Introduction
Schizophrenia is one of the most severe mental disorders (Owen, Sawa, & Mortensen, Reference Owen, Sawa and Mortensen2016), associated with a markedly reduced life expectancy of approximately 13–15 years (Hjorthøj, Stürup, Mcgrath, & Nordentoft, Reference Hjorthøj, Stürup, Mcgrath and Nordentoft2017). Identifying modifiable risk factors and elucidating their underlying mechanisms are essential for informing preventive strategies and potential therapeutic targets. Accumulating evidence indicates that recent stressful life events (SLEs) are a recognized risk factor for schizophrenia (Kilian et al., Reference Kilian, Supanya, Probst, Morgan, Bärnighausen, Kittirattanapaiboon and Reininghaus2021; Ma et al., Reference Ma, Chiappelli, Kvarta, Bruce, van der Vaart, Goldwaser, du, Sampath, Lightner, Endres, Yusuf, Yuen, Narvaez, Campos-Saravia, Kochunov and Hong2023; Oloniniyi et al., Reference Oloniniyi, Weiss, John, Esan, Hibben, Patel, Murray, Cohen, Hutchinson, Gureje, Thara, Morgan and Roberts2022; Pries et al., Reference Pries, van Os, Ten Have, de Graaf, van Dorsselaer, Bak and Guloksuz2020). However, most existing studies rely on retrospective or case–control designs (Kilian et al., Reference Kilian, Supanya, Probst, Morgan, Bärnighausen, Kittirattanapaiboon and Reininghaus2021; Ma et al., Reference Ma, Chiappelli, Kvarta, Bruce, van der Vaart, Goldwaser, du, Sampath, Lightner, Endres, Yusuf, Yuen, Narvaez, Campos-Saravia, Kochunov and Hong2023; Oloniniyi et al., Reference Oloniniyi, Weiss, John, Esan, Hibben, Patel, Murray, Cohen, Hutchinson, Gureje, Thara, Morgan and Roberts2022; Rami et al., Reference Rami, Seo, Kang, Park, Li, Le and Chung2025), limiting temporal and causal inference (Pries et al., Reference Pries, van Os, Ten Have, de Graaf, van Dorsselaer, Bak and Guloksuz2020). Prospective evidence remains scarce, and comprehensive evaluations of distinct types of SLEs in relation to schizophrenia are lacking. More importantly, the psychobiological mechanisms through which recent SLEs contribute to schizophrenia risk remain incompletely understood.
A substantial body of research suggests that recent SLEs may contribute to subsequent increases in inflammatory burden (Bourassa et al., Reference Bourassa, Rasmussen, Danese, Eugen-Olsen, Harrington, Houts, Poulton, Ramrakha, Sugden, Williams, Moffitt and Caspi2021; Kyle Bourassa, Reference Kyle Bourassa2020). For example, a multi-cohort study found that recent SLEs were significantly associated with elevated peripheral C-reactive protein (CRP) levels (Hillmann, Dhingra, & Reed, Reference Hillmann, Dhingra and Reed2024). Moreover, a meta-analysis indicates that recent SLEs can activate the hypothalamic–pituitary–adrenal (HPA) axis and the sympathetic nervous system (SNS), triggering the release of pro-inflammatory factors and potentially contributing to neuroinflammatory processes and stress-related inflammatory responses (Vafadari, Reference Vafadari2021). In addition, evidence from another meta-analysis suggests that SLEs can influence metabolic processes, including insulin resistance, glucose metabolism, and lipid homeostasis (Kivimäki, Bartolomucci, & Kawachi, Reference Kivimäki, Bartolomucci and Kawachi2023). Supporting this, a cross-sectional study observed a positive association between SLEs and serum triglyceride levels in middle-aged adults (Anni et al., Reference Anni, Jung, Shim, Jeon, Lee and Kim2021). According to the HPA axis dysregulation hypothesis, psychological stress can disrupt HPA function, leading to elevated cortisol levels, which in turn increase glucose and insulin levels, thereby potentially contributing to insulin resistance and dyslipidemia (Bouillon-Minois & Dutheil, Reference Bouillon-Minois and Dutheil2022). Such a low-grade inflammatory state, characterized by elevated peripheral pro-inflammatory markers (e.g. CRP and IL-6) and metabolic abnormalities, is increasingly recognized as an important factor implicated in the pathophysiology of schizophrenia (Rantala, Luoto, Borráz-León, & Krams, Reference Rantala, Luoto, Borráz-León and Krams2022; Rawani, Chan, Dursun, & Baker, Reference Rawani, Chan, Dursun and Baker2024; Saada & Stern, Reference Saada and Stern2025). Consistent with this perspective, prospective evidence indicates that elevated inflammatory biomarkers are persistently associated with an increased risk of schizophrenia (Zeng et al., Reference Zeng, Chourpiliadis, Hammar, Seitz, Valdimarsdóttir, Fang, Song and Wei2024). Furthermore, a cross-sectional multi-omics study revealed that alterations in lipid molecules are most prominent in patients with schizophrenia, and functional network analyses support a close link between circulating lipid dysregulation and the disease (Campeau et al., Reference Campeau, Mills, Stevens, Rossitto, Meehan, Dorrestein, Daly, Nguyen, Gonzalez, Jeste and Hook2022).
