Altered brain and physiological stress responses in early psychosis

Abstract

Stress is proposed to be a crucial factor in the onset and presentation of psychosis. The early stage of psychosis provides a window into how stress interacts with the emergence of psychosis. Yet, how people with early psychosis respond to stress remains unclear. The current study examined how stress responses (brain, physiological, self-report) differ in early psychosis. Forty participants (20 early psychosis [EP], 20 healthy controls [HC]) completed a stress task in the scanner that involved viewing stressful and neutral-relaxing images. Physiological responses (cortisol, heart rate) and self-report of stress were also assessed. Region of Interest analyses were conducted with brain regions previously shown to be activated during the stress task (amygdala, hippocampus, striatum, hypothalamus, prefrontal cortex [dorsolateral, ventrolateral, medial orbital]). Linear mixed models were used to test for effects of group (EP, HC) and emotion (stress, neutral-relaxing). HC had higher hippocampus activation to stress versus neutral-relaxing conditions while EP did not show a difference (group × emotion interaction, p=0.04). There were also significant main effects of group with EP having higher amygdala activation (p=0.01), ventrolateral prefrontal cortex activation (vlPFC, p=0.03), self-report of stress (p=0.01), and heart rate (p<0.001). Our study found preliminary evidence that people with early psychosis showed heightened response to stressful and non-threatening situations, especially in the hippocampus, amygdala, and vlPFC. Our findings suggest a novel perspective on stress alterations in early psychosis and highlight the importance of considering both stressful and non-stressful situations.

Introduction

Numerous psychosis models highlight the important role of stress in psychosis (Corcoran et al., 2003; Pruessner et al., 2017; Walker & Diforio, 1997; Zubin & Spring, 1977). For example, the neural diathesis-stress model of schizophrenia proposes that the onset and progression of psychosis occurs due to a combination of an individual’s vulnerability and the occurrence of stressors. Accumulating evidence suggests that stress responses are altered in individuals with psychosis (Aas et al., 2019; Ciufolini et al., 2014; Lardinois et al., 2011; Mondelli et al., 2010, 2012; Myin-Germeys et al., 2001; Myin-Germeys & van Os, 2007; Pruessner et al., 2017; van Nierop et al., 2018; Walker et al., 2008; Walker & Diforio, 1997). In addition, studies in people at high clinical risk for psychosis provides evidence that stress contributes to the progression of psychosis with stress responses predicting conversion to psychosis (Aiello et al., 2012; Trotman et al., 2014; Walker et al., 2013). However, less is known about how the brain and body respond to stress in the early phase of illness, which provides an important window into mechanisms underlying the emergence of psychosis and can be predictive of long-term outcomes (Birchwood et al., 1998; Pardo-de-Santayana et al., 2020). Thus, the current study aims to examine stress responses using a novel multi-method approach in the critical phase of early psychosis which may provide unique opportunities for earlier detection of risk for psychosis, improvements in interventions, and identification of novel treatment targets for psychotic disorders.

Typical stress reactions in healthy individuals involve an initial multisystem, coordinated response involving the hypothalamus-pituitary-adrenal (HPA) axis, autonomic nervous system, and downstream brain networks. The HPA axis is a crucial stress response system drives a slower response to stress that involves a cascade of hormones released in response to a stressor including cortisol from the adrenal glands (Smith & Vale, 2006). The autonomic nervous system includes the sympathetic nervous system that initiates the “fight or flight” responses mediated by adrenaline and noradrenaline and leads to physiological changes such as increased heart rate, blood pressure, sweating, and pupil dilation. In addition, a broad network of brain regions is engaged in response to stress including the amygdala, hippocampus, cingulate, anterior insula, striatum, and additional regions within the prefrontal cortex (middle, inferior, ventromedial, orbital; Berretz et al., 2021; Dedovic et al., 2009; Sinha et al., 2016).

