Introduction
Childhood and adolescence are sensitive periods of brain development, during which the body and the brain are subjected to a myriad of different exposures from the external environment. Many of those exposures present as adversity, including both psychosocial (maltreatment, such as abuse and neglect, and family-related adversities) and biological (suboptimal nutrition, pollution, toxins, etc.). Early adversity is particularly impactful on healthy brain developmental trajectory during sensitive periods, such as the period between the ages of 5 and 12, between the ages of 12 and 18, and between 12 and early adulthood, when socioemotional characteristics, executive functions, and self-referential processes develop and mature, respectively (C. A. Nelson, Sullivan, & Valdes, Reference Nelson, Sullivan and Valdes2025). Key brain circuits involving the hippocampus, amygdala, and prefrontal cortex underlie these functions. Their altered developmental patterns can give rise to the development of mood (amygdala and hippocampus) and psychotic (prefrontal cortex) symptoms and eventually to diagnostic psychiatric conditions, such as major depressive disorder and psychosis spectrum disorders (Bora, Yucel, & Pantelis, Reference Bora, Yucel and Pantelis2009; Hastings, Parsey, Oquendo, Arango, & Mann, Reference Hastings, Parsey, Oquendo, Arango and Mann2004; MacQueen et al., Reference MacQueen, Campbell, McEwen, Macdonald, Amano, Joffe and Young2003; Shin & Liberzon, Reference Shin and Liberzon2010). Therefore, it is of key importance to better understand how biological mechanisms may mediate the effects of early adversity, any environmental exposure with negative neurodevelopmental and behavioral impact, on the development of mental health disorders to facilitate early identification of high-risk individuals and to apply appropriate interventions in time.
Early adverse experiences of diverse nature represent deviations from expected inputs from the environment and, therefore, constitute stress for the organism and activate the canonical biological stress response. This includes the rapid activation of the sympathetic nervous system (SNS), resulting in the release of adrenaline from the adrenal medulla and the noradrenaline-mediated effects of the sympathetic nerve terminals throughout the body, followed by the activation of the hypothalamic–pituitary–adrenal (HPA) axis that results in the release of the glucocorticoid hormone cortisol. In addition, immune-inflammatory molecules, such as pro-inflammatory cytokines, are also activated to exert both systemic and brain effects (McEwen, Reference McEwen1998b; McEwen & Stellar, Reference McEwen and Stellar1993). This acute activation of stress response helps the organism to rapidly adapt to the deviations to the expected norms by redistributing blood, and therefore oxygen and energy sources, to tissues essential for the survival (SNS), to mobilize extra energy sources (HPA axis), and to prepare for fighting potentially invading pathogens (pro-inflammatory cytokines), a coordinated biological process conceptualized as allostasis, maintaining stability through change (McEwen, Reference McEwen1998a; McEwen & Stellar, Reference McEwen and Stellar1993). However, as in the case of early adversity, deviations from the expected environmental inputs continue to be present, not allowing the appropriate negative feedback processes to normalize the activity of these biological systems, each producing chemicals with profound effects on fundamental biological processes, with considerable bioenergetic cost over time (McEwen, Reference McEwen1998b, Reference McEwen2007; McEwen & Stellar, Reference McEwen and Stellar1993). The cost of chronic exposure to fluctuating or continually heightened neural and neuroendocrine responses resulting from repeated or chronic environmental challenges is referred to as allostatic load (AL) (McEwen, Reference McEwen1998b, Reference McEwen2007). AL reflects the cumulative effects of experiences in daily life that involve ordinary events, major challenges, such as early adversity and life events, and physiological consequences of health-damaging behaviors (poor sleep, lack of physical activity, and unhealthy diet) (Suvarna et al., Reference Suvarna, Suvarna, Phillips, Juster, McDermott and Sarnyai2020). When these environmental challenges, repeated deviations from the expected norms, exceed the individual’s biological ability to cope, allostatic overload develops, resulting in processes affecting the functional integrity of the body as well as the brain (McEwen, Reference McEwen2007). AL and allostatic overload can be captured by measuring the cumulative impact of stress over time through biomarkers representing the multisystemic nature of the concerted physiological response. This measure, termed AL index, is constructed by identifying, and summing up, the individual biomarkers from the key stress-responsive physiological systems, including the SNS, neuroendocrine, cardiovascular, metabolic, and immune system, that are in either too high or too low, depending on which confers a greater physiological risk for health (McEwen, Reference McEwen1998a; Seeman, Singer, Rowe, Horwitz, & McEwen, Reference Seeman, Singer, Rowe, Horwitz and McEwen1997).