However, the distinct mediating roles of peripheral immune biomarkers and metabolic biomarkers in the association between recent SLEs and schizophrenia in adult populations remain largely unknown. To address these gaps, we used data from the UK Biobank, a large population-based cohort, to: (1) examine the prospective association between recent SLEs and first schizophrenia diagnosis during follow-up; (2) assess the associations of recent SLEs with immunometabolic profiles and of immunometabolic profiles with schizophrenia; and (3) evaluate whether immunometabolic profiles mediate the relationship between recent SLEs and schizophrenia.
Methods
Study design and participants
The UK Biobank is a population-based cohort that recruited over 500,000 participants aged 37 to 73 years from 22 assessment centers between 2006 and 2010 (Sudlow et al., Reference Sudlow, Gallacher, Allen, Beral, Burton, Danesh, Downey, Elliott, Green, Landray, Liu, Matthews, Ong, Pell, Silman, Young, Sprosen, Peakman and Collins2015). The dataset used in this study included demographic characteristics, behavioral assessments, psychosocial factors, schizophrenia diagnoses, blood biochemistry markers, blood cell counts, and nuclear magnetic resonance (NMR)-based metabolic markers. For the primary analyses, we included 494,151 participants who provided data on recent SLEs. After excluding 857 participants with a prior diagnosis of schizophrenia or self-reported schizophrenia during follow-up, the final analytical sample comprised 493,294 participants (Supplementary Figure S1).
Recent stressful life events
Participants were asked whether they had experienced any SLEs within the last 2 years. They were presented with a pre-specified list of six events, including (1) serious illness, injury, or assault to yourself; (2) serious illness, injury, or assault of a close relative; (3) death of a close relative; (4) death of a spouse or partner; (5) marital separation or divorce; and (6) financial difficulties (Supplementary Table S1). An SLE score was calculated based on the number of events reported, ranging from 0 to 6. Participants were classified as non-exposed (SLE score = 0) or exposed (SLE score > 0) (See, Tan, Valmaggia, & Kempton, Reference See, Tan, Valmaggia and Kempton2025).
Schizophrenia diagnoses
Schizophrenia diagnoses were identified using the International Classification of Diseases recorded according to (ICD-10) code ‘F20’. We integrated data from hospital inpatient records, primary care records, and death registers while excluding cases identified solely through self-report. The outcome was defined as the first recorded schizophrenia diagnosis during follow-up, with the diagnosis date assigned as the earliest date documented across these data sources (Zhang et al., Reference Zhang, Meng, Luo, Yu, Li, Zhou, Chen and Kan2024). Given that UK Biobank participants were recruited in mid-to-late adulthood, incident cases in this study represent first clinically recorded schizophrenia diagnoses during follow-up and may not reflect the timing of schizophrenia onset.
Peripheral markers
Blood biochemistry (category ID 17518) and blood count (category ID 100081) data were derived from approximately 480,000 participants at baseline. Blood count samples collected in 4 mL ethylenediaminetetraacetic acid (EDTA) vacuum tubes were analyzed using a Beckman Coulter LH750 instrument. Thirty blood biochemistry markers and 31 blood cell counts were available from these two UK Biobank categories. Of the 61 total markers, 59 were used in the study; nucleated red blood cell count and percentage were excluded due to >90% of values being 0. Blood biochemical markers were classified into ‘liver function’, ‘renal function’, ‘endocrine’, ‘immunometabolic’, or ‘bone and joint’ categories based on prior literature. Blood cell counts were categorized as ‘white blood cell’, ‘red blood cell’, or ‘platelet’.
