Hippocampal Subfields, Daily Stressors, and Resting Cortisol in Individuals at Clinical High-Risk for Psychosis

Abstract

Introduction

The hippocampus, comprised of functionally distinct subfields, both regulates stress and is affected by it during psychosis pathogenesis. Hippocampal abnormalities are evident across psychosis spectrum and are associated with aberrant cortisol levels and greater environmental stressors exposure. These associations, particularly at the subfield-level, are poorly understood in individuals at clinical high-risk (CHR) for psychosis. This represents a significant literature gap given this critical pathogenetic period is characterized by an interplay between environmental stressors and biological susceptibility.

Methods

A total of 121 participants including 51 CHR (mean age=18.61) and 70 healthy controls (HC; mean age=18.3) were enrolled in the study. Participants completed a structural scan, salivary cortisol assays, and a self-report measure assessing distress from daily stressors exposure (DSI). Hippocampal subfield segmentation was conducted using Freesurfer.

Results

Smaller hippocampal subfields were associated with greater stress levels. Greater DSI was associated with lower volumes in CA1 (r=−.38) and CA2/3 (r=−.29), but not in CA4/DG (r=−.28), presubiculum (r=−.09), or subiculum (r=−.17). Higher resting cortisol was associated with lower volumes in presubiculum (r=−.4) but not subiculum (r=−.22), CA1 (r=.08), CA2/3 (r=.1), or CA4/DG (r=−.005). Regressions indicated effects for CA1 and DSI (β=.57, p=.03) and presubiculum and cortisol (β=.61, p=.02) are specific to CHR participants relative to HCs.

Conclusions

The findings provided insights into links between stress and brain vulnerability during psychosis-risk period. Regional differences highlighted potentially different mechanisms by which stress impacts specific subfields. Presubiculum may be more susceptible to the impact of early stress on HPA-axis and cornu amonis to acute stressors.

1. Introduction

A robust literature implicates the hippocampus as an important structure affected by stress in the pathogenesis of psychotic disorders such as schizophrenia. Both biological and environmental indicators of stress are associated with reduced hippocampal volumes in individuals with psychosis (Collip et al., 2013). Evidence suggests that hippocampal subfields particularly affected by stress are cornu amonis (CA)1, CA2/3 and CA4/dentate gyrus (DG) (Haukvik et al., 2018; Vargas et al., 2018). In terms of biological indicators of stress, studies have found that elevated cortisol levels are associated with reduced hippocampal volumes in individuals with psychosis (Collip et al., 2013; Mondelli et al., 2011; Mondelli et al., 2010). Furthermore, exposure to acute psychosocial stressors such as daily hassles has been examined in the context of observed hippocampal atrophy in psychosis. Momentary stress, such as that experienced from exposure to recent stressors, has been associated with reduced hippocampal volumes in schizophrenia (Collip et al., 2013; Mondelli et al., 2011). Evidence suggests that hippocampus is affected (Dean et al., 2016; Vargas et al., 2019), cortisol levels altered (Walker et al., 2013), and exposure to environmental stress prominent (Devylder et al., 2013; Ristanovic et al., 2020) even prior to illness onset. Yet, investigations on the critical links between stress and hippocampal volumes during the psychosis risk period are lacking. Individuals at clinical high-risk (CHR) for psychosis are characterized by attenuated symptoms and emerging functional and cognitive declines and are considered at imminent risk for transitioning to a formal psychotic disorder such as schizophrenia (Cannon et al., 2008). This population is a promising sample for examining stress and brain morphology considering the absence of confounds such as exposure to antipsychotic medications and acute symptoms.

