Usefulness of near-infrared spectroscopy to detect brain dysfunction in children with autism spectrum disorder when inferring the mental state of others

Authors

  • Ryoichiro Iwanaga PhD,

    Corresponding author
    • Division of Physical and Occupational Therapy, Department of Health Sciences, Nagasaki University Graduate School of Biomedical Sciences, Nagasaki, Japan
    Search for more papers by this author
  • Goro Tanaka PhD,

    1. Division of Physical and Occupational Therapy, Department of Health Sciences, Nagasaki University Graduate School of Biomedical Sciences, Nagasaki, Japan
    Search for more papers by this author
  • Hideyuki Nakane MD, PhD,

    1. Division of Physical and Occupational Therapy, Department of Health Sciences, Nagasaki University Graduate School of Biomedical Sciences, Nagasaki, Japan
    Search for more papers by this author
  • Sumihisa Honda PhD,

    1. Division of Physical and Occupational Therapy, Department of Health Sciences, Nagasaki University Graduate School of Biomedical Sciences, Nagasaki, Japan
    Search for more papers by this author
  • Akira Imamura MD, PhD,

    1. Division of Neuropsychiatry, Department of Translation Medical Sciences, Nagasaki University Graduate School of Biomedical Sciences, Nagasaki, Japan
    Search for more papers by this author
  • Hiroki Ozawa MD, PhD

    1. Division of Neuropsychiatry, Department of Translation Medical Sciences, Nagasaki University Graduate School of Biomedical Sciences, Nagasaki, Japan
    Search for more papers by this author

Correspondence: Ryoichiro Iwanaga, PhD, Department Health Sciences, Nagasaki University Graduate School of Biomedical Sciences, 1-7-1 Sakamoto, Nagasaki 852-8520, Japan. Email: iwanagar@nagasaki-u.ac.jp

Abstract

Aims

The purpose of this study was to examine the usefulness of near-infrared spectroscopy (NIRS) for identifying abnormalities in prefrontal brain activity in children with autism spectrum disorders (ASD) as they inferred the mental states of others.

Methods

The subjects were 16 children with ASD aged between 8 and 14 years and 16 age-matched healthy control children. Oxygenated hemoglobin concentration was measured in the subject's prefrontal brain region on NIRS during tasks expressing a person's mental state (MS task) and expressing an object's characteristics (OC task).

Results

There was a significant main effect of group (ASD vs control), with the control group having more activity than the ASD group. But there was no significant main effect of task (MS task vs OC task) or hemisphere (right vs left). Significant interactions of task and group were found, with the control group showing more activity than the ASD group during the MS task relative to the OC task.

Conclusions

NIRS showed that there was lower activity in the prefrontal brain area when children with ASD performed MS tasks. Therefore, clinicians might be able to use NIRS and these tasks for conveniently detecting brain dysfunction in children with ASD related to inferring mental states, in the clinical setting.

INDIVIDUALS WITH AUTISM spectrum disorder (ASD) have been shown to have a consistent deficit in their theory of mind (ToM), the ability to attribute an independent mental state to self and others.[1-3] A number of neuroimaging studies using functional magnetic resonance imaging (fMRI), or positron emission tomography (PET) suggest that several areas of the brain including the medial prefrontal cortex (MPFC) contribute to ToM.[4-9] Lower or abnormal activity in the MPFC during ToM tasks in individuals with ASD has also been reported,[10-12] while activity in the MPFC during the ToM task located in the dorsal subregion in ASD subjects and ventral areas in control subjects, was previously reported.[11] Therefore, lower or abnormal activity in the lower MPFC region would seem to indicate abnormalities related to impairment of ToM in individuals with ASD.

Neuroimaging data from PET and fMRI can provide information to help identify brain dysfunction associated with ASD, but it is difficult or undesirable to apply these techniques to children in the usual clinical setting.[13, 14] To investigate brain activity concerning ToM in children with ASD, neuroimaging techniques that can be used in clinical settings are necessary. Recently, several near-infrared spectroscopy (NIRS) studies have attempted to detect changes in the hemoglobin oxygenation state in the prefrontal region during emotional or mental tasks.[13, 15, 16] Because near-infrared light is non-invasive, repeated measurements are possible. Subjects are also maintained under natural conditions during examinations, can perform the tasks and move naturally, and the apparatus is small and portable.[17] NIRS, therefore, has many advantages for measuring the brain activity of children with ASD. Two previous NIRS studies indicated abnormalities of prefrontal brain activity in ASD patients during a letter fluency task and a self-face recognition task.[14, 17] Because NIRS could measure activities in a part of the MPFC,[18] it might be useful to detect brain activities concerning ToM. Thus, the abnormal activities of prefrontal areas such as the MPFC, related to inferring mental state, in children with ASD might be detectable on NIRS. If the usefulness of NIRS for detecting brain dysfunction related to inferring mental state could be demonstrated, clinicians would be able to use NIRS for examinations in the clinical setting in order to better understand brain dysfunction and to acquire neurophysiological information to diagnose ASD. With this goal in mind, we investigated the usefulness of NIRS to detect brain activity related to inferring mental state.

