Introduction
Psychotic disorders, including schizophrenia and related conditions, are marked by persistent cognitive impairments [Reference McCutcheon, Keefe and McGuire1], neurobiological dysregulation [Reference Rawani, Chan, Dursun and Baker2, Reference Martinez-Cengotitabengoa, MacDowell, Alberich, Diaz, Garcia-Bueno and Rodriguez-Jimenez3], and systemic alterations [Reference Pillinger, D’Ambrosio, McCutcheon and Howes4, Reference Palomino, Gonzalez-Pinto, Martinez-Cengotitabengoa, Ruiz de Azua, Alberich and Mosquera5], along with a profound disruption of personal wellbeing [Reference Fusar-Poli, Estrade, Stanghellini, Venables, Onwumere and Messas6, Reference Fusar-Poli, Estrade, Esposito, Rosfort, Basadonne and Mancini7]. Besides, cognitive dysfunction is a core feature of psychotic disorders [Reference McCutcheon, Keefe and McGuire1], often preceding the emergence of positive symptoms and persisting throughout the illness [Reference Catalan, Salazar de Pablo, Aymerich, Damiani, Sordi and Radua8–Reference Gonzalez-Ortega, Gonzalez-Pinto, Alberich, Echeburua, Bernardo and Cabrera10]. Deficits typically affect attention, working memory, executive functioning, processing speed, and social cognition [Reference Catalan, Salazar de Pablo, Aymerich, Damiani, Sordi and Radua8, Reference Gil-Berrozpe, Segura, Sanchez-Torres, Amoretti, Gine-Serven and Vieta11], with significant consequences for functional outcomes and quality of life [Reference Gonzalez-Pinto, Martinez-Cengotitabengoa, Arango, Baeza, Otero-Cuesta and Graell-Berna12].
A growing body of evidence points to the immune system – and specifically inflammation – as a key factor in the pathophysiology of psychosis [Reference Saether, Ueland, Haatveit, Maglanoc, Szabo and Djurovic13–Reference Fernandez-Sevillano, Gonzalez-Ortega, MacDowell, Zorrilla, Lopez and Courtet15], particularly during its early stages. Among the most studied inflammatory markers are interleukin-6 (IL-6) [Reference Baek, Kim, Kim, Ryu, Lee and Kim16] and tumour necrosis factor-alpha (TNF-α) [Reference Pantovic-Stefanovic, Velimirovic, Jurisic, Puric, Gostiljac and Dodic17–Reference Patlola, Donohoe and McKernan19], due to their central role in immune activation, neuroinflammation, and potential effects on brain function. IL-6 and TNF-α have been implicated in modulating neurocognitive performance [Reference Hope, Hoseth, Dieset, Morch, Aas and Aukrust20], through both direct and indirect pathways, including effects on synaptic plasticity, neurotransmitter systems (e.g., dopamine and glutamate) [Reference de Bartolomeis, Barone, Vellucci, Mazza, Austin, Iasevoli and Ciccarelli21], and oxidative stress [Reference Martinez-Cengotitabengoa, Mico, Arango, Castro-Fornieles, Graell and Paya22]. IL-6 is secreted by immune cells, endothelial cells, and adipose tissue in response to infection, injury, and stress [Reference Kerkis, da Silva and Araldi23], while TNF-α is mainly produced by macrophages and monocytes and plays a key role in immune response, apoptosis, and inflammatory signalling [Reference Parameswaran and Patial24]. Both cytokines can cross the blood–brain barrier and contribute to neuroinflammation, neuronal signalling, and cognitive impairment [Reference Yirmiya and Goshen25]. Elevated peripheral levels have been associated with reduced cortical thickness and brain volume in regions relevant to cognition such as attention, visual learning, and verbal fluency [Reference Patlola, Donohoe and McKernan19]. In parallel, negative correlations have been observed between inflammatory markers and cognitive performance in individuals with psychosis [Reference Hope, Hoseth, Dieset, Morch, Aas and Aukrust20, Reference Ribeiro-Santos, de Campos-Carli, Ferretjans, Teixeira-Carvalho, Martins-Filho, Teixeira and Salgado26].
