit (OEL) years. The number of OEL years is a measure that corresponds to estimated cumulative lifetime exposure transformed into a computational duration of full-time (i.e. 8 h a day, 5 days a week) exposure on Finnish occupational exposure limit, i.e. upper limit of the safe level. At the time of the study, 10 of the subjects were retired after the diagnosis had been confirmed, and three had changed to another kind of work with

no solvent exposure. All patients had been out from solvent exposure at least for 1 year before the onset of the study. 2.2. Control groups We had two groups of healthy control subjects. One control group (N = 14, 2 female, later referred to as controls) was matched for gender, age, and education years. Their age was on average 55.3 (S.D. 4.0, range 51–64) years, and education 9.5 (S.D. 1.5, range 8–13) years. The other control group (N = 8, 1 female, later referred to as educated controls) was also matched for age and gender, but had 8.5 years more formal education. Their mean age was 54.1 (SD 3.9, range 48–60) years. Education level in this group was on average 18 (S.D. 1.9, range 13–21) years. The highly educated controls were included in order to study the effect of education on performance in visual search task used in this study. This is because performance in many neuropsychological tests with similar stimuli, such as Trail Making Test also used in this study, have been shown to be sensitive to education level (for review, see Lezak et al., 2004, p. 373). The visual system of the control subjects was also carefully examined, since they were recruited from a group of people participating in a large normative project for several clinical vision tests. Their performance in visual acuity, contrast sensitivity, colour vision, and visual field sensitivity tests was normal. They also participated in a full clinical ocular examination including assessment of all basic oculomotor functions. In addition, they filled a comprehensive symptom and health questionnaire similar to that used for CSE patients. These data were used to exclude control subjects that had a possible disease or medication affecting the central nervous

1016

H. Ojanpa¨a¨ et al. / NeuroToxicology 27 (2006) 1013–1023

system or the visual system, history of heavy alcohol consumption, or solvent exposure. 2.3. Apparatus and stimuli The stimuli were generated by using a PC computer and a 2100 CRT (Sony TrinitronGDM-F500) colour display. The graphics adapter was used at a resolution of 1152  864 pixels (0.273  0.273 mm/pixel) and a frame rate of 85 Hz. The characters on the screen were dark (0.3–1 cd/m2) on a white background with the photopic luminance of 89 cd/m2. Measurements were performed in a darkened room, where the only light source was the stimulus presentation monitor. The viewing distance was 57 cm. A chin rest was used to stabilise the viewing distance and subject’s head. The stimuli used in the study were upper case letters and numerals presented in rectangular arrays of 3  3, 5  5, 7  7, and 10  10 items (Fig. 1). The typeface used was Courier New at point size 36. Letter height was 18. 2.4. Letter search procedure The task of the subject was to search for and identify a target letter from a set of distracter numerals in a stimulus matrix (see Fig. 1). The outcome measure was a threshold search time, i.e. the presentation duration of the stimulus array required by the

subject for finding the target with a certain probability. The duration of each stimulus presentation was adaptively varied by using a staircase algorithm that increased the subsequent stimulus presentation after incorrect responses, and decreased it after correct answers (Wetherill and Levitt, 1965). Thus, the error level of all subjects was made equal and the performance could be expressed solely as a function of search time. In addition, the procedure excluded the effect of the speed of the manual reactions from search time estimates. The duration of stimulus presentation was controlled in the following way. At first, the stimulus array was visible for 6000 ms. After each correct response, the presentation duration of the next stimulus array was reduced by a factor of 1.26, and after each incorrect response, presentation duration was correspondingly increased. The counting of reversals started after two incorrect responses. After that, the duration of stimulus presentation was shortened after three consecutive correct responses and increased after each incorrect response (i.e. three-down-one-up version of the algorithm). A resulting estimate for threshold search time at the probability level of 0.79 of correct answers was calculated as a mean of eight reversals. The mean number of stimulus presentations needed for a single threshold estimate was around 50. At first the subject rehearsed the procedure with one short threshold measurement for each array size to get familiar with the procedure. After that, the threshold search times were

Fig. 1. The visual search test procedure. Presentation duration of the search array was varied (see text for details). In each array, there was one target letter (A–E, H, K, N, R, U, Vor Z) in a random position among randomly selected numerals (0–9). The stimulus presentation was preceded and followed by masking arrays constructed of question marks. After each stimulus presentation, the subject verbalised the letter he/she recognised and the experimenter indicated the choice by pressing the corresponding graphical button (on the left). The procedure was a 12 alternative forced-choice task, i.e. the subjects were instructed to guess if they did not find the target. After a response and a delay of 500 ms, a new stimulus array was presented. The subjects were instructed to move their eyes freely on the screen while searching the target letter.

