Clinical Implications
• The quantification of cortical thickness expands our understanding of the neurobiological bases of schizophrenia, and therefore provides more robust data for understanding the pathophysiology of the disease.
• No significant correlations between frontotemporal cortical thickness and age of onset and duration of the disease were found.
• There is progressive cortical thinning associated with schizophrenia.
• There is a significant interaction between age and diagnosis, with patients presenting an accelerated cortical thinning in the frontal regions.
Introduction
Schizophrenia is a chronic and debilitating disease that causes significant disruption in daily activities and social relationships, and may result in decreased quality of life.Reference Jung, Kim and Jang1 Magnetic resonance imaging (MRI) studies have evaluated brain volume and, more recently, cortical thickness in schizophrenia patients.Reference Honea, Crow, Passingham and Mackay2–Reference Nesvåg, Lawyer and Varnäs4 There is evidence that patients with schizophrenia present neurostructural abnormalities, such as cortical volume loss in the frontotemporal regions.Reference Kuperberg, Broome and McGuire3, Reference Nesvåg, Lawyer and Varnäs4 These alterations seem to progress over timeReference Wright, Rabe-Hesketh and Woodruff5–Reference Olabi, Ellison-Wright and McIntosh11 and may be attributed to a plastic adaptation of the brain to the environmentReference Weinberger and McClure12 or genetic factors,Reference Kähler, Rimol and Brown13 or they may be related to exposure to antipsychotic drugs.Reference Dorph-Petersen, Pierri and Perel14, Reference Roiz-Santiáñez, Tordesillas-Gutiérrez and Ortíz-García de la Foz15 Cortical thickness has gained attention in recent years. Because the cerebral cortex makes up most of the brain, the extent of its thickness becomes of great interest for neurodegenerative diseases and in psychiatric disorders. The decrease in cortical thickness is usually regionally specific, and the progress of atrophy can reveal much about the evolution of the disease and the factors that cause disease.Reference Fischl and Dale16
Significant thinning of the dorsolateral prefrontal cortex, medial prefrontal cortex, lateral temporal cortices, left entorhinal cortex, posterior cingulate cortex, precuneus, and lingual cortex, bilaterally have been clearly observed in schizophrenia patients in comparison to healthy controls.Reference Schultz, Koch and Wagner17 A recent meta-analysis of longitudinal structural MRI studies has identified significant frontal, temporal, and parietal lobe gray and white matter volume reductions over time in patients with schizophrenia compared with healthy controls. These findings support the hypothesis that there are progressive reductions associated with schizophrenia,Reference Olabi, Ellison-Wright and McIntosh11 possibly associated with the duration of illness in the brain structure. Premkumar et al Reference Premkumar, Kumari and Corr18, reported smaller prefrontal cortical grey matter volumes, but larger premotor cortical and putamen volumes in chronic patients, when compared to first episode patients and healthy controls, suggesting progressive brain changes in schizophrenia. However, another study failed to provide evidence for progressive cortical atrophy in short to medium durations of illness.Reference Schultz, Koch and Wagner17 Age at onset of psychosis was also investigated in one study, and seems to be correlated with superior temporal gyrus volume—a region of the brain that is involved in the pathophysiology of schizophrenia.Reference Matsumoto, Simmons and Williams19 Nevertheless, the relationship between age at onset and brain abnormalities has not been established.