Progressive hippocampal atrophy is well documented in individuals in later stages of and with chronic psychosis indicating it as a prominent marker of illness (Haukvik et al., 2018; Ho et al., 2017b; Velakoulis et al., 2006). Evidence from populations earlier in the course of illness is less resounding. While some studies note subtle reductions in total hippocampal volume in first episode psychosis (Velakoulis et al., 2006), it tends to remain unchanged in the first few years following psychosis onset (Haukvik et al., 2016). Similarly, a meta-analysis of studies on populations at high-risk for psychosis revealed no significant reductions in the overall hippocampal volume, concluding atrophy to the hippocampus is not a risk marker (Walter et al., 2016). However, more recent investigations focusing on hippocampal subfields rather than on total volume indicated that different regions of the structure are affected in individuals in early course psychosis (McHugo et al., 2020) and those at high-risk (Ho et al., 2017a; Vargas et al., 2018). Specifically, CA volumes appear to be particularly affected prior to illness onset as they are consistently reduced compared to healthy controls (Sasabayashi et al., 2021; Vargas et al., 2018). Moreover, progressive reductions of volume in the CA are associated with worsening of illness (Ho et al., 2017a). Taken together, these findings indicate specific parts of the hippocampus may be a risk marker for psychosis and relate to illness progression.

The role of altered HPA-axis functioning, particularly related to the production and regulation of cortisol and other stress-related hormones has also been well established in the psychosis literature (Hubbard and Miller, 2019). Extant literature in individuals at CHR for psychosis indicates significant differences in both basal and acute cortisol levels relative to healthy controls (Walker et al., 2013). For example, when faced with an acute stressor, those at risk for psychosis present with attenuated cortisol response (Pruessner et al., 2013). In terms of resting cortisol, evidence suggests elevated levels compared to HCs (Carol and Mittal, 2015; Cullen et al., 2020; Walker et al., 2013). Moreover, elevated cortisol levels are associated with transition from the high-risk state to formal psychosis (Cullen et al., 2020). In terms of acute environmental stress, exposure to daily stressors has been found to contribute to the observed alterations of the stress system. Compared to HCs, individuals at CHR for psychosis endorse experiencing more frequent daily stressors and elevated distress in response to the exposure (Cullen et al., 2020). Taken together, these findings indicate the stress system is less capable of regulating biological and environmental stressors prior to psychosis onset.

Critically, stress is mechanistically involved in the changes in the hippocampus via dysregulation of the hypothalamus-pituitary-adrenal (HPA) axis, characterized by abnormal levels of glucocorticoids and aberrant stress signaling (Sapolsky, 2000). Evidence from psychosis literature indicates that higher levels of cortisol are associated with lower hippocampal volumes in both first-episode (Mondelli et al., 2010; Pruessner et al., 2015) and later stages of psychosis (Collip et al., 2013). Similarly, environmental stressors are associated with reductions in hippocampal volumes in individuals with early course psychosis (Mondelli et al., 2011) as well as later in the course of illness (Aas et al., 2014; Collip et al., 2013). However, the relationship between stress levels and hippocampal volumes is not well understood prior to psychosis onset. One study examining overall hippocampal volume found that in male participants at CHR for psychosis, blunted cortisol awakening response was associated with smaller volume in left and right hippocampus (Pruessner et al., 2017a). However, no study to date has reported on links between subfields, rather than overall volume, and stress in individuals at CHR for psychosis. Different subfields have varying levels of sensitivity to changes in cortisol levels (Han et al., 2005); therefore, utilizing hippocampal segmentation as a tool for examining associations between hippocampus morphology and indicators of stress can provide unique and more nuanced insights into the relationship between hippocampus and exposure to stress.

To the best of our knowledge, the current study is the first investigation on associations between hippocampal subfield volumes and biological and acute environmental markers of stress prior to psychosis onset. The first aim was to examine the relationships between the volumes and resting cortisol levels and the volumes and distress from exposure to daily stressors in individuals at CHR for psychosis. Evidence from other stress related psychopathology implicates cornu amonis and dentate gyrus as particularly affected by stress (O’Doherty et al., 2015; Santos et al., 2018). However, evidence is not clear in psychosis spectrum disorders. As such, the strongest relationship was predicted between lower volume in CA1, CA2/3, and CA4/DG and both greater cortisol levels and distress from daily stressors. The associations between stress and presubiculum and subiculum subfields were exploratory. The second aim was to examine group differences in the effects of cortisol and environmental stressors on any target regions the results from the first aim yield.

