Oxytocin levels in individuals with schizophrenia are high in cerebrospinal fluid but low in serum: A systematic review and meta-analysis

Schizophrenia is a debilitating mental illness. Levels of oxytocin have been proposed as a biomarker of schizophrenia; however, the observed levels of oxytocin in individuals with schizophrenia have been inconsistent across studies. We performed a meta-analysis to evaluate oxytocin levels in plasma, serum and cerebrospinal fluid to see if there are statistically different concentrations between individuals with schizophrenia and the comparison group. The meta-analysis followed the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) statement. Following the inclusion and exclusion criteria, 14 studies were included in the meta-analysis. The quality of the study was evaluated by the Newcastle–Ottawa Scale (NOS). A random-effects model was performed using the Comprehensive Meta-analysis software with the standardized mean difference (SMD) and 95% confidence intervals (CIs). Serum oxytocin levels in individuals with schizophrenia were significantly lower than that in comparison group (SMD =  − 1.74, 95% CI =  − 3.22 to − 0.26, p = 0.02) but cerebrospinal fluid oxytocin levels in individuals with schizophrenia were significantly higher than those in the comparison group (SMD = 0.55, 95% CI = 0.05 to 1.04, p = 0.03). Our results suggest that oxytocin levels in cerebrospinal fluid are increased in individuals with schizophrenia but decreased in serum. Therefore, the oxytocin system dysregulation may play a role in the pathophysiology of schizophrenia and it should be measured in more populations for a possible implementation as a biomarker of schizophrenia.


Introduction
There are numerous environmental, biological and genetic factors that interact and contribute to the vulnerability or resilience to develop schizophrenia (SCZ). Schizophrenia is a devastating mental illness; it is highly heritable, chronic, severe and disabling. Schizophrenia is characterized by positive, negative and cognitive symptoms. Positive symptoms consist of psychotic symptoms such as hallucinations, delusions that are often paranoid, as well as disorganized speech and behavior. Negative symptoms include flattened affect, poverty of speech, loss of a sense of pleasure and loss of interests and drive; while cognitive symptoms comprise deficits in attention, working memory, and in executive functions (Owen et al., 2016;Perkovic et al., 2017).
Onset symptoms of schizophrenia typically occur in early adult life and occur earlier in men than in women (Nadeem et al., 2004;Schultz et al., 2007). The etiology of SCZ is multifactorial and reflects an interaction between genetic and environmental contributors. Some risk factors include family history, social isolation, urbanicity, pregnancy and birth complications, acute life events, developmental difficulties, central nervous system infections in childhood and substance abuse (Stilo and Murray, 2019;Wahlberg et al., 2004). New antipsychotic medications have been introduced for the schizophrenia treatment; however, the management of this disease remains a substantial clinical challenge. Patients should be treated early and optimally with antipsychotic agents, however, sometimes patient does not respond to or cannot tolerate standard therapy, therefore, combination therapy is often required. For patients with inadequate response in target psychotic or schizophrenia symptoms, adding a second antipsychotic medication has yielded largely negative results based on randomized trials (Ng et al., 2021).
Nowadays, biomarkers and indicators of schizophrenia diagnosis or the severity of the disease are highly necessary. Biomarkers that can be easily available are body fluids that could help in the (1) diagnostic, (2) prognostic, or (3) prediction of individual response to a particular therapy (theranostics) (Weickert et al., 2013). Many biomarkers have been proposed in schizophrenia (Liu et al., 2021), including oxytocin.
Oxytocin (Oxt) is a neuropeptide synthesized primarily in neurons of the hypothalamic supraoptic and paraventricular nuclei. The Oxt system is involved in the neuromodulation of social behaviors, affiliative behaviors and aggression. Altered levels of Oxt have been reported in individuals with SCZ; therefore, the Oxt system could be linked to the pathophysiology of schizophrenia (Aydın et al., 2019;Goldman et al., 2008;Rubin et al., 2010).
Up to today, the mechanism concerning the effect of oxytocin as a risk factor for developing SCZ is unknown. However, many association studies have shown that individuals with schizophrenia have decreased levels of Oxt. Therefore, the aim of this systematic review and metaanalysis of case-control studies was to evaluate the effect Oxt levels on individuals with schizophrenia and to explore differences in Oxt levels between individuals with SCZ and healthy subjects. Additionally, we examined effects of potential confounding factors that may have an impact on these relationships.

