Positive Effects of Uric Acid on White Matter Microstructures and Treatment Response in Patients With Schizophrenia

Abstract

Background and Hypothesis

Schizophrenia involves microstructural changes in white matter (WM) tracts. Oxidative stress is a key factor causing WM damage by hindering oligodendrocyte development and myelin maturation. Uric acid (UA), an endogenous antioxidant, may protect against oxidative stress. We investigated the effect of UA on WM connectivity in antipsychotic-naive or -free patients with early- or chronic-stage schizophrenia.

Study Design

A total of 192 patients with schizophrenia (122 recent-onset [ROS] and 70 chronic [CS]) and 107 healthy controls (HCs) participated in this study. Diffusion tensor imaging data and serum UA levels at baseline were obtained.

Study Results

Fractional anisotropy was lower in the widespread WM regions across the whole brain, and diffusivity measures were higher in both schizophrenia groups than in HCs. The CS group showed lower diffusivity in some WM tracts than the ROS or HC groups. The linear relationship of serum UA levels with axial and mean diffusivity in the right frontal region was significantly different between schizophrenia stages, which was driven by a negative association in the CS group. WM diffusivity associated with serum UA levels correlated with 8-week treatment responses only in patients with CS, suggesting UA to be protective against long-term schizophrenia.

Conclusions

UA may protect against the WM damage associated with the progression of schizophrenia by reducing oxidative stress and supporting WM repair against oxidative damage. These results provide insights into the positive role of UA and may facilitate the development of novel disease-modifying therapies.

Introduction

Schizophrenia is a chronic disorder characterized by a broad spectrum of cognitive, perceptual, and affective impairments, and is suggested to consist of brain dysconnectivity involving widespread microstructural alterations in the white matter (WM).^1–3 The disrupted WM connectivity in patients with schizophrenia may originate from abnormal neurodevelopmental processes.^4,5 However, changes in WM connectivity have also been observed during the chronic course of schizophrenia, suggesting that various neurobiological factors may affect the integrity of WM tracts.^6–8

Oxidative stress is a key contributor to neuronal insults and malfunctioning of the brain.⁹ The brain, particularly its WM components, is susceptible to oxidative stress, given its high energy demand from oxidative glucose metabolism, high iron and lipid contents, and relatively low antioxidant capacity.^10–12 Oxidative stress disrupts WM microstructures by hindering oligodendrocyte development and myelin maturation,^13,14 promoting axonal degeneration, demyelination, and death of neurons and oligodendrocytes.^15,16 Impaired antioxidant defense^17–20 and its association with greater illness severity,²¹ neurocognitive dysfunctions,^22,23 and poor treatment response²⁴ are reported in patients with schizophrenia. Neuroimaging studies have reported that deficient antioxidant defenses are associated with a progressive loss of brain volume²⁵ and impaired WM maturation young patients with early psychosis.²⁶ These findings suggest that sufficient antioxidant capacity is required to maintain the structural and functional integrity of the WM in the brain of patients with schizophrenia.

Uric acid (UA), the end product of purine metabolism in humans and the great apes, is one of the major endogenous antioxidants, with a powerful free-radical scavenging activity²⁷; against reactive oxygen species, it provides up to 60% of the antioxidant capacity of the human body.^28,29 While excessive UA in the serum causes inflammation in the joints in its crystallized form, the soluble form of UA has neuroprotective properties within the central nervous system (CNS) by preserving the peroxidase activity of antioxidant enzymes and inhibiting iron-induced ascorbate oxidation.²⁹ High UA levels are reported as a potential indicator of lower risk and mild progression in neurodegenerative diseases, including Parkinson’s and Alzheimer’s diseases, and amyotrophic lateral sclerosis.³⁰ A similar relationship between UA levels and disease activity has been observed in patients with multiple sclerosis, an autoimmune demyelinating disease of the CNS.^31–33 Plasma UA levels are reportedly lower in patients with major depressive disorder (MDD) and anxiety disorders than in healthy controls (HCs).^34–36 A diffusion tensor imaging (DTI) study of MDD demonstrated a protective role of UA in WM organization, as serum UA levels were positively and negatively correlated with fractional anisotropy (FA) and radial diffusivity (RD), respectively, in WM regions beneath the frontal cortex.³⁷ This finding suggests that serum UA may affect the structural connectivity of WM tracts, possibly by altering myelin compactness.

