Inflammation and cognitive performance in first-episode schizophrenia spectrum disorders: The moderating effects of childhood trauma

In this study, we aimed to determine whether childhood trauma moderated the relationship between inflammation and cognitive functioning in persons with first-episode schizophrenia spectrum disorders (SSDs). We included data from 92 individuals who participated in the nationwide Early-Stage Schizophrenia Outcome study. These individuals completed the Childhood Trauma Questionnaire, provided a fasting blood sample for high-sensitivity C-reactive protein analysis, and underwent extensive neuropsychological testing. The intervening effects of age, sex, education, smoking status, and body mass index were controlled. Results indicated that childhood trauma levels significantly moderated the relationship between inflammation and four cognitive domains: speed of processing, working memory, visual memory, and verbal memory. Inflammation also predicted verbal memory scores irrespective of childhood trauma levels or the covariates. Upon further exploration, the significant moderation effects appeared to be primarily driven by males. In conclusion, a history of childhood trauma may be an important determinant in evaluating how inflammation relates to the cognitive performance of people with first-episode SSDs, particularly in speed of processing, working memory, visual memory, and verbal memory. We recommend that future researchers examining the effect of inflammation on cognitive functioning in SSDs include trauma as a moderating variable in their models and further examine additional moderating effects of sex.


Introduction
The contribution of inflammation to schizophrenia spectrum disorders (SSDs) has become a major area of study.This is justified by the role that inflammation can play in the pathogenesis of these disorders: chronic or elevated levels of inflammation have been linked to the emergence and course of SSDs, as well as increased mortality in affected individuals (Fond et al., 2018;Misiak et al., 2018).Inflammation has also been associated with the heterogeneity of symptoms observed in SSDs, particularly cognitive deficits, which vary widely (Misiak et al., 2018;Rodriguez et al., 2019).
The immune system and cognition are intertwined.In the central nervous system (CNS), immune cells are largely represented by microglia, which perform phagocytotic functions similar to cells in the macrophage/monocyte lineage.When microglia detect an inflammatory insult, they release pro-inflammatory molecules that help coordinate a whole-body response.This can result in short-term behavioural changes such as fatigue and social withdrawal (Bower and Kuhlman, 2023), as well as longer-term changes in neurogenesis that may affect the development and functioning of key brain structures such as the hippocampus (Chesnokova et al., 2016;Nettis et al., 2020).Microglia also support normal neural function and participate in neuronal and synaptic pruning during the prenatal, postnatal, and adolescent periods.Immune activation during these sensitive periods can derail microglia from fulfilling these functions, which can disrupt cognitive development (Bower and Kuhlman, 2023).
Inflammation can be estimated from several blood-based markers, including the C-reactive protein (CRP) or cytokines such as the interleukin-6 and tumour necrosis factor-α.A recent meta-analysis suggested that in SSDs, elevated inflammatory marker levels are weakly associated with deficits in global cognition and visual memory (Morrens et al., 2022).CRP showed slightly larger effect sizes and related to greater impairments in general cognition, visual memory, verbal memory, and attention.An earlier meta-analysis reported significant relationships between CRP and additional cognitive domains but again, the effect sizes were small (Bora, 2019).The authors concluded that further intermediating and confounding factors need examining.
Recently, childhood trauma exposure has emerged as one such factor.Childhood trauma, including forms of physical, sexual, and emotional abuse and neglect (Bernstein et al., 2003), is highly prevalent in SSDs and possibly related to their emergence and course (Pec et al., 2022;Van Dam et al., 2015;Vila-Badia et al., 2021).Childhood trauma exposure may also affect cognitive functioning later in life.Systemic reviews and meta-analytic investigations in SSDs found significant, negative associations between childhood trauma and both general and domain-specific cognitive abilities (Aas et al., 2014;Vargas et al., 2019).However, except for working memory, the effect sizes associated with these relationships tend to be small (Vargas et al., 2019).
A recent review outlined two ways in which trauma could affect inflammation and its relationship with cognition (Bower and Kuhlman, 2023).One possibility is that trauma contributes to chronically increased inflammation, which indirectly accelerates normal cognitive ageing (Danese and J Lewis, 2017;Morrens et al., 2022;Nettis et al., 2020).However, studies show inconsistent elevations of CRP and other inflammatory markers in childhood trauma survivors (Brown et al., 2021), making a mediation process less likely.Indeed, cytokine levels do not appear to mediate the relationship between childhood trauma and general cognitive ability in individuals with SSDs, although mediation effects were observed for emotion recognition tasks (King et al., 2021).
Another possibility is that childhood trauma moderates the relationship between inflammation and cognition (Bower and Kuhlman, 2023).In this view, trauma sensitizes the individual to further stressors, thereby increasing their susceptibility to inflammation-related neural or behavioural changes (Nusslock and Miller, 2016).Corresponding moderation effects have been observed in psychiatrically healthy subjects, where increased levels of trauma are associated with an augmented link between inflammation and symptoms of depression or cognitive impairment (Kuhlman et al., 2020;Manigault et al., 2021).Presently, it is unknown whether similar effects occur for cognitive functioning in SSDs.
The interrelationships between childhood trauma, inflammation, and cognition could also depend on sex.Females with SSDs report greater exposure to childhood trauma than males (Vila-Badia et al., 2021), suggesting that they might exhibit a stronger relationship between inflammation and cognitive functioning if it indeed depends on trauma.Females generally mount stronger immune responses, which can make them more susceptible to certain diseases, including depression or neurodegenerative disorders (Hanamsagar and Bilbo, 2016;Klein and Flanagan, 2016).Conversely, males show greater proinflammatory responses during childhood, which could heighten the impact of stress on their neural development at that time (Hanamsagar and Bilbo, 2016;Klein and Flanagan, 2016).Given their generally weaker immune systems, males may also be more affected by certain inflammatory insults (Klein and Flanagan, 2016).
The present study aims to investigate the moderating effects of childhood trauma exposure on the association between inflammation and cognitive functioning in SSDs, with a secondary focus on the potential contribution of sex to any moderating effects observed.

