Voxel-based global-brain functional connectivity alterations in first-episode drug-naive patients with somatization disorder

Journal of Affective Disorders 254 (2019) 82–89

Contents lists available at ScienceDirect

Journal of Affective Disorders journal homepage: www.elsevier.com/locate/jad

Research paper

Voxel-based global-brain functional connectivity alterations in first-episode drug-naive patients with somatization disorder

T

Pan Pana,b, Yangpan Oua,b, Qinji Suc, Feng Liud, Jindong Chena,b, Jingping Zhaoa,b, ⁎ Wenbin Guoa,b, a

Department of Psychiatry, The Second Xiangya Hospital of Central South University, Changsha, Hunan 410011, China National Clinical Research Center on Mental Disorders, Changsha, Hunan 410011, China c Mental Health Center of the First Affiliated Hospital, Guangxi Medical University, Nanning, Guangxi 530007, China d Department of Radiology, Tianjin Medical University General Hospital, Tianjin 300000, China b

ARTICLE INFO

ABSTRACT

Keywords: Somatization disorder Global-brain functional connectivity Network Functional magnetic resonance imaging

Background: Altered functional connectivity (FC) is associated with the pathophysiology of patients with somatization disorder (SD). However, inconsistent results were obtained due to different selections of regions of interest (ROIs) in previous researches. This study aims to examine voxel-wise brain-wide FC alterations in patients with first-episode, drug-naive SD in an unbiased way. Methods: A total of 25 patients with SD and 28 age-, sex-, and education-matched healthy controls underwent resting-state functional magnetic resonance imaging. Global-brain FC (GFC) was applied to analyze the images. Receiver operating characteristic curves and support vector machine were used to differentiate the patients from the controls. Results: Compared with healthy controls, patients with SD exhibited increased GFC in the right inferior temporal gyrus (t-value = 4.0663, p < 0.001) and left superior occipital gyrus (t-value = 3.8197, p < 0.001). Decreased GFC in the right insula (t-value = ‒4.1667, p < 0.001) was observed in the patients relative to the controls. The GFC values in the right insula of the patients were positively correlated to their scores of the sleep subscale of the Hamilton Depression Scale (r = 0.455, p = 0.022) and the lie subscale of the Eysenck Personality Questionnaire (r = 0.436, p = 0.029). A combination of GFC values in the right insula and left superior occipital gyrus can be applied to discriminate the patients from the controls with optimal sensitivity, specificity, and accuracy of 88.00%, 85.71%, and 86.79%, respectively. Conclusions: Our study indicates that patients with SD show abnormal GFC in the brain areas of insula-centered sensorimotor network, and thus providing a new perspective for understanding the pathological changes of FC in SD. Furthermore, a combination of the GFC values in the right insula and left superior occipital gyrus may be used as a potential biomarker to identify the patients from the controls.

1. Introduction Characterized by recurring, multiple, and clinically significant complaints with somatic symptoms, somatization disorder (SD) is considered as a difficult problem in psychiatric practice. The somatic symptoms often refer to gastrointestinal, cardiorespiratory, urogenital, and other internal systems or musculoskeletal problems (Lipowski, 1988; Zhang et al., 2015). Patients with SD may have associated comorbidities, including personality disorders, major depressive disorder, panic disorder, and substance-related disorders (LaFrance, 2009). This disorder has a prevalence rate of 4%–7% in the general population and occurs more often in females than in males (Rief et al., 2001). ⁎

Compared with other demographic characteristics such as occupation and income situation, individuals who are older, have less years of education, and work at home are more likely to suffer from SD (Krishnan et al., 2013). Patients with SD accompanied with anxiety have the feeling that their demands for further physical examinations and physical therapies are not being met (Rohlof et al., 2014). Hence, they often contact with their general physicians and repeatedly undergo medical examinations. However, the diagnosis of SD is difficult for general physicians, and thus leading to a high rate of health care utilization, a decline of quality of life, absenteeism from work, and reduction in productivity (Barsky et al., 2005; Koch et al., 2007). Despite numerous researches in SD, the pathophysiology of this

Corresponding author at: Wenbin Guo. Department of Psychiatry, The Second Xiangya Hospital of Central South University, Changsha, Hunan 410011, China. E-mail address: [email protected] (W. Guo).

https://doi.org/10.1016/j.jad.2019.04.099 Received 14 February 2019; Received in revised form 27 April 2019; Accepted 30 April 2019 Available online 30 April 2019 0165-0327/ © 2019 Elsevier B.V. All rights reserved.