The purpose of this study was to examine the usefulness of NIRS in identifying abnormalities in prefrontal brain activity concerning the inference of mental state that were previously identified on fMRI and PET in children with ASD.

Methods

Subjects

A total of 16 children with ASD aged between 8 and 14 years old, and 16 age-matched healthy control children participated in this study (Table 1). The children with ASD were recruited from the Nagasaki Autism Society. The subjects were diagnosed at the Nagasaki City Welfare Center or Nagasaki University Hospital. These children were diagnosed according to DSM-IV-TR criteria by expert pediatricians or psychiatrists.[19] Seven of them were diagnosed with autistic disorders, and nine were diagnosed with Asperger disorder. The diagnoses were confirmed by the first author using diagnostic instruments and screening questionnaires including the Pervasive Developmental Disorder–Autism Society Japan Rating Scale (PARS),[20] which is a diagnostic interview scale for ASD developed in Japan, and the Japanese translation of the High Functioning Autism Spectrum Screening Questionnaire (ASSQ-R).[21, 22] All respondents to the PARS and ASSQ-R were mothers. The scores of all children with ASD were above the cut-off scores of the two subscores of the PARS and ASSQ-R. No subjects had any other neurological or psychiatric disorders, and none of subjects was being medicated. For the control group, healthy children were recruited from an elementary school and a junior high school in the same community. There were no significant differences between the two groups regarding age and sex. There were also no significant group differences in the full scale IQ, verbal IQ, or performance IQ on the Wechsler Intelligence Scale for Children–Third Edition (Table 1). Subjects and their parents gave informed consent according to institutional guidelines of the Ministry of Welfare, Japan. This study received prior approval by the Human Investigation Committee of Nagasaki University Graduate School of Biomedical Sciences.

Table 1. Subject characteristics
 ASDControlComparison between ASD and control
  1. ASD, autism spectrum disorder; ASSQ-R, Japanese translation of the High Functioning Autism Spectrum Screening Questionnaire; FIQ, full-scale IQ; PARS, Pervasive Developmental Disorder- Autism Society Japan Rating Scale; PIQ, performance IQ; VIQ, verbal IQ; WISC-III, Wechsler Intelligence Scale for Children–Third Edition.
n1616  
Sex (M : F)14:212:4χ2(1) = 0.821P = 0.365
Age11.5 ± 1.8 (8–14)11.4 ± 1.8 (8–14)t(30) = 0.34P = 0.736
Dominant hand (R/L)15/116/0χ2(1) = 1.032P = 0.310
WISC-IIIFIQ100.1 ± 9.8 (86–121)105.6 ± 5.9 (95–107)t(30) = 1.763P = 0.089
VIQ99.3 ± 13.8 (83–128)107 ± 7.2 (91–120)t(30) = 1.717P = 0.097
PIQ103 ± 12.3 (78–125)105 ± 7.8 (91–119)t(30) = 0.551P = 0.58
PARS-infant13 ± 3.8 (9–19)1.3 ± 1.4 (0–5)t(30) = 11.6P < 0.001
PARS-present20.6 ± 6.3 (14–35)1.2 ± 1.8(0–7)t(30) = 13.0P < 0.001
ASSQ-R26 ± 7.3(19–46)1.9 ± 2.2(0–9)t(30) = 12.5P < 0.001

Tasks and procedures

The experiment consisted of a 30-s pre-task baseline, after which the following sequence was repeated three times: a 20-s expression of a person's mental state task (MS task); a 60-s baseline; a 20-s expression of an object's characteristics task (OC task); and a 60-s baseline.