Chronic stress-related increases in the AL index have been linked to a variety of systemic and mental health disorders. For example, elevated AL has been shown to contribute, in the longer term, to the development of cardiovascular diseases (Gillespie et al., Reference Gillespie, Anderson, Zhao, Tan, Kline, Brock and Joseph2019; Guidi, Lucente, Sonino, & Fava, Reference Guidi, Lucente, Sonino and Fava2020), type II diabetes (Carlsson, Andreasson, & Wändell, Reference Carlsson, Andreasson and Wändell2011; Steptoe et al., Reference Steptoe, Hackett, Lazzarino, Bostock, La Marca, Carvalho and Hamer2014), musculoskeletal disorders (Mori et al., Reference Mori, Karlamangla, Merkin, Crandall, Binkley, Greendale and Seeman2014; Smith, Maloney, Falkenberg, Dimulescu, & Rajeevan, Reference Smith, Maloney, Falkenberg, Dimulescu and Rajeevan2009), cancer (Parente, Hale, & Palermo, Reference Parente, Hale and Palermo2013; Ruini, Offidani, & Vescovelli, Reference Ruini, Offidani and Vescovelli2015), and metabolic syndrome (Almadi, Cathers, & Chow, Reference Almadi, Cathers and Chow2013; Osei, Block, & Wippert, Reference Osei, Block and Wippert2022). AL index has been found to be elevated in psychiatric disorders and associated with psychotic symptoms (Berger, Juster, et al., Reference Berger, Juster, Westphal, Amminger, Bogerts, Schiltz and Sarnyai2018; Nugent, Chiappelli, Rowland, & Hong, Reference Nugent, Chiappelli, Rowland and Hong2015; Piotrowski et al., Reference Piotrowski, Kotowicz, Rymaszewska, Beszlej, Plichta, Samochowiec and Misiak2019) and depression (Arnaldo, Corcoran, Friston, & Ramstead, Reference Arnaldo, Corcoran, Friston and Ramstead2022). With relevance to children and adolescents, a significant association has been shown between AL and cognitive functions, such as attention and working memory deficits (D’Amico, Amestoy, & Fiocco, Reference D’Amico, Amestoy and Fiocco2020; Evans & Schamberg, Reference Evans and Schamberg2009; Rogosch, Dackis, & Cicchetti, Reference Rogosch, Dackis and Cicchetti2011). Furthermore, elevated AL has been associated with mental health symptoms (MHS) in children and adolescents (Nelson, Sheeber, Pfeifer, & Allen, Reference Nelson, Sheeber, Pfeifer and Allen2020; Rogosch et al., Reference Rogosch, Dackis and Cicchetti2011).
Beyond the established associations between elevated AL index and already existing physical and mental health conditions, longitudinal studies have shown that elevated AL also predicts later development of impaired functioning, morbidity, and mortality (Guidi et al., Reference Guidi, Lucente, Sonino and Fava2020; Seeman, McEwen, Rowe, & Singer, Reference Seeman, McEwen, Rowe and Singer2001; Seeman et al., Reference Seeman, Singer, Rowe, Horwitz and McEwen1997). Similarly, elevated AL has been associated with the later emergence of mental health disorders, such as mood and psychotic disorders, and impaired cognition (Booth et al., Reference Booth, Royle, Corley, Gow, Valdes Hernandez Mdel, Munoz Maniega and Deary2015; Friedman, Karlamangla, Gruenewald, Koretz, & Seeman, Reference Friedman, Karlamangla, Gruenewald, Koretz and Seeman2015; Gou et al., Reference Gou, Cheng, Kang, Zhou, Liu, Hui and Zhang2025; Kapczinski et al., Reference Kapczinski, Vieta, Andreazza, Frey, Gomes, Tramontina and Post2008; Li, Yan, & Li, Reference Li, Yan and Li2024; Perlman, Cogo-Moreira, Wu, Herrmann, & Swardfager, Reference Perlman, Cogo-Moreira, Wu, Herrmann and Swardfager2022). Recent attention has been paid to linking AL in childhood and adolescence to later health consequences through longitudinal studies, as summarized in a recent systematic review. This review concludes that greater AL is associated with poorer health outcomes in both clinical and non-clinical pediatric populations, with possible long-term effects (Lucente & Guidi, Reference Lucente and Guidi2023). Specifically, elevated AL was shown to be associated with mental health issues, including depression (Nelson et al., Reference Nelson, Sheeber, Pfeifer and Allen2020; Rogosch et al., Reference Rogosch, Dackis and Cicchetti2011). We have shown that although the AL index in childhood was not directly associated with the later development of psychotic disorder symptoms (PDS) and mood disorder symptoms (MDS) in a longitudinal birth cohort, females in the higher tertiles of the AL index measured at 9 years of age had an elevated risk of MHS as young adults (Finlay, Adegboye, McDermott, Rudd, & Sarnyai, Reference Finlay, Adegboye, McDermott, Rudd and Sarnyai2025). This finding raises the possibility that further accumulation of adverse events during adolescence and their impact on systemic stress biology, and hence on AL, might be required to give rise to MHS in early adulthood.
Puberty and following adolescence are sensitive periods for brain development, especially in the domains of socioemotional, executive, and self-referential processes (Nelson et al., Reference Nelson, Sullivan and Valdes2025). With regard to the changes in the HPA axis activity, research demonstrates that puberty is a period of excessive HPA axis plasticity and that the effects of early adversity on cortisol regulation change with pubertal maturation. This concept is formalized in the ‘pubertal stress recalibration hypothesis’, which suggests that for the HPA axis, puberty may reopen the system and functionally create a second sensitive period (DePasquale, Herzberg, & Gunnar, Reference DePasquale, Herzberg and Gunnar2021; Gunnar, DePasquale, Reid, Donzella, & Miller, Reference Gunnar, DePasquale, Reid, Donzella and Miller2019). Furthermore, the adolescent brain is uniquely sensitive to stress, as demonstrated by differential, stress-induced structural changes in the key brain structures of emotional regulation and executive functions, such as the amygdala, hippocampus, and prefrontal cortex, respectively, potentially giving rise to stress-related psychiatric disorders that emerge post-adolescence (Romeo, Reference Romeo2017). Since most psychiatric disorders have their onset before the age of 25 (Kessler et al., Reference Kessler, Berglund, Demler, Jin, Merikangas and Walters2005), it is imperative to better understand how early biological vulnerability may contribute to long-term psychopathology.
Our working model (Figure 1) is based on the accumulated knowledge that chronic stress (early adversity, physiological stressors, and adolescent psychosocial stress) results in the overactivation of the same physiological processes that initially help adaptation (allostasis) and lead to elevated AL, which the established measure of stress-related multisystem dysregulation, the AL index, can capture.