H. Ojanpa¨a¨ et al. / NeuroToxicology 27 (2006) 1013–1023

estimated for four different stimulus set sizes in a counterbalanced order (3  3, 5  5, 7  7, 10  10, 10  10, 7  7, 5  5, and 3  3). Eye movements were recorded binocularly during the threshold measurements. The duration of the whole procedure was about 90 min, including a longer break in the middle of the experiment, and shorter breaks between the measurements. 2.5. Eye movement recordings Eye movements were recorded simultaneously with threshold measurements by using an SMI (Sensomotoric Instruments Inc.) EyeLink video eye tracker. The subjects’ gaze direction was recorded with miniature infra-red video cameras, while two infra-red LEDs in each camera illuminated the eyes. The cameras were attached to a headband worn by the subject. The sampling rate of the system was 250 Hz. The eye tracking system was controlled by a separate PC computer connected to the previously mentioned stimulus presentation computer via an Ethernet link. The registration of eye movements started instantaneously with each stimulus presentation and was automatically switched off when the stimulus presentation ended, eyes were moved to the button array, or when the mouse button was pressed for response by the experimenter. The purpose of these measures was to ensure that the collected data reflect eye movements made during the visual search. The collection of eye movement data started after the subject had made two errors in his/her responses. Thus, the eye movement data represent the behaviour at near-threshold level. The saccades and fixations were detected automatically using software provided by the manufacturer of the eye tracker. An eye position sample was regarded as belonging to a saccade if either the acceleration or velocity exceeded their respective thresholds (9500 8/s2 for acceleration or 35 8/s for velocity) for that sample. Other samples were considered to belong to a fixation. The mean of the eye position data for the left and right eye were used in further analyses, except for main sequence parameters, that were analysed separately for both eyes. 2.6. Trail Making Test The Trail Making Test, Parts A and B (originally part of Army Individual Test Battery, 1944) was done as a traditional paper-and-pencil version. In Trail Making Test, Part A (later referred to as TMT-A) the stimuli consist of encircled numbers from 1 to 25 printed in a pseudo-random order on a paper sheet (see Lezak et al., 2004). The subject is required to connect the successive numbers as fast as possible but without making errors. In Trail Making Test, Part B (later referred to as TMTB), the stimuli consist of numbers from 1 to 13, and letters from A to L, and the subject is required to connect them in an alternating order (1-A-2-B-3-C, etc.). The total completion time, and the number of errors were measured. In the procedure applied, the subject’s errors were not corrected during the task by the experimenter.

1017

2.7. Statistical procedure The data did not meet all the assumptions for parametric testing and therefore, non-parametric tests were used. We studied the main effects between three groups by using the Kruskall–Wallis test, which corresponds to ANOVA for k independent samples. In pairwise post hoc comparisons, we used the non-parametric Mann–Whitney U-test (Wilcoxon’s Rank Sum Test). The effect of multiple comparisons on post hoc tests was corrected by setting the criterion for statistical significance for these tests to 0.0166 (the Bonferroni procedure). Letter search data were analysed as follows. The set size effect was studied separately. After that, different set sizes were combined to examine the group differences across set sizes. This was done by first converting all data, i.e. patients and controls, within a given set size into rank scores, and then combining the rank scores for different set sizes within groups. This procedure eliminated the effect of set size from this analysis and allowed the group differences in eye movement parameters across different set sizes to emerge. 3. Results 3.1. Performance and eye movements in letter search test There were significant main effects for visual search time (x2(2) = 44.879, p < 0.001) and the number of fixations (x2(2) = 36.896, p < 0.001). However, there was no statistically significant main effect for fixation duration (x2(2) = 5.007, p = 0.082, ns) nor saccade amplitude (x2(2) = 1.887, p = 0.389, ns). Fig. 2 shows the eye movement and performance data. The patients had statistically highly significantly longer search times than either of the control groups (for controls U = 503, p < 0.001, for educated controls U = 233, p < 0.001). Patients also had considerably more fixations than control subjects (for controls U = 670.5, p < 0.001, for educated controls U = 252.5, p <0.001). The two control groups did not differ statistically significantly in performance nor in the number of eye fixations (for performance U = 891, p = 0.965 ns, for number of fixations: U = 707, p = 0.101, ns). Thus, the letter search test does not seem to be sensitive to differences in the level of education. 3.2. Dynamic properties of saccadic eye movements We also analysed the eye movement main sequence, that is, the relationship between average saccade amplitude and peak velocity. Fig. 3 shows 12 data points for each subject, one for each of the two measurements for array sizes 5  5, 7  7, and 10  10 for both eyes. It can be seen that patients and controls do not differ markedly with respect to main sequence (peak velocity/amplitude: x2(2) = 3.067, p = 0.216, ns and x2(2) = 1.239, p = 0.538, ns for left and right eye). For comparison the normative data measured for horizontal saccades of 18, 58 and 108 by Bahill et al. (1981) is also shown. The comparison suggests that the main sequence is in good agreement with the data of Bahill et al. (1981). Only four