In the MS task, three different photographs of a person's eyes only (black and white) were shown (Fig. 1). This task was referred to as the Eye Task in a previous study.[3] Although Baron-Cohen et al. used a method in which the subjects chose between two terms for mental state printed under each picture,[3] we used a method in which the subject could comment on the mental state of the presented eye pictures. In the OC task, three color photographs of objects (a truck, a flower, and a church) were used. While each photograph was shown on a screen, subjects were required to express the mental state of a person or to describe an object's characteristics in the photograph. During the pre-task and post-task baseline periods, the Japanese letter for the vowels /a/, /i/, /u/, /e/, and /o/ were presented on the screen, and the subjects were instructed to repeat those sounds.

Figure 1.

Picture in the mental state task (‘angry’).

The subjects sat on a comfortable chair in a dark room with their eyes open during the measurements. Each photograph was projected on a screen in front of the subject 1.5 m away with the projector (EPSON EB-1725) controlled by personal computer using Power Point 2003 (Microsoft). The size of the projected area was 80 cm (width) × 60 cm (height; visual angle 29.9 × 22.6°; 1024 pixels × 768 pixels). The photographs of the OC task were presented in all of the projected area, while eyes-only photographs were presented in an 80-cm (width) × 34-cm (height; visual angle 29.9 × 12.9°) area on a white background.

Before the study, in order to assess the perceived difficulty of these tasks, we asked the opinions of 33 healthy, non-subject volunteers (12 male, 21 female; mean age 21.4 ± 3 years) about the mental state of the people in the MS-task pictures and the difficulty level of all tasks. The pictures for the MS task were taken from a face model book published for artists.[23] The healthy volunteers’ major comments about each picture were as follows: the first picture (Fig. 1) was ‘angry’ (39%) or ‘impatient (27%), the second was a ‘blank stare’ (38%) or ‘neutral’ (24%), and the third was ‘happy’ (79%) or ‘showing pleasure’ (42%). Because of these results, these pictures were called ‘angry’ (Fig. 1), ‘blank stare’, and ‘happy’, respectively. The volunteers were asked to rate the difficulty in assessing the feelings associated with each photograph on a scale of 1 (very difficult) to 9 (very easy), referring to the rating scale degree of the International Affective Picture System (IAPS).[24] The mean of the rated difficulty levels of the MS task (angry, blank stare, happy) and the OC task (truck, flower, church) were 4.6 ± 1.3 and 4.7 ± 1.2, respectively. There were no significant differences between the difficulty levels of the tasks (t(31) = 0.092, P = 0.927).

Evaluation of responses during tasks

We recorded all comments made by the subjects during the tasks. The words in comments were counted for the MS task and OC task. The first and fifth authors judged whether the responses were appropriate or inappropriate in both tasks, while also determining whether responses were emotional expression or non-emotional sentences in the MS task. We also counted the number of emotional expression and non-emotional sentences in the MS task.

NIRS measurements

A 22-channel (CH) NIRS system (Hitachi ETG-4000; Tokyo, Japan) was used. The absorption of near-infrared light was measured, and the concentrations of oxygenated hemoglobin (oxy-Hb) and reduced hemoglobin (deoxy-Hb) were calculated. The total-Hb was calculated as the sum of oxy-Hb and deoxy-Hb. The probes measured the relative concentration of Hb changes at 22 measurement points in a 6 × 12-cm area with 15 probes (3 × 5) on the subjects’ frontal regions, with an inter-probe distance of 3.0 cm. The lowest probes were positioned along the Fp1-Fp2 line, according to the international 10/20 system. According to Okamoto et al., the measurement area should include the rostral frontal cortex and dorsal medial prefrontal cortex (Brodmann areas 9, 10).[25] The absorption of near-infrared light was measured with a time resolution of 0.1 s. The obtained data were analyzed in the integral mode, which calculates average waveform. Pre-task baseline was determined by the last 10 s of the baseline just before the task period. Moving average methods were used to exclude short-term motion artifacts in the analyzed data (moving average: 5 s). The data that clearly contained motion artifacts, based on observations and on the NIRS recordings, were excluded from further analysis.

Data analysis

The number of generated words during the MS task and OC task in each group, and the number of emotional expression sentences and non-emotional sentences generated in the MS task in each group were analyzed using two-way analysis of variance (ANOVA). If a main effect or a significant interaction was identified on ANOVA, one-way ANOVA with a Tukey's HSD post-hoc test was performed to identify significant differences between any combination of group and generated words, or group and expression sentences. Also, the ratios of non-emotional sentences to total number of comment sentences in the MS task were compared between the ASD group and control group using Mann–Whitney U-test. Differences in the behavioral data between children with an autistic disorder and children with Asperger disorder were analyzed using one-way ANOVA.