Figure 1. A graphic representation showing how chronic stress may result in the overactivation of stress response, leading to an impact on the structural development of key brain regions, which can contribute to psychopathology. Our working model posits that psychosocial and physiological stressors during childhood and adolescence activate a multisystem physiological stress response characterized by the activation of stress mediators, such as cortisol, norepinephrine, and pro-inflammatory cytokines. These mediators, when produced in excess or chronically over a long period, driven by continuing stress and adversity, impact the structural development of brain areas key to emotional regulation, higher cognitive functions, and behavioral control, and therefore, they can contribute to ensuing psychopathology. Based on this, we hypothesized that elevated AL, the physiological hallmark of chronic stress, in late adolescence, predicts the emergence of mental health symptoms in early adulthood.
Furthermore, the physiological stress mediators, such as cortisol, norepinephrine, and pro-inflammatory cytokines, have been shown to impact the structural development of brain areas key in emotional regulation, higher cognitive functions, and behavioral control, and therefore, they can contribute to ensuing psychopathology. Based on this, we hypothesized that elevated AL, the physiological hallmark of chronic stress, in late adolescence predicts the emergence of MHS in early adulthood. To test this hypothesis, we first examine whether the concerted activation of multiple physiological systems, including cardiovascular, lipid metabolism, glucose metabolism, immune, and anthropometric systems at 17 years, as well as the AL index, was associated with the emergence of MHS before age 25 using data from the Avon Longitudinal Study of Parents and Children (ALSPAC). If such an association can be established, we then aim to explore whether the individual physiological systems included in the AL index are differently linked with the emergence of MHS to identify key systemic processes that may drive later psychopathology.
Methods
Description of cohort and sample selection
Pregnant women (n = 14,541) resident in Avon, United Kingdom, with expected delivery dates from April 1, 1991, to December 31, 1992, were invited to take part in the study and sent a questionnaire to complete (Boyd et al., Reference Boyd, Golding, Macleod, Lawlor, Fraser, Henderson and Davey Smith2013; Fraser et al., Reference Fraser, Macdonald-Wallis, Tilling, Boyd, Golding, Davey Smith and Ness2013; Northstone et al., Reference Northstone, Lewcock, Groom, Boyd, Macleod, Timpson and Wells2019). Of these initial pregnancies, there was a total of 14,062 live births, and of those, 13,988 children were alive at 1 year of age. When the oldest children were approximately 7 years old, an attempt was made to increase the initial sample with eligible cases who had failed to join the study initially. The total sample, including the later enrolment phases, is 14,701 children (alive at 1 year of age). The enrolled mothers and their children have been followed continuously since the study’s inception and continue to be monitored today. Study data were collected and managed using REDCap electronic data capture tools hosted at the University of Bristol (Harris et al., Reference Harris, Taylor, Thielke, Payne, Gonzalez and Conde2009). REDCap is a secure, web-based software platform that supports data capture for research studies. Please note that the study website contains details of all the data that are available through a fully searchable data dictionary and variable search tool. Full cohort details appear elsewhere (Finlay et al., Reference Finlay, Juster, Adegboye, Rudd, McDermott and Sarnyai2023). Our sample group consisted of participants with data on 10–13 biomarkers and available data on MHS before age 26 (n = 1479).
Laboratory measures
For this study, data from 13 biomarkers were obtained to index the functioning of the cardiovascular, immune, and metabolic systems. For details regarding biomarker measurements, collection, and assay procedures, see Finlay et al. (Reference Finlay, Juster, Adegboye, Rudd, McDermott and Sarnyai2023). We included biomarkers measuring cardiovascular functioning, including systolic and diastolic blood pressure and heart rate. Measures of immune functioning included C-reactive protein (CRP). Measures of lipid metabolism included cholesterol, high-density lipoprotein (HDL), low-density lipoprotein (LDL), triglycerides, very-low-density lipoprotein (VLDL), and adiponectin. Glucose metabolism was assessed using fasting glucose and insulin levels. For the anthropometric system, we included body mass index (BMI) (Supplementary Table 1).
Homeostatic model assessment for insulin resistance (HOMA-IR)
HOMA-IR analysis was calculated by fasting glucose (mmol/L) and insulin (mIU/L) levels at age 17 years using the following formula: (glucose × insulin)/22.5. A HOMA score between 0 and 1.9 was considered ‘normal’, 2–2.9 was considered ‘early insulin resistance’ (EIR), and above three was considered ‘insulin resistance’ (IR).
Cutoff points for biomarkers
Each biomarker result was calculated using sample quartiles as cutoffs (i.e., upper or lower quartiles depending on whether low or high values confer a greater risk for health). For all participants, each biomarker was assigned a score of 0 (normal value) or 1 (either above the 75th percentile or below the 25th percentile), depending on the biomarker’s risk ranking. The 25th and 75th percentiles were determined based on the data distribution in participants who did not develop MHS (healthy controls [HCs]).
The allostatic load index calculation
The AL index was calculated at age 17 using the ‘group allostatic load index’ approach (R. P. Juster, McEwen, & Lupien, Reference Juster, McEwen and Lupien2010). Like the above scores, each biomarker result uses sample quartiles as cutoffs (i.e., upper or lower quartiles depending on whether low or high values confer a greater risk for health). Scores were then summed together across all biomarkers for a total score ranging from 0 to 13. Higher total scores indicate greater AL or physiological dysregulation because of chronic stress.
Mental health symptoms
Schizophrenia was assessed using the item ‘ever been diagnosed with schizophrenia’ with a dichotomous answer (yes/no) at the age of 22 years. Psychotic symptoms were assessed using the item ‘psychotic symptoms (ever)’ with a dichotomous answer (yes/no) at the age of 24 years. The Psychosis-Like Symptom Interview (PLIKSi) was conducted by trained psychology graduates in assessment clinics. It was coded according to the definitions and rating rules for the Schedules for Clinical Assessment in Neuropsychiatry. Psychotic episodes (PEs) covered the three main domains of positive psychotic symptoms: hallucinations (visual and auditory), delusions (spied on, persecuted, thoughts read, reference, control, and grandiosity), and thought interference (insertion, withdrawal, and broadcasting). After cross-questioning, interviewers rated PEs as not present, suspected, or definitely present. Cases of PEs in this study were defined as individuals with definite PEs.