Metabolic markers were measured in randomly selected EDTA plasma samples using an NMR-based metabolic biomarker profiling platform, which included data from approximately 120,000 UK Biobank participants at the baseline assessment. The NMR metabolomics (category ID 220) comprised 249 metabolic markers (168 direct measurements and 81 ratios of these markers). The biomarkers include a wide range of metabolic pathways, including low-molecular-weight metabolites (e.g. amino acids, ketone bodies, and glycolysis metabolites), lipoprotein lipids (14 subclasses), inflammation-related biomarkers (e.g. glycoprotein acetyls [GlycA]), fatty acids and their compositions, and other small molecules.
Covariates
Consistent with prior studies (Kang et al., Reference Kang, Yang, Jia, Zhang, Wang, Zhao, You, Deng, Ge, Liu, Zhang, Chen, He, Sahakian, Yang, Zhao, Yu, Feng and Cheng2025; Zhao et al., Reference Zhao, Yang, Sahakian, Langley, Zhang, Kuo, Li, Gan, Li, Zhao, Yu, Feng and Cheng2023), we included baseline covariates that could potentially confound the associations of interest, including age, sex, ethnicity, education level, Townsend Deprivation Index (TDI), smoking status, alcohol consumption, physical activity, and body mass index (BMI). The corresponding UK Biobank data fields and detailed definitions for each covariate are provided in Supplementary Table S2. For analyses of peripheral markers, models were additionally adjusted for fasting time and assay batch.
Statistical analyses
Continuous variables are presented as means with standard deviations (SDs) and categorical variables as counts and percentages. Baseline characteristics were summarized by incident schizophrenia status and compared using χ2 tests or analysis of variance, as appropriate. Missing data on covariates were assessed using Little’s MCAR test (Little, Reference Little1988) and were handled using multiple imputation by chained equations via the R package ‘mice’ (Liang et al., Reference Liang, Zhou, He, Zhang, Wu, Luo, Zhang, Jia, Qi, Ai and Zhang2024).
Cox proportional hazard regression models were used to estimate hazard ratios (HRs) and 95% confidence intervals (CIs) for the associations between SLEs and schizophrenia risk. The proportional hazards assumption was evaluated using Schoenfeld residuals and was confirmed to be satisfied. To investigate associations between SLE scores and peripheral biomarkers, separate linear mixed-effects models were fitted for each of the 59 blood markers and 249 metabolic markers. In each model, SLE scores and covariates were included as fixed effects, while the UK Biobank assessment center was treated as a random effect. Standardized β coefficients were extracted and converted to Cohen’s d according to previous studies (Grant, Reference Grant2014; Jiang et al., Reference Jiang, Noble, Sui, Yoo, Rosenblatt, Horien, Qi, Liang, Sun, Calhoun and Scheinost2023). Associations between these biomarkers and schizophrenia were subsequently assessed using Cox proportional hazards regression. Mediation analyses were conducted using the R package ‘mediation’ to examine whether the prospective association between SLE scores and schizophrenia risk was mediated by blood and metabolic markers. A standard three-variable path model was established, where linear regression was used for the associations between SLE scores and biomarkers, and survival regression was used for the associations between SLE scores or biomarkers and schizophrenia (Deng et al., Reference Deng, Li, Huang, Ou, Zhang, Chen, Zhang, Yang, Dong, Feng, Suckling, Smith, Cheng and Yu2022; Jiang et al., Reference Jiang, Geha, Rosenblatt, Wang, Fu, Foster, Dai, Calhoun, Sui, Spann and Scheinost2025). The significance of mediating effects was determined using 5000 bootstrap iterations, with statistical significance of indirect effects defined using false discovery rate (FDR)-corrected P-values (< 0.05).
Subgroup analyses stratified by sex (male and female) were conducted to explore potential variations. Several sensitivity analyses were performed to evaluate the robustness of the primary findings: (1) Fine-Gray subdistribution hazard models were applied to account for death as a competing risk for incident schizophrenia, and (2) landmark analyses excluded participants who developed schizophrenia or died within the first 2 years of follow-up to reduce potential reverse causation.
All analyses were conducted in R version 4.4.0, and two-tailed P values <0.05 were considered statistically significant, with FDR correction applied to analyses involving multiple biomarkers.
Results
Baseline characteristics
We included 493,294 participants (mean [SD] age, 56.55 [8.09] years; 45.56% male; 94.73% were of White ethnicity) free of schizophrenia at baseline. During a median follow-up of 13.91 years, 492 participants developed a first recorded schizophrenia diagnosis. Baseline characteristics of the total cohort, as well as of individuals with and without incident schizophrenia, are presented in Table 1. Compared with participants without incident schizophrenia, those with incident schizophrenia were more likely to be male, slightly younger, have higher BMI, lower educational attainment, higher proportions of current smoking, and a greater number of recent SLEs.