1018

H. Ojanpa¨a¨ et al. / NeuroToxicology 27 (2006) 1013–1023

Fig. 2. Visual search performance (A) and eye movement data during visual search task (B–D). Error bars represent standard error of the mean (SEM). ‘‘Controls’’ refers to age and education matched controls. ‘‘Educated controls’’ refer to controls which have 8.5 years more education but who are matched for age. Data for fixation duration for smallest set size are not shown, since the presentation durations of the array were below the duration of the single fixation.

individual data points fall below the 1 S.D. of their data, and for these three patients the other data points are within normal range. 3.3. Search strategy

Fig. 3. Main sequence, i.e. relationship between average saccade amplitude and peak velocity. 12 data points for each subject are shown. The lines represent the mean 1S.D. of the normative data reported by Bahill et al. (1981). For individual data points below 1S.D. limit of normative data, identification number of the patient is indicated.

Fig. 4 shows the number of saccades made in different angular directions while searching the largest stimulus array (i.e. during about 50 search trials). Individual differences in search strategy are shown as different shapes of the plot. The typical polar diagram is star-shaped, which shows that subjects made generally more horizontal and vertical than oblique saccades. Since absolute values are shown, the size of the plot reflects the number of eye movements needed. It seems that the patients, who needed the largest number of eye movements also tend to show the strongest trend towards using mainly horizontal saccades. This was investigated by calculating the ratio between sums of horizontal and vertical saccades for each subject, which eliminates

H. Ojanpa¨a¨ et al. / NeuroToxicology 27 (2006) 1013–1023

1019

3.4. Trail Making Test (TMT) Table 2 (bottom section) shows the performance in TMT-A and TMT-B for three groups studied. Statistically significant main effect was found for completion times (x2(2) = 9.744, p = 0.008 for TMT-A, and x2(2) = 12.764, p = 0.002 for TMTB). There was no statistically significant main effect in the number of errors between the three groups (x2(2) = 3.484, p = 0.175, ns and x2(2) = 2.601, p = 0.272, ns for TMT-A and TMT-B, respectively). Pairwise post hoc tests revealed that CSE patients were statistically significantly slower in TMT-A than either of the control groups (U = 41, p = 0.014 for controls, and U = 13, p = 0.003 for educated controls). There was no statistically significant difference between the two control groups for TMTA (U = 46.5, p = 0.525, ns). In TMT-B, the controls and patients did not differ statistically significantly due to large variation (U = 43, p = 0.066, ns), but the educated controls were faster than patients (U = 4, p = 0.001) or other controls (U = 21.5, p = 0.016). 3.5. Estimation of the perceptual span

Fig. 4. Number of saccadic eye movements made to different angular directions for CSE patients (A) and controls (B). Each trace represents one subject and shows the number of saccades during the search of 10  10 stimulus array (about 50 search trials). Since absolute values are given, the size of each individual plot also represents the number of saccades needed in search.

the differences in absolute number of saccades. This ratio was on average 2.3 for patients and 1.7 for controls, which is statistically significantly larger for patients (U = 45, p = 0.011).

The number of items that can be processed during a single eye fixation (perceptual span) was estimated from eye fixation data for each subject (see Table 2, top section). We took the value of on average 1.5 fixations/array to represent the limit of the span. Thus, if the subject needed only one fixation on half or more individual trials to perform the task, the matrix was considered to be within the span. We found that there were 4 patients out of 13, who had a span of 3  3 or smaller in letter search task, and 9 patients who had a perceptual span of at least 3  3 but below 5  5 items. None of the patients had a span of 5  5 or above. On the other hand, only 1 of the 22 controls had a span of about 3  3 (1.52 fixations), whereas 5 of 22 controls had a span larger than 5  5 items. None of the subjects reached a span of 7  7 items. In addition, we analysed the performance in TMT with respect to the size of the perceptual span estimated from eye fixation data. When all subjects were grouped into three groups according to the increasing size of their perceptual span in letter search task (span less or equal to 3  3 items, span between 3  3 and 5  5 items, and span between 5  5 and 7  7