Serum and plasma UA levels are reportedly decreased in patients experiencing their first episode of schizophrenia, but not in those with chronic schizophrenia (CS), compared with HCs.^38,39 This suggests that decreased antioxidant capacity, indicated by lower UA levels, may be involved in the early pathogenesis of schizophrenia. UA levels are reported to be elevated in an acute episode and reduced after antipsychotic treatment in chronic patients with schizophrenia,^40,41 and that, 5 years after an acute episode, they are higher in patients with stable schizophrenia than in HCs.⁴² Although these findings are inconsistent, the oxidative status affecting UA levels may change dynamically during the course of schizophrenia. Furthermore, Luo et al⁴³ have demonstrated the causal effect of the genetic risk for schizophrenia on serum UA levels, but not vice versa, implying that alterations in UA levels may be a consequence or epiphenomenon of the pathogenesis of schizophrenia. Therefore, investigating the potential of serum UA levels as a biological indicator is relevant for evaluating the clinical status of patients with schizophrenia.

Considering previous neuroimaging studies showing the association between oxidative stress and WM structures,^25,26,37 we believe that investigating the relationship between UA levels, WM connectivity, and treatment response in patients at different stages of schizophrenia could provide further insight into the disease. Here, we examined the potential effects of UA on WM connectivity in antipsychotic-naive or -free patients with schizophrenia in the early or chronic stages of the disease. We hypothesized that (1) patients with schizophrenia would show impaired WM connectivity in widespread brain regions compared with HCs, and that (2) serum UA levels would affect WM connectivity in patients with schizophrenia, but depending on the clinical stage of the disease. We further examined the associations between serum UA levels, WM connectivity, and treatment response to clarify the potential role of UA in schizophrenia.

Methods

Participants

A total of 192 patients with schizophrenia and 107 HCs participated in this study. Patients with schizophrenia were recruited among those receiving inpatient psychiatric treatment for an acute episode at the Department of Psychiatry, CHA Bundang Medical Center (Seongnam, Republic of Korea). HCs from the local community were enrolled through online and printed advertisements. All participants underwent a diagnostic interview using the Structured Clinical Interview for DSM-IV-TR Axis I Disorder^44,45 or DSM-5⁴⁶ by experienced psychiatrists. HCs were ensured to have no personal or first-degree family history of psychiatric disorders. Patients with schizophrenia were included if they were antipsychotic-naive or had been antipsychotic-free for at least 6 months. Recent onset was defined as duration of the illness of up to 5 years.⁴⁷ One hundred and twenty-two patients (71.3%) were classified as recent onset, and 70 patients were classified as chronic (duration of illness longer than 5 years). The exclusion criteria were as follows: (1) presence of other psychiatric comorbidities, including mood disorders and substance-related disorders; (2) intellectual disability; (3) clinically considerable or unstable medical illness, including inflammation; (4) metabolic syndrome or renal disease that may affect serum UA levels; (5) neurological disorders or traumatic brain injury; (6) pregnancy and lactation; (7) left-handedness (assessed using the Edinburgh Handedness Inventory⁴⁸); and (8) any contraindications for undergoing magnetic resonance imaging (MRI).

All study procedures were reviewed and approved by the Institutional Review Board of CHA Bundang Medical Center following the latest version of the Declaration of Helsinki and the principles of Good Clinical Practice. All participants provided written informed consent after receiving a thorough explanation of the study procedure.

Clinical Assessment

The severity of clinical symptoms for patients with schizophrenia was assessed using the Positive and Negative Syndrome Scale (PANSS)⁴⁹ at baseline. Follow-up assessment was conducted for the 77 patients with recent-onset schizophrenia (ROS) and 58 with CS after 8 weeks of antipsychotic treatment, as recommended in standard practice. The PANSS was administered by well-trained psychiatrists independent of the study, and symptom improvements were quantified as the percentage of reduction in the total score and each of the subscale scores of the PANSS over the 8-week period. As the PANSS adopts a 1- to 7-point scoring system, the percentage of reduction in the PANSS scores could be up to 85.7%.⁵⁰

Biochemical Analysis

Blood samples for serum UA analysis were collected from the antecubital vein of the patients before initiating the antipsychotic treatment. For each patient, venous blood samples (6 ml) were collected in a gel tube and processed immediately after collection at a laboratory of CHA Bundang Medical Center. Serum UA levels were measured using enzyme colorimetry and the Roche UA2 module from the Roche/Hitachi Cobas 8000 c702 Chemistry Autoanalyzer (Roche Diagnostics GmbH), according to the instructions of the manufacturer.