Journal of Affective Disorders 254 (2019) 82–89

P. Pan, et al.

disease remains unclear. Recently, brain structural and functional alterations related to SD have been explored using different neuroimaging techniques, suggesting that SD is not a disease with abnormal activity in one isolated brain region but a chronic mental disorder involving lesions in multiple brain networks (Boeckle et al., 2016). One of the most examined networks is the default mode network (DMN), which includes medial prefrontal cortex, ventral anterior cingulate cortex, posterior cingulate cortex/precuneus, and dorsomedial thalamus. Functional deficits have been observed in brain areas of the DMN (Otti et al., 2013; Su et al., 2014). The dissociation pattern of the DMN may result in somatic symptom complex and emotional, cognitive, behavioral changes (Gong et al., 2011). Patients with SD showed significantly increased functional connectivity strength (FCS) in the right inferior temporal gyrus (ITG), which belonged to the DMN (Su et al., 2015). Other researchers have reported that the patients exhibited significant alterations in white matter integrity in the cingulum, inferior fronto-occipital fasciculus, anterior thalarmic radiation, and corticospinal tract (Zhang et al., 2015). Significant decrease in connectivity was detected in the patients (Wang et al., 2016). For example, decreased FCS in the left pallidum (Guo et al., 2017), decreased fractional anistropy in the right inferior fronto-occipital fasciculus (Zhang et al., 2015), and decreased voxel-mirrored homotopic connectivity in the insula and angular gyrus/supramarginal gyrus (Su et al., 2016) have been reported in patients with SD. However, the abovementioned findings are inconsistent regarding specific brain regions across studies. For example, patients with SD showed increased connectivity between the left/right Crus I and the left/right angular gyrus in one study (Wang et al., 2016), whereas the patients exhibited decreased homotopic connectivity in the insula and angular gyrus in another study (Su et al., 2016). One important factor accounting for the inconsistent findings is that many imaging studies have focused on functional connectivity (FC) between brain areas of a specified network rather on adopting a whole-brain examination (Li et al., 2017). Therefore, researches are biased by the choice of network of interest and may not include the most significantly different areas that may represent the core pathophysiological alterations of SD. Given this background, resting-state functional magnetic resonance imaging (fMRI) was used to obtain a view of global-brain functional connectivity (GFC) in patients with SD. The present study aimed to examine the GFC differences between patients with first-episode, drugnaive SD and healthy controls. Based on the aforementioned studies, we hypothesized that patients with SD would show abnormal GFC in certain brain areas, especially in the DMN. In addition, altered FC is expected to be correlated to clinical measurements in patients with SD. We further examined whether the GFC values in these brain regions might be considered as potential image biomarkers to differentiate the patients from the controls using receiver operating characteristic (ROC) curves and support vector machine (SVM).

depression, anxiety, and somatization. Eysenck Personality Questionnaire (EPQ) (Eysenck and Eysenck, 1972) was used to evaluate personality dimension. Wisconsin Card Sorting Test (WCST) (Greve et al., 2005) and digit symbol coding of Wechsler Adult Intelligence Scale (WAIS) (Kaufman and Lichtenberger, 2005) were used to determine cognitive function. The participants shared the following exclusion criteria: severe medical or neurological diseases; mental retardation; any history of loss of consciousness; substance abuse; other psychiatric disorders, such as schizophrenia, anxiety disorders, bipolar disorders, or personality disorders; and any contraindications for MRI. Comorbidity with major depressive disorder was allowed given that patients with SD had a high rate of comorbidity with depression. However, the onset of depressive symptoms should occur after the emergence of somatic symptoms in the patients. In addition, potential controls with a first-degree relative who has a history of neuropsychiatric disorders were excluded. The study was approved by the Local Ethics Committee of the First Affiliated Hospital of Guangxi Medical University. A written informed consent was obtained from all the participants. 2.2. Image acquisition and preprocessing A Siemens 3T scanner was used to capture resting-state scans. The imaging data were preprocessed using the DPABI software, which is designed to process data automatically (Yan et al., 2016). Details of image acquisition and preprocessing are provided in the supplementary files. 2.3. GFC analysis For each participant, voxel-wise GFC was computed using the FC between a given voxel and all other voxels within a gray matter mask in Matlab (Cui et al., 2018; Ding et al., 2019). We used a threshold of 0.2 to create a gray matter mask in SPM8, which indicated that voxels with the probability of >0.2 would be classified as gray matter. In addition, this threshold was set to remove the weak correlations possibly arising from signal noise (Liu et al., 2015b). Pearson correlation coefficients (r) between time series of voxels of all pairs were calculated to create the whole brain FC matrix for every participant. To improve the normality of the data distribution, we used a Fisher r-to-z to transform individual correlation matrices to a z-score matrix (Wang et al., 2015). The mean coefficient of a given voxel with all voxels was defined as the GFC of this voxel. The frame-wise displacement (FD) value for each participant was calculated according to a previous study (Power et al., 2012). The mean FD, age, and years of education were used as covariates of no interest in group comparisons by using two-sample t-tests. The significance level was set at p < 0.05 corrected by the threshold-free cluster enhancement (TFCE) correction method. Since SD occurs more often in the females, we performed the same analysis on the female subjects (21 female patients with SD and 22 female controls).

