The Impact of Menarche on Hippocampal Mechanisms of Severity of Psychotic-Like Experiences in the ABCD Study

Abstract

Accumulating evidence suggests that estrogens play an important modulatory role in the pathogenesis of psychosis. Estrogens come online within a dynamic developmental context of emerging psychopathology and neurodevelopment. As a result, estradiol (the primary form of estrogen) may influence psychosis lability directly or indirectly through its neurodevelopmental influence on estrogens-sensitive areas like the hippocampus. Understanding this influence may provide novel insight into mechanisms of psychosis lability. This study included baseline and year 2 timepoints from 4422 female participants from the Adolescent Brain Cognitive Development (ABCD) study (age 8–13), who varied in estradiol availability (pre-menarche, post-menarche, pre- and post-menarche timepoints). Estradiol availability was related to psychotic-like experiences (PLE) severity both directly and as an interactive effect with hippocampal connectivity using menarche status (pre/post) in a multilevel model. PLE severity was highest in individuals with early menarche emphasizing the importance of the developmental timing. Although PLE severity decreased over time in the sample, it stayed clinically-relevant over 2 years. Lower hippocampal connectivity was related to elevated PLE severity. This effect was moderated by estradiol; before the availability of estradiol (pre-menarche), lower hippocampal connectivity significantly contributed to the PLE severity, but when estradiol was available (post-menarche) hippocampal dysconnectivity did not account for PLE severity. This moderation suggests that the estrodiol’s influence on hippocampal plasticity also reduced the mechanistic role of the hippocampus on PLE severity. Further, the lack of a significant direct reduction of PLE severity post-menarche, may suggest an increased role for other interacting psychosis lability factors during this critical developmental period.

1. Introduction

Psychosis spectrum disorders represent a significant source of disability and economic burden (Cloutier et al., 2016). Individuals diagnosed with psychosis experience a 15–20 year reduction in life expectancy (Lee et al., 2018). This burden may be mitigated in biological females who have an 8–10 year delay in psychosis onset despite having a similar lifetime prevalence and risk (Ochoa et al., 2012). This difference in onset occurs during adolescence (Walker et al., 2007), which is also a critical period for hormonal and neural development (Vargas and Mittal, 2022). Normative development, e.g. hormonal changes and brain reorganization, in adolescence (Walker et al., 2007) may interact with early lability for psychosis to modulate the emergence of psychotic-like experiences (PLE; Kelleher et al., 2013, 2012)- a premorbid marker of early psychosis risk. In biological females, the introduction and rise in levels of estrogens, particularly estradiol -the primary source of estrogen during reproductive years, plays a significant role in modulating neurodevelopment (Vijayakumar et al., 2018) especially in estrogen-sensitive brain areas such as the hippocampus (Barha and Galea, 2010; Bean et al., 2014). This neurodevelopmental influence may impact psychosis lability but the direct and indirect effects of estradiol in adolescence is poorly understood. Many studies note that late-life estradiol events like menopause are associated with a second window of lability for psychosis onset (Gogos et al., 2015; Hochman and Lewine, 2004). Fewer studies have examined the impact of estradiol events in adolescence, such as menarche, during the traditional window of psychosis lability. Understanding the influence of estradiol availability on PLE severity has the potential to provide a novel understanding of psychosis etiology and may translate into early interventions.

Biological females tend to have a later onset of psychosis, milder symptom course of psychosis diagnoses (Tamminga, 1997), and additional windows of vulnerability for psychosis onset surrounding estradiol events (e.g., post-partum, menopause; Culbert et al., 2022). These sex differences have been attributed to the effects of estradiol on neural mechanisms of psychosis onset and course beginning with the widespread availability of estradiol at menarche (Patel et al., 2021). As noted, estradiol availability occurs at a critical time of neurodevelopmental plasticity and maturation (Patel et al., 2021), and estradiol’s effects on plasticity may reduce the dysconnectivity related to psychosis (Friston et al., 2016). The extant literature has focused on middle age and late-life events (Culbert et al., 2022), where the absence or sudden decrease in estrogen unmask latent vulnerability for psychosis. Potential estradiol-related mechanisms have not been examined in adolescence -the developmental context when estrogens are introduced. This approach limits the current understanding of this hormone’s influence on PLE severity and mechanisms that may be therapeutic targets to delay onset and reduce course severity (Tamminga, 1997).