2. Material and Methods

2.1. Participants

A total of 51 CHR (mean age 18.61, SD=1.83) and 70 HC (mean age 18.30, SD=2.57) participants were recruited at the Adolescent Development and Preventive Treatment Program. All participant completed a written consent form approved by the Institutional Review Board. Participants under the age of 18 provided written assent and their legal guardians written consent. In accordance with standard procedures outlined in the Structured Clinical Interview for Psychosis Risk Syndromes, those at CHR for psychosis met criteria for the psychosis-risk syndrome by one or more of the following: 1) presence of progressing attenuated psychosis symptoms, 2) presence of schizotypal personality disorder with global functioning decline or age <19, and 3) a family history of psychosis with global functioning decline. Exclusion criteria for all participants included age <12 or >24, psychotic disorder diagnosis, history of head injuries and neurological disorders, and a lifetime diagnosis of substance use. Additional exclusionary criteria for HC participants included meeting psychosis-risk criteria and family history of psychosis. Nine participants who reported use of antipsychotic medications were excluded from all analyses because of potential hippocampal volume changes associated with antipsychotic treatment (Rhindress et al., 2017). Presented data are from a larger study examining stress in a CHR sample, and our group has previously reported on group differences in hippocampal subfields (Vargas et al., 2018) and resting cortisol (Carol and Mittal, 2015; Carol et al., 2017). This is the first study linking stress measures and hippocampal subfield volumes.

2.2. Clinical Assessments

The Structured Clinical Interview for Psychosis Risk Syndromes (SIPS) (Miller et al., 1999) was employed to diagnose attenuated psychosis in CHR participants and rule out symptoms in HCs. The Structured Clinical Interview for DSM-IV (SCID) (First et al., 2012) was administered to participants in both groups to rule out psychosis and assess for other psychiatric disorders. Both were administered by trained assessors.

2.3. Daily Stressors

Daily Stress Inventory (DSI) (Brantley et al., 1987) is a 58-item self-report measure of experiencing common daily hassles over the past 48 hours. The measure has 2 dimensions: the frequency of the encountered daily stressors and the subjective distress caused by each event rated on a 7-point Likert scale from 1 (“occurred but was not very stressful”) to 7 (“caused me to panic”). The average distress variable was used in the analyses. One CHR participant was missing DSI data.

2.4. Cortisol

Three saliva samples for cortisol assay were collected at hourly intervals during the morning of the baseline clinical assessments using passive-drool method. To account for diurnal changes in cortisol levels, collection began as early in the morning as possible, but at least an hour after awakening to avoid capturing cortisol awakening response (Walker et al., 2013). Samples (75μL) were immediately stored at −20C in lab freezer and kept frozen until assay. Because diet and activity affect cortisol, participants were given explicit instructions and provided a log to detail these activities. On the evening before saliva sampling, they were asked to avoid alcohol and caffeine after 6:00 PM. On the morning of the assessment, they were asked to refrain from caffeinated beverages and consume only grains (e.g., toast, cereal, pastry), milk, juice, or water. They were also asked to avoid over-the-counter medications and physical exercise. Awakening time, food, liquid, and medication consumption the previous evening and the morning of the assessment were recorded on a Food and Activity Log. For salivary cortisol assay, the Salimetrics (Salimetrics, LLC, College Park, Pa) High Sensitivity Salivary Cortisol Enzyme Immunoassay Kit was used. This assay captures the full range of salivary cortisol levels (0.003 to 3.0μg/dL) requiring only 25 uL of saliva per test. Two CHR and 10 HC participants were missing cortisol data.

