Lowered antioxidant defenses and increased oxidative toxicity are hallmarks of deficit schizophrenia: neurocognitive and symptom correlates (1-3)

Background: There is now evidence that schizophrenia and deficit schizophrenia are neuroimmune conditions and that oxidative stress toxicity (OSTOX) may play a pathophysiological role. Aims of the study: To compare OSTOX biomarkers and antioxidant (ANTIOX) defenses in deficit versus non-deficit schizophrenia. Methods: We examined lipid hydroperoxides (LOOH), malondialdehyde (MDA), advanced oxidation protein products (AOPP), sulfhydryl (-SH) groups, paraoxonase 1 (PON1) activity and PON1 Q192R genotypes, total radical-trapping antioxidant parameter (TRAP) as well as immune biomarkers in patients with deficit (n=40) and non-deficit (n=40) schizophrenia and healthy controls (n=40). Results: Deficit schizophrenia is characterized by significantly increased levels of AOPP and lowered -SH, and PON1 activity, while no changes in the OSTOX/ANTIOX biomarkers were found in non-deficit schizophrenia. An increased OSTOX/ANTIOX ratio was significantly associated with deficit versus non-deficit schizophrenia (Odds ratio=3.15, p<0.001). Partial least squares analysis showed that 47.6% of the variance in a latent vector extracted from psychosis, excitation, hostility, mannerism, negative symptoms, psychomotor retardation, formal thought disorders, and neurocognitive test scores was explained by LOOH+AOPP, PON1 genotype + activity, CCL11, tumor necrosis factor (TNF)-α, IgA responses to neurotoxic tryptophan catabolites (TRYCATs), whereas -SH groups and IgM responses to MDA showed indirect effects mediated by OSTOX and neuro-immune biomarkers. Discussion: Our findings indicate that with increasing overall severity of schizophrenia, neuroimmune and neuro-oxidative (especially protein oxidation indicating chlorinative stress) toxicities become more prominent and together with lowered antioxidant defenses and impairments in innate Preprints (www.preprints.org) | NOT PEER-REVIEWED | Posted: 9 May 2020 doi:10.20944/preprints202005.0145.v1


Introduction
The first comprehensive neuro-immune hypothesis of schizophrenia was published in 1995 [1] as the macrophage-T lymphocyte theory suggesting that activated M1 macrophage and T helper (Th)-1 immunocytes are activated through multiple hits (e.g. maternal infections or later hits in adulthood) leading to neurodevelopmental disorders and (neuro)inflammatory responses with activation of nitro-oxidative stress and the tryptophan catabolite (TRYCAT) pathways ultimately contributing to the onset of schizophrenia [1]. The first publication that schizophrenia is accompanied by a peripheral inflammatory process followed in 1997 reporting increased levels of acute phase proteins (APPS), including haptoglobin (Hp), and complement factors, which coupled with increased pro-inflammatory cytokines indicate inflammation [2].
While different schizophrenia subtypes (including first-episode psychosis, acute schizophrenic episodes, and chronic schizophrenia) show activated IRS and CIRS pathways, one of the subtypes, namely deficit schizophrenia, is characterized by an overwhelmingly activated IRS coupled with severe deficits in CIRS functions [7][8][9][10][11]. The latter phenotype is characterized (as compared with non-deficit schizophrenia and healthy controls) by increased neurotoxicity through elevated IL-1β, TNF-α and CCL11 levels, and signs of breakdown of gut-paracellular and vascular pathways as well as the blood-brain-barrier, increased bacterial translocation with increased IgA/IgM levels to LPS of Gram-negative bacteria, and increased IgA responses to neurotoxic TRYCATs including picolinic and xanthurenic acid [4,6,9,[11][12][13]. Moreover, deficit schizophrenia is also characterized by severe deficits in the CIRS including a) lowered natural IgM-mediated responses to multiple oxidative specific epitopes (OSEs) especially malondialdehyde and azelaic acid [10], which are produced by innate-like B1 and marginal zone B cells and, as a component of the innate immune system, have anti-inflammatory, housekeeping anti-bacterial functions [10,14]; b) lowered IgM responses to TRYCATs [15], and c) lowered activity of paraoxonase 1 (PON1) activity, a strong antioxidant enzyme that is part of innate immunity and has anti-inflammatory and anti-bacterial activities [16]. As such deficit schizophrenia it to a large extent mediated by IRS-mediated neurotoxicity, which if fueled by an impaired resilience of the innate immune system.