Baseline characteristics of participants
Table 1. Long description
The table compares three groups: Total (N = 493,294), No incident schizophrenia (n = 492,802), and Incident schizophrenia (n = 492).
Demographics:
- Sex: The incident schizophrenia group has a higher proportion of males (55.08 percent) compared to the total population (45.56 percent).
- Age: Mean age is slightly lower in the incident group (55.83) compared to the total (56.55).
- Townsend deprivation index: The incident group has a much higher mean score (1.43) than the total population (-1.32), indicating higher deprivation.
- B M I: Mean B M I is higher in the incident group (28.98) than the total (27.42).
- Ethnicity: Non-White participants make up 15.37 percent of the incident group, compared to 5.27 percent of the total.
Lifestyle Factors:
- Alcohol: Current consumption is lower in the incident group (77.05 percent) versus total (92.04 percent).
- Smoking: Current smokers are more prevalent in the incident group (24.64 percent) compared to the total (10.49 percent).
- Physical Activity: Rates of regular activity are similar across groups (approximately 60 percent).
Stressful Life Events:
- Recent stressful events: 60.57 percent of the incident group reported recent stressful events, compared to 44.88 percent of the total.
- Event Scores: The incident group shows higher frequencies for scores of 1, 2, and 3.
- Specific Stressors: The incident group reported higher rates of serious illness/injury to self (23.78 percent vs 9.46 percent), financial difficulties (25.61 percent vs 12.38 percent), and marital separation (6.91 percent vs 3.18 percent).
Note: SD, standard deviation; BMI, body mass index.
a Missing data: 1634 for ethnicity, 88813 for education level, 513 for alcohol consumption, 1768 for smoking status, 48646 for physical activity.
Associations of recent SLEs with subsequent schizophrenia risk
Compared with participants without recent SLEs exposure, those with any SLE exposure had a 49% higher risk of incident schizophrenia during follow-up (HR = 1.49, 95% CI: 1.24–1.79). Further analysis indicated a significant dose–response relationship between SLE scores and incident schizophrenia during follow-up. Compared with individuals with an SLE score of 0, the adjusted HR increased progressively with greater exposure, reaching 7.15 (95% CI: 5.74–9.11) among those in the highest exposure level. In analyses of individual SLE components, all items were associated with increased schizophrenia risk. The strongest associations were observed for ‘serious illness, injury, or assault to yourself’ (HR = 2.30, 95% CI: 1.86–2.84) and ‘financial difficulties’ (HR = 2.07, 95% CI: 1.83–2.37) (Figure 1).
Association of recent stressful life events with schizophrenia risk (n = 493,294). Note: CI, confidence interval; HR, hazard ratio.
Figure 1. Long description
a: We examined the association of experiencing recent stressful life events (SLEs) and per 1-point increase in the recent SLE score with schizophrenia risk, using participants with an SLE score of 0 as the reference group. Model 1 was unadjusted. Model 2 was adjusted for sex, age, ethnicity, education level, body mass index, Townsend deprivation index, alcohol consumption, smoking status, and physical activity. Points represent HRs, and error bars represent 95% CIs. b: We examined the association between each indicator of recent SLEs and schizophrenia risk. Model 1 was unadjusted. Model 2 was adjusted for sex, age, ethnicity, education level, body mass index, Townsend deprivation index, alcohol consumption, smoking status, and physical activity. Bars represent HRs, and error bars represent 95% CIs.
Associations between recent SLEs and peripheral markers
After FDR correction for multiple testing, 43 blood markers and 142 metabolic markers were significantly associated with recent SLEs (Figures 2 and 3; Supplementary Tables S3–S4). Among blood markers, the strongest associations were observed for monocyte percentage (Cohen’s d = 0.046, P = 1.54 × 10−126), cholesterol (Cohen’s d = 0.045, P = 2.02 × 10−148), and CRP (Cohen’s d = 0.035, P = 9.54 × 10−75). For metabolic markers, the most pronounced associations included degree of unsaturation (Cohen’s d = −0.046, P = 4.33 × 10−30), cholesterol to total lipids in IDL percentage (Cohen’s d = −0.043, P = 1.40 × 10−26), triglycerides to total lipids in IDL percentage (Cohen’s d = 0.046, P = 1.23 × 10−30), and triglycerides to total lipids in large LDL percentage (Cohen’s d = 0.042, P = 5.25 × 10−26).
Blood markers associated with recent stressful life events.