Data sources and search strategy
The electronic databases PubMed, Scopus and EBSCO were searched for articles. The keywords used in the databases were: "oxytocin level" AND "schizophrenia". More specifically, the search details specified in PubMed were: ("oxytocin level" [MeSH Terms] OR "oxytocin level" [All Fields] OR "plasma oxytocin" [All Fields] OR "serum oxytocin" [All Fields] OR "CSF oxytocin" [All Fields]) AND ("schizophrenia" [MeSH Terms] OR "schizophrenia" [All Fields] OR "schizophrenia spectrum disorders" [All Fields]. The reference lists of the retrieved articles were then reviewed to identify additional studies. The protocol for this systematic review and meta-analysis was registered in PROSPERO (CRD42020221590) and can be accessed on: (https:// www. crd. york. ac. uk/ PROSP ERO/ displ ay_ record. php? Recor dID= 221590). The meta-analysis was conducted following the Preferred Reporting Items for Systematic Reviews and Metaanalyses (PRISMA) guidelines. We restricted the search to studies in humans published in English.

Study selection
In this systematic review we included studies that evaluated oxytocin concentrations in individuals with schizophrenia; these studies had to fulfill the following criteria: (1) studied individuals diagnosed with schizophrenia spectrum disorders; (2) the study design had to be a case-control study that reported values in mean and standard deviation of oxytocin levels or had enough data to calculate these values; (3) compared the levels of oxytocin between two groups of individuals, those with SCZ and a healthy control group, and (4) publications in peer-reviewed journals.
The following type of studies were excluded: (1) reviews, case reports, case-only studies, animal studies, and simple commentaries; (2) overlapping publications and (3) publications lacking measures of Oxt levels.
The selection process was independently completed by our research team and any discrepancy or disagreement was resolved by discussion until consensus was reached among the reviewers.

Data extraction, assessment of quality and risk of bias
Two authors (YHD and TBGC) extracted data using a structured form not blinded to study results, author names or institutions. This form included study data and participants description. The methodological quality and risk of bias of included studies was evaluated by two investigators (YHD and TBGC). Disagreements were resolved by a third reviewer (CATZ). The Newcastle-Ottawa (NOS) quality scale consists of three parts: selection, comparability and exposure assessment. High-quality response earns a point, totaling up to nine points (the comparability question earns up to two points). The total is calculated by adding the points of all categories. In our meta-analysis, we considered 6 points or higher as a good-quality study. The risk of bias assessment was performed using the ROBINS-I scale that evaluates the risk of bias in "Low risk", "Moderate risk", "High risk", "Critical risk" in seven domains of ROBINS-I criteria.

Statistical analysis
The Comprehensive Meta-analysis version 2 (Biostat, Englewood, NJ, USA) software was used to deal with quantitative data. Standardized mean differences (SMD) were calculated and represented the differences in the mean oxytocin levels between individuals with SCZ and healthy controls; with 95% confidence intervals (CI) and were illustrated by forest plots.
Heterogeneity was evaluated by Q-test and I 2 -test. Generally, if p < 0.05 (Q-test) or I 2 -test > 50%, when heterogeneity was thought to exist, the random-effect models was used. Heterogeneity is the systematic difference between the results of the studies that cannot be attributed simply to chance. We analyzed subgroups of studies to examine the source of potential heterogeneity based on the region, analytical methods and gender.
Potential publication bias was explored by using funnel plot and Egger line regression test (Egger's test). Sensitivity analyses were conducted to evaluate the stability of pooled results. All tests were two-sided and a p value of < 0.05 was considered statistically significant.
The characteristics of the seventeen selected studies are presented in Table 1. All the studies were published between 1985 and 2019. Eight studies were conducted in the United States, three in Germany, two in Turkey, one in Japan, one in Hungary, one in Portugal and one in China.  Glovinsky, D. (Glovinsky et al., 1994) 1994 USA The sample sizes ranged from 15 to 60 in individuals with schizophrenia and from 7 to 66 in controls. Enzyme-linked immunosorbent assay (ELISA) and radioimmunoassay (RIA) methods were implemented for the detection of Oxt levels. We identified three different biomaterials used for Oxt assays: plasma, serum and cerebrospinal fluid (CSF). Therefore, three meta-analysis were carried out, additionally, subgroup analyses were performed for each one of them based on data availability.
Study-specific quality is summarized in Fig. 2. All studies were awarded ≥ 7 points and were defined as goodquality studies according to NOS scale. The risk of bias was evaluated using the ROBINS-I tool. It was found that overall, a low risk of bias was obtained in 65% of the included studies (Fig. 3).