MRI Data Acquisition

The MRI data were acquired using a 3.0-Tesla GE Signa HDxt scanner (GE Healthcare) at the CHA Bundang Medical Center. Patients with schizophrenia and HCs underwent MRI at baseline. MRI was performed within 4 weeks after initiating the antipsychotic treatment in patients with schizophrenia if they were uncooperative because of exacerbated psychotic symptoms. Diffusion-weighted images (DWIs) were obtained using an echo planar imaging (EPI) sequence with the following parameters: repetition time, 17 000 ms; echo time, 108 ms; field-of-view, 240 mm; matrix, 144 × 144; slice thickness, 1.7 mm; and voxel size, 1.67 × 1.67 × 1.7 mm³. A double-echo option was applied to reduce the eddy current-related distortions. An 8-channel head coil was employed for parallel imaging using the Array Spatial Sensitivity Encoding Technique (ASSET; GE Healthcare) with a sensitivity encoding factor of 2 to mitigate EPI spatial distortions.^51,52 Seventy axial slices parallel to the anterior-posterior commissure line were obtained in 51 directions at b = 900 s/mm² with 8 baseline scans at b = 0 s/mm². DTIs were extracted from the DWIs using the least squares method. The total scanning time was 17 minutes.

DTI Analysis

DTI data were analyzed using the Functional MRI of the Brain (FMRIB) Diffusion Toolbox (FDT) and Tract-based Spatial Statistics (TBSS),⁵³ implemented in the FMRIB Software Library (FSL version 6.0; https://fsl.fmrib.ox.ac.uk/fsl/). All data were visually inspected by experienced investigators for major artifacts, such as geometric distortions, signal dropouts, and insufficient image acquisition.⁵⁴ Subsequently, preprocessing was performed according to the standard FSL protocol. Eddy current and motion-related distortions were corrected using FSL’s “eddy_correct” with the first b = 0 image set as a reference for alignment,⁵⁵ rotating the b-vectors accordingly.⁵⁶ The head motion parameters (mean displacement, rotation, and translation) were extracted from the “eddy_correct” logfile (*.ecclog) to ensure minimal head motion (mean displacement <2 mm). Head motion parameters did not differ among the 3 groups (supplementary table S1). Non-brain tissues were stripped using the Brain Extraction Tool.⁵⁷ FA and non-FA (axial diffusivity [AD], RD, and mean diffusivity [MD]) images were created by fitting a tensor model to the corrected diffusion data using FDT’s “dtifit” function. AD and RD represent the magnitude of diffusion parallel and perpendicular to the WM fibers, respectively. MD is a measure of the average amount of diffusion in each direction. FA refers to the degree of anisotropic diffusion, which describes the extent to which water molecules move in organized or random directions. These images were placed in the Montreal Neurologic Institute standard space using FMRIB’s Nonlinear Image Registration Tool.⁵⁸ Converted FA images were combined and applied to the original FA map to create a standard space version of the FA map. A mean FA image was created by averaging all FA images, and only the centers of the WM tracts were used to create an average FA skeleton. The skeleton was thresholded with FA >0.2 (TBSS default) to include only the major fiber bundles; similar TBSS procedures were used to prepare non-FA images.

Statistical analysis of DTI data was performed based on the general linear model (GLM) using “randomise” in FSL.⁵⁹ Voxel-wise comparisons of the DTI maps among the ROS, CS, and HC groups were conducted at the whole-brain level using a 1-way analysis of variance with F-statistics to determine whether there were any significant between-group differences. Age and sex were demeaned and entered as covariates in the design matrix. Post hoc individual t tests were used to determine the direction of differences in DTI measures between each pair in the 3 groups. Differential effects of serum UA levels on DTI measures were tested between the ROS and CS groups. Demeaned serum UA values were entered into the design matrix separately for each group, and differences in the associations between serum UA and DTI measures were compared using t-statistics (supplementary figure S1). Age and sex were controlled for as covariates, as in previous statistical methods. A permutation-based nonparametric inference within the framework of the GLM (number of permutations = 10 000) was performed to achieve an accurate inference. The statistical significance was adjusted for multiple comparisons using the threshold-free cluster enhancement (TFCE) method, avoiding an arbitrary choice of the cluster-forming threshold while preserving the benefits of cluster-wise corrections.