Table 2 Trail Making Test performance and perceptual span size Groups

N Patients

a

Span 3  3 3  3 < span < 5  5 5  5 < span < 7  7 Patients Controlsy Educated controlsz a y z

4 9 –

Trail Making Test (time/s) y

z

Controls

Educated controls

Part A (S.E.M)

Part B (S.E.M)

1 10 3

– 6 2

68.20 (8.67) 45.32 (3.44) 28.4 (2.54)

200 (33.57) 107.96 (11.53) 87.60 (24.28)

8

57.54 (5.46) 41.71 (4.81) 35.5 (4.33)

162.27 (24.77) 107.21 (11.98) 67.88 (6.26)

13 14

The size of the span (items) was estimated from the number of fixations needed in each array size. Controls matched by age and education. Controls matched by age, on average 8.5 years more education.

1020

H. Ojanpa¨a¨ et al. / NeuroToxicology 27 (2006) 1013–1023

items), the TMT performance between groups differed statistically significantly (x2(2) = 13.521, p = 0.001 for TMTA, and x2(2) = 7.142, p = 0.028 for TMT-B). Post hoc tests revealed that all three groups with different span sizes differed from each other with respect to TMT-A ( p-values in U-test were 0.006, 0.008, and 0.006, respectively), whereas performance in TMT-B differed statistically significantly only between the two groups with the smallest span ( p = 0.015, other p-values were 0.063, ns, and 0.222, ns). 4. Discussion Our results showed that the CSE patients were statistically highly significantly slower than healthy controls in visual letter search task (Fig. 2A). The relative difference between the groups was similar across different set sizes. The poor performance of the CSE patients was related to the increased number of eye fixations. The average size and speed of saccadic eye movements were similar between the groups, which suggests similar oculomotor functions in all subjects (Fig. 3). Analysis of the angular directions of saccadic eye movements showed that the patients relied more on horizontally dominated scanning strategy in comparison to control subjects (Fig. 4). The two control groups with different educational backgrounds did not differ from each other in visual search performance or in any eye movement parameter. This suggests that, unlike many traditional paper-and-pencil neuropsychological tests, performance in the letter search test does not depend on education. The patients were statistically significantly slower in TMT-A than controls or educated controls, whereas the two control groups did not differ from each other statistically significantly. In TMT-B, the controls and patients did not differ statistically significantly due to large variation, but the educated controls were significantly faster than patients or less educated controls. The three groups did not differ in the number of errors in TMTA or TMT-B. The size of the perceptual span estimated individually for each subject from eye fixation data was associated to their performance in TMT, more strongly in Part A (Table 2). 4.1. Visual search performance Previously, the effect of solvent exposure on visual search has been investigated in one study only. In this study by Na¨sa¨nen et al. (2005), the visual performance of a group of 14 patients with a history of long-term exposure to organic solvents was measured. They found that patients performed poorly in the visual letter search test as well as in a pre-attentive pop-out search test, while they differed from controls only slightly in a letter contrast sensitivity test. Thus, Na¨sa¨nen et al. concluded that impairment in the visual search performance may reflect the deficit in the attention or in the speed of information processing, rather than deficit in the pattern recognition or contrast perception per se. Our results are in agreement with their data. Both studies report clear, statistically highly significant difference between patients and controls in a letter search task, and demonstrate

