Introduction

Schizophrenia (SCZ) is one of the several psychiatric disorders with a wide range of psychotic symptoms, with a worldwide prevalence of approximately 1.3% [1]. However, most antipsychotic drugs can alleviate the psychotic symptoms of patients, but they cannot be cured with satisfactory efficacy and safety [2]. So far, current antipsychotic treatments do not benefit up to 30% patients with SCZ, because the underlying pathophysiology of SCZ is still unclear [3]. Accumulating studies from animal models, biochemical pathways related to neurodevelopment and functional brain imaging are intended to identify biomarkers to predict individual treatment responses to antipsychotics [4], but reliable biomarkers have not been found that are readily useful in clinical practice [2].

Studies have shown that the underlying pathophysiology of SCZ may be the result of an imbalance in homeostatic signaling in the redox regulatory system during critical periods of early brain development [5, 6]. Basic studies in the animal models of SCZ have found that reactive oxygen species may be produced by enhanced metabolism of neurotransmitters such as dopamine and glutamate. For example, excess dopamine metabolism leads to a production of hydrogen peroxide (H2O2) by auto-oxidation or monoamine oxidase (MAO) [5]. Hydrogen peroxide produces oxidative stress (OxS) on cells, which can be neurotoxic and impair neurotransmission [6]. To prevent this damage, several enzymatic antioxidants attempt to nullify free radical species [7]. Examples of these enzyme antioxidants include superoxide dismutase (SOD), catalase (CAT), and glutathione peroxidase (GPx). The three enzymatic antioxidants, SOD, CAT, and GPx all convert reactive oxygen species and free radicals into water and/or oxygen. Mechanistically, however, these enzymes differ in the way they produce these byproducts and the stages in which they react with the free radical pathways [8]. SOD is one of the components of the antioxidant defense response system against OxS and it is a key enzyme in superoxide radical detoxification. GPx is responsible for the further conversion of hydrogen peroxide to oxygen and water, which is the product of the enzymatic detoxification of superoxide radicals by SOD [8]. Altogether, all the aforementioned antioxidants work synergistically at different stages of the free radical pathway. Although we can obtain effective information by measuring the activity of each of these antioxidant enzymes, the measurement of total antioxidant status (TAS) can also be used as indication of the sum of all antioxidant activities [9]. Also, there is a direct marker of OxS, malondialdehyde (MDA), which is the end product of lipid peroxidation [10].

A large number of clinical studies have shown that the increased OxS markers are closely related to the pathophysiology of antipsychotic-naïve first episode (ANFE) patients and chronic patients with SCZ [11,12,13,14,15,16,17,18]. The main findings of these studies were that the activity of the primary antioxidant defense system is decreased and the pro-oxidative status is increased in ANFE patients [19]. Of note, a few studies even found that abnormal OxS markers were associated with clinical symptoms of ANFE patients. For example, abnormal OxS markers were reported to be correlated with negative symptoms [20], positive symptoms [21], and cognitive dysfunction in ANFE patients [18, 22]. Interestingly, several recent clinical trials found that antioxidants as an add-on treatment can significantly alter the OxS marker levels and the severity of symptoms in SCZ. For example, omega-3 polyunsaturated fatty acid, certain vitamins and minerals, and N-acetylcysteine have all shown to be able to reduce the severity of psychotic symptoms [23,24,25]. Taken together, these studies collectively support the role of OxS in the pathophysiology of this disorder, suggesting that OxS markers may serve as future biomarkers of SCZ.

Clinical and animal studies have shown that typical antipsychotics may increase free radical production by blocking dopamine receptors, increasing dopamine metabolism, and affecting dopamine circulation [26, 27]. However, it remains inconsistent whether risperidone can increase or decrease OxS in patients [28,29,30]. Our previous study showed that risperidone treatment significantly changed the activities of SOD in patients with chronic SCZ [26, 31]. However, other studies reported that risperidone treatment had no effect on OxS markers and did not change the activities of antioxidant enzymes [32, 33]. Based on these previous studies on OxS markers, we proposed that the changes in OxS markers may be associated with the clinical response to antipsychotic treatment in ANFE patients. Therefore, the purposes of this study were to explore (1) whether there were simultaneous changes in levels of MDA and TAS, or activities of SOD, CAT, and GPx in ANFE patients, relative to controls; (2) whether risperidone treatment for 12 weeks impacted OxS markers; and (3) whether there was a relationship between risperidone response and changes in OxS markers during treatment, or between risperidone response and baseline OxS markers.

