- Split View
-
Views
-
Cite
Cite
Yoichiro Takayanagi, Sue Kulason, Daiki Sasabayashi, Tsutomu Takahashi, Naoyuki Katagiri, Atsushi Sakuma, Chika Obara, Mihoko Nakamura, Mikio Kido, Atsushi Furuichi, Yumiko Nishikawa, Kyo Noguchi, Kazunori Matsumoto, Masafumi Mizuno, J Tilak Ratnanather, Michio Suzuki, Reduced Thickness of the Anterior Cingulate Cortex in Individuals With an At-Risk Mental State Who Later Develop Psychosis, Schizophrenia Bulletin, Volume 43, Issue 4, July 2017, Pages 907–913, https://doi.org/10.1093/schbul/sbw167
- Share Icon Share
Abstract
Background: Despite the fact that only a part of the individuals with at-risk mental state (ARMS) for psychosis do develop psychosis, biological markers of future transition to psychosis have not been well documented. Structural abnormality of the anterior cingulate gyrus (ACG), which probably exists prior to the onset of psychosis, could be such a risk marker. Methods: We conducted a multicenter magnetic resonance imaging (MRI) study of 3 scanning sites in Japan. 1.5-T 3D MRI scans were obtained from 73 ARMS subjects and 74 age- and gender-matched healthy controls. We measured thickness, volume, and surface area of the ACG using labeled cortical distance mapping and compared these measures among healthy controls, ARMS subjects who later converted to overt psychosis (ARMS-C), and those who did not (ARMS-NC). Results: Seventeen of 73 (23%) ARMS subjects developed overt psychosis within the follow-up period. The thickness of the left ACG was significantly reduced in ARMS-C relative to healthy subjects (P = .026) while both ARMS-C (P = .001) and ARMS-NC (P = .01) had larger surface areas of the left ACG compared with healthy controls. Conclusion: Further studies will be needed to identify potential markers of future transition to psychosis though cortical thinning of the ACG might be one of the candidates.
Introduction
Psychotic disorders, such as schizophrenia, are characterized by disabling features. Attempts to delay or even prevent overt psychosis among individuals with at-risk mental state (ARMS) for psychosis, who are exhibiting prodromal symptoms, have been made in the last decade.1 However, the rate of the transition to psychosis in this population is up to 36% after a 3-year follow-up,2 implicating that a large number of ARMS patients would not eventually develop psychosis. Therefore, it is crucial to identify biological changes that precede the transition to psychosis. To date, modalities such as neurophysiological measures,3,4 structural magnetic resonance imaging (MRI),5 or functional MRI6 are used to seek for such changes predating the onset of psychosis in ARMS. In the review of event-related potential studies among ARMS by Bodatsch and colleagues, the authors noted that mismatch negativity has produced the most convincing results regarding independent replication of the predictive validity.3 Structural MRI studies have reported gray matter (GM) reductions in the anterior cingulate gyrus (ACG),7 the superior temporal gyrus,7 the insular,7,8 and parahippocampal gyrus9 or cortical thinning of the ACG10 in ARMS subjects who later developed overt psychosis. German group demonstrated that transition to psychosis was accurately predicted using a structural MRI-based multivariate pattern classification method.11
The ACG regulates cognitive and emotional processing in humans.12 The structural anomaly of the ACG in schizophrenia has been repeatedly reported by MRI studies.13–19 A recent meta-analysis of voxel-based morphometry (VBM) studies of genetic high-risk subjects for psychosis has shown decreased ACG GM volume in the high-risk group compared with healthy controls.20 Another meta-analysis of VBM studies of ARMS or first-episode psychosis (FEP) reported smaller GM volume in the ACG in FEP than in ARMS.2 Moreover, a few studies have demonstrated cortical thinning10 or GM reductions of the ACG21,22 preceding the transition to psychosis in ARMS. Thus, the morphology of the ACG could be a potential marker of the future transition to psychosis among individuals at high risk for psychosis.
Labeled cortical distance mapping (LCDM) is a reliable tool, which can characterize the morphometry of the laminar cortical mantle of cortical structures. LCDM data, which represent the distances between labeled GM voxels and the GM/white matter (WM) cortical surface, are local measures characterizing the morphometry of the cortical mantle.23 Particularly, as this method is designed to be applied to specific cortical regions, it offers better tissue segmentation than whole-brain analyses, which are more susceptible to image inhomogeneities. Previously, we reported reduced ACG GM volume and thickness in schizophrenia compared with healthy controls using LCDM,19,24 but no studies to date have applied this technique to the high-risk cohort of psychosis.