Immune activation is frequently accompanied by increased oxidative stress (OS) and lowered levels of antioxidants [17]. For example, schizophrenia is accompanied by increased biomarkers of lipid peroxidation including MDA, and LOOH (lipid hydroperoxides), increased nitrite levels, and decreased levels of antioxidants like glutathione peroxidase and PON1 activity [18][19][20]. Early and later stages of schizophrenia are accompanied by signs of lipid peroxidation and protein oxidation as measured with protein carbonyls [21]. Nevertheless, there are also negative studies. For example, in patients with first-episode psychosis no significant changes were found in serum peroxides, total antioxidant capacity (TAC) and oxidative stress index [22]. In chronic schizophrenia, no significant changes could be observed in LOOH, PON1 activity, nitric oxide metabolites (NOx), total radical-trapping antioxidant parameter (TRAP) and advanced oxidation protein products (AOPP) [23]. Recent meta-analyses showed increased signs of oxidative stress, including increased MDA and NO, and a lowered total antioxidant status including lowered levels of specific antioxidants including the glutathione in plasma and brain although not all studies could detect such differences [24][25][26][27][28][29]. Such differences may be explained by sample characteristics such as clinical features and staging of illness. In this respect we already published that two strong antioxidant systems, namely IgM to MDA and PON1 enzyme activity, are significantly lowered in deficit versus non-deficit schizophrenia suggesting that lowered antioxidant defenses and, consequently, increased oxidative stress may be a hallmark of deficit rather than of non-deficit schizophrenia. However, no previous research effort has examined a comprehensive set of O&NS biomarkers in deficit versus non-deficit schizophrenia.
Thus, the aim of the current study was to delineate a more comprehensive set of O&NS biomarkers in deficit schizophrenia as compared with non-deficit schizophrenia and healthy controls, including TRAP, PON1 activity, LOOH, MDA, AOPP and -SH groups (thiols or sulfhydryl groups). Moreover, we also examined whether these O&NS biomarkers together with IgM response to MDA are associated with the phenome of schizophrenia (neurocognitive test results and symptom domains) and whether O&NS have an effect on the phenome of schizophrenia above and beyond the effects of the established biomarkers of deficit schizophrenia including IL-6, IL-4, TNF-α, CCL11, IgA directed to neurotoxic TRYCATs and IgA directed to LPS of Gramnegative bacteria).

Participants
This study we enrolled 120 participants, namely 80 patients with schizophrenia and 40 healthy controls. All participants were Thai nationals, aged 18-65 years old and of both sexes. All patients were outpatients admitted to the Department of Psychiatry, Faculty of Medicine, Thailand. Patients were excluded for a current or lifetime diagnosis of axis I disorders other than schizophrenia including major depressive episode, generalized anxiety disorder, bipolar disorder, autism spectrum disorders, schizoaffective disorder, and substance use disorders (except tobacco use disorder). We omitted healthy controls when they showed a lifetime or current diagnosis of axis I DSM-IV-TR disorders or a positive family history of psychosis. Patients and controls were excluded when they showed a) medical diseases including chronic obstructive pulmonary disease, diabetes type 1, inflammatory bowel disease, psoriasis, and rheumatoid arthritis; b) neuroinflammatory and neurodegenerative disease including multiple sclerosis, stroke, and Parkinson's disease; c) any use (lifetime) of immunomodulatory drugs including immuosuppressiva and glucocorticoids; and d) use of therapeutic doses of ω3-polyunsaturated fatty acids and antioxidants six months prior to the study.
