Bridging Autism Spectrum Disorders and Schizophrenia through inflammation and biomarkers - pre-clinical and clinical investigations

There is increasing evidence of altered immune function in ASD and schizophrenia, whether it is related to differences in populations of immune cells, in serum levels of antibodies, or levels of cytokines and interleukines. Such changes have been documented in other review papers [10‐14], but evidence of how these changes associate with clinical features, and in particular disease severity, might further elucidate the pathophysiological mechanisms involved. Table 3 summarizes the main evidence from studies relating biomarkers of immune function and biomarkers of oxidative stress to clinical features in both ASD and schizophrenia.

A meta-analysis in drug-naïve patients with schizophrenia [140], reported increased numbers of total lymphocytes, T lymphocytes (CD3 positive), T helper cells (CD4 positive), and a higher ratio between T helper and T cytotoxic cells (CD4/CD8), but a reduced proportion of T lymphocytes. In acutely relapsed patients, the authors also found a higher proportion of CD4 positive and CD56 positive cells (T helper and natural killer, NK, cells, respectively). After treatment, the CD4/CD8 ratio decreased, while the concentration of CD56 positive cells increased, the former being a state marker while the latter a trait marker [140]. In a previous study, Miller et al. [11] suggested that cytokine levels were also associated with clinical status; some cytokines like IL-1β, IL-6, and TGF-β seem to act as state markers for schizophrenia (they are increased during the acute phase of psychosis and normalize with anti-psychotic treatment) while others, such as IL-12, IFN-ϒ, TNF-α, and sIL-2R, appear to represent trait markers as their levels remain elevated in acute exacerbations and do not remit after anti-psychotic treatment.

Several studies have shown that changes in serum levels of antibodies correlate with disease severity in ASD. Children with autistic disorder had significantly reduced levels of plasma IgG and IgM compared to children with other developmental delays (DD) and typically developing controls [141]. This reduction correlated with behavioral severity, i.e., patients with the most reduced levels of IgG and IgM scored highest in behavioral tests. Significantly higher serum levels of anti-ganglioside M1 antibodies and anti-neuronal antibodies have also been correlated with disease severity and Child Autism Rating Scale (CARS) scores [142, 143]. Children with ASD very frequently show alteration of biochemical pathways, such as the methylation pathway, which performs complex functions, vital to overcoming neurological inflammation. Mutations in pathways, such as the methionine cycle, the folate cycle, the BH4 (biopterin) cycle, and the urea cycle (which are all interrelated) can produce biochemical imbalances which are, at least in part, responsible for the immune imbalance found in children with ASD [144].

A number of studies have found that certain cytokines and interleukines correlate with clinical features (type and severity of symptoms, clinical status, course of illness) both in ASD and schizophrenia (Table 3). The comparison of plasma samples of children with ASD with age-matched typically developing children and children with other DD [145] showed significant increases in the plasma levels of a number of cytokines. In the ASD group, increased cytokine levels were predominantly found in children with a regressive form of ASD and were associated with more impaired communication and aberrant behaviors.

Other studies have linked interleukines to symptom dimensions (positive and negative) in schizophrenia. Specifically, IL-2 has been associated with both the negative [146] and positive subscale [147] of the Positive and Negative Syndrome Scale (PANSS). The soluble IL-2 receptor (sIL-2R) has also been associated with the PANSS total scores, negative symptom and general psychopathology subscale scores [148]. On the other hand, IL-6 [149] has been correlated to negative symptoms and duration of illness, while IL1-RA [150] has been significantly correlated with negative symptoms (PANSS). IL-6 and serotonin levels were increased in autistic patients compared to controls and had a positive correlation with autism severity [151]. In another study [152], the same authors found increased levels of IL-6 and TNF-α, and decreased diurnal variation of cortisol (VAR) in patients with autism compared to controls. Further, IL-6 and TNF-α levels positively correlated to CARS scores while VAR negatively correlated to CARS scores.

Other important components of immune function have also been studied in ASD and schizophrenia. Namely, there is evidence that TNF-α, TGF-β, chemokines, and osteopontin (OPN) are also associated with clinical features and disease severity, as indicated below.

TNF-α is associated with several inflammatory conditions. In patients with chronic schizophrenia, its levels were significantly lower [19] and showed correlations with PANSS total score, PANSS positive and general psychopathology subscores, as well as the cognitive subscale. Specifically, patients with lowered levels of TNF-α were more likely to have severe psychopathological symptoms even though no differences were found between schizophrenia subtypes or type of anti-psychotic treatment. TNF-R1 was also significantly correlated with positive symptoms of PANNS (Hope et al. 2013). A deficient production of TNF-α might be associated with dysfunction of thought, perception, and behavior in these patients.

TGF-β is one of the most important regulators of the immune response. [153] found significantly lower plasma TGF-β1 levels in children with ASD compared with typically developing controls and with children with other developmental disabilities. Importantly, such levels were associated with reduced adaptive behaviors and worse behavioral symptoms.