2. Methods and materials 2.1. Participants A total of 26 patients with SD were recruited from the First Affiliated Hospital of Guangxi Medical University, and 30 healthy controls were recruited from the local community. All participants were right-handed and aged from 18 to 50 years old. The diagnosis of SD was made based on the criteria of the Structured Clinical Interview of the Diagnostic and Statistical Manual of Mental Disorders-IV (SCID), patient version (Gorgens, 2011). All patients were first-episode and drugnaive. Healthy controls were screened by SCID, nonpatient edition ((Gorgens, 2011)). All subjects were assessed with the following tests at the scan day: Hamilton Depression Scale (HAMD) (Hamilton, 1960), Hamilton Anxiety Scale (HAMA) (Hamilton, 1959), and somatization subscale of SCL-90 (Derogatis et al., 1976) to assess the symptom severity of

2.4. Correlation analysis Mean z values were extracted from brain clusters with abnormal GFC. Pearson correlation analyses were performed to explore the relationship of GFC values and clinical variables in the SD group after the normality of the data being checked. Specifically, the correlations between the mean GFC values within regions showing significant between-group GFC differences and the somatization subscale of SCL-90, EPQ were analyzed. The correlation results were Bonferroni corrected at p < 0.05. 83

Journal of Affective Disorders 254 (2019) 82–89

P. Pan, et al.

2.5. ROC analysis

Table 1 Characteristics of participants.

A public software for ROC analyses (ROC version 5.07; http://www. stanford.edu/∼yesavage/ROC.html) was used in the present study. This software is specifically designed to assess clinical database for predictive variables (Main et al., 2017). The main analysis tool of ROC is a curve drawn on a two-dimensional plane. The abscissa of the plane is false positive rate and the ordinate is true positive rate. The curve is obtained when subjects acquire different results due to different criteria under specific stimulating conditions. ROC analysis aimed to examine whether abnormal GFC in a single brain region could be used to identify the patients from the controls. 2.6. Classification analysis using SVM SVM using the LIBSVM software package (Chang and Lin, 2011) was employed to examine whether a combination of abnormal GFC values can be applied to distinguish the patients from the controls. Parameter optimization was conducted by using the Grid search approach and the Gaussian radial basis function kernels. A "leave-one-out" cross-validation approach was applied by the LIBSVM software to obtain the highest sensitivity and specificity (Liu et al., 2015a; Yan et al., 2016). 3. Results 3.1. Subjects

Variables

Patients (n = 25)

Controls (n = 28)

p value

Age (years) Sex (male/female) Years of education (years) FD (mm) Illness duration (months) Somatization subscale of SCL90 HAMD Anxiety/somatization Weight Cognition Retardation Sleep HAMA Somatic anxiety Psychotic anxiety Digit symbol-coding of WAIS EPQ Extraversion Psychoticism Neuroticism Lie WCST Number of categories Achieved Number of errors Number of Perseverative Errors

41.00 ± 10.76 4/21 7.72 ± 4.39 0.08 ± 0.03 59.12 ± 62.22 28.48 ± 10.37

38.71 ± 9.59 6/22 7.82 ± 2.59 0.10 ± 0.05

0.42b 0.73a 0.92b 0.02b

14.32 ± 3.44

<0.001b

18.84 ± 7.31 8.36 ± 2.80 0.60 ± 0.71 2.00 ± 1.80 4.20 ± 2.20 3.68 ± 2.32 22.96 ± 10.95 10.80 ± 5.77 12.16 ± 6.11 8.28 ± 2.87

2.60 ± 1.83 1.43 ± 1.03 0.00 ± 0.00 0.07 ± 0.38 0.00 ± 0.00 0.36 ± 0.95 0.53 ± 0.99 0.25 ± 0.52 0.29 ± 0.81 9.64 ± 2.15

<0.001b <0.001b <0.001b <0.001b <0.001b <0.001b <0.001b <0.001b <0.001b 0.06b

46.84 ± 11.02 50.52 ± 9.01 57.36 ± 9.18 49.44 ± 12.31

49.75 ± 9.65 45.00 ± 8.54 46.78 ± 10.24 47.96 ± 11.01

0.31b 0.03b <0.001b 0.65b

3.52 ± 1.76 22.84 ± 9.12 20.04 ± 9.48

3.89 ± 1.66 24.71 ± 8.91 22.82 ± 8.72

0.43b 0.45b 0.27b

FD, framewise displacement; HAMD, Hamilton depression scale; HAMA, Hamilton anxiety scale; SCL-90, symptom checklist-90; EPQ, Eysenck personality questionnaire; WAIS, Wechsler Adult intelligence scale; WCST, Wisconsin card sorting test. a The p value for sex distribution was obtained by a chi-square test. b The p values were obtained by two samples t-tests.

The data of two controls and one patient were excluded from further analysis due to excessive head motion. Therefore, the final sample included 25 patients with SD and 28 controls. No significant differences were observed between the patients and the controls in terms of age, sex ratio, education level, scores of the Psychoticism/Extraversion/Lie subscales of EPQ, digit symbol coding of WAIS, and WCST. However, the patients had higher scores in HAMA, HAMD, somatization subscale of SCL-90, and EPQ neuroticism than the controls (Table 1). Among them, the patients obtained higher scores in items of HAMD, including anxiety/somatization, sleep, weight, cognition, retardation, and somatic anxiety, as well as in psychotic anxiety of HAMA than the controls. The controls obtained higher FD values than the patients (Table 1).