One inherent problem of ARMS study is the difficulty in obtaining sufficient statistical power (ie, a large sample size) at a single site. To overcome this problem, a few “multicenter” structural MRI studies have been conducted in Europe & Australia9,25 and North America,26 but none in Asia, so far.
In this study, we conducted a multicenter investigation of ARMS subjects in Japan. We used LCDM to assess detailed ACG morphology in ARMS patients and healthy controls to clarify structural changes in the ACG that precede the onset of psychosis.
Methods
Participants
All the ARMS subjects (n = 73) were recruited from specialized clinical services for ARMS at Toyama University Hospital, Toho University Hospital, and Tohoku University Hospital.27 For the diagnosis of ARMS, the Comprehensive Assessment of at-risk mental state (CAARMS)28 (University of Toyama and Tohoku University) or the Structured Interview for Prodromal Syndrome/the Scale of Prodromal Symptoms (SIPS/SOPS)29 (Toho University) was used. At each site, all ARMS subjects were followed regularly for at least 2 years after MRI scanning to identify those who would convert to overt psychosis (ARMS-C) or not (ARMS-NC). Transition to psychosis was determined according to the CAARMS criteria (ie, at least one fully positive psychotic symptom several times per week for more than 1 week) or the SIPS criteria (ie, the presence of a positive symptom existing for more than 1 month or accompanying a serious disorganization or danger). Gender- and age-matched healthy controls (n = 74) were also recruited at each site. Healthy subjects consisted of healthy volunteers recruited from the community, hospital staff, and students at each site. Some of the healthy controls were offered compensations for their participation.
All participants were physically healthy at the time of this study. The exclusion criteria were (1) having a lifetime history of serious head injury, neurological illness, or other serious physical disease, (2) fulfilling the criteria for substance abuse/dependence, and (3) having previous psychotic episodes which met the criteria of the Diagnostic and Statistical Manual of Mental Disorders, Fourth Edition (DSM-IV).30 All subjects provided written informed consent. If the participants were minors, a written informed consent was obtained from their parent as well as from the participant. This study was approved by the Committee on Medical Ethics at each site.
MRI Data Acquisition
University of Toyama.
A 1.5-T scanner (Magnetom Vision, Siemens Medical System, Inc, Erlangen, Germany) was used with 3-dimensional gradient-echo sequence FLASH (fast low-angle shots) yielding 160–180 contiguous T1-weighted slices of 1.0-mm thickness in the sagittal plane. The imaging parameters were as follows: repetition time = 24 ms; echo time = 5 ms; flip angle = 40°; field of view = 256 mm; and matrix size = 256 × 256 pixels. The voxel size was 1.0 × 1.0 × 1.0 mm.
Toho University.
Participants underwent magnetic MRI scans using a 1.5-T scanner (EXCELART Vantage, XGV 1.5 T; Toshiba Medical Systems, Tokyo, Japan) yielding 160 contiguous T1-weighted slices of 1.0-mm thickness in the sagittal plane. The imaging parameters were: repetition time = 24.4 ms: echo time = 5.5 ms; flip angle = 35°; field of view = 250 mm; and matrix size = 256 × 256 pixels. The voxel size was 0.98 × 0.98 × 1.0 mm.
Tohoku University.
A 1.5-T scanner (Achieva, Phillips Medical Systems, Best, The Netherlands) was used for MRI scanning with 3-dimensional fast field echo sequencing yielding 200 contiguous T1-weighted slices of 1.0-mm thickness in the sagittal plane. The imaging parameters were as follows: repetition time = 30 ms; echo time = 5 ms; flip angle = 30°; field of view = 256 mm; and matrix size = 256 × 256 pixels. The voxel size was 1.0 × 1.0 × 1.0 mm.