Mood disorder was assessed using the survey item ‘ever been diagnosed with depression’ and ‘ever been diagnosed with bipolar disorder’, with a dichotomous answer at the age of 22 years. Furthermore, depressive symptoms were assessed at the age of 25 years using the 13-item Short Mood and Feelings Questionnaire (SMFQ), designed for the assessment of depression symptoms in children and adolescents (Angold, Costello, Messer, & Pickles, Reference Angold, Costello, Messer and Pickles1995). Each item is rated on a three-point Likert scale (0 = not true, 1 = sometimes true, and 2 = true), resulting in a total score between 0 and 26, with higher scores indicating greater depressive symptoms. In this study, scores of 12 or higher were classified as indicative of clinically relevant depressive symptoms (Thapar & McGuffin, Reference Thapar and McGuffin1998). Individuals with both MDS and PDS will be referred to as those with ‘MPDS’ henceforth.
Confounders
The sex of the participants, mother’s occupation (or social class), and mother’s age at delivery were treated as confounders. Sex was recorded at birth and treated as a binary variable (male or female). Mother’s occupation was recorded at birth using the UK Office of National Statistics’ socioeconomic classification system (class I = professional; class II = managerial and technical workers; class IIIa = skilled non-manual occupations; class IIIb = skilled manual occupations; class IV = partly skilled occupations; class V = unskilled occupations). From this point forward, this will be referred to as ‘social class’.
Statistical analysis
All analyses were performed using SPSS v23 and R version 4.0.3. Descriptive analysis was performed and stratified by ‘MHS’. Differences in prevalence between categorical variables were tested using chi-square tests. In case of small sample sizes (<5), Fisher’s exact test was used. In contrast, differences in prevalence between several independent groups were tested appropriately using analysis of variance (ANOVA) or the Kruskal–Wallis test. Linear regression was used to examine the relationship between HOMA grouping (normal, EIR, and IR) and HCs, MDS, PDS, and MPDS groups. The assumptions of linear regression were subsequently checked. To examine the inter-relationship among the five biological systems (cardiovascular, lipid metabolism, immune, glucose metabolism, and anthropometric), gender, social class, mother’s age at birth, and AL index, we conducted an inter-correlation analysis using Pearson’s correlation coefficient. Before conducting the correlation analysis, we checked the assumptions of normality for continuous variables using the Shapiro–Wilk test. Correlation coefficients were interpreted using the guidelines proposed by Cohen (Reference Cohen2013), where r values ranging from 0.10 to 0.29 represent a small effect size, 0.30–0.49 represent a medium effect size, and > 0.5 represent a large effect size. Lastly, we used logistic regression to estimate odds ratios (ORs) and 95% confidence intervals (CIs) to assess the association between AL at age 17 years and the development of MDS, PDS, and MPDS. To determine statistical significance, we used a significance level (alpha) of 0.05 for all tests.
Results
Descriptive statistics for the total sample, controls, and cases
We included 1479 participants in this study, with more than ten biomarkers collected and reported. Of these, 542 individuals had any MHS (36.64%).
There were more females than males in the MHS group (65% versus 35%). Mother’s age at delivery did not differ between the HCs and the MHS group (29.91 versus 29.88; p = 0.92), and neither did the percentage distribution of individuals in the different social classes (high = 50% versus 46%, middle = 44% vs 45%, and low = 6% and 8.4%, p = 0.21, respectively).
Contribution of different biological systems in the allostatic load index
By using the four biological systems investigated, ‘lipid metabolism’, ‘cardiovascular’, ‘immune’, and ‘HOMA’, we created the system-specific AL indices. In Figure 2, a positive correlation coefficient value indicates that as the value of one variable (such as mother’s age at delivery) increases, the other variable (e.g., AL index) also increases. We observed significant correlations between various biological systems formulating the AL index. For example, ‘lipid metabolism’ exhibited a stronger positive correlation with the AL index compared to ‘cardiovascular system’ (r = 0.777 vs 0.575, respectively).
Figure 2. Correlation coefficient values indicate the independent variable’s positive or negative influence on the dependent variable (AL index at age 17). Significance is shown in the figure as ‘***’ is p < 0.001 and ‘**’ is p = 0.005.
The allostatic load index in MDS, PDS, and MPDS
The AL index at age 17 was significantly higher in individuals who developed MHS by 25 years of age than the HCs (3.96 versus 3.67, p = 0.01) (Figure 3a). In Figure 3b, it can be observed that the AL index is higher at age 17 in individuals who developed either MDS, PDS, or MPDS compared to the HCs and that the AL index is proportionally increasing with the severity of the disorder (from MDS to MPDS). A sub-analysis showed that the odds of developing MDS, PDS, and MPDS in relation to AL were only significantly increased for MPDS (p = 0.03). In contrast, associations with MDS and PDS were not significant (p = 0.12 and 0.21, respectively) (Supplementary Table 2).
Figure 3. (a) Mean AL index at age 17 for healthy controls (HCs) vs individuals with any mental health symptoms (MHS). (b) Mean AL index at age 17 for HCs, individuals with mood disorder symptoms (MDS), individuals with psychotic disorder symptoms (PDS), and individuals with both mood and psychotic disorder symptoms (MPDS). Significant values are shown on the graphs.