Materials and Methods

Subjects

A total of 354 SCZ patients were recruited from three hospitals in China. Experienced psychiatrists made the diagnosis of SCZ at baseline and re-confirmed it after 3–6 months of follow-up according to the DSM-IV criteria. Patients should meet the following inclusion criteria: a SCID diagnosis; informed consent; age 16–45 years; Han nationality; course of disease less than 5 years; no previous treatment with psychotropic medicines or cumulative use of antipsychotic drugs less than 14 days; no major medical comorbidities; without abuse or substance dependence except tobacco. The patients were screened by 4 psychiatrists, including a SCID interview and physical examination.

One hundred and fifty-two unrelated controls were recruited at the same period as the patients. The control group was screened by SCID to exclude any person with current or past psychiatric disorders. We also excluded subjects who were treated with psychotropic drugs (e.g., mood stabilizers, anti-anxiety drugs, antidepressants, or antipsychotics). Patients and controls were excluded if they had physical diseases (i.e., cancer, diabetes, and hypertension) or took over-the-counter antioxidants.

Our study was carried out in accordance with the Declaration of Helsinki and was approved by the Institutional Review Board of Beijing Huilongguan Hospital. All participants provided written informed consent prior to enrollment.

Assessments of Psychotic Symptoms and Clinical Outcomes

ANFE patients received a flexible dose of risperidone monotherapy (4–6 mg/day) for 12 weeks after a stabilization phase of 1 week since admission. Antidepressants and mood stabilizer were not permitted. In this 12-week observable study, all patients were hospitalized and nurses monitored risperidone compliance. The safety and efficacy of treatments were assessed every 2 weeks through clinical interviews. Six psychiatrists who participated in the training course on the use of PANSS assessed patients using the scale [34]. After training, repeated assessment showed that the inter-observer correlation coefficient of the PANSS total score was maintained at > 0.8. At baseline and at follow-up, the psychiatric symptoms of the patients were evaluated by PANSS.

If the patient’s PANSS total score showed a 30% or more reduction from baseline at week 12, we defined him or her as a responder. A non-responder was defined as a patient whose PANSS total score decreased by less than 30%. The percentage decrease of PANSS total score was calculated as d = (PANSSbaseline-PANSSfollow up)/(PANSSbaseline-30). The 30 in the denominator corresponds to the “nonsymptomatic” score of the PANSS total score. According to previous research, the 30% cut-off value was used as the response criterion in this study [35, 36].

Determination of OxS Markers in the Patients and Controls

Plasma OxS markers, such as MDA, and antioxidant enzymes (SOD, CAT, and GPx) were detected by spectrophotometer as described in a previous literature [22]. TAS levels were measured by a commercially available kit as in our previous study [37]. In brief, antioxidants were evaluated as reductants from Fe3+ to Fe2+, which were chelated by TPTZ to form a Fe2+-TPTZ complex absorbed at 593 nm and recorded using a Multiskan microplate reader (FlowLabs, McLean, VA, USA). Activity was expressed in units per milliliter plasma (U/ml), and MDA levels were expressed in nmol/ml.

Statistical Analysis

To compare demographic and clinical variables between groups, we used X2 test and analysis of variance (ANOVA) for categorical and continuous variables, respectively. Multiple analysis of covariance (MANCOVA) was performed to examine whether there were differences in OxS markers between controls and patients at baseline. The covariates in the MANCOVA analysis include sex, smoking status, age, and body mass index (BMI). Then, in patients, we carried out a Pearson correlation analysis to find the relationship between OxS markers and disease severity at baseline, controlling for sex, smoking status, BMI, duration of illness, and onset age.

Last observation carried forward (LOCF) analysis was carried out for patients who dropped out after the second month. We further explored whether risperidone treatment for 12 weeks altered OxS markers and whether there was a significant difference in OxS markers between responders and non-responders. We hypothesized that after 12 weeks of treatment, responders may have lower levels or activities of OxS markers than non-responders. To test the hypothesis, a 2 × 2 (time point by responder) repeated measure MANOVA analysis (RM MANOVA) was performed with OxS markers as the outcome measures. The responder × time interaction received more attention, as this may detect any potential differences in the patterns of OxS marker changes between non-responders and responders. After a statistically significant omnibus multivariate test of significance, individual univariate analyses of OxS markers were performed. After controlling for sex, BMI, illness duration, and onset age, logistic regression analysis was used to detect whether the baseline OxS markers or changes in each OxS marker could predict the response to risperidone. In the logistic regression model, the dependent variable was the response to risperidone. Pearson correlation analysis was performed between the reduction of OxS markers and the improvement of symptoms in responders and non-responders. In addition, multiple linear regression analysis was also carried out to clarify the independent factors that were related to the reduction of symptoms. ANCOVA analysis was carried out to compare the baseline OxS markers and reduction in OxS markers during treatment, including onset age, smoking status, sex, BMI, and illness duration as covariate variables. Bonferroni corrections were used for multiple tests.