Aligning with theories proposing the critical role for stress in the emergence of psychosis, individuals with psychosis demonstrate altered stress responses across the multiple systems including higher perceived stress and altered physiological responses. For example, individuals with psychosis self-report more stress sensitivity, especially in response to daily life stressors (Myin-Germeys et al., 2001; Myin-Germeys & van Os, 2007). In physiological studies, individuals with psychosis have heightened baseline stress responses (heart rate and cortisol), and less change in response to stressors (cortisol; Aas et al., 2019; Ciufolini et al., 2014; Lardinois et al., 2011; Mondelli et al., 2010a, 2010b; Myin-Germeys et al., 2001; Myin-Germeys & van Os, 2007; Pruessner et al., 2017; van Nierop et al., 2018). Granted there are some caveats to these findings, such as the impact of specific assessment of cortisol, antipsychotics, and stage of illness (Pruessner et al., 2017).

Rodent models of stress provide additional evidence for the role of stress in the onset and expression of psychosis. For example, the methylazoxymethanol acetate (MAM) rodent model of schizophrenia demonstrates heightened stress sensitivity and anxiety in adolescence. The occurrence of stress during the adolescent stage of MAM rodents leads to a schizophrenia-like phenotype (Gomes et al., 2019). In another rodent model of schizophrenia, MMP-9 heterozygous knockout mice (who have lower MMP-9 activity) showed increased responses to psychosocial stress—including less mobility during a tail suspension test, shorter time in a social interaction, and decreased preference for exploring novel objects, relative to unstressed MMP-9 mice and both stressed and unstressed wild type mice (Vafadari et al., 2019). Together, these models of psychosis risk provide suggest that stress can exacerbate psychosis-like behaviors in vulnerable rodents.

Although it has been proposed that people with psychosis show different behavioral and physiological responses to stress relative to healthy controls (Myin-Germeys & van Os, 2007; Pruessner et al., 2017; Walker et al., 2008; Walker & Diforio, 1997), it remains unclear whether individuals with psychosis have different brain responses to stress. Neuroimaging studies examining stress responses in psychosis provide initial evidence of brain activation differences between people with psychosis and healthy controls. For example, in a neuroimaging study utilizing a mental arithmetic stress in 42 individuals (21 with schizophrenia), healthy controls had increased activation in the amygdala, hippocampus, and anterior cingulate cortex during stress conditions, while people with schizophrenia did not show these increases (Castro et al., 2015). Another neuroimaging study examined brain responses during the Montreal Imaging Stress Task in 33 people with psychosis (18 first episode, 15 high risk). When comparing stress versus control conditions, individuals with or at risk for psychosis displayed increased activation in the anterior insula, and decreases in the ventromedial prefrontal cortex, ventral anterior cingulate cortex, and hippocampus (Vaessen et al., 2023). However, this study did not include a control group for comparison.

While previous research provides important initial evidence, more studies are necessary to understand how the brain and body responds to stress in psychosis especially in the early stages. Previous studies examining stress in psychosis have primarily focused on individuals with a longer duration of psychosis; however, the early stage of psychosis is crucial because it provides a window into the mechanisms underlying the emergence of psychosis without the confounding effects of age-related degeneration, chronic illness, and long-term antipsychotic treatment (Birchwood et al., 1998; Pardo-de-Santayana et al., 2020). The severity of pathology in the early stage of psychosis is also predictive of long-term prognosis. Therefore, understanding stress responses during early psychosis has the potential to inform earlier detection of risk for psychosis, improve current interventions, and help identify novel treatment targets for during this critical phase of illness. In addition, previous studies of stress in psychosis commonly use one or two measures of stress responses, which may lead to misinterpretations about differences in stress response and may not capture the complex nature of multiple stress response systems. A multimodal approach to measuring stress responses will identify the specific levels of stress responses altered and more precisely inform the mechanisms underlying psychosis. To address these knowledge gaps, the current study aimed to determine how people with early psychosis respond to stress across multiple measures including self-reports of stress, physiological measures (cortisol, heart rate), and brain activation (functional magnetic responses imaging, fMRI). We hypothesized that individuals with early psychosis would display higher stress responses including activation in stress-responsive brain regions, heart rate, and self-report of stress relative to healthy controls, as well as less change in cortisol in response to the stressor.