Separating MDS, PDS, and MPDS
Of the 542 individuals with MHS, 410 participants (75.6%) reported having MDS, 87 participants (16%) reported having PDS, and 45 individuals (8.3%) reported having MPDS (Supplementary Table 3). More females than males developed MDS (67%) before 25 years of age (p = 0.01). A similar, but not significant, number was observed for the MPDS group, with 73% females developing this (p = 0.07). In contrast, more males than females developed PDS (51%) before 25 years of age (p = 0.05).
Mother’s age at delivery did not differ between the HCs, and MDS, PDS or MPDS (29.91 ± 0.14 versus 30.02 ± 0.22 (p = 0.71) versus 29.27 ± 0.48 (p = 0.29) versus 29.82 ± 0.65 (p = 0.98), respectively). Similarly, the percentage of individuals in the three social class groups (low, middle, and high) was not different between the HCs (low = 6%, middle = 44%, and high = 50%), and those with MDS (low = 9.3%, middle = 42.7%, and high = 48%; p = 0.14), PDS (low = 8.2%, middle = 49%, and high = 42%; p = 0.31) and MPDS (low = 0%, middle = 62%, and high = 38%; p = 0.08). HOMA scores were significantly higher in individuals who developed MDS or PDS, compared to the HCs (p = 0.03 and p = 0.01, respectively) (Figure 4).
Figure 4. HOMA index between healthy controls (HCs), mood disorder symptoms (MDS), psychotic disorder symptoms (PDS), and mood and psychotic disorder symptoms (MPDS). Significance is shown on the graph.
The cardiovascular biomarkers were the highest for individuals with MPDS among all groups (MPDS mean = 0.93 ± 0.14 versus PDS mean = 0.89 ± 0.10 versus MDS mean = 0.80 ± 0.04 versus HC mean = 0.78 ± 0.02) (Supplementary Table 3).
The lipid biomarkers were the highest for individuals with PDS among all groups; however, there was no significance between the MPDS, PDS, and MDS groups versus the HCs (MPDS mean = 2.33 ± 0.25 versus PDS mean = 2.48 ± 0.16 versus MDS mean = 2.11 ± 0.07 versus HC mean = 2.06 ± 0.04), p = 0.53, 0.12, 0.62, respectively) (Supplementary Table 3).
The immune biomarker CRP was the highest for individuals with MPDS among all groups. However, there was no significance between the MPDS, PDS, and MDS versus the HCs (MPDS mean = 2.61 ± 1.12 versus PDS mean = 1.41 ± 0.32 versus MDS mean = 1.36 ± 0.21 versus HC mean = 1.46 ± 0.14); p = 0.13, 0.77, 0.21, respectively) (Supplementary Table 3).
The overall metabolic biomarker BMI was the highest for individuals with MDS, closely followed by individuals with MPDS (MPDS mean = 22.62 ± 0.54 versus PDS mean = 22.54 ± 0.44 versus MDS mean = 22.63 ± 0.18 versus HC mean = 22.29 ± 0.11).
Homeostatic model assessment for insulin resistance (HOMA-IR)
Analysis of the HOMA-IR showed a significant difference in the number of individuals with ‘normal’ HOMA levels, HOMA levels indicating ‘EIR’, and HOMA levels indicating ‘IR’ between the HCs and those with MHS (p < 0.001) (Figure 5a,b). The results from the three MHS groups, MDP, PDS, and MPDS, indicated a significant difference between the HCs and the MDS and PDS groups (p = 0.01, 0.01, respectively), but no difference between the HCs and the MPDS group (p = 0.10) (Figure 5c-e). Using linear regression, we analyzed the three levels ‘normal’, ‘EIR’, and ‘IR’ separately between the HCs, and MDS, PDS, and MPDS. There was a significant difference in the distribution of HOMA categories between individuals with PDS versus the HCs (p = 0.01). Compared to the HCs, individuals with PDS had higher rates of IR (16% vs 8%) and EIR (21% vs 16%). Although the MDS group had slightly more individuals in the EIR group (21% vs 16%) and the IR group (11% versus 8%) compared to the HCs, these differences were not significant (p = 0.21 and 0.09). Individuals with MPDS were significantly more likely to fall into the IR and EIR categories compared to HCs (Fisher’s exact test, p = 0.028). Specifically, 31% had EIR compared to 16% among the HCs, and 9.5% had IR compared to 8% among the HCs.
Figure 5. Percentage of individuals with HOMA levels indicating ‘normal’ (white), ‘early insulin resistance’ (light grey), and ‘insulin resistance’ (dark grey) within the groups. (a) HCs, ‘Healthy controls’; (b) MHS, ‘mental health symptoms’. (c) MDS, ‘mood disorder symptoms’; (d) PDS, ‘psychotic disorder symptoms’ and (e) MPDS, ‘mood and psychotic disorder symptoms’.
Discussion
This study investigated whether late adolescence allostatic multisystem dysregulation influenced by a diverse variety of past stressors is linked to MHS in early adulthood in a large population-based prospective birth cohort. We found that: (1) elevated AL at age 17 was significantly associated with the co-occurrence of MPDS in young adulthood; (2) late adolescence AL gradually increased in relation to MHS severity in early adulthood from MDS to PDS to MPDS; (3) the HOMA index at age 17 was significantly elevated in individuals who developed either MDS or PDS compared to the HCs; and (4) IR was significantly more prevalent in the MHS group compared to the HCs. Furthermore, IR was the highest in the PDS group.
This is the first study to investigate the link between late adolescence multisystem dysregulation and MHS in early adulthood.