Figure 2. Long description
Associations were estimated using linear mixed-effects models, adjusted for sex, age, ethnicity, education level, body mass index, Townsend deprivation index, alcohol consumption, smoking status, physical activity, fasting time, and assay batch. A total of 43 blood markers showing significant associations with recent stressful life events (SLEs) at a false discovery rate (FDR) < 0.05 are presented.
Circulating metabolites associated with recent stressful life events.
Figure 3. Long description
Associations were estimated using linear mixed-effects models, adjusted for sex, age, ethnicity, education level, body mass index, Townsend deprivation index, alcohol consumption, smoking status, physical activity, fasting time, and assay batch. A total of 142 metabolites showing significant associations with recent stressful events (SLEs) at a false discovery rate (FDR) < 0.05 were presented.
Associations between peripheral markers and subsequent schizophrenia risk
After FDR correction for multiple testing, 40 blood markers and 66 metabolic markers were significantly associated with incident schizophrenia during follow-up (Figures 4 and 5; Supplementary Tables S5–S6). Among blood markers, CRP (HR = 1.28, 95% CI: 1.11–1.55), cholesterol (HR = 1.26, 95% CI: 1.12–1.53), and high-density lipoprotein (HDL) cholesterol (HR = 0.51, 95% CI: 0.44–0.79) showed the strongest associations. For metabolic markers, the most significant associations were observed for degree of unsaturation (HR = 0.74, 95% CI: 0.61–0.89), linoleic acid to total fatty acids percentage (HR = 0.75, 95% CI: 0.62–0.89), phospholipids to total lipids in medium HDL percentage (HR = 1.25, 95% CI: 1.05–1.50), and saturated fatty acids to total fatty acids percentage (HR = 1.25, 95% CI: 1.06–1.47).
Blood markers associated with schizophrenia.
Figure 4. Long description
Associations were estimated using Cox proportional hazards regression models, adjusted for sex, age, ethnicity, education level, body mass index, Townsend deprivation index, alcohol consumption, smoking status, physical activity, fasting time, and assay batch. Dots represent mean hazard ratios (HRs), and the horizontal lines indicate the corresponding 95% confidence intervals (CIs). Statistical significance is denoted as follows: * FDR < 0.05, ** FDR < 0.01, *** FDR < 0.001.
Circulating metabolites associated with schizophrenia.
Figure 5. Long description
Associations were estimated using Cox proportional hazard regression models, adjusted for sex, age, ethnicity, education level, body mass index, Townsend deprivation index, alcohol consumption, smoking status, physical activity, fasting status, and assay batch. Dots represent mean hazard ratios (HRs), and the horizontal lines indicate the corresponding 95% confidence intervals (CIs). Significance is denoted by: * FDR < 0.05, ** FDR < 0.01, *** FDR < 0.001.
Mediating analyses of peripheral markers
Mediation analyses were conducted on the 29 blood markers and 56 metabolic markers that showed significant associations with both SLE scores and incident schizophrenia during follow-up. Among the 29 blood markers, 25 were found to significantly and partially mediate the prospective association between SLE scores and first schizophrenia diagnoses identified during follow-up, after adjusting for covariates and correcting for multiple testing. Notably, CRP and HDL cholesterol showed the largest proportions mediated, at 9.0% and 3.6%, respectively. Among the 56 metabolic markers, 48 were identified as significant partial mediators of the prospective association between SLE scores and first schizophrenia diagnoses identified during follow-up. The degree of unsaturation and polyunsaturated fatty acids to monounsaturated fatty acids ratio exhibited the largest proportions mediated at 11.7% and 10.8%, respectively (see Figure 6). Detailed results are presented in Supplementary Tables S7–S8.
Mediation analysis of peripheral markers on the associations between recent stressful life events and risk of schizophrenia.
Figure 6. Long description
a: Among 29 blood markers, 25 significantly and partially mediated the prospective association between recent stressful life events (SLEs) and incident schizophrenia, after adjustment for covariates and corrections for multiple testing. Dots represent the mean proportion mediated, with horizontal lines showing 95% confidence intervals (CIs). b: Among 56 metabolic markers, 48 significantly and partially mediated the prospective association between recent SLEs and incident schizophrenia, after adjustment for covariates and corrections for multiple testing. Dots represent the mean proportion mediated, with horizontal lines showing 95% CIs.
Subgroup and sensitivity analyses
As shown in Supplementary Tables S9–S15, sex-stratified analyses showed consistent associations between recent SLEs and incident schizophrenia during follow-up in both males and females. Similarly, sex-stratified mediation analyses demonstrated largely consistent immunometabolic mediation patterns across sexes, although modest differences in the mediation strength of several biomarkers were observed. Sensitivity analyses further confirmed the robustness of the main findings. Similar associations were observed when accounting for deaths as a competing risk using Fine-Gray subdistribution hazard models and when applying a 2-year landmark analysis (Supplementary Tables S16–S17).