2.5. Structural Magnetic Resonance Imaging

Imaging scans were acquired using a 3-Tesla Siemens Trim Trio magnetic resonance imaging scanner (Siemens Heathineers, Erlangen, Germany) with a standard 12-channel head coil. Structural images were obtained using T1-weighted 3D magnetization prepared rapid gradient multi-echo sequence (sagittal plane; repetition time (TR)=2530 ms; echo times (TE)=1.64, 3.5, 5.36, 7.22, and 9.08 ms; GRAPPA parallel imaging factor of 2; 1 mm³ isomorphic voxels, 192 interleaved slices; FOV=256 mm; flip angle=7°, time=6:03 min). To check for any incidental clinical findings, a T2 weighted acquisition (axial oblique aligned with anterior commissure-posterior commissure line; TR=3720 ms; TE=89 ms; GRAPPA parallel imaging factor 2; 0.9 mm × 0.9 mm voxels; FOV=240 mm; flip angle: 120°; 77 interleaved 1.5 mm slices; time=5:14) was acquired. Imaging data was processed using Freesurfer 5.3 software in accordance with previous studies investigating hippocampal subregions in psychosis (Haukvik et al., 2015; Ho et al., 2017a; Mathew et al., 2014). First, preprocessing and quality assurance of structural images was completed. Next, hippocampal subregions (CA1, CA2/3, CA4/DG, presubiculum, subiculum, hippocampal fissure, and fimbria) were extracted using a Freesurfer 5.3 hippocampal segmentation package (Van Leemput et al., 2009). This package uses a Bayesian probabilistic model, and it has been validated against manual morphometric measurements of ultra-high-resolution MRI scans. Voxel subregion measurements were made on images interpolated to 0.5 × 0.5 × 0.5 mm³.

2.6. Statistical Approach

Independent samples t-tests and chi-square tests were used to examine group differences in continuous and categorical variables, respectively. Independent samples t-tests were also used to assess group differences in distress of daily stress. To account for neurodevelopmental and endocrine changes during adolescence and young adulthood, age was used as a covariate in all analyses including hippocampal volumes and cortisol. Univariate analyses of covariance (ANCOVAs), controlling for intracranial volume (ICV) and age, were utilized to determine group differences in hippocampal subfields. Left and right subfields were highly correlated; therefore, the average volume was used for each to reduce the number of comparisons. The subfields included CA1, CA2/3, CA4/DG, presubiculum, and subiculum. Due to difficulties with obtaining reliable volumes when segmenting smaller brain subregions, fimbria and hippocampal fissure were excluded from the analyses (Van Leemput et al., 2009). ANCOVAs, controlling for age, were used to examine differences in cortisol levels between groups. The mean of 3 cortisol samples was used in analyses. Outliers, defined as values above 3 standard deviations from the mean, were removed prior to calculating means. Partial correlations, controlling for ICV, sex and age, were employed to test the association between cortisol and daily stress with subfields in the CHR group. Lastly, multiple linear regressions were used to explore the group differences in the relationship between stress and hippocampal subfield volumes that were found to be significantly associated in the first aim. Z scores were used in all regression analyses.

3. Results

3.1. Sample Characteristics

Participants at CHR for psychosis and HCs were matched on demographic information including age (t(119)=.73, p=.48), sex (χ²(1, N=121)=1.33, p=.25), race (t(119)=.79, p=.44), ethnicity (χ²(1, N=121)=.18, p=.67), and parental education (t(119)=.35, p=.72). As expected, based on our previous findings from an overlapping sample (Vargas et al., 2018), the hippocampal subregion volumes were significantly lower in individuals at CHR for psychosis for CA1 (F(1, 117)=11.56, p=.001, η²_(partial)=.09), CA2/3 (F(1, 117)=6.38, p=.01, η²_(partial)=.05), and CA4/DG (F(2, 117)=7.21, p=.008, η²_(partial)=.06).The groups had comparable subiculum (F(2, 117)=2.8, p=.09, η²_(partial)=.02) and presubiculum volumes (F(1, 117)=.6, p=.79, η²_(partial)=.001). Additionally, contrary to previous studies with an overlapping sample (Carol and Mittal, 2015; Carol et al., 2017), the results indicated no significant differences in resting cortisol (F(2, 105)=.68, p=.41, η²_(partial)=.006). The current sample is larger, and participants using antipsychotics were excluded, possibly accounting for differences in findings. Lastly, in line with previous findings (Devylder et al., 2013), individuals at CHR for psychosis endorsed greater distress of daily stressors (t(118)=3.75, p<.0001). Participant characteristics are available in Table 1.