All participants, as well as the guardians of patients (parents or other close family members) provided written informed consent to take part in the study. Approval for the study was obtained from the Institutional Review Board of the Faculty of Medicine, Chulalongkorn Neuropsychiatric Interview (M.I.N.I.) in a validated Thai translation [31]. Negative symptoms were assessed using the negative subscale of the Positive and Negative Syndrome Scale (PANSS) [32], the Scale for the Assessment of Negative Symptoms (SANS) [33], and the SDS scale [30].
We also assessed the Brief Psychiatric Rating Scale [ memory-recall, and c) Verbal Fluency Test (VFT) to assess fluency, language, cognitive flexibility, and semantic memory. The Mini-Mental State Examination (MMSE) was used to assess overall neuropsychological functioning by testing naming, orientation, concentration, memory, and constructional praxis. We used three CANTAB probes to compute an index of executive functions, namely spatial working memory (SWM) between errors and SWM strategy, which probe task strategy used by the central executive, executive working memory ability, selfmonitoring ability, and maintenance of data in the visuospatial sketchpad; and the one touch stockings of Cambridge probability solved on first choice, which probes spatial planning.
Consequently, we extracted the first principal component of these three tests scores, which explained 81.7% of the variance, and is therefore an indicant of executive functions [9]. DSM-IV-TR criteria were employed to diagnose tobacco use disorder (TUD). Body mass index (BMI) was computed as body weight (kg) / length (m 2 ).

Assays
We sampled blood at 8:00 a.m. after an overnight fast and serum was frozen at −80 °C until thawed for assay of OS biomarkers including -SH groups, TRAP, PON1 CMPAase activity and PON1 Q192R genotypes, LOOH, MDA, and AOPP. The methods for these assays were described Q192R polymorphism) as well as by the rate hydrolysis of phenyl acetate under low salt condition (AREase, which is less influenced by the PON1 Q192R polymorphism). Analysis were conducted in a microplate reader (EnSpire, Perkin Elmer, USA) [46]. Although the PON1 Q192R genotypes were assayed, those data yielded non-significant results and as such the data are not presented here.

Statistical analysis
Analysis of contingency tables (χ 2 -tests) was employed to check associations between nominal variables and analysis of variance (ANOVA) to check differences in continuous variables between categories. We also show Boxplots, which indicate minimum, Q1, median, Q3 and maximum values, out-values (shown as circles) and far-out or extreme values (shown as stars).
Univariate and multivariate GLM analysis was employed to delineate the associations between OS biomarkers and diagnosis while adjusting for background variables (e.g. age, sex, BMI, drug status, TUD). Protected pair-wise comparisons among treatment means were used to check variables between patients with and without deficit schizophrenia and healthy controls. When performing multiple comparisons, we used the false-discovery rate (FDR) procedure to control type I errors [48]. We used automatic stepwise binary logistic regression analysis to assess the biomarkers that significantly predict deficit schizophrenia while allowing for the possible effects of background variables. Automatic stepwise multiple regression analysis was used to delineate the biomarker set (entered as explanatory variables) that best predicted cognitive test scores and Consequently, complete PLS path modelling is performed to compute path coefficients with exact p-values as well as direct and indirect effects. Confirmatory Tetrad analysis (CTA) is performed to evaluate whether the reflective model of the OSOS LV is not mis-specified [49]. Table 1 shows the socio-demographic, clinical and biomarker data in healthy controls and schizophrenia patients with and without increased oxidative stress toxicity (OSTOX), dichotomized using the median-split method. There were no significant differences in age, marital status, BMI, education, and TUD between the three study groups. There were somewhat more males in schizophrenia than in controls. There were no significant differences in illness duration and age at onset between the two schizophrenia subgroups. Table 1 shows also the symptom domains in the three study groups. The SANS, PANSS negative, psychosis, excitement, mannerism, FTD, and PMR scores were significantly different between the three subgroups and increased from controls → schizophrenia with lower OS → schizophrenia with higher OS.

Socio-demographic data
Hostility was significantly increased in schizophrenia relative to controls. These differences remained significant after FDR p-correction.