Study design and participants
The data for this study were collected from October 2016 to November 2021 as part of the 'Early-Stage Schizophrenia Outcome' (ESO) study coordinated by the National Institute of Mental Health (NIMH-CZ).The ESO study comprises the construction of a multimodal and nationwide database of individuals with first-episode SSDs, and psychiatrically healthy controls.It was approved by the NIMH-CZ Ethics Board (protocol number: 127/17).Participants were over 18 years old and signed an informed consent sheet.
The inclusion/exclusion criteria for this study were met by 92 individuals from the ESO database (Fig. 1).Three participants were not receiving any medication at the time of their participation.The remaining 89 were using antipsychotic medication in the form of monotherapy (n = 54, 60.7 %) or polytherapy (n = 35, 38.0 %).Atypical antipsychotics were used in all cases.There were also 6 instances of typical antipsychotic use, always as part of polytherapy.See Table 2 for other sample characteristics.

Childhood trauma
Childhood trauma exposure was assessed using a previously validated Czech version (Kaščáková, 2018) of the brief form of the CTQ (Bernstein et al., 2003).The CTQ measures five types of childhood trauma: emotional abuse (e.g., verbal assault and humiliation), physical abuse (e.g., visible injuries and hospital visits), emotional neglect (e.g., lack of care, love, and support), physical neglect (e.g., lack of food and safety), and sexual abuse (e.g., sexual contact and coercion).Participants rate the applicability of individual items while they were 'growing up' using 5-point Likert scales ranging from '1-Never True' to '5-Very Often True'.The CTQ includes 25 clinical items and three validity items.We only used the 25 clinical items, which we summed to obtain the total and per-scale scores.Missing values (four participants omitted an item) were median-replaced within childhood trauma type.Cronbach's alpha across the 25 items was satisfactory, α = 0.871.

Inflammation
Inflammation was indexed by high-sensitivity CRP.Fasting blood samples were collected between 7:30 and 8:00 am into 4.9 ml tubes containing a clot activator/gel separator.Samples were inverted 6-8 times, placed in a rack, and stored in a shipment box kept at ambient temperatures.Within 4 h, the samples were shipped to an external laboratory (AeskuLab k.s., now Unilabs Diagnostics k.s.), where they were assayed the same day using latex-enhanced immunoturbidimetry.Samples collected between October 2016 and June 2019 (n = 54, 58.7 %) were assayed using the Siemens ADVIA 1800 Clinical Chemistry System following manufacturer instructions.Intra-assay and inter-assay coefficients of variation typically range from 1.8 % to 3.3 % and 1.9 % to 3.5 %, respectively.The detection limits are 0.12 and 164 mg/l.Samples collected from July 2019 onwards (n = 38, 41.3 %) were assayed using the Atellica CH Analyzer from the same manufacturer.Intra-assay and inter-assay coefficients range from 0.7 % to 1.3 % and 0.4 % to 0.8 %, respectively.The detection limits are 0.5 and 156.0 mg/l.Values below the lower detection limits (n = 35, 38.0 %) were replaced with values corresponding to those limits.No values were over the upper detection limits (observed maximum = 16.05 mg/l).Due to high skewness, CRP values were log-transformed prior to analyses.