Table 1.

Participant characteristics by group

	CHR	HC	p
Sex
Males	28	31
Females	23	39
Total	51	70	NS
Age
Mean years (SD)	18.61 (1.83)	18.30 (2.57)	NS
Race – n (%)			NS
First Nations	2 (3.9)	---
East Asian	2 (3.9)	6 (8.6)
Southeast Asian	---	2 (2.9)
West/Central Asian	1 (2.0)	2 (2.9)
Black	1 (2.0)	2 (2.9)
Central/South American	9(17.6)	15 (21.4)
White	35 (68.6)	42 (60.0)
More than one race	1 (2.0)	1 (1.4)
Ethnicity – n (%)
Hispanic	10 (19.6)	16 (22.9)	NS
Parental education
Mean years (SD)	15.89 (2.24)	15.72 (2.88)	NS

Notes: NS = not significant; CHR = Clinical high-risk; HC = healthy controls

3.2. Partial Correlations Between Hippocampal Subregions and Stress in the CHR group

Correlational analyses indicated cortisol and daily stressors are associated with distinct hippocampal subfields (Table 2). Specifically, greater resting cortisol levels were significantly associated with reduced volume in presubiculum (r=−.36, p=.01). Cortisol was not associated with volumes in CA1 (r=.09, p=.59), CA2/3 (r=.09, p=.57), CA4/DG (r=−.01, p=.95), or subiculum (r=−., p=.13). Next, correlational analyses demonstrated significant associations between higher distress of daily stressors and lower volumes in CA1 (r=−.4, p=.006), CA2/3 (r=−.31, p=.03) and CA4/DG (r=−.3, p=.04). Distress of daily stressors was not significantly associated with volumes in presubiculum (r=−.09, p=.55) or subiculum (r=−.17, p=.25). Additionally, cortisol and distress of daily stressors were not significantly correlated (r=−.2, p=.17), indicating them as separate measures of stress.

Table 2.

Partial correlations between hippocampal subfield volumes with resting cortisol and distress of daily stress in the Clinical High-Risk group.

	CA1	CA2_3	CA4_DG	Presubiculum	Subiculum
Cortisol	.08	.09	−.01	−.36*	−.16
DSI	−.4**	−.31*	−.29*	−.09	−.17

Notes: DSI = daily stress inventory;

^**<.01

^*<.05

3.3. Group Differences in Effects of Stress on Hippocampal Subfield Volumes

In the first regression model, cortisol, group, sex, age, ICV, and cortisol by group interaction were entered as predictors of presubiculum volume. The results showed the model fit the data well (R²=.38, F(6, 101)=10.14, p<.0001) and that cortisol (β=−.74, p=.01), group (β=.33, p=.049), and cortisol by group interaction (β=.38, p=.03) significantly predicted presubiculum volume (Figure 1). In the second, third, and fourth regression models, DSI, group, age, ICV, and DSI by group interaction were included as predictors of CA1, CA2/3, and CA4/DG volumes, respectively. In the second model (R²=.38, F(6, 95)=9.88, p<.0001), main effect of distress of daily stressors (β=−.62, p=.02) but not group (β=.24, p=.15) significantly predicted CA1 volume. Their interaction reached a trend level significance (β=.32, p=.06; Figure 2) in predicting CA1 volume such that greater distress was related to lower volumes in the CHR but not HC group. In the third model (R²=.48, F(6, 95)=14.52, p<.0001), distress of daily stress reached trend level significance in predicting CA2/3 volume (β=−.46, p=.06); group (β=.25, p=.1) and their interaction (β=.26, p=.08) were not a significant predictors. Lastly, in the fourth model, distress of daily stressors (β=.08, p=.82), group (β=.13, p=.53), or their interaction (β=−.03, p=.9) were not significant predictors of CA4/DG volume.