Previous studies have investigated the AL index in patients who are already diagnosed with psychotic and mood disorders, such as schizophrenia (Berger, Lavoie, et al., Reference Berger, Lavoie, McGorry, Nelson, Markulev, Yuen and Amminger2018; Nugent, Chiappelli, Rowland, & Hong, Reference Nugent, Chiappelli, Rowland and Hong2015; Piotrowski et al., Reference Piotrowski, Frydecka, Kotowicz, Stańczykiewicz, Samochowiec, Szczygieł and Misiak2020; Savransky et al., Reference Savransky, Chiappelli, Fisseha, Wisner, Xiaoming, Mirmomen and Hong2018), bipolar disorder (Dargel et al., Reference Dargel, Volant, Brietzke, Etain, Olie and Azorin2020; Juster, Sasseville, Giguère, Consortium, & Lupien, Reference Juster, Sasseville, Giguère, Consortium and Lupien2018), and major depressive disorder (Honkalampi et al., Reference Honkalampi, Virtanen, Hintsa, Ruusunen, Mantyselka, Ali-Sisto and Lehto2021; Kobrosly, van Wijngaarden, Seplaki, Cory-Slechta, & Moynihan, Reference Kobrosly, van Wijngaarden, Seplaki, Cory-Slechta and Moynihan2014) and found significantly increased AL compared to controls. Recent studies have investigated the immune biomarkers IL-6 and CRP in childhood (age 9) in the same ALSPAC study and found that increased IL-6, but not CRP, was associated with an increased risk of PDS and MDS in adulthood (Chu et al., Reference Chu, Stochl, Lewis, Zammit, Jones and Khandaker2019; Khandaker, Pearson, Zammit, Lewis, & Jones, Reference Khandaker, Pearson, Zammit, Lewis and Jones2014; Perry, Zammit, Jones, & Khandaker, Reference Perry, Zammit, Jones and Khandaker2021). Another study by Perry et al. (Reference Perry, Upthegrove, Thompson, Marwaha, Zammit, Singh and Khandaker2018) found that IR at the age of 9 significantly increases the risk of developing PEs at age 18. Our recent study on AL in 9-year-olds did not find an overall association between childhood AL and the development of MHS in early adulthood, but it identified that females in the highest AL index tertile had an elevated risk of MHS in young adults in the same longitudinal birth cohort (Finlay et al., Reference Finlay, Adegboye, McDermott, Rudd and Sarnyai2025). However, our present results indicate that individuals with MHS in early adulthood already had a higher mean AL at 17 years of age compared to the HCs. Interestingly, mean AL at 17 years gradually increased with the severity and complexity of the symptoms in adulthood (MDS = 3.89, PDS = 4.03, and MPDS = 4.47).
Although observational studies implicate metabolic dysfunction as a risk factor for psychiatric disorders, the possibility of reverse causation cannot be ruled out. Mendelian randomization studies have provided mixed support, indicating that the genetic liability for depression or schizophrenia may causally influence inflammatory and metabolic biomarkers (Perry, Burgess, et al., Reference Perry, Burgess, Jones, Zammit, Upthegrove, Mason and Jones2021; Zhang, Chen, Yin, Wang, & Peng, Reference Zhang, Chen, Yin, Wang and Peng2021). Perry, Burgess, et al. (Reference Perry, Burgess, Jones, Zammit, Upthegrove, Mason and Jones2021) conducted a bidirectional study and found evidence that suggests schizophrenia may causally influence IR and inflammation. Similarly, the study by Zhang et al. (Reference Zhang, Chen, Yin, Wang and Peng2021) found that metabolic changes could also be a consequence, as well as a cause, of depressive disorder. These results highlight a possible bidirectional relationship between metabolic dysfunction and psychiatric disorders.
Looking at the results for each biological system, individuals who developed MPDS in adulthood had dysregulation in all but the lipid metabolism system. This trend could also be observed in the cumulative AL score, which was higher for the MPDS group than for the MDS, PDS, and HC groups. This suggests that there might be a continuum in the severity of mental health disorders with regard to their relationship with stress-related multisystem physiological dysregulation. The ‘continuum’ model captures this view by showing that mental health disorders exist along a continuum of severity rather than within discrete diagnostic categories. This perspective acknowledges that psychological symptoms vary in intensity and impact, from mild, subclinical distress to severe, chronic illnesses (Keyes, Reference Keyes2002). One end of the spectrum is characterized by optimal well-being and high functioning, whereas the opposite end involves significant functional limitations, chronicity, and greater clinical needs (Widiger & Samuel, Reference Widiger and Samuel2005). Individuals may move along the continuum and experience mood fluctuations and other MHS and, according to our present results, may exhibit a continuously increasing multisystem dysregulation.
Our results indicate, in agreement with the findings by Perry, Stochl, et al. (Reference Perry, Stochl, Upthegrove, Zammit, Wareham, Langenberg and Khandaker2021) and Perry et al. (Reference Perry, Upthegrove, Thompson, Marwaha, Zammit, Singh and Khandaker2018), that the HOMA score, showing the degree of dysregulation of glucose–insulin homeostasis, at the age of 17 years was significantly higher in individuals who developed MDS, PDS, or MPDS in adulthood compared to the HCs. Interestingly, other systems, when looked at separately, did not show a difference between the HCs and individuals who developed MHS. It is conceivable that the dysregulation of the glucose–insulin homeostasis precedes, and perhaps even drives, the abnormalities in other physiological systems. In fact, there is an emerging view that IR can act as an upstream driver of dysfunction in both immune and lipid metabolic systems (Hotamisligil, Reference Hotamisligil2006; Lumeng & Saltiel, Reference Lumeng and Saltiel2011; Samuel & Shulman, Reference Samuel and Shulman2012). The effect of IR on the brain may account for some of the changes seen in patients with MHS. The bidirectional relationship between metabolic dysregulation and mental health disorders has been intensively studied in recent years (Maksyutynska et al., Reference Maksyutynska, Stogios, Prasad, Gill, Hamza, De and Agarwal2024; Penninx & Lange, Reference Penninx and Lange2018; Pillinger et al., Reference Pillinger, Beck, Gobjila, Donocik, Jauhar and Howes2017; Saxena & Marin-Valencia, Reference Saxena and Marin-Valencia2024). Our findings show that EIR and IR at age 17 were more prevalent in individuals who developed MHS in adulthood compared to the HCs. In line with this finding, Watson et al. (Reference Watson, Simard, Henderson, Nutkiewicz, Lamers, Nasca and Penninx2021) measured IR three times over 9 years in adults without a mental health history and found that a moderate increase in the measures of IR would correspond to an 89% increase in the rate of major depressive disorder over the following 9 years.