The same table also shows the neurocognitive test results. MMSE, VFT, WLM and True recall scores were significantly different between the three study groups and decreased from controls → schizophrenia with lower OS → schizophrenia with higher OS. Executive functions were significantly lower in schizophrenia than in controls. These differences remained significant after FDR p-correction. Table 1 shows also the outcome of GLM analyses that examined the association between biomarkers and the three study groups. TNF-α was significantly higher in schizophrenia with OS as compared with the two other groups. IL-6 was significantly higher in schizophrenia with OS relative to controls. IgM directed to MDA was significantly lower in schizophrenia with OS than in the two other study groups. CCL11 and IgA to the NOX/PRO ratio were significantly higher in schizophrenia than in controls. IgA to LPS of Gram-negative bacteria was significantly higher in schizophrenia patients with OS than in those without OS. Table 2 shows the outcome of a multivariate GLM analysis which examines the association between diagnosis (deficit and non-deficit schizophrenia and controls) with TRAP, -SH groups, CMPAse, LOOH, MDA and AOPP. There was a significant association between the biomarkers and diagnosis with an effect size of 0.186. Univariate GLM analysis, table 3 (model-generated estimated marginal means ±SE obtained by the multivariate GLM analysis) and protected LSD tests showed that -SH groups and PON1 CMPAase activity were significantly lower in deficit schizophrenia than in controls and non-deficit schizophrenia. AOPP was significantly higher in deficit schizophrenia than in the two other study groups. Table 2 shows also the results of univariate GLM analysis which examine the associations between diagnosis and the z unitweighted composite scores. All composites (except 8MITOTOX) were significantly higher (3OSTOX, 3OSTOX/3ANTIOX, 8MITOTOX/4PRORESIL) or lower (3ANTIOX, 4PRORESIL) in deficit schizophrenia as compared with controls and patients with non-deficit schizophrenia.

3OSTOX and 3ANTIOX in deficit and non-deficit schizophrenia
The 8MITOTOX composite score was significantly different between the 3 study groups and increased from controls → non-deficit schizophrenia → deficit schizophrenia. Figure 1 and Figure 2 show the boxplots of the 3OSTOX/3ANTIOX ratio and 8MITOTOX/4PRORESIL ratios in those three study groups.  Mannerism (regression #3) was best explained (21.7% of the variance) by TRAP and education (both negative) and male sex. Regression #4 shows that 22.9% of the variance in FTD was explained by CMPAase, education (negative) and LOOH (positive) and male sex. PMR was best explained (28.8% of the variance) by TRAP, CMPAase, education (inversely), and AOPP and LOOH (both positively, see regression #5). Both negative subdomain rating scores were significantly associated with the 3OSTOX/3ANTIOX ratio (positive), education (negative) and male sex. WLM and MMSE (regressions # 8 and 11) were significantly predicted by the 3OSTOX/3ANTIOX ratio (negative), education (positive) and female sex. We found that 24.9% of the variance in VFT (regression #9) was explained by the 3OSTOX/3ANTIOX ratio (negative) and education (positive), while 33.4% in true recall (regression # 10) was explained by AOPP (inverse), TRAP and education (both positive) and female sex. Up to 45.3% of the variance in executive functions (regression #8) was explained by the 3OSTOX/3ANTIOX ratio and age (both negative) and education (positively). Table 6 shows the correlation matrix between 8MITOTOX, 4PRORESIL and 8MITOTOX/4PRORESIL ratio and the symptom subdomains and cognitive tests as well.