Chemokines are small cytokines with crucial roles in homeostasis, as they control the movement of leukocytes into the central nervous system, regulate neuronal cell migration, proliferation and differentiation, and are important in neuronal-microglia communication. Chemokines are frequently deregulated in disease and have been implicated in ASD and schizophrenia [154]. In a study, the diagnosis of schizophrenia was accompanied by a specific cytokine-chemokine profile [154], i.e., increased levels of CCL11, CCL3, soluble TNF receptors 1 and 2 (sTNF-R1 and sTNF-R2), and decreased levels CXCL10 (also known as IFN-ϒ-inducible protein 10 or IP-10), TNF-α, IL-2, and IL-4. Using five of these biomarkers (sTNF-R1 and sTNF-R2, CCL11, CXCL10, IL-4), the authors found a sensitivity of 70% and specificity of 89.4% for the diagnosis of schizophrenia. IL-2 was inversely associated with negative symptoms, while CCL11 was positively associated with negative symptoms. Further, treatment resistance was associated with increased levels of both TNF receptors (sTNF-R1 and sTNF-R2) and CCL2.

OPN is a phosphoprotein with important roles in inflammation, cell adhesion, chemotaxis, immune response, and protection from apoptosis, depending on its location and context. Levels of OPN in autistic children were significantly higher than those of matched healthy children [155] and correlated with disease severity and CARS scores.

Oxidative stress appears to be important in many neuropsychiatric disorders, including schizophrenia, and several biomarkers can be used to assess oxidative stress and anti-oxidative status. Oxidative stress, which leads to damage of proteins, lipids, and DNA, converts glucose and lipids to reactive carbonyl compounds, which in turn are converted to advanced glycation end products. The accumulation of such products is referred to as carbonyl stress.

Katsuta et al. [156] evaluated the role of carbonyl stress markers in acute stage schizophrenia and found that pyridoxal levels (or vitamin B6 which detoxifies reactive carbonyl compounds) were significantly lower in patients with schizophrenia compared to that in controls. Further, the greater the decrease in pyridoxal levels over time (from admission to discharge) the lesser the improvement in symptoms. In chronic schizophrenia, lower pyridoxal levels have shown correlation with more severe symptoms, especially positive symptoms [157]. Reduced pyridoxal levels and an increased in pentosidine have also been correlated with treatment resistance [158]. Glutathione deficit has also been suggested as an indirect biomarker of oxidative stress in early-onset schizophrenia [159]. Another study examined the peripheral blood mononuclear cells (PBMC) gene expression levels of pro-inflammatory transcription regulator NF-kB and the downstream enzymes iNOS and COX-2, important in inflammatory and oxido/nitrosative status, respectively, and found significant increases in first episode psychosis, in relation to healthy controls. Concomitantly, the NFkB inhibitory subunit IkBa was decreased in patients. Together, these results indicate a pro-inflammatory state of PBMC in first episode phychosis [160]. A very recent review and meta-analysis in first episode and early-onset schizophrenia, concluded that although the heterogeneity between studies is too great to allow an effective comparison, oxidative stress and inflammation are likely increased and lead to worse outcomes [161].

Human genome sequencing projects revealed that the large majority of our genome do not encode for proteins [162]. Surprisingly, throughout evolution the number of protein coding genes remained relatively stable in the different species, while the number of non-protein coding genes greatly increased [162]. The importance of these non-protein coding transcripts has recently been explored. Noteworthy, non-coding RNA are implicated in human evolution and cognition [162] and have important biological functions [163]. These observations led to a shift in the potential of human non-coding genes. This class of genes can be divided in long non-coding and small non-coding RNAs, depending on the size of gene transcripts [164]. Among the latter, microRNAs (miRNA), with approximately 22-nucleotides long, are the most commonly studied. MiRNAs are key regulators of gene expression at a post-transcriptional level, by binding to 3′untranslated regions (UTR), coding sequences or 5′UTR of target messenger RNAs (mRNAs), which leads to inhibition of translation or mRNA degradation [163]. Interestingly, a single miRNA is able to regulate the expression of several genes (coding and non-protein coding genes). Therefore, a single miRNA may simultaneously control key mechanisms of schizophrenia or ASD and regulate inflammatory pathways. Moreover, miRNA levels can be detected in body fluids, including blood, saliva, and urine [165]. Consequently, miRNAs have emerged as new, unconventional, and promising biomarkers for diagnosis and prognosis of mental disorders [166]. In this review, we will focus particularly on miRNAs that have been consistently reported as blood-associated biomarkers for schizophrenia or ASD (Table 4), and we will address their link with inflammatory pathways.