Elevated fasting glucose levels have been extensively demonstrated in first-episode, never-medicated individuals with psychotic disorders (Pillinger et al., Reference Pillinger, Beck, Gobjila, Donocik, Jauhar and Howes2017) and in individuals with depression (Kan et al., Reference Kan, Silva, Golden, Rajala, Timonen, Stahl and Ismail2013). It is not known at present whether such elevation in glucose levels is driven by systemic factors or by impaired brain glucose metabolism and brain energy production through an attempted systemic compensatory effort (Sarnyai & Ben-Shachar, Reference Sarnyai and Ben-Shachar2024). Several biological mechanisms could link IR to the onset and progression of depression and psychotic disorders. These include IR-driven neuroinflammation (Leonard & Wegener, Reference Leonard and Wegener2020), central nervous pathways, i.e., brain IR (Kullmann et al., Reference Kullmann, Heni, Hallschmid, Fritsche, Preissl and Häring2016), neurotransmitter dysregulation (de Bartolomeis et al., Reference de Bartolomeis, De Simone, De Prisco, Barone, Napoli, Beguinot and Fornaro2023), and shared genetic vulnerability (Perry et al., Reference Perry, Bowker, Burgess, Wareham, Upthegrove, Jones and Khandaker2022). These mechanisms can be linked to one another as it has been shown that IR appears to promote neuroinflammation by impacting neurotransmitters, such as serotonin, dopamine, and norepinephrine, which are closely involved in regulating mood, behavior, and stress (Gold, Licinio, & Pavlatou, Reference Gold, Licinio and Pavlatou2013; Webb et al., Reference Webb, Davies, Ashra, Bodicoat, Brady, Webb and Khunti2017). IR often leads to chronic inflammation in the brain and body, and cytokines such as IL-6, CRP, and TNF-alpha are often elevated in patients with IR (Phosat et al., Reference Phosat, Panprathip, Chumpathat, Prangthip, Chantratita, Soonthornworasiri and Kwanbunjan2017). These cytokines can cross the blood–brain barrier, affecting brain function and contributing to MHS. Interestingly, our results (Supplementary Table 3) did not indicate a significant dysregulation in the immune system, as measured by possible alterations in CRP levels, in adolescents who later developed MDS, PDS, or MPDS compared to the HCs. In fact, CRP was lower in individuals who later developed MDS or PDS than in the HCs. Other studies have discussed this relationship, but no agreement has been established, as some studies show elevated CRP levels, while others find no difference or, in line with our findings, point toward a negative association (Dessoki et al., Reference Dessoki, Khattab, Moris, Abdelhakim, Lotfy and Salah2023; Liu, Ely, Simkovic, Alonso, & Gabbay, Reference Liu, Ely, Simkovic, Alonso and Gabbay2021). As CRP was the only immune biomarker in this study, it is important to interpret our results cautiously. Given CRP, a key acute-phase protein involved in the inflammatory response and influenced by the actions of cytokines, was the only immune/inflammatory biomarker used, it is impossible to establish whether immune dysregulation was already prevalent at age 17. Elevated IL-6 is often correlated with increased IR, so future studies should include multiple inflammatory biomarkers for a comprehensive understanding of immune changes (Yuan et al., Reference Yuan, Yang, Yao, Song, Zhou, Liu and Zhang2024). IR is also associated with the dysregulation of glucocorticoids, such as cortisol, in the HPA axis, which is shown to be associated with depression (Mansur, Brietzke, & McIntyre, Reference Mansur, Brietzke and McIntyre2015), and IR has been shown in psychotic disorders (Meyer & Stahl, Reference Meyer and Stahl2009; Ray & Khess, Reference Ray and Khess2016). Our findings indicate that 21% of individuals who developed PDS in adulthood had IR at 17 years, compared to 8% of those who did not exhibit any MHS in early adulthood, which translates to an over 2.5 times increased risk of developing PDS in early adulthood, if the individuals already show IR in late adolescence.
In light of the evidence reviewed above, our present results suggest that accumulating stress and adversity during childhood and adolescence may lead to the development of IR early on, possibly contributed by the repeated and long-term exposure to stress-induced increase in glucocorticoids (Kivimäki, Bartolomucci, & Kawachi, Reference Kivimäki, Bartolomucci and Kawachi2023; Pervanidou & Chrousos, Reference Pervanidou and Chrousos2012). The emerging IR, then, in turn, may drive, together with the direct effect of repeated stress, the dysregulation of other major physiological systems, resulting in an increased AL. These factors together may then contribute to changes in adolescent brain development, especially in late-maturing systems, such as the prefrontal cortex, and result in alterations in brain function that can manifest as MHS in later life (Lenart-Bugla et al., Reference Lenart-Bugla, Szcześniak, Bugla, Kowalski, Niwa, Rymaszewska and Misiak2022). Addressing both IR and elevated AL through metabolic interventions, such as diet and lifestyle changes, early psychotherapy, and careful monitoring of health, may help reduce the risk and likelihood of developing MHS and also improve overall mental and physical health (Finlay, Rudd, McDermott, & Sarnyai, Reference Finlay, Rudd, McDermott and Sarnyai2022).