Elevated IL-6 and TNF-α levels have also been consistently observed in individuals at clinical high risk for psychosis (CHR-P) [Reference Mondelli, Blackman, Kempton, Pollak, Iyegbe and Valmaggia27] and in first-episode psychosis (FEP) patients [Reference Fang, Zhang, Fan, Tang and Zhang28], suggesting their involvement in illness onset and progression. In CHR-P individuals, higher cytokine levels have also been associated with an increased risk of transition to psychosis [Reference Zhang, Chen, Wei, Tang, Xu and Cui29], supporting their value as potential biomarkers of disease evolution [Reference Aymerich, Pedruzo, Salazar de Pablo, Labad, McCutcheon and Pillinger30]. In patients with FEP of schizophrenia, higher levels of peripheral proinflammatory cytokines were associated with poorer performance in Theory of Mind tasks, highlighting a potential link between immune dysregulation and social cognition deficits early in the illness [Reference Baek, Kim, Kim, Ryu, Lee and Kim16].
In addition to its link with cognitive dysfunction, inflammation has also been associated with the severity of positive symptoms in early psychosis [Reference Li and Zeng31, Reference Liemburg, Nolte, Klein and Knegtering32]. These associations suggest that immune dysregulation may contribute not only to neurocognitive impairment but also to the clinical expression of psychotic symptoms. Given these findings, IL-6 and TNF-α are increasingly viewed as candidate biomarkers for both psychosis risk and cognitive deterioration. Targeting neuroinflammatory mechanisms may represent a promising avenue for early intervention and preventive strategies and could help deepen our understanding of the potential causal role of cytokines in the pathogenesis of psychosis [Reference Foley, Griffiths, Murray, Rogers, Corsi-Zuelli and Hickinbotham33]. However, the relationship between inflammation and neurocognition remains complex, and further research is needed to clarify these associations.
This study aims to investigate the relationship among IL-6, TNF-α, and neurocognitive performance in early psychosis. Specifically, we examine group differences in cytokine levels and cognitive functioning among healthy controls (HC), CHR-P, and FEP participants, and explore the associations between inflammatory markers and cognitive domains within this transdiagnostic sample.
Materials and methods
Study design and participants
This study was conducted within the framework of the Prebentziorako Gazte Programa (PREGAP), a longitudinal research initiative investigating individuals at CHR-P and those with FEP (founded by the Department of Health of the Basque Country for research and development projects in health – promotion of health research activity). Participants were recruited from a range of clinical settings, including emergency departments, inpatient units, outpatient clinics, and primary care. Recruitment sites included OSI Bilbao-Basurto (Basurto University Hospital), Lehenak Bilbao (the Bizkaia Mental Health Network), and the Psychiatry Department of Santiago Apostol Hospital in Vitoria-Gasteiz. Baseline assessments and scheduled follow-ups were conducted as part of the study protocol.
Inclusion and exclusion criteria
CHR-P individuals were identified based on the Comprehensive Assessment of At-Risk Mental States (CAARMS) criteria [Reference Yung, Yuen, McGorry, Phillips, Kelly and Dell’Olio34], ensuring that they met established criteria for being at clinical high risk for psychosis. FEP patients met DSM-5-TR criteria for diagnosis, and comorbidities were assessed using the MINI International Neuropsychiatric Interview (MINI) [Reference Sheehan, Lecrubier, Sheehan, Amorim, Janavs and Weiller35]. These patients had experienced psychotic symptoms for less than two years since onset, with diagnoses including schizophrenia, schizoaffective disorder, brief psychotic disorder, psychosis NOS, mood disorders with psychotic features, and substance-induced psychotic disorder. Exclusion criteria included: severe neurological conditions, intellectual disability, and major systemic inflammatory or autoimmune diseases that could confound biomarker analyses. HC were defined as individuals with no current or past psychiatric diagnosis, no first-degree family history of psychotic disorders, and no major neurological, inflammatory, or autoimmune conditions.
