Background
Schizophrenia is a severe mental disorder ranked among the top 20 causes of disability worldwide [
1]. The prolonged illness course and impaired cognitive functioning are the main reasons contributing to the disease burden [
2,
3]. Yet, the pathological mechanism of schizophrenia remains to be unclear. Imaging genetics study provides an opportunity to quantify disease-related neuroanatomical deviations and to elucidate the potential biological mechanism underlying these changes. In recent years, imaging genetics analysis has been increasingly applied to schizophrenia to reveal its potential pathological pathway [
4,
5].
The suggestion that immune dysfunction may contribute to the pathophysiology of schizophrenia has a long history [
6]. The vulnerability-stress-inflammation model [
7] proposes that genetic risks and early life exposures, such as stress or infection, may promote a chronic pro-inflammatory state and subsequently lead to neurotransmission disturbances and brain structural changes that are relevant for schizophrenia [
7,
8]. Supporting this hypothesis, therapeutic studies indicated a beneficial effect of using anti-inflammatory medications as an add-on treatment in the early stage of schizophrenia [
7]. Studies of inflammation indicated that genetically predicted IL-6 was associated with brain volume in the middle temporal gyrus and may potentially be involved in the pathology of schizophrenia [
9]. Likewise, studies also found overlapped genetic loci of schizophrenia and cortical morphometry to be enriched in immunologic gene sets [
4]. However, few longitudinal studies have examined whether treatment-associated brain structural changes are related to neuroinflammation or gene expression of inflammatory markers.
Studies have indicated that progressive brain structural changes have been shown in the general population and patients with schizophrenia [
10,
11]. In schizophrenia, previous studies found subcortical enlargement and cortical thinning in widespread brain regions [
11,
12]. The frontal and temporal cortices were the most affected regions with excessive cortical thinning in patients than controls [
11,
13]. Several large-scale, longitudinal studies have been conducted on schizophrenia [
14], and a few studies focused on the early stage of illness when the majority of patients were medication-naïve and experienced substantial symptom alleviation [
15,
16]. However, most studies did not recruit controls for follow-up, and it is difficult to distinguish the effects of disease progression and natural aging.
Considering that surface area and cortical thickness reflect different aspects of neural development processes [
17] and may be differently affected in schizophrenia, we separately analyzed cortical thickness, surface area, and subcortical volume using surface-based morphometry (SBM). The aim of this study is threefold: (1) to determine the early-stage brain structural changes after a short period of treatment in patients with schizophrenia, (2) to investigate factors (inflammation and antipsychotic treatment) that may contribute to brain structural changes by correlating structural changes with transcription level of gene sets of interest, and (3) to evaluate the association of brain structural changes with changes in symptoms and cognitive performances in patients.
Discussion
This longitudinal study examined the patterns of brain structural changes in patients with first-episode schizophrenia and investigated the contributing factors (inflammation and antipsychotic treatment) to brain structural changes. We found that patients had accelerated pallidum enlargement and frontal cortical thinning but preserved parietal and occipital cortical thickness as compared to controls. Among these structural changes, cortical thickness change in the left superior parietal lobule positively correlated with cognitive performance changes in patients. Notably, we observed a positive correlation between the gene expression level of monocyte and cortical thinning in patients. Taken together, our results identified early-stage brain structural changes and their correlation with the transcriptional profile of inflammatory markers, providing preliminary evidence of the underlying biological process that may contribute to longitudinal brain structural changes in schizophrenia.
Immunological abnormalities have long been reported to be involved in schizophrenia. In previous schizophrenia studies, higher peripheral monocyte count could be relevant to the pathophysiology [
36] and has been considered a possible marker of microglia activation [
37]. Inflammation in the brain involved both monocyte and microglia. Microglial cells and chemokines might recruit monocytes circulating in the peripheral blood into the brain [
38], where neuroinflammation was exacerbated by these monocytes. The investigation of monocyte transcriptome demonstrated a shift in monocyte phenotype in different stages of schizophrenia [
39]. In the current study, the transcriptional profile of monocyte was directly related to the longitudinal change of cortical thickness in patients, suggesting an association of monocyte gene expression with the effect of treatment and this disease itself on the brain.