In this study, oxy-Hb, which is the most sensitive indicator of regional cerebral blood flow,[26] was analyzed. First, the average oxy-Hb waveforms of each subject were calculated by averaging each of the three MS-task and three OC-task data sets. The region of interest (ROI) was the central frontal area on and just above the Fp1 and Fp2 line, because this area is positioned on the rostral prefrontal cortex,[25] and might reflect activation in the lower region of the MPFC. The average oxy-Hb change in task segments in the right hemisphere CH (CH3, CH8, CH12), and left hemisphere CHs (CH2, CH6, CH11) were calculated. We next analyzed the data with a three-way ANOVA for group (ASD vs control), task (MS task vs OC task) and hemisphere (right vs left). If a significant interaction was found, one-way ANOVA with a Tukey's HSD post-hoc test was performed to identify significant differences between any combination of the group, task or hemisphere.

In order to determine the relationship between behavior and the brain activity in the prefrontal area: the correlation between the number of words in the MS task and OC task; the number of emotional expressions and non-emotional sentences; and oxy-Hb changes in each hemisphere were analyzed.

All statistical analysis was performed using SPSS for Windows version 19.0 (SPSS, Chicago, IL, USA), and the significance level was set at P < 0.05.

Results

Behavioral data

The number of generated words in the MS task was 5–42 (22 ± 11.7) in the ASD group, and 0–32 (13.9 ± 9.9) in the control group; and 8–55 (26.8 ± 12.6) in the ASD group and 4–32 (15.5 ± 9.7) in the control group for the OC task. Two-way ANOVA indicated a significant main effect by group (F(1,60) = 12.69, P = 0.001) and no significant main effect by task (F(1,60) = 1.381, P = 0.245) and no significant interaction effects (F(1,60) = 0.302, P = 0.585) for the words generated during tasks. One-way ANOVA indicated significant differences in the number of generated words between each group and task (F(3,60) = 4.791, P = 0.005). On post hoc-test, the word number for the OC task in the ASD group was higher than for the MS task and OC task in the control group. The number of emotional expressions and non-emotional sentences generated in the MS task was 0–10 (3.5 ± 2.7) and 0–13 (3.5 ± 3.9) in the ASD group, and 0–10 (4.2 ± 2.7) and 0–4 (1.4 ± 1.2) in the control group. One ASD subject responded with inappropriate emotional responses such as ‘She looks happy’ for the angry face photograph. Another ASD subject commented on only non-emotional things, and one control subject did not comment (but she was looking at a picture) in the MS task. There was no significant main effect of expression sentences (emotional expression sentences vs non-emotional sentences) in the MS task (F(1,60) = 2.675, P = 0.107) or group (ASD vs control) (F(1,60) = 2.675, P = 0.107). Meanwhile, there was a significant interaction (F(1,60) = 5.243, P = 0.026) between expression sentences and group. One-way ANOVA showed significant differences in the number of expression sentences in each group (F(3,60) = 3.531, P = 0.02). On post-hoc test, the number of non-emotional sentences in the control group was lower than the number of non-emotional sentences in the ASD group and of emotional expression sentences in the control group. The ratio of non-emotional sentences to all expression sentences in the MS task for the ASD group (median, 0.473) was significantly higher than in the control group (median, 0.200; Z = −2.415, P = 0.016). Subjects with ASD tended to refer to the non-emotional aspects of the images in the MS task, for instance ‘This is a woman’, ‘She looks down’.

There were no differences in the behavioral data between children with an autistic disorder and children with Asperger disorder.