Meta-analysis of serum oxytocin levels in SCZ
Second, we evaluated oxytocin levels in serum and found that these values were lower in cases than in the control group. This meta-analysis included five studies with 209 individuals with schizophrenia and 221 individuals in the comparison group (Guzel et al., 2018;Liu et al., 2019;Rubin et al., 2013Rubin et al., , 2010Rubin et al., , 2018. Table 1. The meta-analysis showed that serum oxytocin levels in SCZ individuals were significantly lower than in the comparison group (SMD = − 1.74, 95% CI = − 3.22 to − 0.26 Pg/mL, p = 0.02, I 2 = 77 p(Q) < 0.01) ( Table 2 and Fig. 3).
The results of the subgroup analysis are shown in Table 2. The method of evaluation (ELISA) showed a significant association of decreased oxytocin levels in individuals with schizophrenia. Other subgroups did not show statistical differences (Table 2).

Meta-analysis of CSF oxytocin levels in SCZ
Finally, a third meta-analysis was conducted using three studies (Beckmann et al., 1985;Glovinsky et al., 1994;Sasayama et al., 2012) comprising 95 individuals with schizophrenia and 52 individuals in the comparison group (Table 1). The results of the meta-analysis indicated that cerebro-spinal fluid oxytocin levels in individuals with schizophrenia were significantly higher than in individuals of the comparison group (SMD = 0.55, 95% CI = 0.05 to 1.04, p = 0.03, I 2 = 50%, p(Q) = 0.03).

Sensitivity analysis and publication bias
The sensitivity analysis indicated that the exclusion of none of the primary studies influenced the pooled estimates. For publication bias, the funnel plot was symmetric (Fig. 4). Moreover, Egger's tests did not reveal significant evidence of publication bias among the included studies (Table 2 and Fig. 5).