Method

Participants

Participants consisted of 20 individuals in the early stages of psychosis (EP; 5 people with schizophrenia, 6 people with schizophreniform, and 9 people with bipolar disorder with psychotic features) and 20 healthy individuals without a psychiatric disorder. Individuals with psychosis were recruited from the inpatient units and outpatient clinics of Vanderbilt Psychiatric Hospital. Healthy individuals were recruited from advertisements in the community. Participants were considered for study if they were between 18–35 years old, had premorbid IQ > 70 (as determined by the Weschler Test of Adult Reading (Holdnack, 2001), no history of traumatic brain injury, and did not suffer from a chronic medical illness (e.g. HIV, cancer) or a central nervous system disorder. A Structured Clinical Interview of the DSM-IV-TR (SCID; First et al., 2002) was conducted for all participants by trained research staff. For the psychosis group, the SCID was used and supplemented by clinical information to confirm a diagnosis of schizophrenia spectrum disorders or bipolar disorder with psychotic features. Individuals were considered in the early stage of psychosis if they were within three years of the first onset of psychosis. Healthy controls were excluded if they were currently taking any medication for psychiatric symptoms or had any past or current psychiatric illness assessed by the SCID.

Clinical Symptom Measures

In the people with psychosis, symptom severity within the past two weeks was assessed using the Positive and Negative Syndrome Scale (PANSS; Kay, SR, Fiszbein, A, Oplet, 1987). A five-factor model was used to create PANSS factors for Negative, Positive, Disorganized, Excitement and Distress symptoms (Dickinson et al., 2018; van der Gaag et al., 2006; Wallwork et al., 2012). More details on the calulcation of the PANSS five factors are provided in the Supplement.

Stress Task

Participants completed a well-validated stress task (Blaine et al., 2020; Goldfarb et al., 2020; Sinha et al., 2016) in the MRI scanner. The task includes stress and neutral-relaxing blocks that each consisted of baseline (3 minutes of fixation cross), provocation (6 minutes of stress or neutral images), and recovery phases (4 minutes of fixation cross; Figure 1). For baseline and provocation, the phases were broken down into 1.1-minute runs that consisted of 11 images that were each presented for 5 seconds with a 1-second fixation cross in between pictures. More details of the stress task are presented in the Supplement and described in previous papers (see Sinha et al., 2016).

Figure 1:

fMRI Stress Task

The Stress Task contained two blocks: Stress and Neural-relaxing. Each block had a baseline (three 1-minute runs of fixation cross), provocation (six 1-minute runs of images), and recovery phase (4-minute run). Examples of pictures shown for each condition are shown above. The red asterisks represent the three times cortisol was collected throughout each block. Self-report ratings of stress were collected at the beginning of the block, after each scan during baseline and provocation phases, and at the end of the block as indicated by the black lines. Heart rate was collected during each scan during baseline and provocation phases, as indicated by the striped white and black lines.

MRI Data Acquisition, Data Processing and Statistical Analyses

Structural and functional MRI data were acquired on a Phillips 3T Intera Achieva MRI scanner with a 32-channel head coil. Data were processed in SPM12. Individual runs were concatenated across the scan, slice-time corrected, corrected for motion, coregistered, normalized into standard EPI space using the SPM EPI template, high pass filtered, and smoothed (6mm FWHM). First level individual participant GLMs were estimated with baseline stress, baseline neutral-relaxing, provocation stress, and provocation neutral-relaxing conditions. Six motion parameters (3 rotation, 3 translation) were included as covariates of no interest to control for motion. Additional details about data acquisition and processing are provided in the Supplement.

A Region of Interest (ROI) approach was used to measure brain activation in regions that were activated during the stress task in healthy controls (Sinha et al., 2016). ROIs were defined using a 6mm sphere around coordinate with the peak values of coordinates (Supplemental Table 1) in the bilateral hippocampus, bilateral amygdala, bilateral striatum, bilateral dorsolateral prefrontal cortex (dlPFC), bilateral ventrolateral prefrontal cortex (vlPFC), medial orbital frontal cortex, and right hypothalamus. ROIs were averaged across hemispheres when possible (amygdala, hippocampus, striatum, dlPFC, vlPFC; Supplemental Figure 1). To measure brain activation percent signal change was computed for each condition using MarsBar (Brett et al., 2002).

