To our knowledge, this is the first study to investigate structural cortical differences between UHR group without a family history of psychosis, schizophrenia group, and HC group in terms of cortical thickness in a cross-sectional manner. The major findings of the present study were reduced cortical thickness in the ACC, PFC, STG, and inferior parietal regions in the UHR group compared with HC group; these regions were more extended in schizophrenia group. Age in UHR individuals was negatively correlated with cortical thickness in these regions. ANCOVA among the 3 groups revealed significant differences in mean cortical thickness for each hemisphere and regional cortical thickness differences in distributed regions.
Mounting evidence suggests that, in general, schizophrenia subjects have structural GM deficits in widespread regions, particularly in the frontal and temporal cortices, as a psychopathological feature, and frontotemporal dysconnectivity. The occurrence of these deficits may be linked to functional decline at the prodromal stage of schizophrenia. As expected, our UHR group showed cortical thinning in several frontal cortices and in the STG. The pattern of cortical thinning found in this study was in line with previous MRI studies in schizophrenia and UHR groups, compared with HC group.7,10,46,47
Many previous studies have repeatedly reported morphologic abnormalities of the STG and adjacent temporal cortex, particularly the primary auditory cortex and planum temporale in the left hemisphere,9,48
and its functional deficits include auditory hallucinations or thought disorders in schizophrenia.49,50
Volume reductions in the left posterior STG in schizophrenia were highly correlated with the amplitude reduction of MMN51
and the reduction of the magnetic counterpart of MMN (MMNm).52
Disturbances in MMN may index structural abnormalities of the left posterior STG in schizophrenia. In our previous work, the UHR group showed a negative correlation between left MMNm dipole moment and clinical symptoms, in addition to a smaller right MMNm dipole moment than in HC group.6
These results may implicate structural deficits of the STG in UHR individuals. VBM studies demonstrated that UHR individuals had smaller left STGs than HC individuals.7,53
In addition, Takahashi et al17
have recently reported a progressive decrease of the STG in UHR subjects, using detailed ROI analyses on longitudinal MRI data. Our findings and other work suggest that temporal lobe abnormalities may precede the onset of psychosis and progressively worsen over time. In line with this assumption and consistent with previous studies, our schizophrenia subjects had widespread reductions in the STG compared with HC subjects.
In addition to the STG atrophy, our UHR group showed cortical thinning in the frontal lobe including the PFC, medial frontal cortex, and ACC. First, decreased cortical thickness of the PFC in UHR subjects, compared with HC subjects, has been found consistently by other groups.16,54
Impairments in the PFC may be expected to lead to cognitive dysfunction such as in attention and working memory in UHR individuals.3,4
Some neuroimaging studies have found prefrontal hypofunction in UHR groups, which is implicated in the pathophysiology of schizophrenia.55,56
Second, the medial frontal cortex implicated in the default mode network has a role in social cognition.57
Thus, a decreased medial frontal cortex corresponds with our previous assumption, derived from a psychological study using a theory-of-mind task, showing decreased social cognition and implicating a deficit in the medial frontal cortex.58
Third, the UHR group exhibited significantly reduced cortical thickness in the ACC. The ACC is related to impaired cognition (self-monitoring) and disorganization in patients with schizophrenia.59,60
In the present study, decreased GM of ACC in the UHR group is consistent with previous MRI studies.16
In addition, some VBM studies have reported decreased ACC GM in converters, relative to nonconverters.7,16
In comparison analysis between converters (n
= 8) and nonconverters (n
= 22) in the present study (not described), it was observed that converters showed reduced cortical thickness in right ACC in addition to left lingual cortex, right superior temporal cortex, and bilateral inferior temporal cortex (at a more lenient threshold of P
< .05; see Supplementary figure S1
). However, these findings should be considered preliminary given the small sample size and more lenient threshold. Recently, Fornito et al27
have indicated that abnormalities of the ACC precede psychosis onset by applying a cortical surface-based protocol for parcellating the ACC. Abnormalities of these midline brain structures including the medial frontal cortex and ACC are compatible with prior suggestions about neurodevelopmental anomalies in UHR subjects.61
Abnormalities of cavum septi pellucidi and the ACC folding have been reported in UHR subjects, which reflect early neurodevelopmental anomalies,8,62
as were previously observed in schizophrenia patients.63,64
It suggests that early neurodevelopmental anomalies may lead to disturbances on the subsequent brain maturation and produce further late neurodevelopmental changes during the initial stages of a psychotic illness.65
Longitudinal MRI studies of UHR subjects have reported progressive GM loss before and after psychosis onset in the medial temporal and prefrontal regions using voxel-based approach.16,66
In the UHR group, we found negative correlations between age and cortical thickness in some regions, particularly the right PFC (Brodmann area [BA] 10), right inferior parietal cortex, and bilateral ACC. Several longitudinal MR studies have reported progressive loss of GM volume with increasing age, in addition to duration and severity of illness, in first-episode and chronic schizophrenia,67–69
although these findings are inconsistent.70
Studies of normal brain maturation have consistently reported GM volume reductions in frontal and parietal cortices in adolescence.71
O'Donnell et al72
exhibited a significant inverse correlation between age and cortical thickness in the frontopolar cortex (BA 10) through late childhood and adolescence. Thompson et al73
described the trajectory of GM loss in a longitudinal structural study of early-onset schizophrenia, which started in the parietal cortex and then progressed forward to temporal and frontal cortices, reflecting an exaggeration of the normal cortical developmental processes. In this context, structural changes in the UHR group may be interpreted as an abnormal acceleration of normal cortical developmental processes. Further study is needed to verify this interpretation.