Strengths and limitations
The strengths of this work include the longitudinal design and the inclusion of multiple measures that are informative of multisystemic dysregulation within the AL concept and possible confounders for MDS and PDS. A key limitation of the study is the large number of missing values, a common issue for prospective cohort studies (Pham, Pandis, & White, Reference Pham, Pandis and White2024). Our sample population was significantly small, as we only included participants with more than 10 biomarkers reported to represent AL accurately. Despite this, we found a significant difference in AL for the HCs compared to individuals with MHS, both as a group and separated into individuals with MDS, PDS, and co-occurrence of MPDS. A second limitation is related to the timing of the development of MHS, which were collected and reported before the age of 26 years. PDS and MDS are often diagnosed during adolescence and early adulthood, so several individuals may have been wrongly categorized as ‘HCs’ if they did not display symptoms before the age of 26 (Avenevoli, Swendsen, He, Burstein, & Merikangas, Reference Avenevoli, Swendsen, He, Burstein and Merikangas2015; Tandon, Keshavan, & Nasrallah, Reference Tandon, Keshavan and Nasrallah2008). Third, while biomarker cutoffs were derived using the HC group to establish a normative reference for the AL index, we acknowledge this may introduce circularity and reduce generalizability. Future research should consider prospective methods to define biomarker thresholds and include more biomarkers, including stress-related hormonal biomarkers, such as cortisol, and a more extensive profile of immune biomarkers, to accurately examine the neuroendocrine stress system and inflammation prior to MHS, creating a more accurate AL. Lastly, socioeconomic status was based on the mother’s occupation categories at recruitment, which may not reflect broader or more nuanced indicators of socioeconomic status.
Conclusion
In conclusion, our findings indicate a significant association between biomarker changes representing multisystem dysregulation due to past demands, stress, and trauma on the organism over time during childhood and adolescence and the later emergence of MHS in adulthood. Early systemic metabolic dysregulation may be a significant risk factor in the development of mental health disorders, especially for psychotic disorders. Our results indicate that elevated HOMA, indicating IR, and AL may identify individuals at increased risk of developing mental health disorders in later years, even before displaying relevant symptoms.
Supplementary material
The supplementary material for this article can be found at http://doi.org/10.1017/S0033291725102377.
Data availability statement
ALSPAC data access is through a system of managed open access. The steps below highlight how to apply for access to the data referred to in this article and all other ALSPAC data. The datasets presented in this article are linked to ALSPAC project number B3847. The ALSPAC variable codes highlighted in the dataset descriptions can be used to specify the required variables.
-
• Please read the ALSPAC access policy (https://www.bristol.ac.uk/media-library/sites/alspac/documents/researchers/data-access/ALSPAC_Access_Policy.pdf), which describes the process of accessing the data and samples in detail and outlines the costs associated with doing so.
-
• You may also find it useful to browse our fully searchable research proposals database (https://proposals.epi.bristol.ac.uk/), which lists all research projects that have been approved since April 2011.
-
• Please submit your research proposal for consideration by the ALSPAC Executive Committee. You will receive a response within 10 working days to advise you whether your proposal has been approved.
The study website also contains details of all the data that are available through a fully searchable data dictionary: http://www.bristol.ac.uk/alspac/researchers/data-access/data-dictionary/.
Acknowledgments
We are extremely grateful to all the families who took part in this study, the midwives for their help in recruiting them, and the whole ALSPAC team, which includes interviewers, computer and laboratory technicians, clerical workers, research scientists, volunteers, managers, receptionists, and nurses.
Funding statement
The UK Medical Research Council and Wellcome (Grant ref: 217065/Z/19/Z) and the University of Bristol provide core support for ALSPAC. This publication is the work of the authors, and Sabine Finlay will serve as guarantor for the contents of this paper.
A comprehensive list of grant funding is available on the ALSPAC website. This research was specifically funded by Wellcome Trust and MRC (core) (Grant ref: 076467/Z/05/Z), British Heart Foundation (Grant ref: CS/15/6/31468), NIH (Grant ref: PD301198-SC101645), and MRC (Grant ref: MR/M006727/1).
The Australian Institute of Medical Scientists funded this work (Grant ref: REGS-29277-2021).
Competing interests
The authors declare none.
Ethical standard
Ethical approval for the study was obtained from the ALSPAC Ethics and Law Committee and local research ethics committees. Informed consent for data collected via questionnaires and clinics was obtained from participants following the recommendations of the ALSPAC Ethics and Law Committee at the time. Children were invited to give assent where appropriate. Study participants have the right to withdraw their consent for elements of the study or from the entire study at any time. At age 18, study children were sent ‘fair processing’ materials describing ALSPAC’s intended use of their health and administrative records and were given clear means to consent or object via a written form. Data were not extracted for participants who objected or who were not sent fair processing materials. Consent for biological samples was collected in accordance with the Human Tissue Act (2004). Full details of the ALSPAC consent procedures are available on the study website (http://www.bristol.ac.uk/alspac/researchers/research-ethics/).
Ethical approval for the study was obtained from the ALSPAC Law and Ethics Committee and local research ethics committees (NHS Haydock REC: 10/H1010/70).
Further ethics approval was obtained from the Human Research Ethics Committee at James Cook University, Australia (H8535).
Consent for publication
Participants were provided with fair processing materials describing the studies’ use of the data they provided or those collected through record linkage and about the legal basis under which the study operates, including the sharing of deidentified data with researchers and the publishing of research findings. Study members had the right to withdraw from elements of the study or from the entire study at any time. Full details of the ALSPAC consent procedures are available from the study website.
Open access