Physiological Measures of Stress

Heart Rate:

Heart rate was assessed continuously throughout the baseline and provocation phases of both the neutral-relaxing and stress blocks using the MRI pulse oximeter. Average heart rate (beats per minute) was calculated at 1.1-minute intervals. Two individuals were missing heart rate data (1 HC, 1 EP) and were not included in the analysis.

Cortisol:

Cortisol was collected 7 times throughout the task: (1) Arrival at the scanner; (2) Following the structural MRI scan (about 15 min after first sample); (3) After provocation phase of block one; (4) After recovery phase of block one; (5) Before starting the second block; (6) After the provocation of block two; (7) After the scan was complete. One individual in the healthy control group was missing a cortisol sample and one person from the early psychosis group was missing two cortisol samples.

Self-Report of Stress

After each run, individuals rated their stress using an MRI-compatible button box. In order to gauge the subjective experience of stress, we asked participants to “rate their level of tension, anxiety and distress while viewing the stimuli presented in the run”. Rating scales consist of numbers 1 through 9, with a rating of 1 indicating “not at all” and a rating of 9 indicating “very much”. Three individuals were missing self-report ratings of stress and were not included in this analysis (2 HC, 1 EP).

Statistical Analyses

Separate linear mixed models (LMM) were used to test for group differences in stress measures. For brain activation, LMM were conducted with emotion contrast (neutral-relaxing vs. baseline/stress vs. baseline) and group (EP/HC) as fixed factors and subject as the random factor. Contrasts were used for the brain activation analyses to be consistent with prior work using this task comparing stress and neutral-relaxing image viewing to baseline to account for differences in implicit baseline (Blaine et al., 2020; Goldfarb et al., 2020; Sinha et al., 2016). For the physiological measures and self-report ratings LMM were conducted with timepoint, emotion (stress/neutral-relaxing), and group (EP/HC) as fixed factors and subject as the random factor. Age, sex, race, and order (stress/neutral-relaxing block first) were included as covariates of no interest. All analyses were conducted in SPSS (Version 28) using an alpha < .05. Cohen’s d effect sizes are reported for significant findings. FDR multiple comparisons corrections were conducted for the brain activation ROI analyses.

Results

Early psychosis and healthy controls groups do not differ in demographics except education

Early psychosis and healthy control groups were not significantly different on age, sex, or parental education. Individuals with early psychosis had fewer years of education relative to healthy controls (Table 1).

Table 1:

Demographics of participants

Measure	Healthy Controls	Early Psychosis	Group Difference (p value)
N	20	20
Age years	23.45 (2.74)	22.85 (3.44)	0.54
Sex N male (%)	12 (60%)	12 (60%)	1.00
Parental education years	14.20 (4.72)	12.18 (6.26)	0.26
Race White/Nonwhite	16/4	12/8	0.30
Education years	15.50 (4.40)	10.75 (6.50)	0.01*
Diagnosis (Affective/Non-affective)	--	9/11
PANSS Total	--	42.85 (10.99)
PANSS Positive	--	5.10 (1.80)
PANSS Negative	--	8.05 (4.77)
PANSS Distress	--	6.80 (2.72)
PANSS Concrete/ Disorganized	--	5.05 (2.09)
PANSS Excitement/Hostility	--	5.10 (1.71)
Number of individuals taking antipsychotics		10 (8 Non-aff; 2 Aff)
Chlorpromazine equivalent		330.89
Number of individuals taking mood stabilizers		7 (3 Non-aff; 4 Aff)
Number of individuals taking antidepressants		7 (4 Non-aff; 3 Aff)
Number of individuals taking benzodiazepines		1 (Non-aff)
Number of individuals taking stimulants		1 (Aff)

^*Non-aff: Non-affective psychosis; Aff: Affective psychosis.

Early psychosis and healthy controls groups differed in brain activation of the hippocampus, amygdala, and vlPFC

As expected, there were significant main effects of emotion for the majority of the brain regions examined (Supplemental Table 2). Across groups, there was higher activation during the stress relative to neutral-relaxing images in the hippocampus (p = 0.01, d = 2.15), amygdala (p < 0.001, d = 1.22), striatum (p = 0.007, d = 0.93), dlPFC (p = 0.047, d = 0.67), and medial orbital PFC (p = 0.03, d = 0.71).