3.4. ROC and SVM results ROC analyses showed that the GFC values in the right ITG could be applied to identify the patients form the controls with optimal specificity (92.00%) and sensitivity (71.43%) (Table 3 and Fig. 3). Furthermore, SVM analyses exhibited that a combination of the GFC values in the right insula and left superior occipital gyrus can correctly identify the patients from the controls with a sensitivity of 88.00%, specificity of 85.71%, and accuracy of 86.79% (Fig. 4).

3.2. Group differences in GFC among the brain regions As shown in Fig. 1 and Table 2, compared with the controls, the patients exhibited increased GFC in the right ITG (t-value = 4.0663, p < 0.001) and left superior occipital gyrus (t-value = 3.8197, p < 0.001). By contrast, decreased GFC in the right insula (tvalue = ‒4.1667, p < 0.001) was found in the patients relative to the controls. No other differences were observed in the patients. As shown in Supplemented with the attachments, the female patients exhibited increased GFC in the right ITG (t-value = 3.0365, p < 0.001) and left superior occipital gyrus (t-value = 3.9116, p < 0.001) relative to the female controls. Meanwhile, decreased GFC in the right insula (t-value = - 3.3652, p < 0.001) was observed in the female patients compared with the female controls.

4. Discussion In the present study, patients with SD show abnormal GFC in brain areas of insula-centered sensorimotor network. Specifically, significantly increased GFC in the right ITG and left superior occipital gyrus as well as decreased GFC in the right insula are observed in the patients with SD relative to the controls. Another important finding is that the GFC values in the right insula are positively correlated with the scores of the HAMD sleep scores or EPQ lie scores in the patients. Meanwhile, a combination of the GFC values in the right insula and left superior occipital gyrus can correctly identify the patients from the controls with optimal sensitivity, specificity, and accuracy. Previously, the seed-based FC method (region-of-interest, ROI) has been used to explore brain mechanisms in patients with SD (Li et al., 2018). Although the findings from ROI are informative, this method is insufficient. The results may be biased by the selection of ROIs based on prior assumptions, and it is natural for different studies acquiring different results using different ROIs. Furthermore, it may not cover the important brain areas related to the core pathological alterations of SD. By contrast, we examine FC abnormalities in patients with SD in a voxel-wise brain-wide way. GFC is used to analyze the modalities of FC from the perspective of whole-brain FC, providing the understanding on

3.3. Correlations between abnormal GFC and clinical variables As shown in Fig. 2, significantly positive correlations were observed between the GFC values in the right insula and the sleep subscale of the HAMD (r = 0.455, p = 0.022) or the lie subscale of the EPQ (r = 0.436, p = 0.029) in the patients. Meanwhile, no significant correlations were found between the GFC values in the right insula and the sleep subscale of the HAMD (r = ‒0.080, p = 0.687) or the lie subscale of the EPQ (r = 0.085, p = 0.667) in the controls. 84

Journal of Affective Disorders 254 (2019) 82–89

P. Pan, et al.

Fig. 1. Abnormal GFC in patients with somatization disorder relative to controls. GFC, global-brain functional connectivity.

the work mechanism of the brain at a systematic level. The potential FC alterations may be detected by this method, and the findings are reported in an unbiased way. The present study reveals increased GFC in the right ITG of patients with SD, which is consistent with the previous study (Su et al., 2015). The ITG is located at the lateral and inferior surface of the temporal neocortex, and is considered as a major association area that subserves auditory processing and emotion regulation (Su et al., 2015). In addition, neuroimaging data support the linking of the temporal lobe to alexithymia (Aust et al., 2014; Deng et al., 2013). Patients with SD generally have difficulties of expression and cannot distinguish bodily sensations of emotional arousal when being asked to describe their

subjective feelings in clinical settings. Hence, this finding is consistent with the unexplained symptoms in patients with SD. The insula is hidden in the depth of the cerebral hemisphere by the overlying frontal and temporal opercula, which are located deep inside the lateral sulcus of the sylvian fissure (Ture et al., 1999). Previous studies revealed that the insula had a strong reciprocal connectivity with the prefrontal cortex (Cauda et al., 2011), suggesting that the insula played multiple roles in processing visceral sensory, gustatory, olfactory, vestibular/auditory, visual, verbal, pain, sensory/motor information, and inputs related to music and eating, as well as in modulating attention and emotion (Garcíacampayo et al., 2009; Stein and Muller, 2008). Moreover, the insula participates in the conditioned