Discussion
In this cohort study conducted using data from the UK Biobank among middle-aged and older adults, we found that recent SLEs were associated with an increased risk of first recorded schizophrenia diagnosis during follow-up, with a clear dose–response relationship. Recent SLEs were also associated with alterations in a broad range of immunometabolic markers, several of which were independently associated with schizophrenia risk. Mediation analyses further indicated that specific peripheral immune markers (e.g. CRP) and metabolic markers (e.g. degree of unsaturation) partially mediated the association between recent SLEs and subsequent schizophrenia. These findings provide population-level evidence that psychosocial stress may be linked to schizophrenia through immunometabolic alterations, potentially extending prior mechanistic and clinical research. By identifying CRP and lipid unsaturation as specific mediators, this study highlights candidate biomarkers that may be relevant for risk stratification. Collectively, these findings help bridge epidemiological observations and biological correlates, offering directions for future research on stress-related pathways to psychosis.
Although prior studies suggest an association between recent SLEs and schizophrenia (Kilian et al., Reference Kilian, Supanya, Probst, Morgan, Bärnighausen, Kittirattanapaiboon and Reininghaus2021; Ma et al., Reference Ma, Chiappelli, Kvarta, Bruce, van der Vaart, Goldwaser, du, Sampath, Lightner, Endres, Yusuf, Yuen, Narvaez, Campos-Saravia, Kochunov and Hong2023; Oloniniyi et al., Reference Oloniniyi, Weiss, John, Esan, Hibben, Patel, Murray, Cohen, Hutchinson, Gureje, Thara, Morgan and Roberts2022; Rami et al., Reference Rami, Seo, Kang, Park, Li, Le and Chung2025), most were retrospective or case–control, limiting causal inference. By demonstrating a dose-dependent association between recent SLEs and first schizophrenia diagnoses identified during follow-up in a prospective cohort, our study extends the existing evidence and suggests that the cumulative burden of recent psychosocial adversity may be an important correlate of subsequent schizophrenia risk. Consistent with previous evidence (Xu et al., Reference Xu, Cai, Ji, Ma, Liu, Zhao, Chen, Zhao, Zhang, Wang, Guo, Xue, Wang, Liu, Wang, Zhu and Liu2023; Zuo et al., Reference Zuo, Yang, Wu, Fan, Liu, Xiang, Li, Zhao, Liu and Liu2024), among SLEs, ‘serious illness, injury, or assault to yourself’ and ‘financial difficulties’ showed stronger associations, indicating that stressors affecting personal safety and survival resources may have particular relevance to schizophrenia vulnerability. Rather than implicating a single type of stress exposure, these findings underscore the importance of considering the overall accumulation and heterogeneity of stressful life experiences when investigating environmental influences on schizophrenia. The observed epidemiological associations raise the question of the underlying biological mechanisms, prompting us to investigate whether recent SLEs influence peripheral immunometabolic biomarkers.
We identified significant associations between SLEs and immunometabolic biomarkers related to inflammation, lipid metabolism, and energy regulation. These findings are consistent with prior evidence showing that SLEs in adulthood correlate with elevated CRP levels (Hillmann et al., Reference Hillmann, Dhingra and Reed2024) and that life stress can increase pro-inflammatory biomarkers, including CRP (Hori & Kim, Reference Hori and Kim2019). Additionally, a longitudinal study has reported that in women with hyperlipidemia, greater numbers of recent SLEs are associated with subsequent triglyceride elevation (Kuo et al., Reference Kuo, Ersig, Johnson, Brown, Oakley, Hagen, Barnet and Peppard2023), while an animal study observed that chronic psychological stress reduced the proportion of ω-3 polyunsaturated fatty acids in platelet membranes (Bikulčienė et al., Reference Bikulčienė, Baleišis, Mazgelytė, Rudys, Vosyliūtė, šimkūnaitė-Rizgelienė and