Cognitive functioning
Cognitive functioning was assessed using a battery of validated neuropsychological tests.The assessments were typically scheduled for the early afternoon hours and took 150 min to complete, with a break mid-way.Individual test scores were attributed to one of six cognitive domains: speed of processing (SPOP), working memory (WM), visual memory (VISM), verbal memory (VERBM), attention (ATT), and executive function/ abstraction (EXEC/ABSTR) (Table 1).To obtain the cognitive domain scores, we z-transformed the test scores using the Ms and SDs of 102 healthy controls from the ESO database, checked them for extreme values, and computed their average.Averages were not computed when two or more test scores were missing, resulting in two participants with missing VISM scores and 20 participants with missing ATT scores.For ATT, the missing scores were primarily due to invalid Continuous Performance Test (CPT) performance (Conners, 2000).

Other measures
Clinical characteristics, basic demographics, smoking status, and BMI were used to describe the sample or to function as covariates in the statistical models.Clinical characteristics were: diagnosis, age of onset, duration of antipsychotic treatment (DAT), duration of untreated psychosis (DUP), and medication (type and dose, converted to chlorpromazine equivalents (CPZE) in accordance with Gardner et al. (2010)).The Positive and Negative Syndrome Scale (PANSS) (Kay et al., 1987) and the Global Assessment of Functioning (GAF) (American Psychiatric Association, 1994) were also administered.Demographic data included age (to three decimal points), sex assigned at birth (coded as 0 for males, 1 for females), and years of education.Smoking status was coded as true when the participants reported smoking at least one cigarette daily.BMI was computed as kg/m 2 from the participants' height and weight.

Statistical analyses
Statistical analyses were conducted in SPSS v. 28.0.The significance level was set at p < .05.All tests were two-tailed.

Childhood trauma and sample characteristics
The sample was described with respect to the reported types and severity levels of childhood trauma using standard guidelines (Bernstein and Fink, 1998).Descriptive statistics were generated for all sample characteristics, and the associations between them and the total CTQ score were explored using correlation (Pearson's or Spearman's) or an independent-samples t-test (Student's or Welch's).In the case of significant associations, the relationships between the given variable and per-scale CTQ scores were examined, with Holms-Bonferroni corrections applied.

Moderation analyses
The six cognitive domain scores were individually regressed on the CRP variable (Model 1).Moderation models were then constructed using Hayes's PROCESS macro (Hayes, 2022) (Model 2).Predictors were mean-centred: CRP was the focal predictor and the CTQ score was the moderator.Covariates were: education, age, sex, smoking status, and BMI.They were selected based on their previously described relationships to childhood trauma, inflammation, and cognitive functioning (Brown et al., 2021;Fond et al., 2018;Singh and Chaudhuri, 2014).Bootstrap resampling (5000 samples, seed = 172) was used to estimate 95 % confidence intervals.To aid the interpretation of significant interaction effects, simple slopes were produced at the 16th, 50th, and 84th percentile of the moderator variable (hereafter referred to as the lower, moderate, and higher level, respectively), and the range of significance was computed using the Johnson-Neyman technique.The significance of the interaction effects was also explored separately in males and females.Note that these exploratory analyses were relatively underpowered in females, with mild negative auto-correlation also observed for the VISM moderation model in this subgroup (Durbin-Watson value = 2.654).

Childhood trauma and sample characteristics
Total CTQ scores ranged from 25 to 73 points (M = 37.38, SD = 9.37).The most common types of trauma were emotional neglect, followed by emotional abuse, and physical neglect (Fig. 2a).Childhood trauma severity levels were generally minimal to moderate (Fig. 2b).
Of all sample characteristics, only sex was related to the total CTQ score, with females scoring higher than males (d = − 0.529) (Table 2).Females tended to report higher levels of all trauma types other than physical abuse (Fig. 2a), but these differences did not withstand corrections for multiple testing.