Table 2 Regions with abnormal GFC in the patients. Cluster location

Peak (MNI) x

y

z

Right inferior temporal Gyrus Left superior occipital gyrus Right insula

54 −15 42

−24 −90 0

−30 42 −3

Number of voxels

T value

33 25 49

4.0663 3.8197 −4.1667

GFC, global-brain functional connectivity; MNI, Montreal Neurological Institute. a A positive/negative T value represents increased/decreased GFC in the patients relative to the controls. 85

p value <0.001 <0.001 <0.001

Journal of Affective Disorders 254 (2019) 82–89

P. Pan, et al.

Fig. 2. Positive correlations were observed between the GFC values in the right insula and the sleep subscale of the Hamilton Depression Scale or the lie subscale of the Eysenck Personality Questionnaire in patients with somatization disorder. No significant correlations were found between the GFC values in the right insula and the sleep subscale of the Hamilton Depression Scale or the lie subscale of the Eysenck Personality Questionnaire in the controls. GFC, global-brain functional connectivity. Table 3 ROC analyses for differentiating the patients from the controls by using the GFC values. Brain regions

Area under the curve

Cut-off point

Sensitivity

Specificity

Right inferior temporal gyrus Left superior occipital gyrus Right insula

0.813 0.764 0.790

−0.3340a −0.3523 0.0610

92.00% (23/25) 92.00% (23/25) 96.00% (24/25)

71.43% (20/28) 64.29% (18/28) 57.14% (16/28)

a By this cut-off point, the GFC values in the right inferior temporal gyrus could correctly classify 23 of 25 patients and 20 of 28 controls, resulting in a sensitivity of 92.00% and a specificity of 71.43%. The meanings of other cut-off points were similar. ROC, receiver operating characteristic curve; GFC, global-brain functional connectivity.

Fig. 3. Differentiating the patients from the controls by using the GFC values in the A. Right Insula, B. Left Superior Occipital Gyrus, and C. Right Inferior Temporal Gyrus.

86

Journal of Affective Disorders 254 (2019) 82–89

P. Pan, et al.

Fig. 4. Visualization of classifications in SVM by using a combination of the GFC values in the right insula and left superior gyrus. (Left) 3D view of the classified accuracy with the best parameters; (Right) red crosses represent the controls, green crosses represent the patients, and blue circles represent the support vectors. SVM, support vector machine, GFC, global-brain functional connectivity.

for differentiating the patients from the controls with the sensitivity, specificity, and accuracy was more than 0.8. Hence, we inferred that a combination of a decreased GFC value in the right insula and an increased GFC value in the left superior occipital gyrus can be used as a potential biomarker to identify the patients from the controls. The present study has several new points. First, we examine FC abnormalities in patients with SD in a voxel-wise brain-wide way. Many previous studies have focused on FC between preselected brain regions using a ROI method (Wang et al., 2016). The choices of different ROIs will yield different results in different studies. In addition, some studies may not cover the most important brain regions associated with the core pathological alterations of SD. By contrast, the GFC method used in the present study aims to employ the FC abnormalities in a voxel-wise brain-wide way. This method can detect potential FC alterations and report the findings in an unbiased way. Second, SVM analysis was used to detect whether a combination of the GFC values in the right insula and left superior occipital gyrus can serve as potential biomarkers to differentiate the patients from the controls. Finally, drug-naive, firstepisode patients were recruited to avoid possible effects of prior treatment effects on brain function in the present study. Several limitations should be considered in addition to the small sample size. First, patients showed higher levels of depression and anxiety in the present study. Similar results were obtained when we used the HAMA/HAMD scores as covariates in the group comparisons (data were not reported here). Thus, higher levels of accompanied symptoms of depression and anxiety had limited effects on our results. Second, given the fact that SD occurs in female more than in males approximately five times, a higher proportion of females were recruited for the study, which might restrict the generalizability of the findings. However, the same analysis was performed on the females subjects according to the feature that the females were more vulnerable to SD. The result was similar to the original one, indicating that gender had limited effect on GFC. Third, gray/white matter alterations are not examined in the present study. Hence, the mechanism of gray/white matter alterations underlying GFC remains unclear. Finally, the fMRI data were acquired at rest and not with a specific task. Therefore, abnormal GFC in the present study may reflect the pathological changes of SD in general. Despite the limitation, our study is the first to examine voxel-wise brain-wide FC in SD, which indicates that patients show abnormal GFC in the brain areas of insula-centered sensorimotor network. A combination of the GFC values in the right insula and left superior occipital gyrus may be used as a potential biomarker to identify the patients from the controls. The present study thus provides new insights for