Karčiauskaitė2024). A potential explanation for these associations is that SLEs may be associated with activation of the HPA axis and SNS, promoting the release of neuroendocrine mediators such as cortisol and catecholamines (Papargyri et al., Reference Papargyri, Zapanti, Salakos, Papargyris, Bargiota and Mastorakos2018; Picca et al., Reference Picca, Calvani, Coelho-Junior, Landi, Bernabei and Marzetti2020; van den Heuvel et al., Reference van den Heuvel, Suliman, Bröcker, Kilian, Stalder, Kirschbaum and Seedat2022). Prolonged or dysregulated hormone secretion can disrupt glucose and lipid metabolism, modulate immune cell function, trigger NLRP3 inflammasome activation, promote pro-inflammatory cytokine production, and induce microglia-mediated neuroinflammation (Busillo, Azzam, & Cidlowski, Reference Busillo, Azzam and Cidlowski2011; Gülke, Gelderblom, & Magnus, Reference Gülke, Gelderblom and Magnus2018; Kivimäki et al., Reference Kivimäki, Vahtera, Tabák, Halonen, Vineis, Pentti, Pahkala, Rovio, Viikari, Kähönen, Juonala, Ferrie, Stringhini and Raitakari2018, Reference Kivimäki, Bartolomucci and Kawachi2023; Weber et al., Reference Weber, Frank, Tracey, Watkins and Maier2015). These processes create a vicious cycle of neuroendocrine-immune interactions (Kim, Lee, & Park, Reference Kim, Lee and Park2022; Picca et al., Reference Picca, Calvani, Coelho-Junior, Landi, Bernabei and Marzetti2020), ultimately leading to elevated peripheral inflammatory biomarkers and metabolic disturbances (Draganov et al., Reference Draganov, Arranz, Salazar, de Diego-Adeliño, Gallego-Fabrega, Jubero, Carceller-Sindreu and Portella2019; Kivimäki et al., Reference Kivimäki, Bartolomucci and Kawachi2023; Opel et al., Reference Opel, Cearns, Clark, Toben, Grotegerd, Heindel, Kugel, Teuber, Minnerup, Berger, Dannlowski and Baune2019). Together, these findings suggest that recent SLEs are associated with coordinated immunometabolic remodeling rather than isolated alterations in individual biomarkers, highlighting the systemic biological consequences of chronic psychosocial stress.
Our study further suggests that specific peripheral immunometabolic markers, including CRP, degree of unsaturation, and the proportion of linoleic acid relative to total fatty acids, are associated with schizophrenia risk. Consistent with previous evidence, patients with psychotic symptoms exhibited higher CRP levels, reflecting a chronic low-grade inflammatory state that has been suggested to impair neuronal circuits and compromise blood–brain barrier integrity (Almulla & Maes, Reference Almulla and Maes2025; Ozkalayci & Tastekin, Reference Ozkalayci and Tastekin2024; Pape, Tamouza, Leboyer, & Zipp, Reference Pape, Tamouza, Leboyer and Zipp2019). Schizophrenia patients also exhibit reduced levels of polyunsaturated fatty acids (PUFAs) in cell membranes, accompanied by oxidative lipid accumulation and elevated lipid peroxides, which leads to decreased membrane fluidity and indicates lipid metabolic dysregulation (Li et al., Reference Li, Gao, Wang, Hu, Jiang, Qing, Yang, Cui, Wang, Zhang, Sun and Wan2022; Zhang et al., Reference Zhang, Yin, Yan, Yan, Li, Zhang, Li, Liu, Lin, Zhang, Hu and Song2023). These findings suggest that chronic inflammation may drive or exacerbate lipid abnormalities (Sapienza et al., Reference Sapienza, Agostoni, Repaci, Spangaro, Comai and Bosia2025; Wang et al., Reference Wang, Kockx, Pennings, Lambert, Chow and Kritharides2024) and that structural and functional lipid disturbances, particularly those affecting neuronal membrane properties and neuroimmune regulation, may partially link peripheral immunometabolic alterations to central nervous system dysfunction. These interconnected neuroimmune and metabolic processes may increase vulnerability to psychiatric disorders, including schizophrenia (Del Campo, Nicolson, & Sfera, Reference Del Campo, Nicolson and Sfera2024; Mongan et al., Reference Mongan, Raj Susai, Föcking, Byrne, Zammit, Cannon and Cotter2023; Rantala et al., Reference Rantala, Luoto, Borráz-León and Krams2022).