The first hypothesis of miRNA involvement in schizophrenia was launched in 2005 [167], following publications in the cancer research field reporting miRNA direct control over tumor suppressors/oncogenes and the consequent impact on tumor development [163]. However, it was only in 2007 that the first scientific evidence associating miRNAs to schizophrenia emerged [168]. Perkins et al. reported that seven miRNAs were downregulated in postmortem prefrontal cortex tissue of individuals with schizophrenia compared with non-psychiatric disease individuals [168]. Three of those miRNAs, namely, miR-7, -212, and -132, were confirmed by an independent study [169]. Importantly, miR-7 overexpression has been identified in plasma samples from schizophrenic patients, compared with healthy controls [170]. This result has been validated by others, which strengthens miR-7 as a new diagnosis marker for schizophrenia [171, 172]. The role of miR-7 in regulating SHANK3 mRNA was uncovered using a hippocampal neuronal cell line, transduced with lentivirus for expression of miRNA mimics or its inhibitors [170, 173]. SHANK3 protein is expressed in cortex and hippocampus and contributes to synaptic development [173]. Similarly to humans, SHANK3 has been identified in genetic mice models of ASD [40], as highlighted in Section 2.1.1. Functionally, miR-7 overexpression decreased the density of dendritic spines in a SHANK3-dependent manner [40, 174]. This could partially explain miR-7 effect in neuropsychiatric disorders [170, 174]. However, considering schizophrenia as a multifactorial disease, it would not be surprising that miR-7 would also regulate inflammatory states. In dendritic cells cultures, miR-7 was found to be upregulated in activated dendritic cells stimulated with LPS and IFN-γ compared with both immature and tolerogenic dendritic cells, stimulated with GM-CSF and IL-4 or IL-10 and TGF-β, respectively [175]. miR-7 has also been found to be upregulated in chronic inflammatory and autoimmune diseases, including in nasal mucosa of patients with allergic rhinitis [176] and in B cells of systemic lupus erythematosus patients [177]. Wu et al. showed that impairment of B cell regulation is caused by a decrease in phosphatase and tensin homolog (PTEN) expression, which is post-transcriptionally regulated by miR-7 [177]. Notably, PTEN mutations have been widely described in ASD and loss of PTEN in Purkinje cell has been shown to induce autistic-like traits in mice [178]. Also, PTEN conditional KO mice revealed neurons with defective synaptic function, enlarged neuronal somata, and increased dendritic spine density, and also uncovered defects in myelination in the corpus callosum [179]. However, a direct contribution of miR-7/PTEN link to ASD remains to be determined. Deregulation of miR-7 expression levels in plasma or serum of patients with ASD has not yet been described.

Additionally, miR-212 and miR-132, which are transcribed from the same primary-miRNA and share the same seed region and some of the targets [180], have been widely reported as biomarkers for schizophrenia [168, 169]. miR-212 is upregulated in peripheral blood mononuclear cells from schizophrenic patients compared with healthy controls, while increased expression of miR-132 was found in plasma samples of schizophrenic patients [171]. Interestingly, miR-132 levels were decreased upon treatment [172]. Both miR-212 and miR-132 are able to simultaneously regulate neurons morphogenesis and immune processes [180]. On the one hand, mice with floxed-Cre and knockout alleles for miR-212/132 have dendrites with reduced length and branches in newborn neurons in the adult hippocampus [181]. Moreover, these miRNAs participate in the synaptic function [180]. miRNA-212/132 play a role in immune system regulation [180]. For instance, miRNA-212/132 levels are increased in inflammatory diseases such as rheumatoid arthritis and osteoarthritis [182, 183]. Supporting the role of these miRNAs in inflammation, miR-212/132 are induced by ligands of TLR1, TLR2, and TLR5 [182]. Therefore, miRNA-212 and miR-132 are potential schizophrenia markers and may mediate the interplay with inflammation. Regarding ASD, studies performed in lymphoblastoid cell lines derived from autistic patients and controls identified miR-132 as a biomarker [184, 185]. However, results are contradictory as one study reports miR-132 as overexpressed [184] while another shows miRNA-132 as underexpressed in patients versus controls [185].