Statistical Analysis

Demographic and clinical characteristics were compared using analysis of (co)variance or independent t tests for continuous variables and chi-square tests for categorical variables. Partial correlation analysis, including age and sex as covariates, was performed to examine the associations between the mean values of the DTI measures extracted from significant clusters in the TBSS analysis, serum UA levels, and treatment response. Additionally, a mediation analysis was performed to estimate the effect of WM connectivity and serum UA levels on treatment response. Statistical significance is presented uncorrected unless otherwise indicated. Statistical analyses were conducted using the Statistical Package for the Social Sciences (version 27; IBM Corp). The PROCESS macro was used to test the mediation model.⁶⁰ Indirect effects were estimated using bootstrapping with 5000 resamples at 95% CIs.

Results

Demographic and Clinical Characteristics of the Study Participants

Table 1 shows the demographic and clinical characteristics of the study participants at baseline. Patients with ROS were younger than HCs and those with CS, and had a shorter duration of illness than the patients with CS. One hundred and forty-four participants with schizophrenia (75.0%; 113 ROS and 31 CS) were antipsychotic-naive, and the remainder had been antipsychotic-free for at least 6 months. There were 146 participants with schizophrenia who were medicated with antipsychotics before an MRI scan for more than 1 day (81.3%; 99 ROS and 57 CS). Among the medicated participants, MRI was performed within 1 week in 113 participants (77.4%; 75 ROS and 38 CS) and between 1 and 2 weeks in 28 participants (19.2%; 17 ROS and 11 CS). The severity of clinical symptoms was not significantly different between the ROS and CS groups, except for the negative symptoms, which were higher in the CS than in the ROS group. The statistical significance remained unchanged after controlling for age, sex, and the chlorpromazine equivalent dose of antipsychotics on MRI scans.

Table 1.

Demographic and Clinical Characteristics of the Study Participants

	HCs (n = 107)	ROS (n = 122)	CS (n = 70)	Statistics	P
Sex
Male, n (%)	47 (43.9)	38 (31.1)	28 (40.0)	Χ2 = 4.15	.126
Female, n (%)	60 (56.1)	84 (68.9)	42 (60.0)
Age (y, mean ± SD)	39.4 ± 9.8	32.0 ± 11.3	39.6 ± 12.3	F = 16.54	<.001a
Duration of illness (mo, mean ± SD)a		7.4 ± 11.2	136.1 ± 91.6	t = −11.7	<.001b
Duration of antipsychotics before MRI scan (d, mean ± SD)		5.1 ± 5.7	6.6 ± 7.7	t = −1.54	.126
Chlorpromazine equivalent dose of antipsychotics at MRI scan (mg/d, mean ± SD)		410.2 ± 305.7	541.5 ± 383.9	t = −2.45	.016b
PANSS at baseline (mean ± SD)c
Total		110.8 ± 26.0	116.3 ± 24.8	t = −1.43	.154
Positive symptom		28.8 ± 6.5	28.4 ± 6.5	t = 0.43	.666
Negative symptom		26.2 ± 8.9	28.8 ± 8.5	t = −1.99	.048
General psychopathology		55.8 ± 13.6	59.1 ± 14.0	t = −1.59	.112

Note: CS, chronic schizophrenia; HCs, healthy controls; MRI, magnetic resonance imaging; PANSS, Positive and Negative Syndrome Scale; ROS, recent-onset schizophrenia.

^aPairwise comparisons: ROS < HCs (Bonferroni-corrected P < .001), ROS < CS (Bonferroni-corrected P < .001).

^bLevene’s test for equality of variances indicated that the variances were not assumed to be equal between the 2 groups.

^cThe results remained the same after controlling for age, sex, and chlorpromazine equivalent dose as covariates.