increase in search times with increasing stimulus array size (set size effect). However, in the data of Na¨sa¨nen et al. patients and controls had generally longer threshold search times than in the present data. Overall quantitative differences in search times between the two studies are probably related to the differences in measurement settings.1 The patient group in Na¨sa¨nen et al. (2005) study also included four patients who did not have the CSE diagnosis, and they found that exceptionally poor performance in all vision tests was associated in aetiology other than CSE. Therefore, these patients may have acted as outliers in letter search results, and may partially explain the worse average search times of Na¨sa¨nen et al. (2005) patient group compared to the patient group in this study. 4.2. Main sequence of saccadic eye movements as an indicator of oculomotor functions To screen some aspects of the basic oculomotor functions of the CSE patients, we calculated the relationship between amplitude and peak velocity of saccades. This eye movement main sequence parameter has been regarded as clinical measure of the general oculomotor functioning, more specifically the pulse component of the pulse-step neural control signal for saccades (Ciuffreda and Tannen, 1995). The dynamic properties of saccades are generated by a network consisting of premotor burst neurons and inhibitory omnipause cells in the brainstem reticular formation and midbrain, and certain areas in the cerebellum (see Kandel et al., 2000; Leigh and Kennard, 2004, and, e.g. Noda and Fujikado, 1987; Noda and Suzuki, 1979; Noda et al., 1988). The lesions or diseases affecting these brain structures cause slowing down of saccadic velocity (brainstem reticular formation) or loss of saccadic accuracy (cerebellum) (Leigh and Kennard, 2004). Main sequence can be used to assess the integrity of the eye velocity, whereas the accuracy of saccades cannot be estimated from our visual search data. To our knowledge, there are no earlier studies where eye movements of the CSE patients have been monitored by using a video eye tracker. Nor have any oculomotor parameters of the CSE patients been investigated from numerical eye movement data measured during dynamic information processing task. For the patients in our study, the main sequence parameters were highly similar to controls, and mostly within the 1S.D. range provided by Bahill et al. (1981) for horizontal saccades. Also the main sequences measured from normal subjects in studies of Garbutt et al. (2001, Fig. 1), Harwood et al. (1999, Fig. 6), and Lebedev et al. (1996, Fig. 2) are in good agreement with our data. This suggests that the basic control of the saccadic eye movements is intact in the CSE patients of our study. In addition, our results support the ecological validity of main sequence measurements in earlier studies, since the data measured during dynamic information processing seems to be similar to earlier data measured for plain reflexive saccades. 1 There was no prior rehearsal session in the Na¨sa¨nen et al. (2005) experimental setting, and the number of visual array repetitions was substantially smaller than in the present study. Three CSE patients participated in both studies.

H. Ojanpa¨a¨ et al. / NeuroToxicology 27 (2006) 1013–1023

Previously, some groups of CSE patients were studied by using otoneurological tests. In these studies some of the CSE patients showed prolonged latencies in voluntary saccades, or reduced ability to suppress vestibulo-ocular reflex in rotating chair, whereas no significant pathology in velocities of saccades, smooth pursuit gain, or accuracy of saccades was found (Mo¨ller et al., 1990; Niklasson et al., 1997). Our oculomotor data are in agreement with their observations on saccadic velocity. In summary, it seems that the differences in eye movements and poor performance of the CSE patients in visual search tasks cannot be explained by any specific pathology in the oculomotor system that affects the velocity of the eye movements. Therefore, the higher level processes of visual perception, such as limitations in attentional capacity or speed of information processing may be affected. 4.3. Search strategy Typically the subjects scanned the stimulus arrays in a relatively irregular fashion. We investigated the general search strategy of the subjects by calculating the number of saccades made to different angular directions. The polar diagram shows that subjects made generally more horizontal and vertical than oblique saccades, which results in a star-shaped polar plot (Fig. 4). This strategy is probably related to the grid-shaped configuration of the stimulus matrix. The magnitude of interindividual differences can be investigated by comparing the angular uniformity of the plots in Fig. 4A and B. The patients in general, and especially the patients that needed the highest number of eye movements (i.e. have the largest plots) show the strongest inclination of using mainly horizontal saccades. This may reflect their general inefficiency in the use and aquisition of effective information processing strategies. In addition, the observed strategy may be a consequence of their attentional limitations (see section below), which force them to scan the matrix with relatively small steps in order to find the target. It should be noted that the subjects were told that they are allowed to move their eyes freely in all directions while searching for the target. 4.4. What do eye movements tell about information processing in patients with CSE? Since the eye movements were not measured in the earlier study of Na¨sa¨nen et al. (2005), the poor performance of CSE patients in visual letter search could not be further characterised. Our data show that the largest difference between the CSE patients and the controls is in the number of fixations needed to find the target letter in the stimulus array (Fig. 2B). Since saccadic eye movements for CSE patients were normal, the increased search times seem to be related to visual information processing during eye fixations, and not to oculomotor slowness of eye movements. The number of eye fixations that are needed to find the target depends on the size of the perceptual span, i.e. the number of items that can be processed during a single eye fixation. If