miR-181b was shown to be upregulated in postmortem cortical gray matter from the superior temporal gyrus in schizophrenia [186]. This miRNA directly targets glutamate ionotropic receptor AMPA type subunit 2 (GRIA2), which has been implicated in development of synaptic plasticity, and the calcium sensor protein named visinin like 1 (VSNL1) [186]. Shi et al. reported miR-181 was upregulated in plasma samples of patients with schizophrenia compared with healthy controls [187]. This result was further confirmed by Song et al. [188]. Importantly, decreased expression of this miRNA was associated with symptomatology improvement after anti-psychotic treatment [188]. Furthermore, after patients’ stratification, it was found that serum levels of miR-181b were significantly increased in family schizophrenia than in sporadic schizophrenia [187]. In one of the first studies analyzing miRNAs as ASD biomarkers in plasma, samples of 55 autistic children and equivalent number of controls were tested [189]. In contrast to reports in schizophrenia, authors found miR-181b as significantly downregulated in autistic children compared with control group, and this result was independent of the method used for data normalization [189]. Therefore, miR-181b may have distinct roles in schizophrenia and ASD. However, the role of miR-181 in ASD is still inconsistent. Olde et al. used a pharmaceutical rat model of ASD, based on prenatally delivery of valproic acid, to study differently expressed miRNAs in amygdala, which is known to be enlarged in ASD patients [190]. Authors found significantly elevated levels of miR-181c, but not miR-181b, in the amygdala of valproic acid-exposed rats [190]. Experiments performed in microglia cell culture shown that miR-181c levels are decreased, while TNF-α levels are increased, when cells are cultured under oxygen-glucose deprivation [191]. Future studies on biomarkers should include all mature forms of miR-181 family. Hutchison et al. reported that miR-181 directly targets MECP2, a protein whose deregulation causes behavior disorders [192] (details on MECP2 in vivo animal studies are described in Section 2.1). Validation of miR-181:MECP2 targeting in independent studies is still missing. Mutations causing decreased levels of MECP2 have been associated with patients with Rett syndrome, which often exhibits ASD-like behaviors, while extra copies of this gene cause MECP2 duplication syndrome characterized by autistic behaviors [193]. Recently, Sztainberg et al. showed that symptoms of MECP2 duplication in mice were highly improved by restoration of normal MeCP2 levels using antisense oligonucleotides, which act through a base-pair complementarily mechanism similar to miRNAs [193]. Therefore, future studies using miR-181 for diagnosis or as a tool to understand molecular mechanisms in ASD and schizophrenia should be considered. In inflammation, the role of miR-181 also seems to be dependent on tissue/cell type. miR-181a, closely related to miR-181b, increases number of B cells by targeting a pro-apoptotic gene, BIM, in lymphoma cells [194], while in T cells it impairs cell sensitivity, and positive and negative selection [195, 196]. In fibroblasts, miR-181a directly regulates CXCL8, a pro-inflammatory cytokine [197]. Furthermore, in endothelial cells, miR-181b acts as an anti-inflammatory mediator by inhibiting NF-κB signaling pathway through targeting of importin-α3 [195, 198]. In vivo studies revealed that miR-181-deficient mice showed impaired lymphoid development and T cell homeostasis, and absence of mature NKT cells [199]. These effects in miR-181-KO mice were mediated by deregulation of PTEN levels [199], the gene commonly associated with comorbid ASD [200]. Finally, miR-181c has been implicated in neuroinflammation by targeting TNF-α in microglial cells [191]. The potential of miR-181 as a biomarker and as a molecular player in schizophrenia and ASD and the common traits to inflammation are worthy to be explored.

Another miRNA biomarker commonly described in schizophrenia is miR-30e, which is upregulated in plasma samples of schizophrenia patients compared to controls [171]. The same research group further validated miR-30e as a biomarker in a second cohort of patients and demonstrated its implication in the clinical outcome as miR-30e levels were decreased upon treatment [172]. Moreover, levels of this miRNA in peripheral blood mononuclear cells collected from schizophrenia patients were also increased, although miR-30e plasma levels have a higher sensitivity for diagnosis [171]. In ASD, miR-30a*, a miR-30 family member, was found to be differently expressed between cultured lymphoblastoid cells derived from autistic patients or their unaffected siblings, but so far, it has never been reported as an ASD biomarker in circulating plasma [201]. Importance of miR-30 for ASD is reinforced by the finding that miR-30 levels are increased in the amygdala following exposure to valproic acid, a rat model of ASD [190]. Furthermore, miR-30a is present in human prefrontal cortex and it targets BDNF, a crucial protein for cortical development and maturation [202]. Besides the role in neurons [202], miR-30 family is a player in the regulation of inflammation-associated genes [203, 204]. miR-30b attenuates phagocytosis and modulates secretion of TNF-α, IL-6, and IL-12p40 [203], while miR-30b contributes to B-cells dysfunction [205].

Independent studies reported increased levels of miR-34a in plasma samples of schizophrenia patients compared with healthy controls [172, 188]. High levels of this miRNA have also been found in peripheral blood mononuclear cells isolated from patients with schizophrenia [214]. Interestingly, levels of miR-34b, which mature sequence differs from miR-34a in only few nucleotides, are increased in peripheral blood of autistic patients [206]. However, the effect of miR-34a in inflammation is still controversial. On the one hand, miR-34a is associated with an anti-inflammatory profile by driving macrophage polarization into the M2 phenotype. This miRNA decreases levels of pro-inflammatory cytokines, including TNF-α and IL-6, and it inhibits pro-inflammatory response by decreasing notch1 levels and NF-kB activation in macrophages upon stimulation with lipopolysaccharide [207]. On the other hand, in a mouse macrophage cell line, miR-34a targets Twist-2 [208], which promotes production of the anti-inflammatory cytokine IL-10 in myeloid cells [209].