FreeSurfer-Initialized LCDM
We detailed the FreeSurfer-Initialized LCDM pipeline previously.19 This pipeline provides measures for cortical thickness, volume, and surface area of the ACG. Briefly, the FreeSurfer software suite (version 5.3)31 generates the GM/WM cortical surface and parcellates the brain into regions of interest (ROIs) and then LCDM segments the cortical ROI and computes the distance of each GM voxel from the FreeSurfer surface. Before initiating the LCDM pipeline, all images preprocessed by FreeSurfer were carefully inspected and any errors were manually corrected.
The FreeSurfer-developed region labels are used to extract the ROI from the original MRI. Next, the ROI is segmented into WM, GM, and cerebrospinal fluid (CSF) using a mixture model averaging method.32,33 The method for generating LCDMs previously developed for the ACG34–36 was applied. To generate a distance map for the GM, the distance between each GM voxel and the closest GM/WM surface vertex is calculated at a 1 × 1 × 1 mm resolution. Voxels in the range of −2 to 8 mm were used for the analysis. The data give information on the probability distribution function of the GM distance from the GM/WM surface. Then a cumulative distribution function (CDF) was calculated, and the cortical thickness was determined using the distance where the CDF reaches 95%. Due to outlier voxels at distances greater than 6 mm, the area under the CDF up to this 95% was used as the volume of the ACG. The surface area of WM was calculated from the triangulated surface.
LCDM analyzes a subvolume encompassing the ROI. The boundary between WM and CSF is often indistinct, but by applying LCDM, this challenge can be partially overcome by viewing the ROI as a laminar structure composed of GM voxels and a local surface coordinate system based on an anatomically defined GM/WM cortical surface. Therefore, LCDM differs from global approaches such as FreeSurfer31 which averages point-to-point distances between outer and inner cortical surfaces; it should be noted that VBM can be used to generate similar data but for whole brains.37
Statistical Analysis
Demographic and clinical variables were compared among groups via 1-way analysis of variance (ANOVA) or a chi-square test. Cortical thickness, volume, and surface area of the ACG in each hemisphere were compared among the diagnostic groups by calculating the analysis of covariance (ANCOVA) with diagnosis (ie, controls, ARMS-NC, and ARMS-C) as the between-subject factor, and age, intracranial volume (ICV), and scanning site as covariates. To evaluate the laterality of ROI measures, we added repeated measures of ANCOVA with the hemisphere (left and right) as a within-subject factor using the same covariates. A Bonferroni’s correction was used for post hoc pairwise comparison. The associations of the ACG measures and clinical variables (ie, weeks between scanning and transition, and medication dosage) were examined by calculating partial correlation coefficients controlling for age, ICV, and site. As we used different versions of CARRMS criteria depending on the site, we conducted the correlational analyses to examine the associations between the ACG measures and the severity of prodromal positive symptoms independently for each site (ie, Tohoku or Toyama), by calculating partial correlation coefficients controlling for age and ICV. The sum of the severity of 3 (Toyama) or 4 (Tohoku) items of positive symptoms was used for the correlational analyses. Since the raw data of the SOPS evaluated at Toho University were not available, the subjects recruited at Toho University were excluded from these correlational analyses. To see if the ACG measures predict the transition to psychosis, we conducted a Cox regression analysis with the development of overt psychosis as the dependent variable, and the Z scores of the ACG measures adjusted for age, ICV, and site as the independent variables. The statistical significance level was set as P < .05 (2-tailed).
Results
Demographic and Clinical Characteristics
Of the 73 ARMS subjects, 17 (23%) developed psychosis within the follow-up period. The psychosis diagnoses of ARMS subjects who converted to psychosis (ARMS-C) based on the criteria of DSM-IV were as follows: 11 schizophrenia cases, 1 delusional disorder case, 1 schizophreniform disorder case, and 4 cases of psychotic disorder not otherwise specified. There were no statistical significant differences regarding age and gender among healthy controls, ARMS-NC, and ARMS-C groups. Of 73 ARMS subjects, 29 (38%) were taking antipsychotics at the time of MRI scanning. Of 57 ARMS-NC subjects, 6 (11%) were taking typical antipsychotics, whereas 16 (29%) were taking atypical antipsychotics. Of 17 ARMS-C subjects, 7 (41%) were taking atypical antipsychotics. The mean dosage of antipsychotics (chlorpromazine equivalent) was 55 ± 94 mg/d and 92 ± 136 mg/d for ARMS-NC and ARMS-C subjects, respectively. Healthy controls had significantly higher educational attainment than ARMS-NC (P < .001) and ARMS-C (P < .001) groups (table 1).