Our study is also consistent with a potential biological pathway between recent SLEs and schizophrenia. Specifically, CRP reflects a persistent state of low-grade systemic inflammation (Rantala et al., Reference Rantala, Luoto, Borráz-León and Krams2022), whereas changes in the degree of unsaturation may indicate alterations in lipid composition associated with membrane fluidity and neuronal function (Li et al., Reference Li, Gao, Wang, Hu, Jiang, Qing, Yang, Cui, Wang, Zhang, Sun and Wan2022; Zhang et al., Reference Zhang, Yin, Yan, Yan, Li, Zhang, Li, Liu, Lin, Zhang, Hu and Song2023). These inflammatory and metabolic abnormalities may increase schizophrenia risk by impairing neuronal membrane function, blood–brain barrier integrity, and neuroimmune regulation (Del Campo et al., Reference Del Campo, Nicolson and Sfera2024; Mongan et al., Reference Mongan, Raj Susai, Föcking, Byrne, Zammit, Cannon and Cotter2023; Rantala et al., Reference Rantala, Luoto, Borráz-León and Krams2022). Sex-stratified exploratory analyses further suggested that these immunometabolic pathways were largely shared across males and females. Although slight differences in effect estimates were observed across sexes, these differences were modest and should be interpreted with caution. These findings suggest that integrating psychosocial and biological assessments may help improve risk stratification for schizophrenia. Future studies are warranted to explore whether combining recent SLE assessments with peripheral immunometabolic profiles (e.g. CRP and lipid markers) could help identify individuals at higher risk who may benefit from targeted monitoring or preventive interventions.
Our study possesses several strengths. First, its large sample size and long follow-up period enhance statistical power and enable the investigation of long-term associations. Second, by examining the mediating role of immunometabolic markers in the association between recent SLEs and schizophrenia risk, the study provides novel evidence supporting potential biological pathways underlying this association. Several limitations should also be acknowledged. First, information on recent SLEs was primarily self-reported, which may introduce measurement error. Second, as a volunteer-based cohort, UK Biobank participants are not fully representative of the general UK population, tending to be healthier and more socioeconomically advantaged. This selection bias may limit the generalizability of the findings and may also contribute to survival bias, particularly in relation to severe psychiatric disorders (Fry et al., Reference Fry, Littlejohns, Sudlow, Doherty, Adamska, Sprosen, Collins and Allen2017). Third, despite adjustment for multiple covariates, residual confounding cannot be excluded due to unmeasured or imprecisely measured factors. Fourth, although sensitivity analyses excluding participants who developed schizophrenia or died within the first 2 years of follow-up yielded consistent results, reverse causality cannot be fully excluded. Fifth, the age structure of the cohort should be considered when interpreting findings. Most participants were in midlife at baseline (mean age 56.55 years), substantially beyond the typical peak age of schizophrenia onset. Consequently, individuals with early-onset schizophrenia are likely to have developed the disorder prior to study entry and would not be captured as incident cases during follow-up. Therefore, incident cases identified in this study more likely reflect first recorded clinical diagnoses of schizophrenia in adulthood rather than true age at onset across the full life course, potentially leading to under-ascertainment of early-onset cases. Sixth, schizophrenia cases were identified using linked health records, which may not fully capture prior psychiatric history and may therefore introduce outcome misclassification.
Conclusion
Based on prospective data from the UK Biobank, this study shows that recent SLEs are significantly associated with an increased risk of subsequent schizophrenia in a dose–response manner. Mediation analyses further suggest that specific immunometabolic markers partially account for this association, supporting a potential biological pathway linking stress exposure to schizophrenia risk. Given that UK Biobank participants are recruited in mid-to-late adulthood, these findings are applicable to schizophrenia with onset in later life and are not directly generalizable to typical early-onset schizophrenia. Within this context, these findings highlight the importance of identifying and managing major stressors at the population level. Future interventions targeting both SLEs and the associated immunometabolic pathways may offer promising strategies for schizophrenia prevention.
Supplementary material
The supplementary material for this article can be found at http://doi.org/10.1017/S0033291726105595.
Acknowledgments
This research was conducted under UK Biobank application 95114.
Author contribution
Study design and concept: Lan Guo, Xiali Zhong, and Yannan Guo. Data curation: Wanxin Wang and Lijie Li. Formal analysis: Yannan Guo and Xuanni Lu. Methodology and supervision: Yannan Guo, Wanxin Wang, and Ciyong Lu. Interpretation of data: Yannan Guo, Wanxin Wang, and Lijie Li. Writing – original draft: Yannan Guo. Writing – review and editing: Lan Guo, Xiali Zhong, and Yannan Guo. All the authors read and approved the final manuscript as submitted and agree to be accountable for all aspects of the work.
Funding statement
This study was supported by a grant 2025A1515012576 from the Guangdong Basic and Applied Basic Research Foundation. The funding agencies had no involvement in the study design, data collection, data interpretation, report preparation, or manuscript submission-related decisions.
Competing interests
The authors declare no competing interests.
Ethical standard
The authors assert that all procedures contributing to this work comply with the ethical standards of the relevant national and institutional committees on human subjects experimentation and with the Helsinki Declaration.
Open access