NIRS data

The top half of Figure 2 shows grand mean oxy-Hb change during the MS task (black line) and the OC task (dotted line) in the ASD group. The mean oxy-Hb change in the ASD group was <0 during the MS-task period in both hemispheres (right, −0.011 ± 0.028 mmol/L·mm; left, −0.028 ± 0.033 mmol/L·mm), while oxy-Hb change was >0 in both hemispheres (right, 0.013 ± 0.03 mmol/L·mm; left, 0.012 ± 0.032 mmol/L·mm) during the OC-task period. The bottom half of Figure 2 shows the grand mean changes in oxy-Hb concentration during the MS task (black line) and the OC task (dotted line) in the control group. The control group had higher oxy-Hb change in both hemispheres during the MS-task period (right, 0.11 ± 0.034 mmol/L·mm; left, 0.083 ± 0.034 mmol/L·mm) and the OC-task period (right: 0.039 ± 0.032 mmol/L·mm, left; 0.037 ± 0.028 mmol/L·mm). For oxy-Hb change, significant main effects for group were observed, for which the control group had higher activity than the ASD group (F(1,120) = 15.432, P < 0.001). No significant effects were seen for either ‘task’ (F(1,120) = 0.534, P = 0.466) or ‘hemisphere’ (F(1, 120) = 0.419, P = 0.518). For oxy-Hb change, ANOVA indicated no significant interaction of task × group × hemisphere (F(1,120) = 0.013, P = 0.911), task × hemisphere (F(1,120) = 0.291, P = 0.590), or group × hemisphere (F(1,120) = 0.026, P = 0.873). There was, however, a significant interaction of task × group (F(1,120) = 6.368, P = 0.013). One-way ANOVA showed significant differences in oxy-Hb change between each task in each group (F(3,60) = 3.902, P = 0.013). On post-hoc test, oxy-Hb change in the MS task of the ASD group was lower than in the control group (P = 0.01), but there were no significant differences among the others.

Figure 2.

Mean change in Hb concentration during the verbal explanation of (image) a mental state and (image) an object's characteristics in (a) the autism spectrum disorder group and (b) the control group. A total of 22 channels were mounted on the frontal region (inset). Task periods are shown between the two vertical parallel bars in each graph.

There were no correlations between oxy-Hb change in each hemisphere and behavioral data.

Discussion

This study explored the usefulness of NIRS for detecting differences in brain activity between children with ASD and children without developmental disorders during the inference of mental states of others.

The MS task and OC task were the same in that they involved expression of opinions about photographs. In addition, there were no significant differences in difficulty levels between the tasks as rated by non-subject controls. The tasks differed, however, in that one required inference of a person's mental state and the other did not. Therefore, the differences in oxy-Hb changes between tasks would seem to indicate cortex activation related to inference of the mental state of others.

The behavioral data indicated that ASD subjects referred more to the non-emotional aspects of the images of a person's eyes. This might be explained by previous studies that noted a deficit in facial emotion recognition in individuals with ASD.[27] ASD subjects tended to comment more in the OC task than the control group in both tasks. These differences, however, were not consistent with the oxy-Hb change being lower in the ASD group than the control group in the MS task. Furthermore, behavioral data showed no significant correlation with oxy-Hb change in each hemisphere. This suggests that there is no relationship between differences in behavior and the differences in oxy-Hb change between both groups in the MS task. We believe that the lower oxy-Hb change in the ASD group than the control group was not caused by brain activity related to word or sentence production.

The interaction of task × group for oxy-Hb change on post-hoc test indicates that activation in the prefrontal region was lower in the ASD group relative to the control group during the MS task. Lower activity in the prefrontal area during the MS task in the ASD group might suggest that there is a deficit in prefrontal region activity related to inferring the mental state of others in children with ASD, as reported in previous studies.[10-12] Because the CH in ROI were placed in the central position within 3 cm above the Fp1 and Fp2 line, oxy-Hb change in the ROI should reflect activation in and above the Brodmann 10 area, including the lower region of the MPFC.[25] Therefore, the significant interaction of task × group for oxy-Hb change in ROI might reflect the differences in activity in the lower region of the MPFC related to ToM between children with ASD and controls. This result would be consistent with a previous study using fMRI in adults with ASD.[11] Although some previous studies noted differences in activity in the prefrontal area between hemispheres during ToM tasks,[8, 9] there were no hemispheric differences in the present study. We had thought that the children with ASD would show lower activity in both prefrontal regions when inferring the mental state of another person while looking at the eye area only.

Therefore NIRS might be able to identify abnormal activity as reported in previous studies with ASD individuals using fMRI.[10-12] NIRS combined with the present tasks could be useful for identifying brain dysfunction related to inferring mental state, and for acquiring neurophysiological information to diagnose ASD in the usual clinical settings.

This study had some limitations. The MS tasks and OC tasks show differences not only in ‘inferring mental state or not’, but also ‘colored or not’ and ‘facial or not’. Therefore, the results might be affected by other factors unconnected with inferring mental states. Further study excluding these possible affective factors should be conducted.

Acknowledgment

This study was supported by a Grant-in-aid for Young Scientists (B) no. 17790712 from the Japan Society for the Promotion of Science. All authors declare that they have no conflict of interest.

Ancillary