Variables . | Group . | ||||
---|---|---|---|---|---|
HC . | ARMS-NC . | ARMS-C . | Statistics . | P . | |
Number of subjects (total) | 74 | 56 | 17 | ||
Scanning site 1 (Toyama) | 52 | 11 | 5 | ||
Scanning site 2 (Toho) | 5 | 19 | 4 | ||
Scanning site 3 (Tohoku) | 17 | 26 | 8 | ||
Age (mean ± SD) | 22.6 ± 4.3 | 22.3 ± 6.4 | 19.9 ± 4.4 | F = 1.8 | .17 |
Gender (male/female) | 37/37 | 22/34 | 5/12 | χ2 = 3.0 | .22 |
Handednessa (right/both/left) | 58/0/0 | 34/8/2 | 10/2/2 | ||
Education yearsb (mean ± SD) | 14.8 ± 1.9 | 12.2 ± 2.6 | 12.1 ± 2.1 | F = 25.6 | <.001 |
Parental education yearsc (mean ± SD) | 12.9 ± 2.2 | 13.5 ± 2.0 | 13.2 ± 1.5 | F = 0.7 | .49 |
On typical antipsychotics (n [%]) | 6 (11) | 0 (0) | |||
On atypical antipsychotics (n [%]) | 16 (29) | 7 (47) | |||
Weeks between scanning and transition (mean ± SD) | 39 ± 33 | ||||
Antipsychotics dosaged (mg/d, mean ± SD) | 55 ± 94 | 92 ± 136 |
Variables . | Group . | ||||
---|---|---|---|---|---|
HC . | ARMS-NC . | ARMS-C . | Statistics . | P . | |
Number of subjects (total) | 74 | 56 | 17 | ||
Scanning site 1 (Toyama) | 52 | 11 | 5 | ||
Scanning site 2 (Toho) | 5 | 19 | 4 | ||
Scanning site 3 (Tohoku) | 17 | 26 | 8 | ||
Age (mean ± SD) | 22.6 ± 4.3 | 22.3 ± 6.4 | 19.9 ± 4.4 | F = 1.8 | .17 |
Gender (male/female) | 37/37 | 22/34 | 5/12 | χ2 = 3.0 | .22 |
Handednessa (right/both/left) | 58/0/0 | 34/8/2 | 10/2/2 | ||
Education yearsb (mean ± SD) | 14.8 ± 1.9 | 12.2 ± 2.6 | 12.1 ± 2.1 | F = 25.6 | <.001 |
Parental education yearsc (mean ± SD) | 12.9 ± 2.2 | 13.5 ± 2.0 | 13.2 ± 1.5 | F = 0.7 | .49 |
On typical antipsychotics (n [%]) | 6 (11) | 0 (0) | |||
On atypical antipsychotics (n [%]) | 16 (29) | 7 (47) | |||
Weeks between scanning and transition (mean ± SD) | 39 ± 33 | ||||
Antipsychotics dosaged (mg/d, mean ± SD) | 55 ± 94 | 92 ± 136 |
Note: ARMS, at-risk mental state; C, converters; HC, healthy controls; NC, non-converters; SD, standard deviation.
aData missing for 31 subjects.
bData missing for 7 subjects.
cData missing for 38 subjects.
dChlorpromazine equivalent.