Moderation analyses
Significant effects of CRP, CTQ, or their interaction were observed for four domain scores: SPOP, WM, VISM, and VERBM (Table 3).Full results of the exploratory analyses in males and females are available in supplementary material (Tables S1-4).
For SPOP, significance was observed for the moderation model (Model 2a), which explained 22.1 % variance in the domain score.While neither predictor variable was independently associated with the score, their interaction was, and the addition of this interaction term significantly improved model fit, ΔR 2 = 0.085.Simple slopes (Fig. 3a) showed that a crossover interaction likely explained this.A significant relationship between CRP and SPOP was only observed at higher levels of trauma, B = − 0. A similar pattern emerged for WM and VISM: the moderation models (Models 2b and 2c) were significant, explained a substantial percentage of variance in the domain scores (25.3 % and 23.0 %, respectively), and benefitted from the addition of the interaction terms (ΔR 2 = 0.109 and 0.090, respectively).Signs of cross-over interactions were again observed (Figs.3b-c).For WM, CRP predicted the domain score at higher levels of childhood trauma, B = − 0.854 [95 % CIs: − 1.364-− 0.344], SE = 0.257, t = − 3.329, p = .001,and significance was typically achieved at CTQ scores of 1.696 or higher (33.7 % of the sample).The moderation effect was only significant in males, B = − 0.093 [95 % boot CIs: VERBM was the only domain score that was significantly predicted by CRP in the base model (Model 1d), with 6.0 % of variance explained.CRP also maintained significance in the moderation model (Model 2d), which accounted for 23.9 % of the variance observed in VERBM, and still benefitted from the inclusion of the interaction term, ΔR 2 = 0.038.The relationship between CRP and VERBM was significant at both moderate, B = − 0.513 [95 % CIs: − 0.913-− 0.114], SE = 0.201, t = − 2.557, p = .012,and higher levels of childhood trauma, B = − 1.159 [95 % CIs: − 1.828-− 0.490], SE = 0.336, t = − 3.446, p < .001(Fig. 3d).Judging from the Johnson-Neyman output, significant relationships between CRP and VERBM started to occur at CTQ total scores of − 5.289, which were obtained or surpassed by 60.9 % of the sample.After sexstratification, the moderation effects were non-significant, although the conditional effects maintained significance in males (Table S4).