In another longitudinal study, results of the comparison between patients with schizophrenia and healthy controls showed reduced cortical thickness in frontal-temporal areas in patients. This study also reported an association between cortical thinning and the use of antipsychotics, demonstrating cortical changes in patients with schizophrenia during the course of the disease.Reference van Haren, Schnack and Cahn20 Van Haren etal Reference van Haren, Cahn, Hulshoff Pol and Kahn21 revised the confounding effects of brain volume loss in schizophrenia, and showed that the use of cannabis and medication intake are important confounding factors on the interpretation of changes in brain volume. However, the volume changes in the brains of patients with schizophrenia appear to be related to the outcome of the disease, even after considering medication intake.Reference van Haren, Cahn, Hulshoff Pol and Kahn21
Bose etal Reference Bose, Mackinnon and Mehta22 conducted a study to evaluate the effect of age on brain volume in patients with schizophrenia. In this study, a reduction in the gray matter volume related to age was observed in patients with schizophrenia when compared to healthy controls. Also, an increased loss of white matter in schizophrenic patients was also observed, suggesting that at disease onset or during neurodevelopment, there is an initial reduction in gray matter, followed by a subsequent effect on the white matter.Reference Weinberger and McClure12
The aim of this study was to evaluate the effect of age of onset and duration of illness on cortical thickness in patients with schizophrenia in comparison to healthy controls.
Methods
Subjects
One hundred fifty-seven patients with schizophrenia were recruited from the Schizophrenia Program (PROESQ) at the Federal University of São Paulo (UNIFESP). The Structured Diagnostic Interview for Diagnostic and Statistical Manual of Mental Disorders, 4th ed. [DSM-IV (SCID)], applied by trained psychiatrists, was used to confirm the DSM-IV criteria for schizophrenia. Age at onset of illness was defined as onset of psychotic symptoms according to any available source. Duration of illness was defined as the difference in years between age at onset and age at investigation. Eighty-seven healthy controls without any psychiatric disorders and family history of psychotic illness, assessed through a modified version of SCID, were enrolled in this study. The groups were matched according to education, age, and gender (see Table 1 for details). No subjects had a history of neurological illness, substance abuse, or traumatic brain injury. All participants signed an informed consent approved by the Research Ethics Committee of UNIFESP.
Table 1 The socio-demographic and clinical characteristics of the sample
SD: standard deviation; N/A: not applicable; M: male; F: female.
MRI data acquisition
All subjects were submitted to a brain MRI scan at the Department of Diagnostic Radiology (UNIFESP) on a 1.5 T Siemens (Magnetom Sonata AG, Medical Solutions, Erlangen, Germany) with an 8-channel head coil Siemens (MAGNETON Sonata) scanner. A series of exploratory sagittal images (9 to 11 slices of 5 mm with 1 mm spacing) was performed in order to evaluate the image quality and positioning of the heads of the subjects. T1 images were acquired sequentially using a pulse sequence (SPGR) with the following parameters: TR = 2000 ms, TE = 3.42 ms, matrix size = 256 × 256, FOV = 245 mm, flip angle = 15°, NEX = 1, 1.0-mm slice thickness with no gaps, yielding 192 slices). Nine images were excluded because of poor-quality MRI data (due to movement).
Image processing
Cortical surface modeling and volumetric estimation of brain regions (based on T1 images) were performed by using the Freesurfer image analysis package (http://surfer.nmr.mgh.harvard.edu/). In summary, this automated processing includes (implemented in recon-all pipeline): imaging intensity normalization, removal of nonbrain tissues, segmentation of the gray/white matter and subcortical volumetric structures, tessellation of the gray/white matter boundary, topology correction, spherical surface-based intersubject registration based on the cortical surface curvature (ie, sulcus and gyri), and automated parcellation of brain regions. Further technical details of these procedures are described in prior publications.Reference Fischl, Salat and Busa23–Reference Fischl and Dale25 A postprocessing visual inspection for quality was conducted, and all processed images were considered in further analysis.
Statistical analysis
The Query Design Estimate Contrast (QDEC) interface of FreeSurfer was used to carry out a general linear model (GLM) analysis at each vertex of the cortical surface. Cortical thickness was considered as the dependent variable; group (patients vs healthy controls), age, and their interaction were explanatory variables, and intracranial volume was a nuisance variable. Results were corrected for multiple comparisons at the cluster level using the Monte Carlo approach for p-cluster < 0.05 (z-vertex > 1.3).