Correlations symptoms, cognitive tests and the more comprehensive composite scores
8MITOTOX was correlated (partial correlations after adjusting for age, sex and education and after p-correction for FDR) with all symptom subdomains and all neurocognitive tests. The 4PRORESIL composite score was significantly associated with all symptom domains and cognitive test results except hostility and VFT. The 8MITOTOX/4PRORESIL ratio was significantly associated with all symptoms and cognitive tests scores. Figure 3 shows the partial regression of the total SANS score on the 8MITOTOX/4PRORESIL ratio. and IgM to MDA. In addition, PON1 genotype was a significant predictor of -SH groups and the latter was significantly associated with zLOOH+zAOPP. Sex was also associated with TRAP and zLOOH+zAOPP and age with TRAP. CMPAase and MDA were not significant in this PLS analysis. There were significant specific indirect effects of -SH groups (t=-2.59, p=0.010) and sex (t=2.47, p=0.014) on OSOS both mediated by zLOOH+zAOPP. There were significant total effects of PON1 genotypes, -SH groups, TRAP, IgM to MDA, and zLOOH+zAOPP on OSOS. and the genotype-CMPAase supervariable, while the latter was also a predictor of -SH groups.

Results of smart-PLS analysis
Moreover, IgM directed to MDA was associated with CCL11, zLOOH+AOPP, TNF-α and IgA to TRYCATs. There were specific indirect effects of IgM to MDA on OSOS mediated by TNF-α (t=-2.33, p=0.020) and of -SH groups on OSOS mediated by IgA to TRYCATs (t=-2.28, p=0.02).

Discussion
The first major finding of this study is that deficit schizophrenia is accompanied by highly significant disorders in oxidative stress toxicity including significantly increased AOPP levels and lowered antioxidant defenses as indicated by lowered -SH groups and PON1 activity. Moreover, multivariable analysis showed that increased MDA and lowered TRAP were associated with deficit schizophrenia. In the Introduction, we reviewed some original papers and meta-analyses reporting contradictory results on oxidative stress toxicity and lowered antioxidant defenses in schizophrenia. However, it is difficult to discuss our results with respect to these and other studies as the current study found that changes in OS biomarkers are confined to deficit schizophrenia. Therefore, our results show that the selection of specific phenotypes may determine the results on oxidative stress in schizophrenia.
To the best of our knowledge there is only one preliminary study which reported increased levels of one type of reactive oxygen species (ROS), namely peroxide levels, and total antioxidant potential in deficit schizophrenia as compared with controls [50]. In addition, those authors reported that the ratio of total peroxides versus antioxidants potential was significantly higher in patients with deficit schizophrenia. Nevertheless, our results show that deficit schizophrenia is accompanied by increased oxidative stress toxicity, which attributable to increased AOPP levels (an index of protein oxidation) rather than to increased levels of lipid hydroperoxides (indicating lipid peroxidation following increased ROS) and MDA (indicating increased aldehyde formation following increased lipid peroxidation). Increased AOPP production is the consequence of the formation of dityrosine residues in proteins as a consequence of increased ROS attacks (including by hydrogen peroxides) and neutrophil-associated chlorinated oxidants including chloramines or hypochlorous acid (hypochlorous or chlorinative stress), which, in turn, is the consequence of One of the trigger factors of hypochlorous stress is LPS-stimulation of neutrophil phagosome Toll-Like Receptors [66,67]. This is important, as deficit schizophrenia, but not nondeficit schizophrenia, is accompanied by signs of breakdown of paracellular and vascular barriers in the gut and increased serum levels of IgA directed against LPS of Gram-negative bacteria indicating leaky gut and increased bacterial translocation [12,13]. It is interesting to note that major depression and bipolar disorder 1 are accompanied by highly increased MDA, LOOH and AOPP levels [63,64] and that mesial temporal sclerosis is accompanied by highly increased MDA levels, whereas AOPP levels are less severely disordered [62]. Therefore, it appears that increased protein oxidation without severe lipid peroxidation and aldehyde formation is a more specific OS biomarker profile of deficit schizophrenia.