aversive learning, perception of affective and motivational components in pain, sleep, mood stability, stress-induced immunosuppression and language (Flynn, 1999). In addition, the insula plays an important role in detecting, integrating, and filtering information connected with somatic, autonomous, and emotional states (Hoeft, 2012). Decreased FCS in the insula have been reported in SD (Su et al., 2016). The results of the present study showed that patients with SD had decreased GFC in the right insula. The GFC values in the right insula were positively correlated to HAMD sleep and EPQ lie scores in the patients. Abnormal GFC in the insula may affect the sleep of patients with SD and provide theoretical basis for the pathophysiological of sleep disorder in SD. In addition, reduced ability to conceal and weakening ability to defend in patients with SD may be related to the decreased GFC values in the right insula. Previous neuroimaging studies show that the right insula plays an important role in sensorimotor network, which exhibits main functions including activity inhibition and interoceptive awareness (Uddin, 2015). Therefore, the insula may act as an integrate center of the sensorimotor network, and decreased GFC in the right insula may disrupt its integrated role in patients with SD. Notably, increased GFC values in the left superior occipital gyrus were observed in SD in the present study. The occipital region is associated with visual processing. The occipital cortex seems to play its supplementary role in the attention (Ahrendts et al., 2011; Wang et al., 2007). Moreover, the left superior occipital gyrus plays an important role in the visual network, which belongs the sensor cortex systems (Mantini et al., 2007). Increased GFC in the left superior occipital gyrus has been detected in the patients, suggesting that the present patients may have trouble in processing visual information. Abnormal GFC values might be used as potential markers to discriminate the patients from the controls. First, ROC analysis was used to detect whether abnormal GFC in a single brain cluster can be utilized as potential markers to differentiate the patients from the controls. Theoretically, the area under the curve of these regions was more than 0.7, which was an acceptable accuracy for established diagnostic indicators (Swets, 1988). However, sensitivity or specificity less than 0.75 seemed to be an indicator with poor accuracy (Gong et al., 2011). In this study, ROC analysis showed that the GFC values in these brain regions, including right ITG, left superior occipital gyrus, and right insula, might not be used as optimal markers to individually identify the patients from the controls. The SVM analyses were further conducted to determine whether a combination of the GFC values of two brain clusters could discriminate the patients from the controls with optimal sensitivity and specificity. The SVM results indicated that a combination of the GFC values in the right insula and left superior gyrus 87

Journal of Affective Disorders 254 (2019) 82–89

P. Pan, et al.

understanding the pathological changes of FC in SD.