Findings
Overall, the cortex was significantly thinner in patients with schizophrenia compared to healthy controls, mainly in the temporal and prefrontal cortical regions, and also parietal and occipital cortices.
There was a significant negative correlation between duration of illness and the bilateral prefrontal (superior, inferior, and rostral middle gyri) and the left medial orbitofrontal cortical thickness. Since significant correlations were found between duration of illness and cortical thickness in patients with schizophrenia, regression analyses were performed to further evaluate the relationship between age and duration of illness. In this model, only age remained significant.
A significant age effect on cortical thickness was observed in the bilateral middle frontal, rostral middle frontal, and left superior frontal cortices (Figure 1).
Figure 1 Statistical maps showing age-related differences in cortical thickness between patients with schizophrenia and healthy control subjects. The colourbar represents the z-statistics value of the cluster.
A significant age by group interaction was observed for cortical thickness, with patients with schizophrenia having a thinner left superior frontal cortex with increasing age to a greater degree than did the controls (Figure 2).
Figure 2 Cortical thickness maps showing interaction effects between the between age and group (schizophrenia patients vs healthy control) at the left superior frontal gyrus (in orange).
No significant correlations between cortical thickness and age of onset were observed.
Discussion
Our main findings indicate that there is a pattern of loss of cortical thickness in the frontal region that is more pronounced in patients with schizophrenia than in healthy controls (Figure 3). In addition, there is a correlation between age in schizophrenia and thinning of the cortex in the frontal regions of patients. Significant differences were reported in the bilateral middle frontal, rostral middle frontal, and left superior frontal cortices. Furthermore, the results of our regression analysis demonstrated that age contributed more to the variation of cortical thickness than the duration of illness.
Figure 3 Scatter-plots of age and cortical thickness (local maxima of interaction effect) for the two groups.
Changes in cortical thickness are of great interest because they may suggest different neurobiological processes. Cortical thickness is able to reflect the dendritic arborization or changing myelination at the gray/white matter interface.Reference Panizzon, Fennema-Notestine and Eyler26–Reference Sowell, Thompson and Leonard29 The geometric differences are predominantly linked to the development of neuronal connections and cortical patterns of connectivity, and are thus markers of brain development.Reference Rakic30, Reference Armstrong, Schleicher, Omran, Curtis and Zilles31
A previous study that evaluated the correlation between age and cortical thickness in patients with schizophrenia and healthy controls showed a similar pattern of loss of cortical thickness among schizophrenic patients, suggesting that the reduction in cortical thickness may not be progressive throughout the course of the disease, but a pathological process occurring in a limited time related to disease onset.Reference Kubota, Miyata and Yohida32 These results are inconsistent with our findings.
A recent study on the effects of aging on cortical thickness in healthy subjects demonstrated that cortical regions are not affected equally by aging, suggesting that the frontal regions are susceptible to atrophy.Reference McGinnis, Brickhouse, Pascual and Dickerson33 Another important finding of this study is that the standard loss of cortical thickness varies with age: In old–old age, the greatest loss in primary sensory/motor cortices and regions of low postnatal surface area expansion were found. This result supports our main findings that demonstrate a loss of cortical thickness related to age in the frontal region in patients with schizophrenia, which does not follow the pattern of normal aging.