Clinical and cognitive assessments
Each participant underwent a comprehensive assessment covering multiple domains. Clinical data were collected using a comprehensive set of standardized assessments to evaluate symptom severity, functioning, and psychiatric history. Psychotic symptomatology was assessed with the Positive and Negative Syndrome Scale (PANSS) [Reference Kay, Fiszbein and Opler36] in FEP individuals, while individuals at CHR-P were classified using the CAARMS criteria [Reference Yung, Yuen, McGorry, Phillips, Kelly and Dell’Olio34]. Depressive symptoms were measured with the Calgary Depression Scale (CDS) [Reference Addington, Addington and Maticka-Tyndale37], and overall functioning was evaluated using the Social and Occupational Functioning Scale (SOFS) [Reference Birchwood, Smith, Cochrane, Wetton and Copestake38], and the Global Assessment of Functioning (GAF) [Reference Jones, Thornicroft, Coffey and Dunn39]. The Clinical Global Impression (CGI) [Reference Guy40] scale was included to provide a clinician-rated measure of illness severity. Additionally, childhood adversity was assessed using the Childhood Trauma Questionnaire (CTQ) [Reference Bernstein and Fink41].
Cognitive and social cognitive abilities were assessed using a battery of well-established neuropsychological tests. General cognitive functioning was evaluated with selected subscales of the Wechsler Adult Intelligence Scale-IV (WAIS-IV) [Reference Wechsler42]. Executive functioning and cognitive flexibility were measured using the Wisconsin Card Sorting Test (WCST) [Reference Heaton, Chelune, Talley, Kay and Curtiss43] and the Stroop Test [Reference Stroop44], while processing speed and cognitive flexibility were further assessed with the Trail Making Test A and B (TMT-A, TMT-B) [Reference Reitan45]. Verbal fluency was examined through the COWAT task [Reference Ross46], visuospatial memory and organization were assessed using the Rey Complex Figure Test [Reference Spreen and Benton47], and verbal learning and memory were measured with the Hopkins Verbal Learning Test (HVLT) [Reference Brandt48].
Social cognition was evaluated through multiple tasks assessing different domains. Facial emotion recognition was measured using the PERE Facial Recognition Task [Reference Gil-Sanz, Fernández-Modamio, Bengochea-Seco, Arrieta-Rodríguez, González-Fraile and Pérez-Fuentes49], while the Movie Assessment for Social Cognition (MASC) [Reference Dziobek, Fleck, Kalbe, Rogers, Hassenstab, Brand and Convit50] assessed mentalizing abilities and theory of mind.
Biomarker collection
A fasting blood sample was collected between 8:00 and 10:00 AM following overnight fasting, in order to minimize potential circadian variation. Samples were used for inflammatory marker analysis, including IL-6 and TNF-α, and broader biometric parameters such as cardiometabolic and hormonal markers (e.g., prolactin, lipid profile, and glucose metabolism). Samples were stored and processed at Laboratory Medicine Department of Basurto University Hospital.
Statistical analysis
Baseline differences among CHR-P, FEP, and HC were analysed using ANOVA or Kruskal-Wallis tests for continuous variables, and Chi-square tests for categorical variables. Pearson correlation analyses were first conducted to examine the bivariate associations among IL-6, TNF-α, and neurocognitive variables. Given the risk of inflated type I error due to multiple comparisons, these correlations were not interpreted independently. Instead, only cognitive outcomes showing nominally significant associations (p < 0.05) with either cytokine were entered into multiple linear regression models. Similar approaches have been employed in previous research investigating inflammation and cognition in psychosis [Reference Saether, Ueland, Haatveit, Vaskinn, Barthel Flaaten and Mohn51].