Significant between group differences were observed in three of the brain regions: the hippocampus, amygdala, and vlPFC (Figure 2; Supplemental Table 2). For the hippocampus there was a group × emotion interaction (p = 0.04, d = 0.68). Post hoc analyses revealed that the healthy control group showed a significant main effect of emotion with higher activation to stress versus neutral-relaxing (p = 0.002) but there was no significant difference in the early psychosis group (p = 0.70). For the amygdala and vlPFC, there were main effects of group, with higher activation across emotion conditions in the people with early psychosis relative to healthy controls (amygdala: p = 0.01, d = 0.84; vlPFC: p = 0.03, d = 0.71). Of note, the significant interactions did not withstand corrections for multiple comparisons.

Figure 2:

Groups differed in brain activation of hippocampus, amygdala, and vlPFC

Bar graphs display the predicted values of percent signal change for the three brain regions with significant group effects: hippocampus, amygdala, and ventrolateral prefrontal cortex (vlPFC). The stress condition (stress versus baseline) is shown in red, and the neutral-relaxing condition (neutral-relaxing versus baseline) is in blue. Healthy controls are shown on the left and the early psychosis group on the right. Standard error bars are displayed.

Early psychosis group has elevated heart rate relative to healthy controls but do not differ in cortisol

Heart Rate:

There was a main effect of group for heart rate (Figure 3; p < .001, d = 1.50; Supplemental Table 3). People with early psychosis displayed higher heart rate across emotion and timepoints relative to the healthy controls.

Figure 3:

Elevated heart rate in early psychosis group relative to healthy controls

Average heart rate is displayed for each group for each condition across time. The stress condition is shown on the left and the neutral-relaxing is shown on the right. Healthy controls are shown in blue, and the early psychosis group is in red. Standard error bars are displayed.

Cortisol:

There were no significant group effects for cortisol. There was a significant main effect of timepoint (Supplemental Figure 2, Supplemental Table 4, p < 0.001, d = 1.04). Overall, the cortisol levels decreased over time.

Early psychosis group had higher self-report of stress compared to healthy controls

There was a main effect of group with higher ratings in the early psychosis group relative to healthy controls (F (1,33) = 6.14, p = 0.02, d = 0.86; Supplemental Figure 3; Supplemental Table 5). There was a significant main effect of emotion (p < 0.001, d = 6.33), main effect of timepoint (p < 0.001, d = 1.42), and emotion × timepoint interaction (p < 0.001, d = 1.86). Post hoc analyses demonstrated significant emotion × timepoint interactions for ratings during provocation (all p < 0.001).

Discussion

The current study aimed to determine whether people with early psychosis have alterations in brain activation, physiological responses, or perceived stress during a stressful task. Using a multimodal approach to investigate stress responses, we made two important discoveries. First, people with early psychosis showed altered brain activation in regions of stress modulation (hippocampus, amygdala, and vlPFC). Second, people with early psychosis also had higher stress ratings and heightened heart rate across stress and neutral-relaxing conditions. Our findings provide preliminary evidence on how individuals with early psychosis respond to stress and suggest that differences in stress responses in psychosis lie in the heightened responses to everyday encounters, rather than being limited to situations that would be normally considered stressful. These findings propose a different perspective of the role of stress in early psychosis and may suggest a shift in focus for stress research in psychosis.

The main finding of the study was that people with early psychosis have altered brain activation in the hippocampus, amygdala, and ventrolateral prefrontal cortex relative to healthy controls. For the hippocampus, the healthy controls showed higher responses to stressful relative to neutral relaxing images, as previously shown (Sinha et al., 2016) but the early psychosis group did not show this difference. This finding was due to stronger activation during the neutral-relaxing condition in the early psychosis group compared to the healthy controls. Importantly, the two groups did not differ in responses to stressful stimuli in the hippocampus. Our results align with a previous study in psychosis showing elevated hippocampus activation across neutral and stress conditions in individuals with schizophrenia (Castro et al., 2015). A previous meta-analysis also highlighted heightened responses to neutral stimuli in schizophrenia in the hippocampus, but there were no significant group differences when comparing the emotion versus neutral conditions (Dugré et al., 2019). The hippocampus plays a role in emotional learning and is involved in episodic and contextual memory (Wiltgen et al., 2010) and is particularly vulnerable to the effects of stress (Alfarez et al., 2009; Gould & Tanapat, 1999; Lupien et al., 1998; McEwen, 1999, 2007, 2013; Pruessner et al., 2017). The hippocampus is also affected in psychosis as people with early psychosis with smaller hippocampal volume and altered hippocampal connectivity (Avery et al., 2022; Haijma et al., 2013; Heckers & Konradi, 2010; McHugo et al., 2018). Together, our findings suggest that the hippocampus may be showing a stress-like response to neutral, non-stressful situations in early psychosis, which may interfere with emotional learning and functional memory formation (Tyng et al., 2017).