Intell. Syst. Technol. 2, 1–27. Cui, X., Liu, F., Chen, J., Xie, G., Wu, R., Zhang, Z., Chen, H., Zhao, J., Guo, W., 2018. Voxel-wise brain-wide functional connectivity abnormalities in first-episode, drugnaive patients with major depressive disorder. Am. J. Med. Genet. B. Neuropsychiatr. Genet. 177, 447–453. Deng, Y., Ma, X., Tang, Q., 2013. Brain response during visual emotional processing: an fMRI study of alexithymia. Psychiatry Res. 213, 225–229. Derogatis, L.R., Rickels, K., Rock, A.F., 1976. The SCL-90 and the MMPI: a step in the validation of a new self-report scale. Br. J. Psychiatry 128, 280–289. Ding, Y., Ou, Y., Su, Q., Pan, P., Shan, X., Chen, J., Liu, F., Zhang, Z., Zhao, J., Guo, W., 2019. Enhanced global-brain functional connectivity in the left superior frontal gyrus as a possible endophenotype for schizophrenia. Front. Neurosci. 13, 145. Eysenck, S.B., Eysenck, H.J., 1972. The questionnaire measurement of psychoticism. Psychol. Med. 2, 50–55. Flynn, F.G., 1999. Anatomy of the insula functional and clinical correlates. Aphasiology 13, 55–78. Gorgens, K.A., 2011. Structured Clinical Interview For DSM-IV (SCID-I/SCID-II). In: Kreutzer, J.S., DeLuca, J., Caplan, B. (Eds.), Encyclopedia of Clinical Neuropsychology. Springer, New York, New York, NY, pp. 2410–2417. Garcia-Campayo, J., Fayed, N., Serrano-Blanco, A., Roca, M., 2009. Brain dysfunction behind functional symptoms: neuroimaging and somatoform, conversive, and dissociative disorders. Curr. Opin. Psychiatry 22, 224–231. Gong, Q., Wu, Q., Scarpazza, C., Lui, S., Jia, Z., Marquand, A., Huang, X., McGuire, P., Mechelli, A., 2011. Prognostic prediction of therapeutic response in depression using high-field MR imaging. Neuroimage 55, 1497–1503. Greve, K.W., Stickle, T.R., Love, J.M., Bianchini, K.J., Stanford, M.S., 2005. Latent structure of the Wisconsin Card Sorting Test: a confirmatory factor analytic study. Arch. Clin. Neuropsychol. 20, 355–364. Guo, W., Liu, F., Chen, J., Wu, R., Li, L., Zhang, Z., Chen, H., Zhao, J., 2017. Anatomical distance affects cortical-subcortical connectivity in first-episode, drug-naive somatization disorder. J. Affect. Disord. 217, 153–158. Hamilton, M., 1959. The assessment of anxiety states by rating. Br. J. Med. Psychol. 32, 50–55. Hamilton, M., 1960. A rating scale for depression. J. Neurol. Neurosurg. Psychiatry 23, 56–62. Hoeft, F., Gabrieli, J.D., Whitfield-Gabrieli, S., Haas, B.W., Bammer, R., Menon, V., Spiegel, D., 2012. Functional brain basis of hypnotizability. Arch. Gen. Psychiatry 69, 1064–1072. Kaufman, A.S., Lichtenberger, E.O., 2005. Assessing Adolescent and Adult Intelligence, 3rd Edition. John Wiley Professio. 69, 1293–1297. Koch, H., van Bokhoven, M.A., ter Riet, G., van der Weijden, T., Dinant, G.J., Bindels, P.J., 2007. Demographic characteristics and quality of life of patients with unexplained complaints: a descriptive study in general practice. Qual. Life Res. 16, 1483–1489. Krishnan, V., Sood, M., Chadda, R.K., 2013. Caregiver burden and disability in somatization disorder. J. Psychosom. Res. 75, 376–380. LaFrance Jr., W.C., 2009. Somatoform disorders. Semin. Neurol. 29, 234–246. Li, T., Wang, Q., Zhang, J., Rolls, E.T., Yang, W., Palaniyappan, L., Zhang, L., Cheng, W., Yao, Y., Liu, Z., Gong, X., Luo, Q., Tang, Y., Crow, T.J., Broome, M.R., Xu, K., Li, C., Wang, J., Liu, Z., Lu, G., Wang, F., Feng, J., 2017. Brain-wide analysis of functional connectivity in first-episode and chronic stages of schizophrenia. Schizophr. Bull. 43, 436–448. Li, R., Liu, F., Su, Q., Zhang, Z., Zhao, J., Wang, Y., Wu, R., Zhao, J., Guo, W., 2018. Bidirectional causal connectivity in the cortico-limbic-cerebellar circuit related to structural alterations in first-episode, drug-naive somatization disorder. Front. Psychiatry 9, 162. Lipowski, Z.J., 1988. Somatization: the concept and its clinical application. Am. J. Psychiatry 145, 1358–1368. Liu, F., Guo, W., Fouche, J.P., Wang, Y., Wang, W., Ding, J., Zeng, L., Qiu, C., Gong, Q., Zhang, W., Chen, H., 2015a. Multivariate classification of social anxiety disorder using whole brain functional connectivity. Brain. Struct. Funct. 220, 101–115. Liu, F., Zhu, C., Wang, Y., Guo, W., Li, M., Wang, W., Long, Z., Meng, Y., Cui, Q., Zeng, L., Gong, Q., Zhang, W., Chen, H., 2015b. Disrupted cortical hubs in functional brain networks in social anxiety disorder. Clin. Neurophysiol. 126, 1711–1716. Mantini, D., Perrucci, M.G., Del Gratta, C., Romani, G.L., Corbetta, M., 2007. Electrophysiological signatures of resting state networks in the human brain. Proc. Natl. Acad. Sci. U S A 104, 13170–13175. Main, K.L., Soman, S., Pestilli, F., Furst, A., Noda, A., Hernandez, B., Kong, J., Cheng, J., Fairchild, J.K., Taylor, J., Yesavage, J., Wesson, A.J., Kraemer, H., Adamson, M.M., 2017. DTI measures identify mild and moderate TBI cases among patients with complex health problems: A receiver operating characteristic analysis of U.S. veterans. Neuroimage. Clin. 16, 1–16. Otti, A., Guendel, H., Henningsen, P., Zimmer, C., Wohlschlaeger, A.M., Noll-Hussong, M., 2013. Functional network connectivity of pain-related resting state networks in somatoform pain disorder: an exploratory fMRI study. J. Psychiatry Neurosci. 38, 57–65. Power, J.D., Barnes, K.A., Snyder, A.Z., Schlaggar, B.L., Petersen, S.E., 2012. Spurious but systematic correlations in functional connectivity MRI networks arise from subject motion. Neuroimage 59, 2142–2154. Rief, W., Hessel, A., Braehler, E., 2001. Somatization symptoms and hypochondriacal features in the general population. Psychosom. Med. 63, 595–602. Rohlof, H.G., Knipscheer, J.W., Kleber, R.J., 2014. Somatization in refugees: a review. Soc. Psychiatry Psychiatr. Epidemiol. 49, 1793–1804. Stein, D.J., Muller, J., 2008. Cognitive-affective neuroscience of somatization disorder and functional somatic syndromes: reconceptualizing the triad of depression-anxietysomatic symptoms. CNS Spectr. 13, 379–384. Su, Q., Yao, D., Jiang, M., Liu, F., Jiang, J., Xu, C., Dai, Y., Yu, M., Long, L., Li, H., Liu, J.,