Linear regression models were used to assess the association between cytokine levels and cognitive performance, adjusting for potential confounders including age, sex, IQ, psychosis risk group, and symptom severity (positive symptom z-score). The models were estimated using ordinary least squares regression, and assumptions (normality, linearity, and homoscedasticity of residuals) were verified. Model fit was evaluated using the F-statistic, with significance set at p < 0.05.
To minimize the risk of overfitting, we restricted the number of predictors per model and ensured an adequate participant-to-variable ratio, following established guidelines [Reference Green52]. Only a small set of theoretically relevant covariates was included, and dependent variables were selected based on prior bivariate associations. This strategy was designed to increase model robustness and reduce the likelihood of spurious associations. To control for multiple testing, we applied the Benjamini–Hochberg False Discovery Rate (FDR) [Reference Benjamini and Hochberg53] correction to the significant associations identified in the regression models.
All analyses were conducted using R software (version 2024.04.1 + 748) [54], and results are reported with standardized coefficients, confidence intervals, and effect sizes where applicable.
Ethical considerations
The study was approved by the Ethics Committee of the Basque Country, and all participants provided written informed consent prior to enrolment.
Results
Socio-demographic and clinical characteristics of the sample
The demographic and clinical characteristics of the sample are summarized in Table 1. The study included 33 HC, 35 CHR-P, and 39 FEP. CHR-P group (mean = 22.4 years; SD = 5.55) was younger than HC (mean = 27.5 years; SD = 3.77) and FEP subjects (27.9 years; SD = 9.27) (p < 0.05). IQ scores showed significantly a declining trend across groups, with HC having the highest mean IQ (97.4, SD = 17.3), followed by CHR-P (96.7, SD = 19.8), and FEP presenting the lowest IQ (85.1, SD = 18.6). In terms of sex distribution, no significant differences were found between groups (χ2 = 4.01, p = 0.13), although descriptively, the FEP group had a higher proportion of male participants (64.1%) compared to CHR-P (42.9%) and HC (45.5%). Regarding ethnicity, most participants in all groups identified as Caucasian, with the highest proportion in HC (93.9%), followed by CHR-P (82.9%) and FEP (71.8%). The proportion of Latin participants increased across groups, from 6.1% in HC to 14.2% in CHR-P and 23% in FEP. Arab and other ethnicities were less represented in the sample.
Table 1. Socio-demographic characteristics of the sample
* Significant difference among HC, FEP, and CHR-P.
** Significant difference among HC, CHR-P, and FEP.
*** Significant difference between FEP and CHR-P.
Regarding employment status, the HC group had the highest proportion of employed individuals (60.61%), compared to 17.14% in CHR and 23.08% in FEP. The CHR-P and FEP groups had a higher proportion of students (48.57 and 35.90%, respectively) compared to HC (27.27%). Notably, temporary work disability was reported exclusively in CHR-P (14.29%) and FEP (23.08%), while no participants in the HC group reported disability.
Regarding IL-6 levels, individuals in the FEP group showed the highest mean concentration (mean = 3.51, SD = 4.07), followed by healthy controls (HC) (mean = 2.56, SD = 0.98) and the CHR-P group (mean = 2.34, SD = 1.29). However, these differences were not statistically significant. For TNF-α, the mean level was highest in the FEP group (mean = 7.94, SD = 2.64), followed by HC (mean = 7.51, SD = 1.54) and CHR-P (mean = 6.62, SD = 1.28). Only the difference between FEP and CHR-P was statistically significant (p = 0.0251).
In the FEP group, DSM 5-TR schizophrenia was the most common diagnosis, accounting for 25.6%. Schizoaffective disorder was present in 12.8%, while brief psychotic episodes and psychosis NOS were diagnosed in 10.3%. Bipolar disorder and psychosis induced by substances represented 5.1 and 2.6%, respectively. Delusional disorder and PTSD were less frequent, each comprising 2.6% of the sample.