Figure 1.

Group by mean resting cortisol levels predicting presubiculum volumes.

Note: β=.38, p=.03; CHR = clinical high-risk for psychosis, HC= healthy control

Figure 2.

Group by mean distress from daily stressors predicting CA1 volumes.

Note: β=.32, p=.06; CHR = clinical high-risk for psychosis, HC= healthy control, CA = cornu amonis

4. Discussion

The current study provided important initial insights into the relationship between basal biological and environmental indicators of stress with the hippocampal subfields during the psychosis-risk period. The correlational analyses indicated that elevated psychosocial stress and resting cortisol were linked to smaller volumes in distinct subfields. Specifically, lower volumes in cornu amonis and dentate gyrus subfields were correlated with greater distress of daily stressors and lower volumes in presubiculum with higher resting cortisol levels. Moreover, regression analyses indicated that these associations were specific to the CHR group.

The links between psychosocial and biological stress with the hippocampus helped to shed light on a growing literature that suggests effects of stress and brain vulnerability play an important synergistic role in the progression and maintenance of psychosis. The inverse relationship between distress of daily stressors and CA1, CA2/3, and CA4/DG is consistent with investigations in other stress-related psychopathology (O’Doherty et al., 2015; Santos et al., 2018). Additionally, it is in line with broader findings in the psychosis literature showing that exposure to environmental stressors is related to reductions in the overall hippocampal volume. The existing studies demonstrated the associations of these reductions with either exposure to early trauma (Aas et al., 2014) or increased reactivity to daily stressors (Collip et al., 2013) in individuals with psychosis. Next, the correlation between higher resting cortisol levels and presubiculum is in part consistent with the finding of links between elevated resting cortisol with reduced left hippocampal volume in individuals with first-episode psychosis (Mondelli et al., 2010). However, studies looking at the effect of cortisol on the subfields in depression and PTSD suggested only CA and DG subfields are affected (O’Doherty et al., 2015; Santos et al., 2018). Therefore, the observed relationship between cortisol and presubiculum and subiculum in the current study could be reflective of pathogenic processes specific to psychosis. The current study extended the existing literature in several ways. First, it demonstrated these associations are observable in the CHR group. Second, it provided initial support for the importance of examining subfield volumes as opposed to overall hippocampal volume in relation to stress in psychosis-spectrum disorders. Further, it provided novel insights into the effects of recent stressors on the hippocampal subfields in individuals at CHR for psychosis. Lastly, it implicated presubiculum and subiculum as potentially relevant subfields affected by stress. Taken together, these findings suggest associations between stress and the hippocampus are important for understanding pathogenic processes prior to psychosis onset.