Other circulating miRNA biomarkers in schizophrenia and autism have been reported, but more consistent data is still needed to determine their significance as circulating biomarkers in the blood. For instance, the levels of miR-195 should be further investigated. This miRNA has been identified as upregulated in plasma of schizophrenia patients [171] and in serum of ASD patients [189], but other studies show it is decreased in serum of schizophrenia patients [187] or do not found a significant association [172]. Also, miR-195 levels are increased in plasma of schizophrenia patients resistant to treatment and decreased in patients responsive to treatment when compared with health donors [210] Within the brain, expression of this miRNAs is more associated with neurons rather than with glial cells [211]. Similarly to miR-30a, miR-195 directly targets BDNF [202].

Moreover, miR-346 was identified as a schizophrenia biomarker in plasma samples [172, 187], while its levels were only slightly increased in serum samples [187]. This miRNA regulates the release of the pro-inflammatory cytokine TNF-α, following stimulation of a monocytic cells line with LPS [212]. Additional potential miRNA biomarkers in schizophrenia include miR-219-2-3p, miR-1308, miR-92a, miR-17and let-7g in serum [187], miR-130b and miR-193a-3p in plasma [213], and miR-449a, miR-564, miR-432, miR-548d, miR-572, and miR-652 in monocytes [214]. Regarding ASD blood-associated biomarkers, further candidates include miR-151a-3p, 320a, miR-328, miR-433, miR-489, miR-572 as underexpressed, and miR-663, miR-101-3p, miR-106b-5p, miR-130a-3p, and miR-19b-3p as overexpressed in plasma samples [189]. Finally, a study in the Chinese population revealed that levels of let-7a, let-7d, miR-103a, and miR-1228 are reduced in peripheral blood of ASD patients compared with healthy controls [215]. Studies reporting circulating miRNAs as biomarkers for mental diseases are mainly restricted to plasma/serum or blood cell samples. Analysis of miRNA levels in urine as diagnostic or prognosis tool for schizophrenia or ASD has not been reported so far. However, a recent study was conducted in saliva of ASD patients, which showed differences in levels of a set of miRNA for ASD children compared with control group [216]. miRNA stability through association with proteins, and miRNA protection through encapsulation in extracellular vesicles (e.g., exosomes) are essential to avoid miRNA degradation and to allow the detection of miRNA in cell-free liquid biopsies [217].

Non-coding RNA research field is expected to continue to grow in the upcoming years and may shed light into the common traits between inflammation and behavioral disorders, including schizophrenia and ASD. Validation of reported miRNA biomarkers as well as search for novel and reliable miRNA candidates for schizophrenia and ASD diagnosis or prognosis are essential. Importantly, studies screening for inflammatory biomarkers should consider patients’ pharmacological treatments, as detailed in the next sections.

The neurotransmitter dopamine not only regulates numerous bodily functions (behavior, movement, endocrine, cardiovascular, renal, and gastrointestinal function) but also constitutes an important bridge between the nervous and immune systems [219] since dopamine receptors are present in almost all immune cell subpopulations [220]. Dopamine (or dopamine agonists) have been shown to modulate the activation, proliferation, and cytokine production in immune cells [220, 221]. In a recent study by Yan et al. [219], the authors found that dopamine inhibits the NLRP3 inflammasome through the dopamine D1 receptor. The NLRP3 inflammasome is a cytosolic protein complex which is assembled in response to microbial infection or endogenous danger signals, thereby promoting the maturation and release of pro-inflammatory cytokines like IL-1β or IL-18. It is crucial in the initiation of inflammation and development of immune responses and has been implicated in diseases such as type 2 diabetes, atherosclerosis, and gout [222]. Yan et al. [219] not only demonstrated that dopamine is an endogenous inhibitor of NLRP3 activation but also suggested that dopamine is a potent anti-inflammatory agent.

On the other hand, serotonin (5-hydroxytryptamine (5-HT)) not only regulates many important physiologic aspects in the central nervous system (mood, aggression, sleep, appetite) but also peripherally (pain sensation, bone mass, tissue regeneration, platelet coagulation, and gastrointestinal function) [223]. Only 5% of the body’s 5-HT is produced in the brain. Platelets carry 5-HT, and th us are the major source of 5-HT for immune cells and lymph tissue, but monocytes, mast cells, and T cells also appear able to synthesize small amounts of 5-HT [223]. There is evidence that 5-HT has diverse signaling roles in immune cell function and is important in both innate and adaptive immune response. Further, 5-HT has been proposed as a T-cell modulator for many years now, and there is increasing evidence supporting the role for 5-HT signaling in B lymphocytes [224].

As described above, it is easily anticipated that antipsychotic drugs will most certainly affect immune function and the effects of anti-psychotics on cytokine levels have been documented in several studies Table 5 summarizes the main evidence from studies that relate biomarkers of immune function and oxidative stress to anti-psychotic treatment.