Within the CHR-P group, most participants, 71.4%, were classified with attenuated psychosis syndrome (APS), followed by BLIPS (brief limited intermittent psychotic symptoms) and APS & BLIPS, which each accounted for 11.4% of cases. A small proportion was categorized as having genetic risk and deterioration (GRD) or GRD & APS, at 2.9% each.
The comorbidity in CHR-P group was as follows: the most prevalent comorbid diagnosis was affective disorders, at 23.8%, followed by anxiety disorders, OCD, depression, and borderline personality disorder were each present in 14.27% of cases. ADHD, adjustment disorder, polysubstance use disorder, ASD, and brief psychotic episodes each accounted for 4.8% of the CHR-P group.
Neurocognitive characteristics of the sample
Table 2 presents the neurocognitive performance across groups. Overall, individuals with FEP showed the weakest performance in most cognitive domains. Significant differences were found between the FEP group and both CHR-P and HC in verbal fluency (COWAT), visual memory (Rey memory), verbal learning (HVLT), and social cognition (PERE correct total). Processing speed and cognitive flexibility, assessed through TMT A and B, also differed significantly, with FEP participants performing worse than both comparison groups. In the Stroop Word and Stroop Word-Colour conditions, group differences were also significant, with FEP individuals showing greater interference effects. Mentalization abilities measured by the MASC task revealed significantly lower undermentalization scores in the FEP group compared to HC, while the CHR-P group differed from HC in the WCST total score. No significant group differences were found in Rey copy or Stroop Colour.
Table 2. Neurocognitive performance characteristics of the sample
* Significant difference between CHR-P and FEP.
** Significant difference between FEP and CHR-P; and significant difference between FEP and HC.
*** Significant difference between all groups.
+ Significant differences between HC and the other groups.
++ Significant differences between HC and CHR-P.
Relationship between neurocognition and IL 6 and TNF-
$ \boldsymbol{\unicode{x03B1}} $
The heatmap (Figure 1) illustrates the strength and direction of the bivariate correlations, with red tones indicating positive correlations and blue tones representing negative correlations. Overall, IL-6 exhibited weak to moderate associations with some of the cognitive performance tests. IL-6 showed positive correlations with TMT A (r = 0.33, p = 0.0018) and TMT B (r = 0.30, p = 0.0049), although these associations were of small magnitude. Additionally, a negative correlation was observed between IL-6 and MASC total (r = −0.25, p = 0.0389) and IQ (r = −024, p = 0.0257), while a positive correlation was found between IL-6 and MASC undermentalization (r = 0.34, p = 0.0041). Regarding TNF-α, a stronger negative correlation was found with PERE test (r = −0.39, p = 0.0002).
Figure 1. Correlation matrix of inflammatory markers (IL-6 & TNF-α) and cognitive tasks. COWAT, Controlled Oral Word Association Test; HVLT, Hopkins Verbal Learning Test; MASC, Movie for the Assessment of Social Cognition; PERE, Emotion Recognition Task; TMT, Trail Making Test; WCST, Wisconsin Card Sorting Test.
Multiple linear regression models were conducted to examine the relationship among IL-6, TNF-α levels and cognitive performance while adjusting for IQ, sex, age, psychosis risk group (HC, CHR, and FEP), and positive symptom severity. We introduced into the model as dependent variables those with positive associations in the bivariate correlations.
IL-6 exhibited no association with TMT-A (p = 0.075). The FEP group demonstrated a significant positive association with TMT-A (β = 12.23, p = 0.0013), indicating that individuals in this group had significantly longer completion times compared to the reference group. Neither sex nor positive symptoms were significantly associated with TMT-A completion time. The model accounted for 24% of the variance in TMT-A (R 2 adjusted = 0.24).
The association between TMT-B performance and IL-6 levels was significant overall (F (7, 75) = 9.52, p < 0.001), explaining approximately 47% of the variance in TMT-B scores (adjusted R 2 = 0.42). Higher IL-6 levels were not significantly associated with TMT-B performance (p = 0 .19). In contrast, lower IQ (p < 0.001), older age (p = 0.038), and FEP group (p = 0.001) were significant predictors of poorer performance. Sex and positive symptoms did not show significant effects.