Serum UA Levels in Patients With Schizophrenia

Serum UA levels assessed before the initiation of antipsychotic treatment were lower in the ROS than in the CS group, after controlling for age and sex (F = 5.00, P = .027; figure 1). Using 1-sample t tests, serum UA levels in both ROS and CS groups were comparable (ROS: t = −1.09, P = .278; CS: t = 1.545, P = .127) to the age-standardized mean serum UA levels in the general Korean population (5.1 mg/dl).⁶¹ In the ROS and CS groups, 17 (13.9%) and 13 (18.6%) patients had hyperuricemia (men, >7 mg/dl; women, >6 mg/dl), respectively, which were higher than the prevalence of hyperuricemia in the general Korean population (11.4%). The proportion of patients with hyperuricemia was not significantly different between the ROS and CS groups (Χ2 = 0.73, P = .394).

Fig. 1.

Comparison of serum UA levels between ROS and CS groups after controlling for age and sex. Note: CS, chronic schizophrenia; ROS, recent-onset schizophrenia; UA, uric acid.

Baseline Comparison of WM Connectivity Among Patients With ROS, CS, and HCs

Whole-brain voxel-wise comparisons among patients with ROS, CS, and HCs, after controlling for age and sex, showed significant differences in WM connectivity across widespread brain regions in each of the FA, AD, RD, and MD maps (TFCE-corrected P < .05; supplementary figure S2). Upon pairwise comparisons (TFCE-corrected P < .05/6 = .008), both ROS and CS groups showed significantly lower FA values than HCs in the WM tracts connecting various association cortices and subcortical areas, while no significant differences between the ROS and CS groups were observed (figure 2A). Diffusivity measures were higher in the ROS and CS groups than in HCs (figures 2B–D). Notably, patients with CS showed a distinctive pattern of between-group differences in several brain regions compared with patients with ROS or HCs. Patients with CS had higher AD in a localized region of the splenium of the corpus callosum, whereas patients with ROS had higher AD in larger areas across the temporo-parietal and subcortical regions. Patients with CS had lower AD in the fronto-temporo-parietal areas than the other 2 groups, and the region size was greater than that of patients with ROS. The results were broadly similar to those of the other diffusivity measures.

Fig. 2.

Pairwise comparisons for each DTI measurement (A: fractional anisotropy [FA]; B: axial diffusivity [AD]; C: radial diffusivity [RD]; D: mean diffusivity [MD]). The statistical significance was corrected for multiple comparisons (TFCE-corrected P < .05/6 = .008). Note: CS, chronic schizophrenia; DTI, diffusion tensor imaging; HCs, healthy controls; ROS, recent-onset schizophrenia; TFCE, threshold-free cluster enhancement.

Clinical Symptom Improvements After 8 Weeks of Antipsychotic Treatment in Patients With Schizophrenia

After 8 weeks of antipsychotic treatment, 77 and 58 patients in the ROS and CS groups, respectively, were followed up to assess their treatment response. Table 2 shows the characteristics of the 8-week follow-up period. Patients with CS had higher PANSS total and subscale scores than those with ROS. Statistical significance remained after controlling for age, sex, and the chlorpromazine equivalent doses of antipsychotics at 8 weeks. The percentage of reduction in the PANSS for the total and subscale scores of positive symptoms and general psychopathology was significantly greater in patients with ROS than in those with CS. The improvement in general psychopathology was not significant after controlling for age, sex, and chlorpromazine equivalent dose of antipsychotics at 8 weeks.

Table 2.

Demographic and Clinical Characteristics of Patients With Schizophrenia Who Were Followed Up After 8 Weeks of Antipsychotic Treatment