Results are somewhat heterogeneous across studies, and this may result from differences in stage of illness (first episode psychosis or chronic schizophrenia), schizophrenia subtypes, gender, smoking habits, age, medical comorbidities, or differences in anti-psychotic treatment (type of anti-psychotic, dosage and duration of treatment) [154].

Witte et al. [225] found a mixed pro- and anti-inflammatory profile in drug naïve FEP patients, that is, levels of interleukin IL-1RA, IL-10, and IL-15 were increased significantly compared to controls. The levels of IL-1RA and IL-10 decreased after treatment with atypical anti-psychotics, and the changes in IL-10 levels were significantly correlated with improvements in negative, general, and total symptom scores. The finding that only IL-10 responded to treatment in parallel with symptom improvement suggests that it could be used as a potential treatment response biomarker in future studies of schizophrenia.

In another study with drug naïve FEP patients [226] treatment with risperidone had a significant suppressant effect on several serum cytokine levels, namely, IL-6, IL-10, and TNF-α. Results suggest that risperidone seems to normalize a specific cytokine profile in first episode psychosis, characterized by monocytic and T-regulatory cell responses and additionally decrease Th2 functions. Curiously, the improvement in psychopathology with risperidone treatment was not related to risperidone-induced changes in cytokine levels from baseline to endpoint. Also, treatment with risperidone significantly decreased IL-4 and TNF-α but only in patients with depressive symptoms [226]. In contrast, Song et al. [227] found that in a sample of drug-naïve FEP, IL-6 returned to baseline levels but TNF-α increased following treatment with risperidone. Other studies have also found different effects on cytokine levels following risperidone treatment which might be explained by differences in media (immune cells versus microglia), study type (in vitro versus in vivo), stages of illness, and schizophrenia subtypes [154].

A recent meta-analysis on the effect of anti-psychotic drugs on blood levels of cytokines in patients with schizophrenia [228] found that haloperidol and clozapine treatment increases peripheral sIL-2R levels [11, 229], but leads to decreases in IL-1β and IFN-ϒ, and possibly increases in IL-12. Generally unaffected by anti-psychotic treatment are IL-2, IL-4, IL-6, IL-10, IL-1RA, sIL-6R, TGF-β1, and TNF-α, with the exception of clozapine which appears to increase IL-6 and sTNF-R. Typical anti-psychotics have been shown to suppress plasma IL-6, soluble IL-6 receptors, and transferrin receptor levels [230] whereas clozapine has been shown to increase sIL-2 receptor, IL-6, and IL-1RA [231, 232]. However, Borovcanin et al. [233] reported that treatment with anti-psychotics might decrease IL-4, IL-6, and IL-27 levels in schizophrenia.

Anti-psychotic drugs have also been noted for their potential anti-inflammatory role in schizophrenia. Consistent with the hypothesis that schizophrenia is associated with an exacerbation of the inflammatory response is the observation that the long-term administration of anti-psychotic drugs increases the level of anti-inflammatory cytokines, such as IL-10, and decreases the concentration of pro-inflammatory cytokines, namely, IL-1beta and TNF-alpha [234]. It is important to clarify if there is a difference between the effect of first and second generation anti-psychotic drugs [229]. Common to both groups appears to be the decrease of IL-2 levels (in vivo, ex vivo, and in vitro studies) [229] and a discernable difference between both groups on the serum levels of cytokines does not appear to exist. Clozapine, contrary to typical (haloperidol), and other atypical (olanzapine and risperidone) anti-psychotics stimulates the levels of IL-6 and TNF-alpha [229], which might be related to some of its adverse effects. In summary, anti-inflammatory and anti-psychotic drugs appear to act as modulators of the inflammatory response, presumably by decreasing the activity of microglia, which would manifest itself by an inhibition of pro-inflammatory cytokines and the prostaglandin PGE2 production.

Only two anti-psychotics (risperidone and aripiprazole) have been approved by the FDA for treatment of irritability in ASD, but neither drug shows improvement in the core symptoms of the disorder, and the long-term consequence of their use in children remains unknown. Therefore, few studies are available on the effects of anti-psychotics on biomarkers of immune function in ASD. Tobiasova et al. [235] evaluated whether risperidone-associated improvement was related to changes in concentrations of inflammatory cytokines and growth factors (EGF, IFN-ϒ, IL-13, IL-17, CCL2, IL-1, IL-1-RA) in children with ASD. This study found that although ASD patients had increased levels of EGF and decreased levels of IL-13 compared to controls, all inflammatory serum markers remained stable over a period of 8-week treatment with risperidone. Interestingly, even though treatment induced clinical improvement in aberrant and maladaptive behavior, this was not associated with changes in serum levels of these inflammatory biomarkers. However, in a study by Choi et al. [236], there was a significant decrease in CCL11 (also known as Eotaxin) and CCL2 levels after 8 weeks of risperidone treatment in children with ASD. Also, mean values of IL-5 were significantly higher in the responder group compared to non-responders.