For MASC Total (F (7, 57) = 6.84, p < .001), IL-6 was not significantly associated with overall mentalizing performance (β = −0.25, p > 0.05). However, IQ was positively associated with mentalizing performance (β = 0.069, p = 0.015), indicating that higher IQ scores relate to better mentalizing ability. HC outperformed other groups in mentalizing accuracy (β = 5.71, p = 0.0018, R 2 adjusted = 0.40).
In the case of MASC undermentalization (F (7, 57) = 4.12, p = 0.001), IL-6 showed a significant positive association with undermentalization (β = 0.28, p = 0.0337), suggesting that higher IL-6 levels may be linked to a tendency to under-attribute mental states to others. IQ showed a trend toward significance (β = −0.034, p = 0.054), implying that lower IQ may be related to increased undermentalization. HC exhibited lower undermentalization scores compared to other groups (β = −2.48, p = 0.031). The model explained 25% of the variance (R 2 adjusted = 0.25).
Regarding the PERE recognition test (F (6, 77) = 3.60, p = 0.003), TNFα showed a significant negative association with performance (β = −1.37, p = 0.0022), suggesting that higher TNFα levels may be linked to poorer facial emotion recognition. No other variables showed significant effects in this model. The model explained 16% of the variance (R 2 = 0.16).
After applying the Benjamini–Hochberg FDR correction for multiple comparisons to the two significant associations identified in the regression models. Both the association between TNF-α and emotion recognition (PERE) (adjusted p = 0.0044) and the association between IL-6 and undermentalization (MASC) (adjusted p = 0.0337) remained statistically significant after correction.
Discussion
In this study, we examined the associations between inflammatory markers and neurocognitive performance across individuals at CHR-P, patients with FEP, and HC. As expected, the FEP group exhibited the most pronounced cognitive impairments, particularly in verbal fluency, memory, processing speed, and social cognition. Adjusted analyses revealed two significant associations between inflammation and social cognition. Higher IL-6 levels were independently associated with increased undermentalization on the MASC task, suggesting a link between systemic inflammation and difficulties in inferring others’ mental states. Additionally, elevated TNF-α levels were negatively associated with performance on the PERE test, indicating a potential impact on facial emotion recognition. These findings remained significant after accounting for key confounders, highlighting that specific aspects of social cognition may be particularly sensitive to peripheral immune dysregulation in early psychosis.
Given prior evidence linking inflammatory markers with the severity of positive symptoms in early psychosis [Reference Liemburg, Nolte, Klein and Knegtering32], we included positive symptom severity as a covariate in our models. Although our study was not primarily designed to test this association, exploratory analyses in the FEP group supported its relevance, and future studies should consider this dimension when examining immuno-cognitive interactions.
The neurocognitive profile observed in our sample reinforces the notion that cognitive impairments are a core feature of psychotic disorders [Reference McCutcheon, Keefe and McGuire1], with the FEP group showing the most pronounced deficits across domains such as verbal fluency, memory, processing speed, and cognitive flexibility [Reference Jonas, Lian, Callahan, Ruggero, Clouston and Reichenberg55]. These findings are consistent with previous literature indicating that such deficits are already present at illness onset and tend to persist over time [Reference Catalan, McCutcheon, Aymerich, Pedruzo, Radua and Rodriguez56]. The intermediate performance of the CHR-P group suggests that subtle cognitive alterations may emerge even before the onset of frank psychosis [Reference Catalan, Salazar de Pablo, Aymerich, Damiani, Sordi and Radua8]. In mentalization abilities, significant group differences were observed in the MASC task: the HC group demonstrated better accuracy in identifying mental states, while FEP individuals showed a greater tendency toward undermentalization [Reference Catalan, Angosto, Diaz, Martinez, Guede and Pereda57], a pattern that has been linked to social cognition deficits in schizophrenia. The absence of significant differences in tasks such as Rey copy and Stroop Colour further highlights the domain-specific nature of these alterations, underlining the importance of targeted cognitive and social-cognitive assessment in early detection strategies.