Our results on lowered TRAP antioxidant defenses extend the findings of Albayrak et al. [50] who reported that serum total antioxidant potential was specifically lowered in deficit schizophrenia. Lowered TRAP is an index of lowered antioxidant defenses that comprises the effects of hydrophobic antioxidants including vitamin E, and hydrophilic antioxidants including vitamin C, uric acid, and bilirubin [68]. Interestingly, in pregnant women we observed that lowered TRAP is strongly associated with increased AOPP production, indicating that lowered TRAP may contribute to protein oxidation [65]. Nevertheless, in the current study we found that the lowered levels of PON1 CMPAase activity and -SH groups were much more specific for deficit schizophrenia than lowered TRAP. SH-disulphide homeostasis is a key component of antioxidant defenses of the body and additionally plays a role in cell signaling, detoxification, protein regulation, apoptosis and transcription [69,70]. Some studies indicated lowered brain and serum GSH levels or related enzymes (e.g. GSH-peroxidase) in schizophrenia [71][72][73]. Nevertheless, -SH groups are part of protein (e.g. albumin) and non-protein compounds (e.g. free cysteine and GSH) while the protein -SH groups are abundant in plasma and GSH is more abundant in red blood cells [44,45]. Usually, there is little difference between the levels of total -SH groups and protein-bound -SH because the levels of GSH in plasma are comparatively very low.
Recently, we reviewed the role of PON1 enzyme activities in schizophrenia and discussed that in medicated and unmedicated schizophrenia patients, PON1 AREase and CMPAase/paraoxonase activities are frequently reduced [74]. Nevertheless, in the current study, lowered PON1 CMPAase activity was confined to deficit schizophrenia. In the plasma, PON1 binds to HDL and this functional PON1-HDL complex protects against macrophage-mediated lipid oxidation, including that of LDL and displays peroxidase activity [75][76][77]. Elevated PON1 enzyme activity displays anti-inflammatory effects by inhibiting the production of LPS-induced pro-inflammatory cytokines through regulation of MAPK and NF-B pathways, and by attenuating macrophage activities including production of monocyte chemoattractant protein-1 [74]. In first-episode psychosis, lowered activity of PON1 AREase is accompanied with increased plasma levels of M1 (IL-6), Th-2 (IL-4) and T regulatory (IL-10) cytokines [78]. Apart from antioxidant and anti-inflammatory properties, PON1 has also metabolic effects including effects on glycolysis and the Krebs cycle and, thus, energy metabolism, as well as stimulating insulin production and secretion [79]. While part of these effects are associated with PON1-related -SH groups, PON1 also shows thiolactonase activity whereby PON1 degrades homocysteine thiolactone, which may induce protein N-homocysteinylation as a consequence of an interaction between a free thiol derived from a protein cysteine residue and the free thiol group of homocysteine [74]. As such, lowered PON1 thiolactonase activity may play a role in the -SH group-related redox status of functional proteins, which may result in increased nitro-oxidative stress, production of protein adducts, immune-inflammatory and autoimmune responses and, consequently cellular toxicity [74,80]. Moreover, the PON1-HDL complex may be damaged and inactivated by peroxynitrite formed by nitric oxide and ROS, myeloperoxidase, and increased levels of IL-1 and TNF-α, which inhibit the synthesis of PON1 in the liver [74]. These processes may attenuate the regulatory effects of PON1 on myeloperoxidase activity thereby causing more protein oxidation, nitrosylation and formation of peroxynitrite. All in all, the loss of antioxidant defenses in deficit schizophrenia through lowered -SH groups, TRAP, and PON1 CMPAase activity, which is at least partly related to the QQ and QR genotypes, appears to be a key factor in the neuro-oxidative pathophysiology of deficit schizophrenia.
The second major finding of this study is that increased oxidative stress toxicity, lowered levels of antioxidant defenses and increases in the OSTOX/ANTIOX ratio are significantly associated with all symptom domains of schizophrenia as well as with the neurocognitive test scores indicating impairments in episodic and semantic memory and executive functions.