Role of the funding source This study was supported by grants from the National Key R&D Program of China(2016YFC1307100 and 2016YFC1306900) and the National Natural Science Foundation of China (Grant Nos. 81571310, 81630033, 81771447, and 81471363). Ethica approval All procedures performed in studies involving human participants were in accordance with the ethical standards of the institutional and/ or national research committee and with the 1964 Helsinki declaration and its later amendments or comparable ethical standards. Author contributions The manuscript was written through contributions of all authors. Among them, Jingping Zhao, Jindong Chen and Wenbin Guo provided the conception of the work. Qinji Su, Pan Pan and Yangpan Ou completed the data collection. Feng Liu was responsible for guiding the collection of imaging data. Then, Qinji Su and Pan Pan carried out data analysis and interpretation. The manuscript was drafted by author Pan Pan and critically revised by Wenbin Guo. Final, all authors have given approval to final version of the manuscript. Conflict of interest Author Pan P declares that he/she has no conflict of interest. Author Liu F declares that he/she has no conflict of interest. Author Ou Y declares that he/she has no conflict of interest. Author Su Q declares that he/she has no conflict of interest. Author Chen J declares that he/she has no conflict of interest. Author Zhao J declares that he/she has no conflict of interest. Author Guo W declares that he/she has no conflict of interest. Acknowledgements The authors also thank all individuals who served as the research participants Informed consent Informed consent was obtained from all individual participants included in the study. Supplementary materials Supplementary material associated with this article can be found, in the online version, at doi:10.1016/j.jad.2019.04.099. References Ahrendts, J., Rusch, N., Wilke, M., Philipsen, A., Eickhoff, S.B., Glauche, V., Perlov, E., Ebert, D., Hennig, J., van Elst, L.T., 2011. Visual cortex abnormalities in adults with ADHD: a structural MRI study. World J. Biol. Psychiatry 12, 260–270. Aust, S., Alkan, H.E., Koelsch, S., Heekeren, H.R., Heuser, I., Bajbouj, M., 2014. How emotional abilities modulate the influence of early life stress on hippocampal functioning. Soc. Cogn. Affect. Neurosci. 9, 1038–1045. Barsky, A.J., Orav, E.J., Bates, D.W., 2005. Somatization increases medical utilization and costs independent of psychiatric and medical comorbidity. Arch. Gen. Psychiatry 62, 903–910. Boeckle, M., Schrimpf, M., Liegl, G., Pieh, C., 2016. Neural correlates of somatoform disorders from a meta-analytic perspective on neuroimaging studies. Neuroimage Clin. 11, 606–613. Cauda, F., D'Agata, F., Sacco, K., Duca, S., Geminiani, G., Vercelli, A., 2011. Functional connectivity of the insula in the resting brain. Neuroimage 55, 8–23. Chang, C.C., Lin, C.J., 2011. LIBSVM: A library for support vector machines. ACM Trans.

88

Journal of Affective Disorders 254 (2019) 82–89

P. Pan, et al. Zhang, Z., Zhang, J., Xiao, C., Guo, W., 2014. Dissociation of regional activity in default mode network in medication-naive, first-episode somatization disorder. PLoS One 9, e99273. Su, Q., Yao, D., Jiang, M., Liu, F., Jiang, J., Xu, C., Dai, Y., Yu, M., Long, L., Li, H., Liu, J., Zhang, Z., Zhang, J., Xiao, C., Guo, W., 2015. Increased functional connectivity strength of right inferior temporal gyrus in first-episode, drug-naive somatization disorder. Aust. N Z J Psychiatry 49, 74–81. Su, Q., Yao, D., Jiang, M., Liu, F., Long, L., Dai, Y., Yu, M., Zhang, Z., Zhang, J., Liu, J., Xiao, C., Zhao, J., Guo, W., 2016. Decreased interhemispheric functional connectivity in insula and angular gyrus/supramarginal gyrus: Significant findings in first-episode, drug-naive somatization disorder. Psychiatry Res. Neuroimaging 248, 48–54. Swets, J.A., 1988. Measuring the accuracy of diagnostic systems. Science 240, 1285–1293. Ture, U., Yasargil, D.C., Al-Mefty, O., Yasargil, M.G., 1999. Topographic anatomy of the insular region. J. Neurosurg. 90, 720–733. Uddin, L.Q., 2015. Salience processing and insular cortical function and dysfunction. Nat. Rev. Neurosci. 16, 55–61.

Wang, H., Guo, W., Liu, F., Chen, J., Wu, R., Zhang, Z., Yu, M., Li, L., Zhao, J., 2016. Clinical significance of increased cerebellar default-mode network connectivity in resting-state patients with drug-naive somatization disorder. Medicine (Baltimore) 95, e4043. Wang, J., Jiang, T., Cao, Q., Wang, Y., 2007. Characterizing anatomic differences in boys with attention-deficit/hyperactivity disorder with the use of deformation-based morphometry. AJNR Am. J. Neuroradiol. 28, 543–547. Wang, L., Xia, M., Li, K., Zeng, Y., Su, Y., Dai, W., Zhang, Q., Jin, Z., Mitchell, P.B., Yu, X., He, Y., Si, T., 2015. The effects of antidepressant treatment on resting-state functional brain networks in patients with major depressive disorder. Hum. Brain. Mapp. 36, 768–778. Yan, C.G., Wang, X.D., Zuo, X.N., Zang, Y.F., 2016. Dpabi: data processing & analysis for (resting-state) brain imaging. Neuroinformatics 14, 339–351. Zhang, J., Jiang, M., Yao, D., Dai, Y., Long, L., Yu, M., Liu, J., Zhang, Z., Xiao, C., Guo, W., 2015. Alterations in white matter integrity in first-episode, treatment-naive patients with somatization disorder. Neurosci. Lett. 599, 102–108.

89