Our finding of a robust association between higher TNF-α levels and IL-6 and poorer performance in facial emotion recognition and undermentalization aligns with experimental evidence suggesting a direct impact of systemic inflammation on social cognitive processes. For instance, experimental studies show that mild immune activation, such as with typhoid vaccine or IF-α, can impair social cognition and induce negative emotion processing biases, even in the absence of mood changes [Reference Balter, Hulsken, Aldred, Drayson, Higgs and Veldhuijzen van Zanten58, Reference Cooper, Godlewska, Sharpley, Barnes, Cowen and Harmer59]. While our study used a cross-sectional design, the results are consistent with the idea that inflammation can selectively impair the neural mechanisms involved in decoding emotional expressions, particularly in vulnerable clinical populations. The robustness of this association, in contrast to the nonsignificant findings for broader cognitive tasks, may reflect a particular vulnerability of socio-affective processing systems to inflammatory dysregulation in early psychosis.
One possible explanation for these specific associations with social cognition, is that complex cognitive functions – such as mentalization and emotion recognition – rely on distributed neural networks that are especially vulnerable to inflammatory disruption. Proinflammatory cytokines like IL-6 and TNF-α can impair synaptic plasticity, reduce prefrontal connectivity, and disrupt neurotransmitter regulation [Reference Miller, Haroon, Raison and Felger60-Reference Williams, Burgess, Suckling, Lalousis, Batool and Griffiths63], all of which are essential for social cognitive processing. Experimental findings have shown that inflammatory mediators (e.g., COX-2) can inhibit synaptic strength and modulate performance in cognitive tasks, particularly in the prefrontal cortex [Reference Garcia-Alvarez, Garcia-Portilla, Gonzalez-Blanco, Saiz Martinez, de la Fuente-Tomas and Menendez-Miranda64].
Moreover, glial cells – particularly microglia – play a critical role in mediating the effects of inflammation on the brain [Reference Colonna and Butovsky65, Reference Gutierrez-Fernandez, Gonzalez-Pinto, Vega, Barbeito and Matute66]. Upon activation, microglia release proinflammatory cytokines, which can initiate and sustain neuroinflammatory cascades. These cytokines have been shown to induce oxidative stress, alter synaptic architecture, and compromise blood–brain barrier integrity, all of which may disrupt the finely tuned networks underlying social cognitive processes [Reference Kraft and Harry67]. Cytokine release and glial activation may jointly drive neuronal dysfunction in psychiatric and neurodegenerative disorders, underscoring their relevance in early psychosis [Reference Smith, Das, Ray and Banik68].
TNF-α and IL-6 showed specific associations with emotion recognition and undermentalization, respectively, suggesting that proinflammatory cytokines may selectively disrupt prefrontal–limbic circuits underlying social cognition in early psychosis [Reference Tripoli, Quattrone, Ferraro, Gayer-Anderson, La Cascia and La Barbera69–Reference Haroon, Raison and Miller71].
Strengths and limitations
This study presents several strengths. It includes a well-characterized clinical sample covering the psychosis spectrum, with both CHR-P and FEP participants, enabling meaningful comparisons across illness stages. A key procedural strength of this study is the standardized timing of blood collection, which was conducted between 8:00 and 10:00 AM after overnight fasting. This protocol aimed to minimize circadian variation in cytokine levels, particularly for IL-6 [Reference Nilsonne, Lekander, Akerstedt, Axelsson and Ingre72] and TNF-α, which are known to follow diurnal secretion patterns. Controlling for timing reduces potential measurement noise and improves the comparability of inflammatory marker data across participants. The use of a comprehensive cognitive battery, alongside the assessment of inflammatory markers, allowed for an integrative analysis of neuroimmune-cognitive interactions. Importantly, multivariate models were employed to control for key confounding variables, including IQ, age, sex, and symptom severity, enhancing the reliability of the observed associations.