In addition, people with early psychosis also had heightened amygdala and vlPFC activation to both the stress and neutral-relaxing conditions relative to healthy controls. The amygdala, which projects to the hypothalamus, is a central brain region for stress responses and threat responses more broadly (Shin & Liberzon, 2010). Individuals with psychosis have smaller amygdala volume (van Erp et al., 2018) and heightened amygdala activation to neutral stimuli (Anticevic et al., 2012; Dugré et al., 2019). In the current study, people with early psychosis also displayed stronger activation in the ventrolateral PFC, a region that contributes to emotion processing, emotion regulation, and cognitive control (Badre & Wagner, 2007; Chick et al., 2020; Li et al., 2022). Individuals with psychosis also have decreased volume and activation in the vlPFC during a variety of verbal, emotion, and cognitive tasks (Buchy et al., 2015; Jeong et al., 2009; Kaladjian et al., 2007; Van Erp et al., 2018).

To summarize the brain findings, the early psychosis group showed heightened stress-related activity in the amygdala, but not in the hippocampus, showing no difference between stress versus neutral-relaxing conditions, which could interfere with proper emotional memory formation and retrieval (Wiltgen et al., 2010). Of note, the early psychosis group had elevated hippocampus responses to neutral-relaxing, relative to healthy controls. Early psychosis is marked by heightened levels of negative emotions, stress sensitivity, and paranoia that may be driven by altered evaluation of situations (Freeman et al., 2002; Pruessner et al., 2017; Strik et al., 2018). These symptoms may partly be driven by altered activity amygdala and hippocampus shown in our study. The heightened level of negative emotions is also supported by altered vlPFC activity in early psychosis, a region of cognitive integration and evaluation of emotion. Together these findings suggest that early psychosis show specific patterns of activity in the vlPFC-limbic regions, across neutral and stressful situations, which provides insights into altered emotional processing in early psychosis.

We also observed an overall higher self-report of stress and heightened heart rate in early psychosis relative to healthy controls. More specifically, individuals with early psychosis had a higher self-reported stress and heart rate across both stress and neutral conditions, reflecting upregulated autonomic nervous system (ANS) function. These findings align with previous work showing that people with psychosis have increased stress sensitivity and perceive daily stressors as more stressful than healthy controls (Myin-Germeys et al., 2001; Myin-Germeys & van Os, 2007). People with psychosis also have higher heart rate and decreased heart rate variability, a measure of how well a person adapts to changes in the environment (Clamor et al., 2016; Haigh et al., 2021; Quintana et al., 2016). Increased ANS responses across conditions have been interpreted as basal-state hyperactivity in neurophysiological functions (Hwang et al., 2022). This suggests that individuals with psychosis have heightened general arousal (or stress responses), which was not specific to stressful stimuli or situations. Our brain findings also indicate hyperactivity across conditions in the amygdala and vlPFC in the early psychosis group, suggesting that early psychosis may be characterized by hyperactive basal state. However, there are also other factors that can influence heart rate that are unrelated to the stressor task that may have contributed such as body size, medications, physical activity, and nicotine use (Palatini & Julius, 1997). Future studies should aim to clarify the factors that contribute to heart rate differences in early psychosis.