Discussion
The major finding of this study is that childhood trauma significantly moderates the association between inflammation and cognitive functioning in first-episode SSDs.Specifically, elevated levels of highsensitivity CRP predicted worse performance in SPOP, WM, and VISM only in participants reporting higher childhood trauma levels on the CTQ (Bernstein et al., 2003).We also observed a more robust relationship between inflammation and VERBM, which was augmented by the presence of childhood trauma but not entirely dependent on it.Notably, the moderating effects of childhood trauma seemed stronger in males than females.These results support the idea that inflammation and childhood trauma play an interactive role in the development of cognitive deficits in SSDs, and open up avenues for further research.
While these findings are novel in the context of SSDs, they parallel research on different populations or symptom classes.For example, childhood trauma survivors exposed to further immune stimulation show an association between inflammation and the symptoms of depression or cognitive difficulty, which does not occur in individuals without childhood trauma (Kuhlman et al., 2020).A prospective study of female adolescents found that those who experienced childhood adversity and later transitioned to depression showed a concurrent increase in CRP and interleukin-6, and a subsequent "coupling" of these inflammatory markers and depression (Miller and Cole, 2012).The authors hypothesized that childhood adversity potentiates a phenotype characterized by the co-occurrence of depression and inflammation.Expanding on this idea, Nusslock and Miller (2016) proposed the neuroimmune network hypothesis.The hypothesis is based on the idea that childhood adversity sensitizes the CNS and cells in the monocyte/ macrophage lineage to further stressors.The exaggerated responsiveness of these systems generates increased bidirectional crosstalk between them, contributing to their coupling.While speculative, the central tenets of the hypothesis are grounded in theory.Notably, previous reports have shown that the priming of microglia and other sensitization processes-sometimes referred to as "kindling" -play a role in the symptomatology and pathogenesis of SSDs (Müller, 2018;Pec et al., 2022) and age-related cognitive decline (Sparkman and Johnson, 2008).
As a system becomes sensitized, it gradually learns to respond more strongly to lower-threshold stimuli until it reaches a state of constant activation despite the absence of a stimulus.Correspondingly, the neuroimmune network hypothesis predicts chronic low-grade inflammation (Nusslock and Miller, 2016).We did not observe elevated CRP levels in individuals with higher childhood trauma exposure, consistent with previous findings (Brown et al., 2021).While this suggests that our findings are indeed better explained by a moderation rather than a mediation process, it also indicates that the neuroimmune network hypothesis might be an incomplete explanation of our findings.The chronic elevations of inflammatory marker levels may be lower than we were able to detect.Inflammation may also increase in response to greater trauma severity than present in our sample, or only following certain trauma types, such as childhood abuse, as opposed to neglect (Brown et al., 2021).Sexual abuse may be an appropriate focus of future studies given its reported relationship with SSD symptoms, including cognitive deficits (Aas et al., 2012;Nettis et al., 2020).However, caution is necessary when examining the impact of individual trauma types, given the substantial likelihood of their overlap (Vargas et al., 2019).
Pending replication, our results have important treatment implications.Prior investigations into the efficacy of anti-inflammatory agents on cognition in SSDs have yielded inconsistent results (Çakici et al., 2019), and the lacking consideration of childhood trauma could explain this inconsistency.If so, individuals with childhood trauma might be the appropriate targets for anti-inflammatory therapies, from which others might not benefit.In addition to medications, psychosocial treatments should be offered.In randomized clinical trials, psychosocial treatments such as cognitive behavioural therapy are associated with lasting improvements in immune system function (Shields et al., 2020).The presence of cognitive deficits and other SSD symptoms can diminish the psychosocial functioning of affected individuals (Kalisova et al., 2023), further emphasizing the importance of scheduling such treatments.
In our sample, CRP coupled with four cognitive domain scores: SPOP, WM, VISM, and VERBM.This partially aligns with the results of the latest meta-analysis, which found a link between CRP and general cognition, VISM, VERBM, and ATT (Morrens et al., 2022).The ATT domain in this meta-analysis included scores from tests that we assigned to SPOP (e.g., Digit-symbol coding) or WM (e.g., Digit span).These domain composition differences might partially explain the differing results.Indeed, no relationship between inflammation and ATT was Note.CTQ = Childhood Trauma Questionnaire; DUP = duration of untreated psychosis; DAT = duration of continual antipsychotic treatment; CPZE = antipsychotic medication converted to chlorpromazine equivalents (two missing due to examiner error/ no dose data); PANSS = The Positive and Negative Syndrome Scale (one missing due to examiner error/ no general psychopathology data); GAF = Global Assessment of Functioning Scale; BMI = body mass index; CRP = (high-sensitivity) C-reactive protein.Diagnosis was coded as 0 for schizophrenia, 1 for acute and transient psychotic disorder.Sex was coded as 0 for males, 1 for females.Smoking status was coded as 0 for non-smokers, 1 for smokers (daily).
B. Keřková et al. found in a study which used a variant of the CPT test that we opted for (Dal Santo et al., 2020).However, we caution that a large number of invalid CPT assessments were noted, which could have affected our results.The partial discrepancy between our results and the metaanalytic findings by Morrens et al. (2022) is also likely explained by the aforementioned inclusion of childhood trauma in our statistical models.In fact, VERBM was the only domain score that was significantly predicted by CRP irrespective of trauma levels or other variables.This could be due to the effects of inflammation on the development and functioning of the hippocampus, which is critical for memory and learning (Chesnokova et al., 2016;Nettis et al., 2020;Ritchie et al., 2020).Childhood trauma exposure significantly differed by sex, being more prevalent in females.A higher prevalence of childhood trauma in