Although regional specificity to stress was excepted, the differential associations with cortisol and daily stress across subfields were somewhat surprising. Several interpretations might be relevant here. First, it is possible that these distinct effects may be in part due to the sensitivity of different parts of the hippocampus to corticoid (cortisol) receptors. In the present study, distress of daily stressors is a measure of more acute stress as it assesses exposure to recent environmental stressors while resting cortisol is a measure of more tonic conditions. Two types of corticoid receptors, mineralocorticoid (MR) and glucocorticoid (GR), are present in the hippocampus in high levels, and they mediate the effects of cortisol in the brain. These receptors have vastly different levels of affinity for cortisol and become activated under different conditions (Wong and Herbert, 2005). GRs are only occupied during times when circulating levels of cortisol increase, such as following an acute daily stressor (Wong and Herbert, 2005). Critically, a post-mortem study including brains of individuals with schizophrenia demonstrated a regional specificity of reductions in GR mRNA levels to CA and DG regions (Webster et al., 2002). In contrast, MRs have high affinity for cortisol and become occupied under and regulate basal physiological conditions (ter Heegde et al., 2015). Evidence from non-human primate studies indicate that levels of MR mRNA are low in subiculum and presubiculum, particularly compared to other parts of the hippocampus (Pryce et al., 2005). Therefore, these regions could be more impacted by resting cortisol levels. Importantly, dysregulation (including reductions and lower expression) of the corticoid receptors is well evidenced in psychotic disorders (Sinclair et al., 2011; Webster et al., 2002). While these factors have not been investigated prior to psychosis onset, the distinct associations observed in the current CHR sample may be related to dysregulation of the stress signaling system (i.e., activity and functionality of corticoid receptors both under stressful -affecting CA and DG- and basal -affecting presubiculum and subiculum-conditions).

Another possibility is that the timing of development in the adolescent and young adult period characteristic of the CHR population along with the early pathogenic stage play a role. As the illness progresses, the putative effects of psychosocial and biological stress may impact the hippocampus more broadly. Indeed, studies in later phases of psychosis find progressive volume loss that defuses from CA1 to others subfields (Ho et al., 2017b) and that overall hippocampal volume is associated with stress (Collip et al., 2013; Mondelli et al., 2011; Mondelli et al., 2010). Notably, in the current sample, the presubiculum and subiculum volumes were not significantly lower relative to healthy controls. However, atrophy to these two subfields is evident in later stages of the illness (Haukvik et al., 2015; Vargas et al., 2018). Therefore, the inverse association between resting cortisol and volumes may be a marker of illness progression and subsequent damage to presubiculum. However, it is puzzling that the outer pre-subiculum layer is most strongly linked with cortisol. This might relate to measurement sensitivity unique to discrete biological measures of stress. Alternatively, it could reflect a process through which stress related factors affecting more core subfields are first detectable in outer layers of the structure. Lastly, these findings could be related to the cascading cell migration (parahippocampus > presubiculum > subiculum > CA1/2 > CA3 > DG) during early development and maturation of the hippocampus (Gomez and Edgin, 2016). Considering presubiculum is one of the subfields that develops the earliest, the observed association with resting cortisol could reflect an impact of early stress on the HPA-axis function.

Further, multiple regressions provided evidence for group differences in these associations. First, group by cortisol interaction predicted presubiculum volume such that higher cortisol levels were predictive of lower volume in the CHR but not HC group (Figure 1). Similarly, group by DSI interaction predicted CA1 volume at trend level (Figure 2) such that greater DSI was predictive of lower volume in the CHR but not HC group. The results are in line with previous reports on individuals with psychosis showing that biological stress and hippocampal volumes are associated in patients but not in controls (Collip et al., 2013; Mondelli et al., 2010). There is also evidence to suggest sex differences, where the effects were observable in males only (Pruessner et al., 2015). Similar results were reported prior to psychosis onset. Specifically, the only study directly examining associations with cortisol and the hippocampus in this group suggested that blunted awakening cortisol response is associated with lower overall volumes in males at CHR for psychosis but not healthy controls (Pruessner et al., 2017a). Notably, literature thus far has only investigated overall hippocampal volumes in relation to stress in psychosis spectrum disorders, possibly diminishing more nuanced group differences. The current study extended these findings by elucidating associations with the subfields. Moreover, in terms of the inverse relationship between the hippocampal volumes and greater exposure to environmental stressors, group differences have been reported only in individuals with psychosis (Aas et al., 2014; Collip et al., 2013). Current findings highlighted that recent stressors are relevant during the high-risk period, possibly further contributing to hippocampal atrophy prior to illness onset. Overall, these results indicated that the effects of stress on the hippocampal subfields are not a normative consequence of exposure elevated stressors, but they are related to psychosis pathogenesis.