Discussion
A biomarker refers to a molecular change in body tissues and fluids that can be used as a clinical indicator of a disease; in this case, the biomarker will allow better diagnostic procedures and therapeutic strategies in individuals with schizophrenia (Perkovic et al., 2017;Vargas, 2014).
The use of biomarkers that can be measured or determined from easily available body fluids such as serum or cerebrospinal fluid is highly recommended. As oxytocin acts centrally within the brain to control behavior, it has been suggested as biomarker of mental disorders. Oxytocin is a highly conserved neuropeptide produced in the hypothalamus, it has modulatory effects on a variety of social and non-social behaviors such as affiliation, stress, memory and learning (Churchland and Winkielman, 2012;Jones et al., 2017).
In the meta-analysis, we included fourteen studies that comprised 518 cases and 409 controls. Interestingly, we found that Oxt levels were lower in serum samples of individuals with SCZ but higher in CSF samples of individuals with SCZ.
A possible explanation for the increased Oxt levels in patients with schizophrenia could be that the release of Oxt contributes to the modulation and maintenance of other hormones such as cortisol, which could increase the response to the stressful stimuli (Donadon et al., 2018;Liu et al., 2019). Then, stressful experiences might alter the functioning of the suprachiasmatic nucleus altering the release of endogenous Oxt (Donadon et al., 2018;Liu et al., 2019). On the other hand, it has been proposed that low levels of serum oxytocin in individuals with acute schizophrenia are associated with the functionality of the patient, the severity of the disease and with the severity of the positive symptoms (Guzel et al., 2018). This could explain the contrasting observations regarding oxytocin concentrations in the central system (high in schizophrenia) and in blood (low in schizophrenia); nevertheless, more studies are necessary in order to confirm and understand what causes those differences. As we observed in our meta-analysis, oxytocin concentrations in cerebrospinal fluid indicated that were significantly higher in individuals with schizophrenia than in controls. Beckmann et al. (Beckmann et al., 1985) reported increased concentrations of oxytocin in schizophrenic patients, and they were higher in those receiving neuroleptic treatment. In addition, oxytocin concentrations increased after three weeks of neuroleptic treatment. However, another study did not fully support these results (Sasayama et al., 2012); therefore, more data are required to fully explore this relationship.
In our meta-analysis we evaluated several confounding factors in order to identify those that could impact the relationship between Oxt levels and SCZ. For instance, different analytical methods to measure oxytocin concentrations were used among the studies. Subgroup analyses revealed that Oxt levels were significantly lower in patients with SCZ compared to controls when the concentrations were measured through ELISA; suggesting that ELISA assays may have been more sensitive in detecting Oxt levels. All in all, we need more studies to confirm the effect of Oxt levels on individuals with schizophrenia, using unit methods in order to improve the accuracy and reliability of the results.
The most frequently used biomarkers in schizophrenia are those associated with the neuroimmune and neuroendocrine system, metabolism, different neurotransmitter systems and neurotrophic factors. Nevertheless, to date none of the biomarkers has achieved levels of precision that would allow it to be used in the diagnosis and treatment of patients with schizophrenia. However, blood based biomarkers as oxytocin has high specificity and sensitivity in distinguishing SCZ patients from healthy controls, moreover, cerebrospinal fluid Oxt allow the analysis of biochemical changes that reflect pharmacological activity and this may be clinical accessibility in the living patient (Harris et al., 2012;Wan et al., 2012).
Our study had some advantages. First, the sensitivity analyses indicated that the removal of individual studies did not alter the final results, which increased the robustness of our conclusions. Second, no significant publication bias was detected, suggesting that the results are also reliable. On the other hand, our meta-analysis also presents some limitations. Firstly, the number of studies included was small and stratified analyses might have lower power to identify potential confounding factors. Secondly, there was significant heterogeneity in this analysis, which may have affected the precision of the overall results. Thirdly, we observed opposing patterns of findings observed for serum vs. CSF. This is important, because according the source and origin of oxytocin, its mechanism can be different. Finally, the present meta-analysis focused only on studies published in English, the ones reported in other languages might have changed the present results.
The results of this meta-analysis could be potentially very relevant for patients and health care providers due to that the oxytocin system is regarded as being of relevance for social interaction. Several researching suggest that the Oxt system may play a role in regulating the expression of schizophrenia-spectrum disorders. Studies have identified that Oxt levels are associated with more severe symptoms in all three domains of SCZ. Respect to cognition, Oxt levels have been associated with enhanced social cognition including more prosocial behavior and superior emotion recognition Shilling and Feifel, 2016). Evidence for a relationship between Oxt levels and symptom suggests that Oxt may contribute to the etiology of SCZ, therefore, targeting the central Oxt system may yield novel treatments to address these symptoms, moreover, can contribute to diagnostic and prognostic to this disease improving the quality of life in patients with schizophrenia and decrease health costs worldwide (Ng et al., 2021).

Conclusions
In summary, the present meta-analysis found increased levels oxytocin in CSF but decreased levels of oxytocin in serum. These results suggests that the oxytocin system dysregulation may play a role in the pathophysiology of schizophrenia. Oxt could be an interesting candidate in the search for biomarkers that could improve diagnosis and evaluation of patients with schizophrenia. Nonetheless, more well-designed and large-scale studies are needed to