Table 3
The relationships between inflammation and cognitive functioning and their moderation by childhood trauma exposure.Note.SPOP = speed of processing; WM = working memory; VISM = visual memory; VERBM = verbal memory; CRP = (high-sensitivity) C-reactive protein (logtransformed, mean-centred); CTQ = Childhood Trauma Questionnaire (total score, mean-centred), BMI = body mass index.Sex was coded as 0 for males, 1 for females.Smoking status was coded as 0 for non-smokers, 1 for smokers (daily).females has been reported previously, although it is less clear whether some types of trauma are consistently more frequent in females than others (Vila-Badia et al., 2021).Despite their generally higher childhood trauma exposure, females appeared less affected by it in the relationship between inflammation and cognitive functioning.This was most evident in the "core cognitive domains" of SPOP and WM, which are thought to give rise to individual differences in general intelligence (Frischkorn et al., 2019;Schubert et al., 2017).We speculate that the higher interdependence between inflammation, childhood trauma, and functioning in these domains might be due to early-development sex-differences in immune signalling.Neonate males have more microglia than females and also manifest greater pro-inflammatory responses up until puberty (Hanamsagar and Bilbo, 2016;Klein and Flanagan, 2016).In preclinical studies, stress affects microglial function, neuronal development, and consequently behaviour primarily in males, which could explain sexdifferences in the prevalence of disorders that develop earlier in life, such as autism spectrum disorders and SSDs (Bower and Kuhlman, 2023;Hanamsagar and Bilbo, 2016).However, the absence of significant moderation effects in females could also be attributed to the relatively lower power of tests conducted in this subgroup.While we only observed marginally stronger relationships between inflammation and VISM scores in females exposed to higher trauma levels, others found female-specific associations between CRP and visual learning (Dal Santo et al., 2020).Overall, investigating sex-differences in the relationships between trauma, inflammation, and cognitive functioning holds promise for future research.Childhood trauma was independent of all other sample characteristics measured, suggesting that their intervening effects were likely low.Nonetheless, it may be beneficial to consider the contribution of additional variables in more powered samples.For instance, individuals with retest-confirmed first-episode schizophrenia generally show worse cognitive performance than those with a brief psychotic disorder (Ayesa-Arriola et al., 2016).Some antipsychotic medications, such as risperidone, can also have anti-inflammatory properties, although these might only transpire in more chronic patients (Patlola et al., 2023).
One of the limitations of this study is its cross-sectional design.As a result, our interpretations are speculative and require support from prospective studies, similar to those conducted by Miller and Cole (2012).The extent to which our findings can elucidate the development of cognitive deficits in SSDs is also limited by the present lack of a psychiatrically healthy control group.Blood samples were not collected at three separate time points, as is usually recommended to improve the reliability of measurement (Koenig et al., 2003).Additionally, blood samples and neuropsychological test data were collected up to 14 days apart, though CRP and cognitive performance tend to remain stable over short periods (Calamia et al., 2013;Danesh et al., 2004;Ockene et al., 2001).We encourage future researchers to extend our results using additional inflammation markers and childhood trauma measures, along with more standard neuropsychological test batteries.For example, while the CTQ is a widely-used and sensitive measure of childhood trauma that can thus provide good comparability and reliability of findings (Bernstein et al., 2003;Brown et al., 2021), it could be useful to trace childhood trauma exposure using measures that afford more detail (e.g., a semi-structured interview) or external validation (e.g., register data on treatments and therapies performed during childhood and adolescence).
Our results hold implications for contemporary research in the wake of the COVID-19 pandemic.Infection with the SARS-CoV-2 virus can result in long-term cognitive impairment (Alnefeesi et al., 2021;Guo et al., 2022;Ritchie et al., 2020;Tavares-Júnior et al., 2022).Several pathways can lead to this effect, including an increase in inflammation and subsequent hippocampal dysfunction (Alnefeesi et al., 2021;Ritchie et al., 2020).Correspondingly, persistent deficits in verbal memory are among the most frequently observed in recovered patients (Guo et al., 2022;Tavares-Júnior et al., 2022).Individuals with psychiatric disorders such as SSDs can be more likely to become infected with SARS-CoV-2, develop more severe disease, and have poorer outcomes (De Hert et al., 2022;Taquet et al., 2020 but see also Canal-Rivero et al., 2021).Due to psychopathology and environmental stressors from childhood or adult life, individuals with SSDs may also have a pre-existing vulnerability for inflammation-related cognitive dysfunction.Monitoring cognitive trajectories in this population could therefore be of great interest to researchers and clinicians alike.

Fig. 1 .
Fig. 1.Data were extracted from the ESO database, with a total of 92 participants meeting the inclusion/exclusion criteria for this study.Note.ESO = Early-Stage Schizophrenia Outcome study; CTQ = Childhood Trauma Questionnaire; CRP = (high-sensitivity) C-reactive protein.

Table 1
Neuropsychological tests and the domains they measured.

Table 2
Sample characteristics (n = 92) and their relationships with the total CTQ score.