	ROS (n = 77)	CS (n = 58)	Statistics	P
Baseline
Sex
Male, n (%)	21 (27.3)	26 (44.8)	Χ2 = 4.49	.034
Female, n (%)	56 (72.2)	32 (55.2)
Age (y, mean ± SD)	32.5 ± 11.5	38.6 ± 12.4	t = −2.97	.004
Duration of illness (mo, mean ± SD)	8.7 ± 13.2	138.1 ± 98.5	t = −9.93	<.001a
Duration of antipsychotics before MRI scan (d, mean ± SD)	5.5 ± 5.7	6.6 ± 7.0	t = −0.97	.335
Chlorpromazine equivalent dose of antipsychotics at MRI scan (mg/d, mean ± SD)	443.4 ± 327.1	589.3 ± 388.2	t = −2.37	.019
PANSS (mean ± SD)b
Total	109.0 ± 25.6	119.8 ± 22.7	t = −2.54	.012
Positive symptom	29.0 ± 6.3	28.8 ± 6.3	t = 0.13	.900
Negative symptom	25.3 ± 9.0	29.9 ± 8.1	t = −3.05	.003
General psychopathology	54.8 ± 13.6	61.1 ± 13.1	t = −2.72	.007
Serum UA (mg/dl, mean ± SD)c	4.9 ± 1.7	5.5 ± 1.9	F = 3.82	.053
After 8 weeks of antipsychotic treatment
Chlorpromazine equivalent dose of antipsychotics (mg/d, mean ± SD)	630.1 ± 315.2	858.6 ± 487.8	t = −3.11	.002a
PANSS (mean ± SD)b
Total	64.7 ± 20.7	79.4 ± 22.9	t = −3.98	<.001
Positive symptom	14.1 ± 4.9	17.3 ± 5.2	t = −3.52	<.001
Negative symptom	16.6 ± 6.9	21.0 ± 7.5	t = −3.53	<.001
General psychopathology	34.0 ± 10.9	41.2 ± 12.0	t = −3.65	<.001
% PANSS reduction (mean ± SD)d
Total	40.3 ± 14.0	33.3 ± 14.7	t = 2.81	.006
Positive symptom	50.6 ± 14.7	39.3 ± 16.2	t = 4.23	<.001
Negative symptom	31.8 ± 19.8	29.0 ± 17.5	t = 0.86	.389
General psychopathology	37.3 ± 15.0	32.1 ± 14.9	t = 2.02	.045

Note: CS, chronic schizophrenia; MRI, magnetic resonance imaging; PANSS, Positive and Negative Syndrome Scale; ROS, recent-onset schizophrenia; UA, uric acid.

^aLevene’s test for equality of variances indicated that the variances were not assumed to be equal between the 2 groups.

^bThe results remained the same after controlling for age, sex, and chlorpromazine equivalent dose as covariates.

^cAge and sex were controlled for as covariates.

^dThe percentage reduction in the PANSS for general psychopathology was not significant after controlling for age, sex, and chlorpromazine equivalent dose as covariates. The results of the other variables were unchanged.

Relationship Between Serum UA, WM Connectivity, and Treatment Response in Different Stages of Schizophrenia

Serum UA levels were significantly correlated with AD in the fronto-parietal regions of patients with CS, while patients with ROS showed no significant correlations after controlling for age and sex (figure 3A); other DTI measures were not correlated with serum UA levels. The linear relationship with serum UA levels differed significantly according to the disease stage in the AD and MD maps (figure 3B). Significant clusters commonly found in the AD and MD maps consisted of WM tracts in the right frontal area, including the anterior and superior corona radiata, superior longitudinal fasciculus, anterior limb of the internal capsule, and external capsule. This interaction effect was driven by a significant negative association in patients with CS, which was absent in patients with ROS (figure 3C).

Fig. 3.

Relationship between serum UA levels, WM connectivity, and treatment response in different stages of schizophrenia. (A) WM regions show a significant inverse correlation between serum UA levels and AD in patients with CS. (B) The interaction effect between the serum UA levels and stages of schizophrenia was found in the right frontal region in AD and MD maps (TFCE-corrected P < .05). (C) The interaction effect was driven by a significant negative association in patients with CS, which was absent in patients with ROS (TFCE-corrected P < .05). (D) The mean AD and MD values extracted from the WM region showing the interaction between serum UA levels and diffusivity measures were negatively associated with the 8-week treatment response in patients with CS. (E) Mediation models indicating that the effect of serum UA on treatment response was mediated completely by the alterations in AD and MD in the right frontal region. Note: AD, axial diffusivity; CS, chronic schizophrenia; MD, mean diffusivity; ROS, recent-onset schizophrenia; TFCE, threshold-free cluster enhancement; UA, uric acid; WM, white matter.