Variables . | Group . | ||||
---|---|---|---|---|---|
HC . | ARMS-NC . | ARMS-C . | Statistics . | P . | |
Number of subjects (total) | 74 | 56 | 17 | ||
Scanning site 1 (Toyama) | 52 | 11 | 5 | ||
Scanning site 2 (Toho) | 5 | 19 | 4 | ||
Scanning site 3 (Tohoku) | 17 | 26 | 8 | ||
Age (mean ± SD) | 22.6 ± 4.3 | 22.3 ± 6.4 | 19.9 ± 4.4 | F = 1.8 | .17 |
Gender (male/female) | 37/37 | 22/34 | 5/12 | χ2 = 3.0 | .22 |
Handednessa (right/both/left) | 58/0/0 | 34/8/2 | 10/2/2 | ||
Education yearsb (mean ± SD) | 14.8 ± 1.9 | 12.2 ± 2.6 | 12.1 ± 2.1 | F = 25.6 | <.001 |
Parental education yearsc (mean ± SD) | 12.9 ± 2.2 | 13.5 ± 2.0 | 13.2 ± 1.5 | F = 0.7 | .49 |
On typical antipsychotics (n [%]) | 6 (11) | 0 (0) | |||
On atypical antipsychotics (n [%]) | 16 (29) | 7 (47) | |||
Weeks between scanning and transition (mean ± SD) | 39 ± 33 | ||||
Antipsychotics dosaged (mg/d, mean ± SD) | 55 ± 94 | 92 ± 136 |
Variables . | Group . | ||||
---|---|---|---|---|---|
HC . | ARMS-NC . | ARMS-C . | Statistics . | P . | |
Number of subjects (total) | 74 | 56 | 17 | ||
Scanning site 1 (Toyama) | 52 | 11 | 5 | ||
Scanning site 2 (Toho) | 5 | 19 | 4 | ||
Scanning site 3 (Tohoku) | 17 | 26 | 8 | ||
Age (mean ± SD) | 22.6 ± 4.3 | 22.3 ± 6.4 | 19.9 ± 4.4 | F = 1.8 | .17 |
Gender (male/female) | 37/37 | 22/34 | 5/12 | χ2 = 3.0 | .22 |
Handednessa (right/both/left) | 58/0/0 | 34/8/2 | 10/2/2 | ||
Education yearsb (mean ± SD) | 14.8 ± 1.9 | 12.2 ± 2.6 | 12.1 ± 2.1 | F = 25.6 | <.001 |
Parental education yearsc (mean ± SD) | 12.9 ± 2.2 | 13.5 ± 2.0 | 13.2 ± 1.5 | F = 0.7 | .49 |
On typical antipsychotics (n [%]) | 6 (11) | 0 (0) | |||
On atypical antipsychotics (n [%]) | 16 (29) | 7 (47) | |||
Weeks between scanning and transition (mean ± SD) | 39 ± 33 | ||||
Antipsychotics dosaged (mg/d, mean ± SD) | 55 ± 94 | 92 ± 136 |
Note: ARMS, at-risk mental state; C, converters; HC, healthy controls; NC, non-converters; SD, standard deviation.
aData missing for 31 subjects.
bData missing for 7 subjects.
cData missing for 38 subjects.
dChlorpromazine equivalent.
ACG Measures
ANCOVAs adjusted for age, gender, ICV, and site revealed a significant main effect of diagnosis for the thickness (F2,139 = 3.972, P = .021) and surface area (F2,139 = 8.218, P < .001) of the left ACG. Post hoc tests showed that (1) the thickness of the left ACG was reduced in ARMS-C subjects compared with healthy controls (P = .025) and (2) ARMS-C and ARMS-NC subjects had larger left ACG surface area than controls (P = .001 and P = .008, respectively) (figure 1). Repeated measures of ANCOVA also demonstrated a significant main effect of the hemisphere for the thickness (F2,139 = 5.366, P = .022) and volume (F2,138 = 6.586, P = .011) of the ACG, and a significant hemisphere × diagnosis interaction for the surface area (F2,138 = 4.006, P = .02). Post hoc tests showed that the thickness of the ACG was smaller in the right hemisphere (P = .001) among all groups while the ACG surface area was larger in the left hemisphere only in ARMS-NC and ARMS-C patients (P = .003 and P = .014, respectively) (table 2). There were no significant correlations among the ACG measures and clinical variables (ie, weeks between scanning and transition, and medication dosage) after controlling for age, ICV, and site. No significant associations of the ACG measures and the severity of prodromal positive symptoms were found. In the Cox regression analysis, ACG measures did not predict the transition to psychosis (supplementary table 1).