On the other hand, cortisol levels did not differ between people with early psychosis and healthy controls. Previous studies show that individuals with psychosis have elevated baseline levels of cortisol and less change in cortisol in response to stressors (Aas et al., 2019; Ciufolini et al., 2014; Lardinois et al., 2011; Mondelli et al., 2010, 2012; Pruessner et al., 2017; van Nierop et al., 2018). However, our study did not find a significant group differences in cortisol. This indicates that psychosis in the early phase may be more characterized by brain and autonomic nervous system (ANS) arousal, rather than HPA axis activity. However, it is possible the lack of differences in cortisol response were because of our task design. First, our stressor of threatening images did not effectively evoke a cortisol response, but there was a significant cortisol difference between stress and neutral-relaxing in initial in healthy controls using this task (Sinha et al., 2016). One difference from the previous study is that cortisol was measured using cortisol in the current study rather than using blood. Blood samples of cortisol have higher concentrations than saliva and there may be a greater difference between blood and saliva samples when assessing cortisol in response to a stressor versus baseline levels (Hellhammer et al., 2009). Second, saliva assessments of cortisol could be different when collected in the MRI scanner. For example, other studies have shown cortisol responses to stressors are different in the MRI scanner compared to cortisol responses outside the scanner because the scanning environment may be considered a stressor (Tessner et al., 2006). Third, another potential issue is our baseline assessment was at the MRI scanner, which is an innately stressful situation. In the future it may be helpful to have a period of relaxation prior to the baseline similar to the design of the Triers Social Stress Task (TSST), a well validated stressful task design that reliability induces cortisol responses (Kirschbaum et al., 1993). Future research will be crucial to clarify how cortisol responses differ in individuals with early psychosis.

The current study had limitations. First, the sample size was modest. To mitigate the effects of low statistical power, we focused on a priori regions of interest to increase power and limit the number of tests conducted. However, the significant brain findings did not withstand the FDR correction for multiple comparisons. Despite the preliminary nature of our study, we still believe our findings represent an important initial step in understanding stress in early psychosis using a multi-method approach. In the future, it will be important to replicate these findings with larger samples. Second, the psychosis group included both affective and non-affective psychosis. Future studies should aim to determine if there are differences in stress responses between these two groups. Third, many of the individuals with psychosis were taking medications that may impact stress responses, such as antipsychotics. We provide a detailed account of the medications use in our sample and hope that future studies will determine the potential impact of medication on stress responses in early psychosis. Fourth, the current study did not examine sex differences in stress responses. Sex differences are shown in stress responses in general (Bangasser et al., 2019) and some evidence points to sex differences in stress responses in psychosis (Goldstein et al., 2015). However, our sample was not powered to examine sex differences. Future studies should aim to examine sex differences in stress responses in psychosis as this may lead to important new discoveries. Fifth, the current study excluded individuals with cannabis use disorder, however we did not collect a dimensional assessment of cannabis use. Future studies should aim to examine the impact of cannabis use on stress responses in early psychosis. Sixth, all participants completed the SCID either to diagnosis or rule out the presence of psychiatric disorders. However, the current study did not include a specific diagnostic measure of personality disorders, such as the Structured Interview for DSM-5 Personality Disorders (Frist et al., 2016), or self-report measures of schizotypal personality traits. Future studies should aim to examine how stress responses vary in relation to schizotypal personality traits.

Taken together, the current study provides preliminary evidence on how individuals with early psychosis respond to stress and suggests that early psychosis is characterized by a generally heightened response to both stressful and neutral or non-threatening situations, especially in the ventrolateral-limbic and autonomic nervous systems. This is an important addition to current theories that have suggested alterations in responses to stressors that most people find to be threatening, or an event that challenge homeostasis (Pruessner et al., 2017; Walker & Diforio, 1997). Intriguingly, this perspective aligns with theories of aberrant salience in psychosis that propose heightened stress state overtime alters dopamine levels and mundane, non-threatening information can be marked as salient or threatening (Howes & Nour, 2016; Kapur, 2003; Nudelman & Waltz, 2022). These findings highlight the importance of considering both stressful and non-stressful situations in studying stress responses in early psychosis, as overall heightened stress responses may be a crucial factor in the development and maintenance of psychotic symptoms. These findings may also have clinical implications for context of interventions on stress responses that may suggest coping and regulation techniques be also focused on relaxing in neutral situations. Further research is needed to fully understand the implications of altered stress responses in early psychosis and to develop effective interventions for regulating stress in individuals with early psychosis.