Serum UA levels were not correlated with treatment response, presented as the percentage of reduction in the total PANSS scores, in either the ROS or CS groups, after controlling for age, sex, and chlorpromazine equivalent dose of antipsychotics after 8 weeks (ROS: r = 0.041, P = .730; CS: r = 0.126, P = .360). The mean AD values extracted from the WM regions that showed a negative correlation with serum UA levels in the CS group were also not associated with the treatment response (r = −0.154; P = .261). However, the mean AD and MD values in the WM regions showing different linear relationships with serum UA levels between the ROS and CS groups significantly correlated with the treatment response in patients with CS (figure 3D). The mediation analysis (figure 3E) showed that serum UA levels were negatively associated with AD (β = −0.000028, P < .001) and MD (β = −0.000015, P < .001), which in turn correlated with the treatment response in each model (AD: β = −46 827.56, P = .029; MD: β = −76 953.05, P = .032). The indirect effects of serum UA on treatment response were significant in both models. However, the residual direct effects were not significant, indicating that AD and RD fully mediated the relationship between serum UA levels and treatment response.

Discussion

In this study, the relationships between serum UA levels, WM connectivity, and 8-week treatment response were investigated in patients with ROS and CS. Although the serum UA levels in patients with schizophrenia were not significantly different from that in the general population, the CS group had higher serum UA levels than the ROS group. These patients had lower FA and higher diffusivity measures in a wide range of WM regions across the whole brain than the HCs. However, the CS group showed a unique pattern of differences distinct from the ROS group and HCs, in terms of diffusivity measures. Furthermore, the association of serum UA levels with WM connectivity and treatment response differed, depending on the clinical stage of schizophrenia. Higher serum UA levels were associated with lower AD and MD in the WM tracts of the right frontal region, whereas no significant association was observed in patients with ROS. The mean AD and MD values from the WM regions also had significant interaction between serum UA and the schizophrenia group, showing an inverse correlation between the serum UA levels and the 8-week treatment response in patients with CS.

Serum or plasma UA levels are reportedly decreased in first-episode patients^38,39 and increased in patients with CS 5 years after an acute episode.⁴² Consistently, our results showed higher serum UA levels in patients with CS than in those with ROS. Schizophrenia has been suggested to be associated with oxidative damage to the brain, which is linked to other pathophysiological processes, including neuroinflammation, neuronal and mitochondrial dysfunction, and excitation-inhibition imbalance.¹¹ In studies using animal experimental models, UA demonstrated neuroprotective effects by reducing oxidative stress, inhibiting neuroinflammation, and altering blood-brain barrier permeability.^62–65 As an endogenous free-radical scavenger, UA levels increase as a compensatory mechanism to counteract excessive oxidative stress in several medical conditions such as atherosclerosis and neuroinflammatory diseases.^66–68 Considering that higher serum UA levels indicate an increased antioxidant capacity of the serum,⁶⁹ patients with CS may have a larger antioxidant capacity to counterbalance the accumulated oxidative burden and restore brain damage associated with the chronicity of the illness.

The widespread impaired WM connectivity found in both ROS and CS groups compared with HCs is consistent with previous literature.^1,2,70–72 Longitudinal studies that followed individuals at ultra-high risk for psychosis have demonstrated that WM volume and connectivity measures decrease with the progression of psychosis.^73–76 This suggests that the pathogenic alterations in WM tracts occur before the development of full-blown psychosis, even though these are less severe than those in patients with schizophrenia. Lower values of FA, which quantifies the degree of water diffusion along the neuronal fibers, indicate structural damage to myelin integrity, which is determined by the structural characteristics of the myelin sheath, axonal membrane, and neurofibrils.⁷⁷ FA changes are usually accompanied by alterations in other diffusivity measures as FAs are calculated based on water diffusivity parallel and/or perpendicular to neuronal fibers.^78,79 Water diffusivity parallel to WM tracts (AD) is associated with axonal density or diameter, while that perpendicular to the WM tract (RD) is associated with myelin compactness.^77,80 Therefore, the overall finding of higher AD, RD, and MD values in patients with ROS and CS than in HCs suggests that WM changes in patients with schizophrenia may be due to decreased axonal density and myelin compactness.⁸¹