The evidence of microglial activation in schizophrenia has been addressed in Section 3 of this review, and the indirect assessment of glial activation via peripheral blood markers is important for understanding the processes that are occurring in vivo. Glial cell activation stimulates astrocytes to produce SB100, which is a marker of inflammation. Increased S100B levels in the early stages of schizophrenia support the idea of neurodegenerative process and anti-psychotics such as haloperidol and clozapine have been shown to decrease S100B release from glial cells [237]. S100B levels were assessed in the serum of drug-naïve early-stage, medicated chronic schizophrenia patients, and healthy controls [237]. Results showed significantly increased serum S100B levels in both drug-naïve and medicated patients compared to controls, and also in drug naïve compared to medicated patients.

In vitro studies regarding the release of inflammatory markers, under the influence of anti-psychotic drugs, shed some light on the above results. For instance, LPS-activated glial cells showed a reduced level of IL-1beta and IL-2 after administration of chlorpromazine [238]. Risperidone appears to be particularly effective in inhibiting the activation of microglia by iNOS [239], which, as discussed earlier, is detrimental to neurogenesis. Haloperidol and risperidone inhibited the secretion of S100B by glioma cells, after stimulation by IL-6 [240]. This is in accordance with the finding that S100B levels were significantly higher in drug-naïve patients than in those that were medicated, as discussed earlier.

While there is evidence that oxidative stress pathways are involved in the pathophysiology of schizophrenia, there is less information regarding the effects of anti-psychotics on these pathways. Zhang et al. [241] showed that both risperidone and haloperidol reduce superoxide dismutase levels in schizophrenia patients, even though they do not normalize initially increased levels of plasma nitric oxide levels. In this study, lower superoxide dismutase (SOD) levels at baseline predicted greater symptom improvement, and greater change in SOD levels correlated with greater symptom improvement.

Paroxonase 1 (PON1) is an antioxidant enzyme synthesized in the liver which protects high density lipoprotein against oxidative stress damage. In a study by Noto et al., drug naïve FEP patients showed decreased PON1 activity and increased total radical trapping antioxidant parameter (TRAP) values [242]. TRAP levels are determined by the effects of specific non-enzymatic antioxidants in the plasma such as uric acid, albumin, bilirubin, protein-bound SH (thiol) groups, and vitamin E and C [243]. Treatment with risperidone increased PON1 and decreased lipid hydroperoxides (LOOH) levels, suggesting that risperidone may modulate oxidative stress pathways possibly through anti-oxidative activities by increasing PON1 activity and lowering a marker of lipid peroxidation (LOOH). Also, a study of the oxidative stress in first episode psychosis showed that the total level of peroxides was higher in these patients than in normal controls, and that difference resolved with anti-psychotic treatment [244].

The ability to predict if a patient will respond to a particular treatment by the use of quantifiable biomarkers would allow individualized treatment options. This would enable the selection of treatments that are more likely to succeed for a particular patient, avoiding multiple and unnecessary trials, which can lead to patient non-adherence and relapse. In Table 6, we summarize existing evidence from studies of predictors of response to anti-psychotic treatment.

The contribution of genes in schizophrenia and ASD has been the focus of several studies which have been subject of other reviews [245, 246]. However, the role of gene expression in predicting treatment response has been less studied so far Correia et al. [247] explored the effects of multiple candidate genes on clinical improvement and occurrence of adverse drug reactions with risperidone, in children with ASD. The study included genes involved in the pharmacokinetics (CYP2D6 and ABCB1) and pharmacodynamics (HTR2A, HTR2C, DRD2, DRD3, and HTR6) of the risperidone, and the BDNF gene. Results showed that the HTR2A c.-1438G > A, DRD3 Ser9Gly, HTR2C c.-995G > A, and ABCB1 c.1236C > T polymorphisms were predictors of clinical improvement with risperidone therapy. Likewise, Lit et al. found that expression of a group of five genes (GBP6, RABL5, RNF213, NFKBID, and RNF40) prior to treatment initiation was correlated with response to risperidone [248]. Interestingly, RNF40 is located at 16p11.2, a region implicated in both ASD and schizophrenia, and RNF40 and RNF213 have RING domains (Really Interesting New Gene domains (RING)) which contain zinc binding sites. In a study by Arnold et al. [249], in children with autistic disorder, decrease in body zinc status while taking risperidone was strongly associated with greater improvement in irritability. Another study [14] refers that multiple genes associated with neuronal cell growth are associated with a positive response to risperidone in children with ASD.

Several studies have explored polymorphisms of genes coding for dopamine and serotonin receptors of anti-psychotic drugs and treatment response in patients with schizophrenia. Signal transduction genes have also been explored, as have cytochrome polymorphisms, and polymorphisms associated with side effects. A significant association was found between the A-2518G polymorphism of the CCL2 gene and treatment resistance, with resistant patients more frequently carrying the G-allele [250].