However, we did not observe a significant association between structural changes and inflammation markers. Previous studies revealed elevated peripheral inflammatory levels in patients with schizophrenia and their correlation with increased basal ganglia volume [
40]. The effect of peripheral inflammatory markers on brain structural changes was further supported by genetic studies implicating multiple biological processes such as neural development and synaptic transmission [
9]. Leukocyte count, even within the normal range, was correlated with polygenic scores of schizophrenia [
41]. Since the antipsychotic medication was related to the decrease of cytokines [
42], and basal ganglia volume change was quite common after medication [
43], one possible mechanism of antipsychotics in the treatment of psychosis could be mitigating neuroinflammation.
Although antipsychotics can alleviate symptoms of schizophrenia, some studies raised concern that they may contribute to additional cortical thinning [
44]. Subsequent studies argued that cortical thinning after medication was not linked to clinical or cognitive deterioration [
45] and might even improve patients’ prefrontal functional activity and cognitive control ability [
13]. In a macaque monkey study, antipsychotic-induced volume reduction was related to reduced glial cell number, whereas the number of neurons remained unchanged [
46]. In our study, we found the transcriptional profile of monocyte positively correlated with cortical thickness changes after treatment, indicating the important role of inflammation, especially monocyte, in cortical thickness changes in the early stage of schizophrenia.
Several other factors may contribute to structural changes in patients with schizophrenia, such as medication and neuromodulation treatment. Studies found that medication duration and dosage correlated with cortical thinning in patients [
11,
45], with effects varied across different types of antipsychotics [
14]. As a non-invasive adjuvant therapy, repetitive transcranial magnetic stimulation (rTMS) may prevent volume reduction or cortical thinning [
47] and increase brain metabolism at the stimulation site [
48]. The temporoparietal junction is a commonly applied target site.
We also analyzed the factors that might contribute to these structural changes, including medication, times of rTMS, and peripheral inflammatory markers. Although given the fact that treatment factors were able to cause structural changes, we did not find a significant correlation between brain changes and the medication dose, duration, or times of rTMS as former studies reported [
11,
45]. Three reasons might contribute to this. First, it might be due to our lack of comparison to unmediated patients, since the disease course would also lead to cortical thinning even in prodromal high-risk individuals [
49]. Thus, when calculating the effect of treatment, the effect was not linearly related to structural changes due to the interference of the disease course. Second, it could also be on account of the combined treatment of multiple antipsychotics for each patient, because some studies argued that different types of antipsychotics had varied influences on structural changes as well [
14]. Lastly, different patients had different responses to antipsychotics, and our limited sample size might not suffice to prove the correlation with treatment factors.
Among the brain regions showing significant group or group by time interaction effects, we found a positive correlation between superior parietal lobule thickness change and WAIS digital span score change in patients, suggesting the cognitive relevance of this region. The superior parietal lobule belongs to the association cortex and is involved in high-order processes like cognition, information integration, and self-awareness [
50]. Structural and functional abnormalities of the frontal-parietal network were frequently reported and have been related to cognition impairments in schizophrenia. Previous studies have reported that low-frequency rTMS could regulate functional connectivity within the frontal-parietal network and improve cognitive functioning [
47]. Our study supported the role of the superior parietal lobule in working memory and suggested this potential region as a stimulation target site for cognitive remediation.
In the study, we found a regional-specific effect of accelerated frontal cortical thinning and pallidum enlargement as well as preserved parietal and occipital cortices that may be due to medication and rTMS, respectively. Our results were consistent with former cross-sectional studies [
11‐
13,
43] and provided a longitudinal view in patients with first-episode schizophrenia who were unmedicated and responsive to treatments. Remarkably, consistent with previous meta-analysis [
11] and longitudinal study [
45], cortical thickness represented more prominent changes in regional analysis rather than surface area, which might reflect different mechanisms underlying these two measurements. It is assumed that the developmental trajectories of surface area were predominantly influenced by genetic factors [
11,
17]. Meanwhile, cortical thickness can be affected by additional environmental or neurodegenerative factors (disease, inflammation, treatment, aging, etc.) even in adulthood [
17]. Our study supported that cortical thickness might be more sensitive to treatment and inflammation effects.
Several limitations need to be considered when interpreting our results. First, transcriptomic data were not collected in this sample but retrieved from the external AHBA dataset, leading to false-positive effects in micro–macro association due to spatial autocorrelation. Using generative null models, the correlation results between brain and transcriptional maps were verified. Second, our findings of brain structural changes reflect mixed effects of treatment and disease progression on the brain, which cannot be easily separated apart. However, we tried to control for the time effect by including follow-up MRI measurements of healthy controls in linear mixed-effect models. Third, the sample size of this study was moderate. Future studies with a larger sample size or using a meta-analytic approach could further confirm our findings.
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.