Based on dynamic neuroanatomical changes across clinical conditions, structural neuroimaging studies have recently tried to identify UHR individuals or schizophrenia individuals and predict conversion using neuroanatomical pattern classification. Koutsouleris et al74
have distinguished UHR group from HC group by depending on structural between-group differences with respect to GM volume. Cortical thickness analysis using principal component analysis discriminated between schizophrenia and HC groups.24
However, a recent study has reported that cortical thickness asymmetry analysis, which evaluated the cortical asymmetry of corresponding regions between left and right hemispheres, contributes to detection of UHR and first-episode psychosis but not cortical thickness analysis.75
The discrepancies between these results may result from differences in analysis approaches and sample characteristics.
There are notable differences in sample characteristics between our dataset and previously published research. The patients with schizophrenia who participated in this study were in maintenance therapy periods after the recovery from their first psychotic episodes, and the clinical status of the patients was relatively stable. It was thought that chronic patients with schizophrenia were less suitable for comparison of brain structures between UHR and schizophrenia groups because there may be a potential for underlying structural brain abnormalities occurred from the chronic symptoms. Thus, the chronic schizophrenia patients were not appropriate for the purpose of the study that was to investigate dynamic brain changes preceding onset of psychosis. Our UHR individuals had no family history of psychosis, while schizophrenia has a strong genetic component.76
It is thus hard to generalize our results to schizophrenia. A potential explanation for structural differences we found is that these differences may be associated with environmental or epigenetic factors rather than genetic factors. Genetic factors may not be necessary to produce structural changes in UHR subjects. Among the 8 converters, 5 individuals converted to schizophrenia and others converted to other forms of psychiatric disorders. Therefore, our results may be related to psychosis rather than schizophrenia.
Several limitations in the present study should be taken into account. First, all schizophrenia patients and 10 UHR subjects were taking atypical antipsychotics, raising the possibility of a medication confound. Recent studies suggest that antipsychotic treatment may contribute to the changes of cortical thickness,77,78
although some studies have not found these effects.46,47
However, in this study, the UHR subjects took psychotropic medications only after the complete clinical evaluations, and the time intervals between the beginning of medications and the MRI scans were relatively short (mean = 4.64 ± 4.48 d). Additional analysis revealed no significant differences in cortical thickness between 10 medicated and 19 unmedicated subjects in the UHR group. Comparisons between the 19 unmedicated UHR subjects and HC subjects remained significant in terms of differences in ROIs. However, we still cannot rule out the confounding effects in comparisons between schizophrenia and other groups. Further research is needed to examine more precisely how the effects of treatment affect these changes in cortical thickness. Second, the present study was not able to directly compare longitudinal brain changes in converters with those in nonconverters. So, it is not clear whether the structural changes found led to the progressive effects of the disease or other factors. However, a previous study using a similar methodology lacked a HC group and so did not address the deviation from normal brain changes.54
The present study contained HC, schizophrenia, and UHR groups without a family history of psychosis, and the sample size for each group was modest as these subjects were relatively difficult to recruit.
In summary, we demonstrated a pattern of cortical thinning in the UHR group compared with HC group, particularly in the ACC, PFC, STG, and inferior parietal regions, and these changes were further extended for schizophrenia by applying surface-based method. These structural abnormalities could be potential candidates for vulnerability markers for the development of schizophrenia. Further longitudinal investigation of these groups will clarify the issue of progressive brain changes related to the illness process.