1021

stimulus visibility is heavily reduced, e.g. by very low stimulus contrast or low visual acuity of the subject, the size of the perceptual span is reduced (Legge et al., 1997). However, when visibility is good, the higher-level factors, such as task difficulty or capacity of spatial visual attention, affect the size of the perceptual span. The low-level factors were not likely to restrict the visibility of the stimuli in our study, since the patients and controls were comparable on visual acuity and contrast sensitivity, the stimuli were large, and stimulus contrast was high. Thus, the difference between the groups in perceptual span does not seem to reflect the low-level signal-to-noise ratio, but is rather suggested to reflect the differences in the capacity of spatial visual attention. One way to assess the size of the perceptual span is to gradually increase the amount of stimuli and investigate, when a single fixation is not sufficient to perform the task anymore (see, e.g. Na¨sa¨nen et al., 2001; Ojanpa¨a¨ et al., 2002 for words). If it is not possible to manipulate the set sizes individually, the number of fixations needed for information processing in different fixed sizes of the stimulus matrix can be investigated. We found that 4 of the 13 patients had a span below 3  3 items in this task, and none of the patients had a span of 5  5 or more. On the other hand, only 1 of the 22 controls had a span of about 3  3 (1.52 fixations), whereas 5 of 22 controls had a span even larger than 5  5 items. Thus, the perceptual span of the CSE patients was limited also in individual level when compared to controls. For some patients, the reduced sensitivity of the peripheral visual field may also contribute to the size of the span. However, this does not explain the effect, since 8 of the 13 patients had a normal visual field. Moreover, only one of the patients that had a span of less than 3  3 (patients 2, 4, 9, and 11), had a reduced light sensitivity in the peripheral visual field. Thus, the increased number of eye fixations (limited perceptual span) suggests that the capacity of spatial visual attention in CSE patients of this study is reduced. Interestingly, some previous studies using completely different methods in investigating CSE have reached a similar conclusion. In the studies of Morrow et al. (1992), Stollery (1996), and Akila et al. (in press), neuropsychological tests were used to characterise the cognitive functions impaired in CSE. The authors concluded that the deficient allocation of attentional resources may explain the decrements that the CSE patients have in various neuropsychological tests. Our data suggest that the attentional limitations in CSE patients include a reduced capacity of spatial visual attention. 4.5. Perceptual span and Trail Making Test in investigating CSE In several previous studies, the inferior performance in the TMT has been shown to be associated to chronic solvent exposure (see, e.g. Cherry et al., 1985; Dick et al., 2000; Morrow et al., 1990; Nilson et al., 1999). Nilson et al. (1999) studied the TMT performance in symptomatic exposed workers, patients with CSE diagnosis, and healthy controls. For CSE patients our results are in good agreement with their

1022

H. Ojanpa¨a¨ et al. / NeuroToxicology 27 (2006) 1013–1023

data. For healthy control subjects the results of the two studies are qualitatively similar, although the completion times in our data are somewhat longer for TMT-B, especially for less educated controls. Nilson et al. (1999) used a procedure, where errors were corrected by the experimenter during the task, which may explain the faster completion times. The effect of education level of controls on performance in TMT is in agreement with previous findings (for review, see Lezak et al., 2004, p. 373). We compared the TMT performance of our subjects to the size of their perceptual span estimated from eye fixation data (Table 2). When all subjects were grouped into three groups according to the size of their perceptual span, the three groups differed from each other with respect to TMT-A. For TMT-B, only the two groups with the smallest span differed from each other significantly. Thus, the size of the perceptual span is associated to performance in TMT-A, whereas TMT-B seems to tap other cognitive functions more than visual search, at least for subjects who do not have exceptionally limited perceptual span. This view is further supported by the results of Crowe (1998), who used linear regressions to determine the differential contributions of several isolated measures in explaining the variance observed in TMT performance of healthy subjects. He found that measures of ‘‘visual search’’ and ‘‘motor speed’’ were the main determinants of the variance in TMT-A, whereas for TMT-B, the measures of ‘‘cognitive alternation’’ and ‘‘visual search’’ contributed most strongly to performance. This is in accordance with our finding, that TMTA and letter search task were both insensitive to the effects of education, whereas for the same subjects the performance in TMT-B was affected by education. Thus, performance in letter search task seems to be more similar with the TMT-A than TMT-B. However, more studies are needed to confirm the present results due to the small number of subjects. The dominance of visual search component in TMT-A showed by the present data and Crowe (1998) study makes it a useful tool in neuropsychological assessment of CSE patients, especially if more specific visual search tests or possibilities for eye movement measurement are not available. The disproportionally poor performance in TMT-A, especially with other tests showing adequate (or proportionally less impaired) motor speed, can serve as an indication of possible limitations in the size of the CSE patient’s perceptual span. 5. Conclusions We studied a small but well characterised group of CSE patients, and found that they were clearly slower than controls, and needed more eye fixations in a task requiring quick visual search and information processing. The reduced amount of items that the patients could process during a single eye fixation is suggested to reflect a reduced capacity of their spatial visual attention. The visual acuity, contrast sensitivity, and the velocity of saccadic eye movements were normal, which suggest that low-level visual factors or oculomotor defects do not explain the results. The size of perceptual span estimated from eye fixation data was associated to the performance in