Some studies have evaluated the effect of duration of illness in schizophrenia, and the results are somehow controversial. Velakoulis etal Reference Velakoulis, Wood and Smith34 assessed disease duration and volume of gray matter, and reported volume reduction in the mean right temporal regions, anterior hippocampus, uncus, and parahippocampal gyrus, whereas Premkumar etal Reference Premkumar, Kumari and Corr18 reported a reduction in prefrontal cortex, parietal-occipital cortex gray matter, thalamus, and putamen volume. In another study, Premkumar etal Reference Premkumar, Fannon and Kuipers35 reported reduced volume in the right middle frontal, the left fusiform gyrus, and the prefrontal cortex. The cortical thickness of patients with schizophrenia was evaluated by another study,Reference Kuperberg, Broome and McGuire3 which reported a significant negative correlation of cortical thickness with duration of illness in the inferior frontal cortex, the dorsolateral prefrontal cortex, and the anterior cingulate gyrus, while Schultz etal Reference Schultz, Koch and Wagner17 and Nesvåg etal found no correlation.Reference Nesvåg, Lawyer and Varnäs4 Thus, our findings of duration of illness as it did not relate to cortical thickness are in agreement with previous studies.Reference Nesvåg, Lawyer and Varnäs4, Reference Schultz, Koch and Wagner17
Our results also demonstrated widespread cortical thinning in patients with schizophrenia compared to healthy controls, which corresponds nicely to the findings of previous studies showing the involvement of the frontal and temporal regions, among others, in schizophrenia.Reference Honea, Crow, Passingham and Mackay2–Reference Nesvåg, Lawyer and Varnäs4
Concerning the age at onset analysis, we found no correlation, whereas Matsumoto etal,Reference Matsumoto, Simmons and Williams19 in a smaller sample (40 patients with early-onset schizophrenia), reported a positive correlation between volume and age at onset time. Nevertheless, the previous study investigated an early-onset schizophrenia sample, which could be a more severe form of the disease associated with a greater genetic predisposition,Reference Kumra and Schulz36 and, hence, differing from our approach. Another important factor is the results of that study did not undergo repair statistics, which increases the probability of finding a random association.Reference Fischl and Dale16
Neuropsychological studies have demonstrated the role of the frontotemporal regions in the physiopathology of schizophrenia. The frontal lobe features in the structure of the prefrontal cortex make up a large portion of the total cortex. They establish connections with other brain structures such as anatomic, motor, sensory, and limbic areas, possibly serving as a “central executive.” The frontal lobe also contains the orbitofrontal cortex, which, according to some studies, plays a role in affective and motivational behavior. Furthermore it has been shown that abnormalities in these regions lead to affective and cognitive symptoms of schizophrenia.Reference Berberian, Trevisan and Moriyama37, Reference Goldman-Rakic and Selemon38
One of the main characteristics of this study that increase its validity is the large number of participants, with a sample that was properly paired and very homogeneous. The results underwent a conservative correction, using a threshold of p = 0.05, therefore controlling for false-positive results. A possible bias of this study is that the sample was selected from a referral center for the treatment of schizophrenia, which can make the sample unrepresentative of the community. Moreover, we have to consider the small sample of healthy controls, which could be a bias for the patients versus controls analysis. Also, an effect of medication intake and cannabis use should also be considered, as they could exert a confounding effect.Reference Lieberman, Chakos and Wu6 Another confounding effect is that, in this study, patients and controls were matched on level of education. Education has a significant effect on neuroanatomy, and its effect can be confounded with illness characteristics.
The cross-sectional nature of this study is a limitation on detecting effects of age on brain morphology, because it complicates the interpretation of the data. Moreover, the association between exposure and disease refers only to the time of data collection.
Therefore, more longitudinal studies are needed to evaluate a possible progression that is caused by schizophrenia.
Conclusion
Our findings suggest that cortical thinning observed in patients with schizophrenia is related to age, suggesting a possible progressive role of this disorder or an effect of long-term treatment with antipsychotics.
Disclosures
Idaiane Batista Assunção Leme, Vanessa Kiyomi Ota, João Ricardo Sato, Maria Isabel Melaragno, Sintia Iole Nogueira Belangero, Marilia de Arruda Cardoso Smith, and Andrea Parolin Jackowski do not have anything to disclose. Ary Gadelha: speaker's honoraria, Janssen. Jair de Jesus Mari: speaker's honoraria, Eli-Lilly. Rodrigo Affonseca Bressan: research grants, Roche and Eli-Lilly; speaker's honoraria, Eli-Lilly.