Patients with CS showed not only higher but also lower diffusivity in several WM regions than patients with ROS or HCs, which was more prominent in AD than in RD and MD. The regions showing significantly lower AD in patients with CS compared with those with ROS overlap with those showing a significant inverse correlation with serum UA levels in patients with CS. Therefore, the differences in diffusivity measures, particularly AD, between the ROS and CS groups could have resulted from the effect of serum UA levels on the WM tracts. Oxidative stress is a pathogenic factor that disrupts WM connectivity due to axonal damage and demyelination in patients with schizophrenia.^26,39,82,83 Although both ROS and CS groups showed overall impairment in WM microstructures, patients with CS with higher serum UA levels may benefit from the antioxidant function of UA in facilitating WM repair against oxidative stress-induced axonal damage and demyelination.⁸⁴ The potential positive effect of UA is reflected in our finding of the negative correlation between serum UA levels and AD and MD in the right frontal WM tracts, which correlated negatively with the 8-week treatment response in patients with CS. Although no direct association was found between serum UA levels and treatment response, mediation analysis revealed an indirect effect of serum UA on treatment response via alterations in AD and MD in patients with CS. Patients with ROS did not show any significant relationship between serum UA levels, WM connectivity, and treatment response. RD did not correlate with serum UA levels; thus, the association between serum UA levels and MD may be affected by alterations in AD. In addition, treatment response was correlated with AD in WM tracts, showing differential patterns of association with serum UA levels between the ROS and CS groups, and not in the entire regions showing a correlation in patients with CS. As redox reactions equilibrium is a dynamic process that maintains tissue homeostasis and protects against pathological events, compensatory UA changes may be more prominent in patients with long-term illness than in those who have recently developed schizophrenia.

Our finding of an inverse correlation between serum UA levels and AD, but not RD, in patients with CS, indicates that UA may have a primary influence on axonal structure rather than on myelin.⁸⁵ The region showing a significant association with treatment response was localized in the right frontal area and contained WM fibers connecting the fronto-subcortical and fronto-posterior cortical areas. Frontal dysconnectivity has been suggested as a critical factor from which a broad range of schizophrenia symptoms arises.⁸⁶ The association between UA and AD may be because of the exceptional susceptibility of myelin-producing cells to oxidative stress, which requires a more powerful antioxidant potential than endogenous UA can compensate.¹² Oligodendrocytes and their precursor cells are particularly vulnerable to oxidative stress as they contain higher levels of free iron than that required for myelin synthesis and maintenance.⁸⁷ Furthermore, oligodendrocytes themselves produce reactive oxygen species owing to the intensive metabolic activity that occurs during myelination.⁸⁸ Since myelin abnormalities are central to the development of brain dysconnectivity in schizophrenia,^89,90 further research is necessary to elucidate how the pathogenesis of schizophrenia related to WM interacts with UA and other antioxidant mechanisms.

This study had several limitations. First, UA levels were measured in peripheral blood. Although peripheral blood has commonly been used as a proxy for the brain, whether UA levels in the CNS are comparable to those in peripheral blood is to be determined. Further studies are required to verify the close association between UA levels in the serum and other body fluids. Second, as UA levels were not measured in HCs, we could not demonstrate the differences in UA levels between patients with schizophrenia and HCs. In addition, the cross-sectional nature of this study limited the inference of the longitudinal course of biological changes and the causal relationship between changes in UA levels and WM connectivity. Future longitudinal studies using comparative data are necessary to advance our understanding of the neurobiology of schizophrenia concerning oxidative stress and WM connectivity. Third, DTI data were acquired after the initiation of antipsychotic treatment in some patients with schizophrenia owing to their prior uncooperativeness. Although the mean duration of antipsychotic treatment before the DTI scan was not sufficient to affect WM structures (5.6 ± 6.5 days), investigations with drug-naive patients are required to exclude the confounding effect of antipsychotic medication.

In conclusion, we demonstrated disrupted WM connectivity, which resulted from decreased axonal density and myelin compactness throughout the entire brains of patients with schizophrenia in both the early and chronic stages. Serum UA levels correlated positively with the degree of axonal coherence in the right frontal area and with the 8-week treatment response in the CS, but not in the ROS group. As an endogenous antioxidant, UA may protect against the WM damage associated with the progression of schizophrenia by improving the microenvironment of neuronal cells and by supporting remyelination. Our findings may expand the understanding of the positive role of UA and provide knowledge for the development of novel disease-modifying therapies for patients with schizophrenia.

Supplementary Material

Supplementary material is available at https://academic.oup.com/schizophreniabulletin/.

Funding

This study was supported by the Basic Science Research Program through the National Research Foundation, funded by the Ministry of Science and ICT, Republic of Korea (NRF-2021R1C1C1012901).