. | HC (n = 74) . | ARMS-NC (n = 56) . | ARMS-C (n = 17) . | ANCOVAa . | Post hoc . | ||||||||
---|---|---|---|---|---|---|---|---|---|---|---|---|---|
Diagnosis . | Hemisphere . | Hemisphere × Diagnosis . | . | ||||||||||
Measures . | Mean . | SD . | Mean . | SD . | Mean . | SD . | F . | P . | F . | P . | F . | P . | . |
Left ACG thickness (mm) | 3.76 | 0.56 | 3.65 | 0.61 | 3.45 | 0.60 | 3.972 | .021 | 5.366 | .022 | 1.663 | .193 | HC > ARMS-C |
Right ACG thickness (mm) | 3.90 | 0.53 | 3.84 | 0.50 | 3.78 | 0.52 | 0.233 | .792 | Left < right | ||||
Left ACG volume (mm3) | 5360 | 1067 | 5570 | 1084 | 5766 | 1208 | 1.415 | .246 | 6.586 | .011 | 1.037 | .357 | |
Right ACG volume (mm3) | 5440 | 1516 | 5284 | 1336 | 5225 | 1266 | 0.099 | .906 | NS | ||||
Left ACG area (mm2) | 1861 | 599 | 2024 | 634 | 2315 | 874 | 8.218 | <.001 | 2.081 | .151 | 4.006 | .020 | ARMS-NC, ARMS-C > HC |
Right ACG area (mm2) | 1805 | 602 | 1801 | 588 | 1894 | 901 | 0.245 | .783 | Left > right in ARMS-NC and ARMS-C |
. | HC (n = 74) . | ARMS-NC (n = 56) . | ARMS-C (n = 17) . | ANCOVAa . | Post hoc . | ||||||||
---|---|---|---|---|---|---|---|---|---|---|---|---|---|
Diagnosis . | Hemisphere . | Hemisphere × Diagnosis . | . | ||||||||||
Measures . | Mean . | SD . | Mean . | SD . | Mean . | SD . | F . | P . | F . | P . | F . | P . | . |
Left ACG thickness (mm) | 3.76 | 0.56 | 3.65 | 0.61 | 3.45 | 0.60 | 3.972 | .021 | 5.366 | .022 | 1.663 | .193 | HC > ARMS-C |
Right ACG thickness (mm) | 3.90 | 0.53 | 3.84 | 0.50 | 3.78 | 0.52 | 0.233 | .792 | Left < right | ||||
Left ACG volume (mm3) | 5360 | 1067 | 5570 | 1084 | 5766 | 1208 | 1.415 | .246 | 6.586 | .011 | 1.037 | .357 | |
Right ACG volume (mm3) | 5440 | 1516 | 5284 | 1336 | 5225 | 1266 | 0.099 | .906 | NS | ||||
Left ACG area (mm2) | 1861 | 599 | 2024 | 634 | 2315 | 874 | 8.218 | <.001 | 2.081 | .151 | 4.006 | .020 | ARMS-NC, ARMS-C > HC |
Right ACG area (mm2) | 1805 | 602 | 1801 | 588 | 1894 | 901 | 0.245 | .783 | Left > right in ARMS-NC and ARMS-C |
Note: Abbreviations are explained in the first footnote to table 1. Bold values indicate P < .05. ACG, anterior cingulate gyrus.
aAge, intracranial volume, and scanning site were entered as covariates.
. | HC (n = 74) . | ARMS-NC (n = 56) . | ARMS-C (n = 17) . | ANCOVAa . | Post hoc . | ||||||||
---|---|---|---|---|---|---|---|---|---|---|---|---|---|
Diagnosis . | Hemisphere . | Hemisphere × Diagnosis . | . | ||||||||||
Measures . | Mean . | SD . | Mean . | SD . | Mean . | SD . | F . | P . | F . | P . | F . | P . | . |
Left ACG thickness (mm) | 3.76 | 0.56 | 3.65 | 0.61 | 3.45 | 0.60 | 3.972 | .021 | 5.366 | .022 | 1.663 | .193 | HC > ARMS-C |
Right ACG thickness (mm) | 3.90 | 0.53 | 3.84 | 0.50 | 3.78 | 0.52 | 0.233 | .792 | Left < right | ||||
Left ACG volume (mm3) | 5360 | 1067 | 5570 | 1084 | 5766 | 1208 | 1.415 | .246 | 6.586 | .011 | 1.037 | .357 | |
Right ACG volume (mm3) | 5440 | 1516 | 5284 | 1336 | 5225 | 1266 | 0.099 | .906 | NS | ||||
Left ACG area (mm2) | 1861 | 599 | 2024 | 634 | 2315 | 874 | 8.218 | <.001 | 2.081 | .151 | 4.006 | .020 | ARMS-NC, ARMS-C > HC |
Right ACG area (mm2) | 1805 | 602 | 1801 | 588 | 1894 | 901 | 0.245 | .783 | Left > right in ARMS-NC and ARMS-C |
. | HC (n = 74) . | ARMS-NC (n = 56) . | ARMS-C (n = 17) . | ANCOVAa . | Post hoc . | ||||||||
---|---|---|---|---|---|---|---|---|---|---|---|---|---|
Diagnosis . | Hemisphere . | Hemisphere × Diagnosis . | . | ||||||||||