Trail Making Test, Part A. Results suggest that visual search tasks are useful in the characterisation of the neurotoxic effects of industrial solvents. However, due to the small number of subjects, further studies are needed to confirm the present results. Acknowledgements The authors wish to thank Ph.D., Prof. Veijo Virsu and Ph.D. Taina Nybo for their valuable comments on the manuscript, M.Sc. Kaarina Rantala for assessing the patients’ solvent exposures, and Ms. Outi Fischer, Ms. Hannele Kataja, and Ms. Seija Karas for their help in handling the patients’ medical records and assisting in the color vision and contrast sensitivity measurements. References Akila R, Mu¨ller K, Kaukiainen A, Sainio M. Memory performance profile in occupational chronic solvent encephalopathy suggests working memory dysfunction. J Clin Exp Neuropsychol, in press. Anger WK. Worksite behavioral research: results, sensitive methods, test batteries and the transition from laboratory data to human health. Neurotoxicology 1990;11:629–720. Army Individual Test Battery. Manual of directions and scoring. Washington, DC: War Department, Adjutant General’s Office; 1944. Bahill AT, Brockenbrough A, Troost BT. Variability and development of a normative data base for saccadic eye movements. Invest Ophthalmol Vis Sci 1981;21:116–25. 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(10):1079–92. Cherry N, Hutchins H, Pace T, Waldron HA. Neurobehavioural effects of repeated occupational exposure to toluene and paint solvents. Br J Ind Med 1985;42:291–300. Chouanie´re D, Cassitto MG, Spurgeon A, Verdier A, Gilioli R. An international questionnaire to explore neurotoxic symptoms. Environ Res 1997;73:70–2. Ciuffreda KJ, Tannen B. Eye movement basics for the clinician. St. Louis: Mosby-year Book, Inc.; 1995 p. 53–5. 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. Crowe SF. The differential contribution of mental tracking, cognitive flexibility, visual search, and motor speed to performance on parts A and B of the Trail Making Test. J Clin Psychol 1998;54(5):585–91. Curcio CA, Allen KA. Topography of ganglion cells in human retina. J Comp Neurol 1990;300:5–25. Curcio CA, Sloan KR, Kalina RE, Hendrickson AE. Human photoreceptor topography. J Comp Neurol 1990;292:497–523. Dick F, Semple R, Chen R, Seaton A. Neurological deficits in solvent-exposed painters: a syndrome including impaired colour vision, cognitive defects, tremor and loss of vibration sensation. Q J Med 2000;93:655–61. Donoghue AM, Dryson EW, Wynn-Williams G. Contrast sensitivity in organicsolvent-induced chronic toxic encephalopathy. J Occup Environ Med 1995;37:1357–63. Garbutt S, Harwood MR, Harris CM. Comparison of the main sequence of reflexive saccades and the quick phases of optokinetic nystagmus. Br J Ophthalmol 2001;85:1477–83. Gobba F, Cavalleri A. Color vision impairment in workers exposed to neurotoxic chemicals. Neurotoxicology 2003;24:693–702. Harwood MR, Mezey LE, Harris CM. The spectral main sequence of human saccades. J Neurosci 1999;19:9098–106. Iregren A, Andersson M, Nyle´n P. Color vision and occupational chemical exposures. I. An overview of tests and effects. Neurotoxicology 2002;23: 719–33.