Taken together, these findings are in line with the diathesis-stress and stress-cascade models of schizophrenia which posit that existing vulnerability interacts with environmental stress and normative and pathogenic neurobiological processes to ultimately drive illness onset (Corcoran et al., 2003; Pruessner et al., 2017b; Walker et al., 2008). Specifically, the study highlighted the dynamic relationship between the hippocampus morphology, exposure to environmental stress, and stress signaling system in individuals identified at-risk for developing psychosis. This period is particularly relevant because it coincides with adolescence and young adulthood, a developmental time characterized by neuromaturational changes (Fuhrmann et al., 2015; Keshavan et al., 2014) and increasing environmental challenges (Sawyer et al., 2018). Individuals at CHR for psychosis endorse greater exposure to environmental stressors relative to healthy controls and increasing sensitivity to stress (Devylder et al., 2013; Ristanovic et al., 2020). These factors are related to aberrant HPA axis functioning, as evidenced by abnormal cortisol levels, which can have a downstream effect on an already vulnerable hippocampus (Collip et al., 2013; Pruessner et al., 2017b). In turn, the hippocampus is less capable of regulating the HPA axis related negative feedback loop contributing to more stress sensitivity. This stress cascade further confers risk and likely contributes to psychosis onset (Corcoran et al., 2003). Notably, considering the hippocampus has a critical role in the stress response system, the directionality of effects observed in the current study and broader literature is not conclusive. It is possible individuals at CHR for psychosis present with lower hippocampal volume early on in childhood and adolescence, contributing to subsequent aberrations in cortisol levels and stress sensitivity. Nevertheless, these conclusions are in line with a broader literature linking increasing hormone dysregulation and stress-related changes in hippocampal morphology to other factors related to psychosis pathogenesis such as altered dopaminergic function (Howes et al., 2017) and brain derived neurotropic factor gene expression (Aas et al., 2014; Mondelli et al., 2011).

Lastly, the current study presents with some limitations. First, the correlational analyses did not account for multiple comparisons. Future studies with larger samples should model analyses targeted towards investigating unique contributions of acute versus basal levels of stress to hippocampal subfields atrophy. The analyses also did not account for the directionality of effects. Longitudinal studies with multiple time points are needed to establish a temporal association between lower subfield volumes and exposure to stressors and yield more conclusive inferences about causality. Next, while DSI is a good measure of subjective experience of acute environmental stressors, future investigations would benefit from examining biological measures of acute stress in relation to hippocampal subfields. This approach would provide further clarity on the distinct mechanisms by which basal versus elevated stress conditions may impact specific subfields. Additionally, DSI was normed for individuals 17 and over, and the current study includes several participants ages 12–16. Similarly, SCID is most appropriately used with adults; however, the American Psychiatric Association acknowledges it can be used with adolescents (APA, n.d.). Further, the present sample had missing data for 10 HC participants which could have impacted the results of the regression analyses. Lastly, hippocampal atrophy as a result of exposure to stressors and dysregulation of the stress signaling system is observed in other psychiatric conditions such as major depressive disorder (Santos et al., 2018) and posttraumatic stress disorder (O’Doherty et al., 2015). Future studies should compare individuals at CHR for psychosis to other clinical groups to determine if the pattern observed in the current study is specific to emerging psychosis rather than a result of broader stress-related psychopathology.

Highlights.

• Morphology of the hippocampal subfields is associated with stress in individuals at CHR for psychosis.

• Lower volumes in the subiculum is associated with greater resting cortisol levels.

• Lower volumes in CA1 and CA2/3 is associated with greater distress form exposure to daily stressors.

• These associations are specific to individuals at CHR relative to healthy controls.