Measures . | Mean . | SD . | Mean . | SD . | Mean . | SD . | F . | P . | F . | P . | F . | P . | . |
Left ACG thickness (mm) | 3.76 | 0.56 | 3.65 | 0.61 | 3.45 | 0.60 | 3.972 | .021 | 5.366 | .022 | 1.663 | .193 | HC > ARMS-C |
Right ACG thickness (mm) | 3.90 | 0.53 | 3.84 | 0.50 | 3.78 | 0.52 | 0.233 | .792 | Left < right | ||||
Left ACG volume (mm3) | 5360 | 1067 | 5570 | 1084 | 5766 | 1208 | 1.415 | .246 | 6.586 | .011 | 1.037 | .357 | |
Right ACG volume (mm3) | 5440 | 1516 | 5284 | 1336 | 5225 | 1266 | 0.099 | .906 | NS | ||||
Left ACG area (mm2) | 1861 | 599 | 2024 | 634 | 2315 | 874 | 8.218 | <.001 | 2.081 | .151 | 4.006 | .020 | ARMS-NC, ARMS-C > HC |
Right ACG area (mm2) | 1805 | 602 | 1801 | 588 | 1894 | 901 | 0.245 | .783 | Left > right in ARMS-NC and ARMS-C |
Note: Abbreviations are explained in the first footnote to table 1. Bold values indicate P < .05. ACG, anterior cingulate gyrus.
aAge, intracranial volume, and scanning site were entered as covariates.
We also examined the site effects or site × diagnosis interactions for ACG measures via ANCOVAs to see the potential confounding by using multiple scanners and uneven proportions of diagnostic groups in each site. There were significant site × diagnosis interactions for ACG volume and area as well as significant site effects for thickness and area in ANCOVAs. Particularly, the significant site × diagnosis interaction found in the analysis of ACG surface area indicates that our results of ACG area may have been confounded by the difference in MRI scanners and the uneven proportion of ARMS-C, ARMS-NC, and healthy controls across the sites (supplementary table 2).
Discussion
We found cortical thinning of the left ACG in ARMS patients who later developed overt psychosis relative to healthy controls. Our findings of altered ACG morphology in ARMS-C subjects collaborate with previous literature.10,21,22 To our knowledge, this is the first study that has used a recently developed powerful neuroimaging tool, namely LCDM, to examine ACG morphology among individuals with ARMS. Furthermore, as far as we are aware, this is the first multicenter structural MRI study employing individuals with ARMS in Asia. As we did not detect significant structural changes of the ACG between ARMS-C and ARMS-NC groups, we are unable to state that our finding (ie, cortical thinning of the left ACG) is a potential candidate for an ARMS prognostic marker though the results may reach statistically significant levels with a larger sample size (figure 1).
Our results are partly in line with the previous study by Fornito and colleagues10 who found cortical thinning of the bilateral rostral paralimbic ACG in ARMS-C patients compared with healthy controls. The effect size of the cortical thinning in ARMS-C as compared with healthy controls was −0.528 (Cohen’s d). Fornito et al10 demonstrated that the effect size of the cortical thinning of the bilateral rostral paralimbic ACC in ARMS-C was −0.596. Therefore, the magnitude of the ACC thinning found in this study is comparable to the previous ROI-based study.10 However, we did not find “cortical thickening” of the ACG in ARMS-NC patients relative to controls, which was reported by Fornito and colleagues.10 Methodological differences between these studies, particularly the difference in the ROI definitions, may account for these discrepancies.