Results

Demographic Data and Abnormal OxS Markers in Patients and Controls at Baseline

Table 1 shows demographic and clinical data for ANFE patients and healthy controls. The BMI of the patients was significantly lower than that of the controls (21.6 ± 3.6 vs 23.9 ± 4.9, p < 0.01).

Table 1 Demographic characteristics and clinical data in antipsychotics-naive first episode (ANFE) patients with schizophrenia and healthy controls
Full size table

Relative to controls, ANFE patients showed lower MDA level and GPx activity, but higher activities of SOD, CAT, and TAS (all p < 0.05 after Bonferroni corrections, effect size from 0.21 to 0.98) (Table 2).

Table 2 Levels and activities of oxidative stress markers in ANFE patients with SCZ and healthy controls
Full size table

Follow-Up Comparisons of OxS Markers Before and After Treatment

After 12 weeks of treatment with risperidone, the PANSS total score and 3 subscores were significantly decreased than those of the baseline (all p < 0.01). Risperidone monotherapy had no effect on CAT and SOD enzyme activities in patients after 12 weeks of treatment (all p > 0.05). However, the patients showed a significant decrease in GPx activity (t = 3.9, p < 0.001), MDA level (t = 2.4, p = 0.019), and an increase in TAS level (t = 6.1, p < 0.001) after treatment (Fig. 1). After Bonferroni corrections, the differences in GPx and TAS activities remained significant (both p < 0.05).

Fig. 1
figure 1

Levels of malondialdehyde (MDA) and total antioxidant status (TAS), activities of glutathione peroxidase (GPx) enzyme, superoxide dismutase (SOD), and catalase (CAT) were altered after risperidone monotherapy for 12 weeks in both responders and non-responders.

Full size image

Associations Between OxS Markers and Response to Risperidone

In the initial study of 354 patients, 293 patients agreed to receive risperidone treatment for 12 weeks, and 211 patients completed the study. For various reasons, 82 patients dropped out of this study, including withdrawing their consent, changing the treatment plan, discharging from the hospital without following the doctor’s advice, and unknown reasons. After 12-week risperidone treatment, we found 168 responders and 50 non-responders with OxS markers. No serious adverse effects were noticed during the 12-week treatment.

To test whether risperidone treatments altered OxS markers and whether there were differences between responders and non-responders, a RM MANOVA was performed with OxS markers as the outcome indicators. The results revealed that there was a significant time effect (Wilks’ lambda F = 3.4, p = 0.007), but no responder × time effect (Wilks’ lambda F = 1.1, p = 0.37) or responder effect (Wilks’ lambda F = 1.5, p = 0.18). Univariate analyses showed that there was a significant interaction effect of responder × time for GPx (F = 4.9, p = 0.029, Table 3 and Fig. 1) and a significant main effect of time for GPx (F = 18.5, p < 0.001), but there was no main effect of responder (p > 0.05). At the end of treatment, there were marginally significant differences in GPx activity (44.0 ± 11.8 vs 50.0 ± 14.7; F = 3.6, p = 0.05) as well as in the reduction of GPx from baseline to post-treatment (12.4 ± 16.1 vs 4.3 ± 12.4; F = 3.7, p = 0.05) between non-responders and responders. Furthermore, we found that there were significant main effects of time (F = 32.7, p < 0.001) and responder (F = 5.6, p = 0.02) on TAS, but there was no interactive effect of responder × time on TAS. There was no significant main effect of responder and time, as well as the interaction of responder × time on MDA, CAT, and SOD (all p > 0.05). Since responders and non-responders had different sex and education, the models were adjusted for these confounding variables.

Table 3 Oxidative stress markers in responders and non-responders
Full size table

Further logistic regression analysis was performed to investigate the association between OxS markers at baseline, changes in OxS markers after treatment, and the response to risperidone. The results showed that the TAS level at baseline was a predictor of response to risperidone treatment (Wald χ2 = 5.6, p = 0.018). Multiple regression analysis confirmed the above result (β = − 0.24, t = − 2.27, p < 0.05), with PANSS total reduction from baseline to post-treatment as an independent factor and baseline OxS markers, sex, education as dependent factors.