Moreover, these OSTOX/ANTIOX biomarkers explained a large part of the variance in a general factor extracted from the symptom domains and neurocognitive tests, reflecting OSOS. As reported previously, this single latent trait is essentially unidimensional and underpins the key domains of schizophrenia which are, therefore, manifestations of this common underlying construct [4-6,36]. This latent OSOS factor, which reflects the late phenome of schizophrenia, indicates disorders in neuronal circuits including in the "prefronto-striato-thalamic, prefrontoparietal, prefronto-temporal, and dorsolateral prefrontal cortex, as well as hippocampus and amygdala" [4,5, 81,82]. By inference, our results suggest that OSTOX and ANTIOX biomarkers impact OSOS through the multiple neurotoxic effects of protein oxidation (see above) and lowered antioxidant defenses, which increase the propensity towards aberrations in immune-inflammatory and nitro-oxidative stress pathways and cell signaling, apoptosis, transcription, detoxification, and protein regulation processes [69,70,74]. Moreover, PON1 genotypic distribution may be one of the genetic drivers leading to increased OSTOX and thus increments in OSOS.
The third major finding of this study is that the OSTOX and ANTIOX biomarkers have significant effects on OSOS above and beyond the effects of neuro-immune biomarkers including a) increased TNF-α, IL-6, CCL11, IgA directed to neurotoxic TRYCATs and PLS of Gramnegative bacteria and b) reductions in lowered natural IgM responses to MDA, which indicate lowered resilience of the innate immune system against inflammation, oxidative stress and bacterial stressors. Nevertheless, we found that a large part of the variance (47.6%) in the general latent trait OSOS, reflecting the late phenome of schizophrenia, is explained by a combination of neuro-immune and neuro-oxidative stress toxicity (TNF-α, IL-6, IgA to TRYCATs, LPS, CCL11, LOOH, AOPP, MDA) and lowered protection and resilience against these neurotoxic compounds through lowered PON1 CMPAase activity, TRAP, -SH groups and natural IgM to MDA. Thus, the current study indicates that OSOS increases along a continuum from normal controls → nondeficit schizophrenia → deficit schizophrenia in association with increasing neuro-immune and neuro-oxidative stress toxicity, while lowered antioxidant defenses and lowered natural IgM to MDA are a hallmark of deficit schizophrenia only. As such, the results of the present study indicate that the combination of neurotoxicity, which is more abundant when OSOS increases, with lowered antioxidant protection and natural IgM shape a distinct nosological entity, namely deficit schizophrenia.
The results of the present study should be discussed with respect to its limitations. First, it would have been more interesting if we had assayed a broader panel of OS and antioxidant biomarkers, especially MPO, as well as more neurotoxic cytokines including IL-17 and interferonγ. Second, this is case-control study which does not allow to draw firm causal inferences.
In conclusion, deficit schizophrenia is characterized by significantly increased levels of oxidative toxicity coupled with lowered TRAP, -SH, and PON1 activity levels, while these changes are not evident in non-deficit schizophrenia. A large part of the variance in the latent construct extracted from all symptom domains and neurocognitive scores was explained by a combination of neurotoxic compounds and lowered antioxidant defenses and natural IgM. These aberrations shape deficit schizophrenia as a distinct nosological entity.

Conflicts of interest
The authors have no conflict of interest with any commercial or other association in connection with the submitted article.     PANSSnegative: negative subscale of the Positive and Negative Syndrome Scale SANS: The Scale for the Assessment of Negative Symptoms *Sex: male = 1, female =0 OR: Odds ratio, 95% CI: 95% confidence intervals. LOOH: lipid hydroperoxides, PON1: paraoxonase activity, TRAP: total radical-trapping antioxidant parameter, AOPP: advanced oxidation protein products. 3OSTOX: sum of z LOOH + z malondialdehyde + z AOPP, reflecting oxidative stress toxicity. 3ANTIOX: sum of z PON1 CMPAase + z sulfhydryl groups + z TRAP, reflecting antioxidant defenses.
3OSTOX/3ANTIOX: z 3OSTOXz 3ANTIOX ; reflecting the oxidative stress toxicity / antioxidant ratio. Listed are partial correlation coefficients after adjusting for age, sex and education. * All p<0.01, except hostility (p<0.05) after p-correction for false discovery rate ** All p<0.01, except MMSE, WLM, and executive functions (p<0.05) and hostility and VFT (not significant) after p-correction for false discovery rate *** All p<0.01, except hostility (p<0.05) after p-correction for false discovery rate N=107 and n=108 for symptom domains and cognitive tests, respectively.