Nonetheless, some limitations should be considered. The cross-sectional nature of the study prevents conclusions regarding causality or the temporal relationship between inflammation and cognitive functioning. Cytokine levels were measured at a single time point, limiting the ability to capture fluctuations in inflammatory status. The relatively small sample size, particularly in subgroup analyses and regression models with multiple covariates, may have reduced statistical power to detect subtle effects. In addition, missing data in some variables led to reduced effective sample sizes, which further constrained analytical precision. While the cognitive battery covered key domains, certain aspects of social cognition, such as theory of mind and attributional style, were not extensively explored. Finally, the use of bivariate correlations and stepwise regression models may have increased the risk of overfitting and biased parameter estimation. Although efforts were made to reduce this risk by limiting predictors and adjusting for key covariates, more robust statistical approaches (e.g., penalized regression or cross-validation) may strengthen future analyses.
Conclusion
Our findings suggest that TNF-α and IL-6 are selectively associated with impairments in social cognition – specifically emotion recognition and undermentalization – in early psychosis, highlighting their potential as biomarkers and therapeutic targets. Clinically, incorporating inflammatory markers into early assessments may help identify patients who could benefit from tailored interventions aimed at improving social functioning.
Data availability statement
The data underlying this article are not publicly available due to the confidential nature of the clinical information. These data can, however, be made available from the corresponding author upon reasonable request, subject to approval by the relevant ethics committee and data-sharing agreements.
Acknowledgements
We would like to extend our thanks to Miren Legido for her assistance in data collection, and to the participants who generously offered their time to contribute to this study.
Financial support
This study was supported by a research grant from the Department of Health awarded through the 2022 call for research and development projects in health – promotion of health research activity. The funded project, PREGAP: Prebentziorako Gazte-Programa Psikosian (Prevention Program for Psychosis in Young People (grant number 2022111036). PFP is supported by #NEXTGENERATIONEU (NGEU), funded by the Ministry of University and Research (MUR), National Recovery and Resilience Plan (NRRP), project MNESYS (PE0000006) – A Multiscale integrated approach to the study of the nervous system in health and disease (DN. 1553 11.10.2022). AGP thanks the support of the Spanish Ministry of Science, Innovation and Universities, integrated into the Plan Nacional de I + D+ I y cofinanciado por el ISCIII-Subdirección General de Evaluación y el Fondo Europeo de Desarrollo Regional (FEDER (PI18/01055; PI21/00713) CIBERSAM, the Basque Government 2022111054, and the University of the Basque Country IT1631–22.
Declaration of generative AI and AI-assisted technologies in the writing process
During the preparation of this work, the authors used ChatGPT (OpenAI) to support the editing and refining of the English language and style of the manuscript. After using this tool, the authors reviewed and edited the content as needed and take full.
Competing interests
Dr. Catalan reports having received support to attend scientific meetings from Janssen, ROVI, and Lundbeck in the last five years. She is also supported by the Instituto de Salud Carlos III, Spanish Ministry of Economy and Competitiveness. Dr. Salazar de Pablo has received personal fees from Janssen Cilag and Menarini. Dr. Aymerich is supported by the Alicia Koplowitz Foundation and has received personal fees or grants from Janssen Cilag and Neuraxpharm, outside the current work. Dr. Gonzalez-Pinto has received grants and served as a consultant, advisor or CME speaker for Janssen-Cilag, Lundbeck, Otsuka, Alter, Angelini, Novartis, Rovi, Takeda, the Spanish Ministry of Science and Innovation (CIBERSAM), the Carlos III Institute, the Basque Government, and the European Framework Program of Research. All other authors report no biomedical financial interests or potential conflicts of interest.
Open access
Comments
No Comments have been published for this article.