H. Ojanpa¨a¨ et al. / NeuroToxicology 27 (2006) 1013–1023 Juntunen J. Neurotoxic syndromes and occupational exposure to solvents. Environ Res 1993;60:98–111. Kandel ER, Schwartz JH, Jessell TM. Principles of neural science. New York: McGraw-Hill; 2000 p. 789–92. Kaukiainen A, Riala R, Martikainen R, Akila R, Reijula K, Sainio M. Solventrelated health effects among construction painters with decreasing exposure. Am J Ind Med 2004a;46:627–36. 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 2004b;77:39–46. Lebedev S, Van Gelder P, Tsui WH. Square-root relations between main saccade parameters. Invest Ophthalmol Vis Sci 1996;37:2750–8. Legge GE, Ahn SJ, Klitz TS, Luebker A. Psychophysics of reading - XVI. The visual span in normal and low vision. Vision Res 1997;37:1999–2010. Leigh RJ, Kennard C. Using saccades as a research tool in the clinical neurosciences. Brain 2004;127:460–77. Lezak MD, Howieson DB, Loring DW. Neuropsychological assessment. 4th ed. New York: Oxford University Press; 2004 p. 371–74. Mergler D, Blain L, Lagace´ J-P. Solvent related colour vision loss: an indicator of neural damage? Int Arch Occup Environ Health 1987;59:313–21. Mergler D, Huel G, Bowler R, Frenette B, Cone J. Visual dysfunction among former microelectronics assembly workers. Arch Environ Health 1991;46: 326–34. Morrow LA, Robin N, Hodgson MJ, Kamis H. Assessment of attention and memory efficiency in persons with solvent neurotoxicity. Neuropsychologia 1992;30:911–22. Morrow LA, Ryan CM, Hodgson MJ, Robin N. Alterations in cognitive and psychological functioning after organic solvent exposure. J Occup Med 1990;32(5):444–50. Muttray A, Wolff U, Jung D, Konietzko J. Blue-yellow deficiency in workers exposed to low concentrations of organic solvents. Int Arch Occup Environ Health 1997;70:407–12. ¨ dkvist L, Larsby B, Tham R, Ledin T, Bergholz L. Otoneurological Mo¨ller C, O findings in workers exposed to styrene. Scand J Work Environ Health 1990;16:189–94. ¨ dkvist LM, Ekberg K, Flodin U, Dige N, Sko¨ldestig Niklasson M, Mo¨ller C, O ˚ . Are defecits in the equilibrium system relevant to the clinical investigaA tion of solvent-induced neurotoxicity? Scand J Work Environ Health 1997;23:206–13. Nilson L, Barrega˚rd L, Ba¨ckman L. Trail Making Test in chronic encephalopathy: performance and discriminative potential. Clin Neuropsychol 1999;13:314–27.

1023

Noda H, Fujikado T. Topography of the oculomotor area of the cerebellar vermis in macaques as determined by microstimulation. J Neurophysiol 1987;58:359–78. Noda H, Murakami S, Yamada J, Tamaki Y, Aso T. Saccadic eye movements evoked by microstimulation of the fastigial nucleus of macaque monkeys. J Neurophysiol 1988;60:1036–52. Noda H, Suzuki DA. The role of the flocculus of the monkey in saccadic eye movements. J Physiol 1979;294:317–34. Na¨sa¨nen R, Kaukiainen A, Hero V, Pa¨a¨llysaho J, Mu¨ller K, Hari R, Akila R, Sainio M. Effects of long-term occupational solvent exposure on contrast sensitivity and performance in visual search. Environ Toxicol Pharmacol 2005;19:497–504. Na¨sa¨nen R, Ojanpa¨a¨ H, Kojo I. Effect of stimulus contrast on performance and eye movements in visual search. Vision Res 2001;41:1817–24. Ojanpa¨a¨ H, Na¨sa¨nen R, Kojo I. Eye movements in the visual search of word lists. Vision Res 2002;42:1499–512. O’Regan JK, Le´vy-Schoen A, Jacobs AM. The effect of visibility on eyemovement parameters in reading. Percept Psychophys 1983;34:457–64. Rayner K. Eye movements in reading and information processing: 20 years of research. Psychol Bull 1998;124:372–422. Reinvang I, Borchgrevink HM, Aaserud O, Lie V, Malt UF, Nakstad P, Larsson PG, Gjerstad L. Neuropsychological findings in a non-clinical sample of workers exposed to solvents. J Neurol Neurosurg Psychiatry 1994;57:614–6. Ross J, Morrone MC, Goldberg ME, Burr DC. Changes in visual perception at the time of saccades. Trends Neurosci 2001;24:113–21. Skeel R, Nagra A, VanVoorst V, Olson E. The relationship between performance-based visual acuity screening, self-reported visual acuity, and neuropsychological performance. Clin Neuropsychol 2003;17:129–36. Stollery B. Long-term cognitive sequelae of solvent intoxication. Neurotoxicol Teratol 1996;18:471–6. van der Hoek JAF, Verberk MM, Hageman G. Criteria for solvent-induced chronic toxic encephalopathy: a systematic review. Int Arch Occup Environ Health 2000;73:362–8. ¨ ber die indirekte Sehscha¨rfe. Zeitschrift fu¨r Psychologie und Wertheim TH. U Physiologie der Sinnesorganen 1894; 7, 172–87. Translated in English by Dunsky IL. Peripheral visual acuity. Am J Optom Physiol Opt 1980; 57, 915–24. Wetherill GB, Levitt H. Sequential estimation of points on a psychometric function. Br J Math Stat Psychol 1965;18:1–10. White RF, Proctor SP. Solvents and neurotoxicity. Lancet 1997;349:1239–43. WHO. Organic solvents and the central nervous system. Environ Health 1985;5 Copenhagen.