Discussion

To our knowledge, this is the first study to examine how risperidone treatment changed several types of OxS markers and how these changes in OxS markers were correlated with risperidone response in ANFE patients. This study found that the SOD and CAT activities and TAS levels in patients were higher than those in healthy controls. The study also showed that ANFE patients had lower MDA levels and GPx activity than healthy controls. Twelve weeks of risperidone treatment increased TAS levels but decreased GPx activity. Also, the GPx activity of non-responders decreased more than that of the responders, suggesting that antipsychotics treatment can partially regulate the abnormalities of the redox system. Although the hypothesis that SCZ patient exhibited a wide range of abnormal OxS markers has been verified so far [19], the novelty of this study stems from the simultaneous measurement of 3 types of antioxidant enzymes, total antioxidants levels, and lipid peroxidation end products in ANFE patients at their first episode.

The findings of these widespread changes in antioxidant enzymes in ANFE patients suggest that the remodeling of the redox regulation system occurs in the early stage of the disease. We cannot provide a reasonable explanation for the increase of SOD and CAT enzyme activities, the increase of TAS levels, as well as the decrease of MDA levels and GPx in these patients. Antioxidant enzyme activities may be overstressed to protect against oxidative damage, which is a compensatory mechanism, suggesting that antioxidant system may be activated in the early stage of this disease. The current study was consistent with our previous studies and other studies conducted in ANFE patients, which showed higher activities of antioxidant enzymes in peripheral blood [38,39,40]. However, our results contradict some recent studies in ANFE patients [41, 42]. In particular, some studies have shown no difference between ANFE patients and controls [43,44,45]. Some factors, such as age, sex, genetic variables, smoking, obesity, dietary habits, activation of endocrine stress axis, interethnic differences in gene polymorphisms related to redox system regulation, and the measurement methods (levels or activities) of OxS indicators used in the assays, may contribute to the discrepancy [10, 46,47,48]. In particular, different stages of disease progression, disease duration, and differences in sample sources (serum, red blood cells, or plasma) may have an important impact on the discrepancies between these results [10, 48, 49]. Therefore, the different results between different studies may be due to the various methods of sampling and measuring OxS markers. Despite the discrepancies between the various studies, our findings revealed that in the early stages of the disease, there is a severe dysregulation of redox system and subsequent OxS persistence.

Our study further showed that risperidone significantly increased TAS levels after 12 weeks of treatment, suggesting that the elaborately regulated oxidant detoxification system may be overstressed during risperidone treatment. At present, the mechanism of risperidone’s elevated TAS levels at first episode and during treatment is unclear. Since TAS levels are the sum of all antioxidants that defend against pro-oxidants and free radicals, we speculate that the increase in TAS levels in patients may be a response to increased free radicals and pro-oxidants as a compensatory mechanism [5, 50]. Therefore, elevated TAS levels at baseline and at follow-up in patients indicate that OxS may occur in the initial stages of disease and during the progression of the disease. Clinical studies and animal studies have found that antipsychotic drugs enhanced the production of free radicals through drug metabolism and increased catecholamine turnover. These events lead to an increase in antioxidant enzymes [50, 51]. Although we expected that patients who responded to risperidone treatment may have reduced TAS levels, in our longitudinal study, the results showed that there was no significant difference in TAS levels before and after treatment between responders and non-responders. Interestingly, our logistic regression analysis found that baseline TAS levels were an effective predictor of the response to risperidone over 12 weeks of treatment, showing that poor responders had higher baseline TAS levels. However, schizophrenia is a brain disorder. At present, it is hard to connect peripheral antioxidant measures to neural effects of antipsychotics. Therefore, the changes in TAS levels from baseline to follow-up cannot fully reflect the changes in the brains of SCZ patients. Consistent with our findings, the results of the central antioxidant studies also showed abnormal oxidative stress markers in brain regions, such as the medial prefrontal cortex, anterior cingulate cortex, and striatum [52, 53]. However, whether peripheral antioxidant levels may represent similar levels in the brain and how peripheral antioxidant measures are related to neural effects of antipsychotics deserves further research.