The SNAP-25 is a synaptosomal-associated protein directly involved in the release of neurotransmitters. Muller et al. [251] studied patients with schizophrenia or schizoaffective disorder, with prior suboptimal response to anti-psychotic treatment, and found that patients with the SNAP-25 gene variant (Mnll polymorphism) had significant changes in PANSS scores after a 14-week treatment with anti-psychotics. In another study [252] with drug naïve FEP patients, the authors found significant associations between 5-HTT-LPR variants and early negative symptom response to treatment. Genetic variations in the 5-HTT-LPR polymorphism had previously been associated with variations in anti-psychotic drug response [253‐255] but mainly to clozapine in samples of chronic patients [252]. In summary, so far, there is no predictor as to which patient (with a first psychotic episode) will respond to which treatment based on genetic biomarkers [9]. In a study with schizophrenia patients who were initially unmedicated or anti-psychotic naïve [256], molecular signatures that could predict symptom improvement over the first 6 weeks of treatment were described. Lower insulin levels at T0 were predictive of symptom improvement after anti-psychotic treatment, and the three molecules with the greatest differences between the short- and long-term relapse groups were leptin, proinsulin, and TGF-α. Further, leptin, insulin, and C-peptide increased significantly in patients who relapsed later but showed no change in the group that relapsed earlier.

The function of hypothalamic–pituitary–adrenal (HPA) axis in schizophrenia patients has been extensively reviewed by Bradley and Dinan [257]. Even though the authors conclude that there seems to be clinically relevant HPA axis dysfunction in patients with schizophrenia, results are heterogenous across studies and should be interpreted taking into account possible confounders. There seems to be evidence of elevated basal cortisol in some but not all patients compared to controls. However, basal cortisol secretion is influenced by psychological stress which undermines a fair comparison between patients and controls, in that psychological stress derived from the illness itself (and the presence of psychotic symptoms) and hospitalization by mental illness is irreproducible. Also, anti-psychotic drugs can influence cortisol secretion in several ways, either by reducing psychotic symptoms (and reducing the psychological stress associated with them) or by a more direct result of their pharmacological action.

Some studies have investigated the function of the HPA axis in predicting response to anti-psychotic treatment. Mondelli et al. 2010, in a study with FEP patients, reported that anti-psychotics normalized diurnal cortisol hypersecretion but failed to normalize the blunted CAR [258]. In a later longitudinal study with FEP patients [259], non-responders to 12-week treatment with anti-psychotics showed persistently lower CAR and higher IL-6 and IFN-ϒ levels when compared to responders. Blunted CAR and the reduced HPA axis reactivity to stress have also been associated with more severe symptoms and worse cognitive function in patients with psychosis [260]. Further, more blunted CAR but not more elevated cortisol levels during the day appears to be associated with cognitive dysfunction in these patients [260]. Persistent non-suppression of cortisol levels following the dexamethasone test after 4 weeks of anti-psychotic treatment has also been associated with poor clinical outcome [261].

Taking into consideration that the therapeutic effect of anti-psychotic drugs is attributed to their capacity to block dopamine and serotonin receptors, it seems likely that neurotransmitter metabolites should be investigated as potential predictors of treatment response. Two studies in the 90s [262, 263] reported elevated levels of dopamine (plasma homovanillic acid and pHVA) and decreased levels of noradrenalin (plasma 3-methoxy-4-hydroxyphenylglycol and pMHPG, respectively) metabolites in the plasma of patients with FEP who responded to treatment with anti-psychotics. A more recent study [264] found that treatment resistance in schizophrenia was associated with relatively normal striatal dopamine synthesis but elevated glutamate levels in the anterior cingulate cortex [264]. Another study [265] found that increased pretreatment prolactin response to the D-fenfluramine test, an indirect measure of serotonin activity, was associated with non-response to haloperidol in FEP patients.

More recently, Condray et al. [266] found that levels of 3-hydroxykynurenine (3-OHKY) quinolinic acid, a product of tryptophan metabolism (which is neurotoxic), not only predicted severity of clinical symptoms but also the degree of clinical improvement following brief treatment with anti-psychotics in patients with first-episode psychosis. However, studies are needed to replicate these findings in larger samples and further elucidate the value of such metabolites as treatment predictors.

It is nowadays evident that patients do not respond in the same way to anti-psychotic drugs. While intervention in dopaminergic neurotransmission seems to work for many patients, it is clearly insufficient for clinical response in others. Howes et al. [267] propose a subtyping of schizophrenia based on the underlying neurobiological mechanism rather than the current phenomenological approach which would be more useful to guide diagnostic testing and treatment. In type A schizophrenia hyperdopaminergia underlies the onset of the disorder and symptoms, and these patients show good response to dopamine-blocking anti-psychotics. In contrast, type B schizophrenia patients have normal dopaminergic function and symptoms that are unrelated to dopaminergic function, with glutamatergic alterations being a probable candidate among others [267]. Further studies are needed to help clarify the underlying biological mechanism of illness and to help determine which patients would benefit from which type of anti-psychotic treatment.