The findings of previous studies examining the entire cortex with voxel-based or surface-based approaches in ARMS-NC and ARMS-C subjects are mixed with regard to ACG morphology,9,21,22,25,26,38 probably due to the differences in sample size, analytic software, or voxel threshold. Those studies also reported GM reductions in medial temporal, temporal, or prefrontal regions,9,21,22,38 although some reported no significant GM changes in ARMS-C patients relative to ARMS-NC patients or healthy controls.25,26 LCDM offers better tissue segmentation as compared with VBM or standard FreeSurfer method, while it is not designed to conduct whole-brain analysis. Regional specificity of our findings should be tested in future studies.
Among several studies that measured cortical thickness and surface area of the ACG simultaneously in schizophrenia, most studies reported cortical thinning19,24,39 and a nonsignificant change of surface area in schizophrenia relative to healthy controls.19,24,40 Taken together with our findings, cortical thinning of the ACG in schizophrenia may emerge before the onset of psychosis and continue to exist thereafter.
The rate of transition to psychosis among ARMS subjects in our study in about 2-year follow-up (23%) is comparable to the result of the meta-analysis by Fusar-Poli et al (29%).2 Given the disabling features of psychotic disorders such as schizophrenia in many aspects,41,42 attempts to delay or prevent full-blown psychosis in individuals with ARMS have been tested using therapeutic interventions such as cognitive behavioral therapy, atypical antipsychotics, or omega 3 fatty acids (reviewed by Stafford et al).1 Although some studies demonstrated the effectiveness of these interventions in terms of prevention of developing psychosis, surrogate markers to identify ARMS individuals who have higher risk of developing psychosis would be necessary, considering the potential adverse effects of the interventions and the transition rates demonstrated in the literature (ie, less than 40% in ARMS).
We should consider possible problems stemming from the use of different MRI scanners with different acquisition parameters. In addition, the proportion of control and ARMS subjects in each site is not comparable. Indeed, the significant site × diagnosis interaction found in the analysis of ACG surface area indicates that our results of ACG area may have been confounded by the difference in MRI scanners and the uneven proportion of ARMS-C, ARMS-NC, and healthy controls across the sites. However, we believe that we have minimized the problem of using multiple scanners since we used only T1-weighted images scanned by 1.5-T scanners, and we treated scanning sites as covariates in the statistical models. Furthermore, as site × diagnosis interactions were not found for ACG thickness, we believe cortical thinning of the left ACG in ARMS-C is not merely an artifact due to using different scanners or uneven proportions of diagnostic groups among 3 sites.
In addition to using different scanners and uneven proportions of diagnostic groups in each site, there are several limitations that should be taken into account. First, although we were able to accumulate a relatively large sample size of ARMS patients (n = 73) by collecting subjects at multiple sites, the number of ARMS-C (n = 17) is moderate and may not be sufficient. The negative results of the Cox regression analysis might partly be due to the small sample size of ARMS-C group. Second, our data might have been confounded by using different criteria (ie, SIPS/SOPS or CAARMS) for the diagnosis of ARMS. Although these 2 measures substantially overlap, SIPS/SOPS is generally more restrictive.29 Third, as some of the ARMS subjects had taken antipsychotics, we could not exclude the influence of antipsychotics on brain morphology. For instance, a recent meta-analysis has demonstrated a different and contrasting moderating role of medication intake on cortical GM changes depending on whether the patients were treated with typical antipsychotics (more progressive GM loss correlated with higher daily dosage) or atypical antipsychotics (less progressive GM loss with higher daily dosage).43 Finally, we lack sufficient follow-up (ie, longitudinal) MRI data which would allow us to see whether ACG measures are stable or not, through the course of early psychosis.
In conclusion, our findings suggest that the morphology of the ACG may distinguish ARMS subjects who will develop overt psychosis from those who will not in the next couple of years, and thus can be a potential marker for the transition to psychosis, although further studies are needed to confirm these points.
Supplementary Material
Supplementary data are available at Schizophrenia Bulletin online.
Funding
This study was supported by grants to Y.T. (Kiban C No. 26461738), T.T. (Kiban C No. 26461739), and M.S. (Kiban B No. 24390281) from the Japanese Society for the Promotion of Science, and Health and Labour Sciences Research Grants for Comprehensive Research on Persons with Disabilities from the Japan Agency for Medical Research and Development (AMED) to K.M., M.M., and M.S. J.T.R. and S.K. were supported by NIH grants (R01MH105660 and P41EB015909).
Acknowledgments
The authors have declared that there are no conflicts of interest in relation to the subject of this study.
References