Our third finding was that GPx activity was reduced at the first episode and further decreased after 12 weeks of treatment, consistent with a previous study [54]. Both our study and previous studies have shown that peripheral GPx activity in ANFE patients was lower than that in controls [42, 55,56,57]. Other studies using postmortem samples of ANFE patients also showed a decrease in GPx activity in the caudate nucleus and in the prefrontal cortex compared with healthy controls [58], in particular a recent review in GPx [11]. The effect of treatment on GPx activity has also been seen in several other studies. For example, a study reported that risperidone decreased GPx activity in PC12 cell line [59]. Moreover, two previous studies reported that GPx activity was lower in long-term hospitalized chronic patients with SCZ treated with risperidone [31, 60]. However, Stojkovic et al. found that risperidone monotherapy reversed the decrease of GPx activity in thalamus, nucleus caudate, and hippocampus of perinatal phencyclidine-induced animal models of SCZ [61]. The results were contrary to ours, and they found that risperidone increased GPx activity to control levels in the hippocampus [61]. A few of studies even showed no changes in GPx after long-term treatment with risperidone [62, 63]. We speculate that the difference in the results between the studies is due to the different participants in these studies. In our current study, the subjects were first episode patients with SCZ, but the participants in other studies were either chronic patients or animals. We have no reasonable explanation for the increase in activity of almost all antioxidant enzymes, but the decrease in GPx activity. This finding may be a compensatory response to the increase of several antioxidant enzymes in patients; however, this is only our speculation, and further research is necessary to reveal the underlying mechanisms.

Our results showed that after 12 weeks of treatment, responders had higher GPx activity than non-responders and the decrease in GPx activity in responders was less than that in non-responders, indicating that GPx activity may be a prognostic feature for risperidone monotherapy. At present, we do not know why risperidone can block the decrease of GPx activity in responders, while GPx activity in patients who did not respond to risperidone continued to decrease during the progression of the disease. However, it is well known that GPx is a critical enzyme in the GSH system, which can counteract the destructive effects on brain tissue by excessive glutamate and oxidative metabolism [52]. Also, the decrease in GPx activity is closely related to the severity of the disease [42], which may be an important part of the pathogenesis of SCZ. Previous studies have shown that some of the components of the GSH system have emerged as promising therapeutic targets for SCZ patients. For example, using a longitudinal design and ultrahigh field 7 T magnetic resonance spectroscopy (MRS) protocol, Dempster et al. reported that higher GSH was associated with shorter response time of early antipsychotics in ANFE patients, while higher glutamate was associated with more severe dysfunction [64]. Limongi et al. found that higher levels of GSH can reverse the downstream pathophysiological effects of hyperglutamatergic state using ultra-high field (7 T) MRS and resting state functional magnetic resonance imaging (fMRI) [65]. Further, the studies in rat brain showed that chronic risperidone treatment can restore the region-specific changes in the activities of GPx enzymes involved in glutathione synthesis and metabolism [61]. Overall, our study showed that GPx activity was closely related to the risperidone response in ANFE patients, and the interventions that increase brain GPx activity may improve the early intervention results of SCZ.

Several limitations should be noted. First, it is uncertain whether peripheral levels and activities of OxS markers are correlated to their respective CNS counterparts. Second, the current clinical study cannot eliminate the influences of different doses of risperidone, although there was no difference in the type and dose of risperidone between responders and non-responders. Third, in this study, only the activities of several antioxidant enzymes and TAS levels were measured, and we did not take into account the other markers of the antioxidant defense system and OxS. Considering the complexity of antioxidant protection system with antioxidant enzymes and non-enzyme antioxidant molecules, this study provides limited insights into free radical-mediated central nervous system dysfunction. Fourth, the effect of disease (SCZ) on outcomes cannot be fully evaluated using the current design. To examine this, a comparison of ANFE patients with and without risperidone and controls with and without risperidone would need to be made in the future study. Fifth, in this study, the BMI of the control subjects recruited in this study was higher than that of ANFE patients. However, most of previous studies found no significant difference in BMI between ANFE patients and controls, and only a few studies reported that ANFE patients had lower BMI [66]. In order to control the influence of BMI on oxidative stress markers, BMI was added as a covariate to all analyses.

In conclusion, the blood MDA, TAS levels and SOD, GPx, and CAT activities were significantly altered in ANFE patients with SCZ, suggesting that abnormalities of OxS markers may have a critical role in the pathophysiology of this disease. Risperidone treatment for 12 weeks did not completely normalize the redox system in SCZ patients. Compared with non-responders, responders showed less reduction in GPx activity. Patients who responded to risperidone treatment displayed more “normalized” GPx activity at baseline, with greater improvement in symptoms. These results suggest that changes in GPx activity may be useful for predicting the antipsychotic response of ANFE patients.