Estrogens have a neuroprotective effect- promoting healthy connectivity during development (Biegon and McEwen, 1982; Gould et al., 1990; McEwen, 2001). Although estrogen impacts many brain regions, the hippocampus is a particularly important site of estrogen impact, receiving multiple benefits from the effects of estrogen, including increased hippocampal plasticity via synaptogenesis (Woolley, 1998), spine density (Gould et al., 1990), and neuroprotective influences (Diaz Brinton et al., 2000). The hippocampus is also of particular interest in psychosis risk, reflecting the dose-dependent effect of psychosis risk on hippocampal volumes (Vargas et al., 2018) as well as structure and function predicting psychosis course (Bähner and Meyer-Lindenberg, 2017; Benetti et al., 2009; Dean et al., 2016). As a result, the effects of estrogens may culminate in a reduction of hippocampal dysconnectivity and degeneration that has been previously observed in psychosis (Bähner and Meyer-Lindenberg, 2017; Benetti et al., 2009; Damme et al., 2020; Dean et al., 2016; Mittal and Walker, 2011; Vargas et al., 2018). This hormonal impact has been studied retrospectively (Damme et al., 2020; Maric et al., 2005), limiting insight into the developmental context in which these effects occur. However, examining individuals during their transition from late childhood into early adolescence may provide greater insight into questions regarding the importance of developmental timing and relevance to early psychosis lability- PLEs (Damme et al., 2020). To date, there have been no studies describing the impact of estradiol events, like menarche, on psychosis lability over time or within the developmental context of adolescence.

Longitudinal data available from Adolescent Brain Cognitive Development (ABCD) study was used to model the relationship between estradiol availability (marked by menarche) and PLE severity directly and through interactive effects on hippocampal volume and hippocampal connectivity in this critical window of neurodevelopment. We expected that estrogen availability (post-menarche status) would be related to lower PLE severity. Prior to menarche (a low estrogen availability period), we expected that decreased hippocampal volume and connectivity would be associated with increased PLE severity. When estradiol is available (post-menarche), we expected that hippocampal volume and/or plasticity may be improved and related to lower PLE severity. Finally, as an exploratory aim, we will examine whether this impact is specific to PLE severity or may reflect a general benefit to psychopathology by applying the same model to depression symptom severity.

2. Materials and Methods

2.1. Participants.

The ABCD Study (release 4.0) included 21 sites across the United States that collected participants with a broad demographic diversity range recruiting through public and private schools. Each site employed a standardized probability sampling aimed to minimize systemic bias to ensure outreach to historically underrepresented groups and to maximally reflect the distribution of the demographic and socioeconomic characteristics of the U.S. population. The current subsample included 4422 individuals who were assigned female at birth; they ranged in age from 8.92 to 13.5 years of age M(SD)=10.59(1.10), and included a wide racial and socioeconomic makeup, Table 1. All research protocols were approved by each respective institutional review board, which included obtaining parents’ informed consent and child’s assent. In the ABCD Study, clinical assessments included current psychopathology symptoms via a self-report version of the Prodromal Questionnaire-Brief Child Version (PQ-CB; Karcher and Barch, 2021) and the Childhood Behavioral Checklist (CBCL).

Table 1.

Whole Sample Demographic Metrics

Parameter		Mean (s.d.)
Age		10.59(1.10)
BMI		19.49 (4.29)
PQ-CB Severity		5.07(9.34)
Race/Ethnicity		%
	White	52%
	Black	14%
	Hispanic	21%
	Asian	2%
	Other	11%
Household Income		%
	<$50,000	29.83%
	50,000–100,000	28.99%
	>$100,000	41.17%
Pubertal Status	Baseline	Follow-Up
Pre-Menarche	96.67%	64.83%
Post-Menarche	3.33%	35.17%

2.2. Clinical Assessments.

2.2.1. Psychotic-like experiences symptoms severity.

The Prodromal Questionnaire-Brief Child version (PQ-BC) metric was taken from the curated and validated severity scores provided by the ABCD Study (Loewy et al., 2011). PLE severity as assessed using the PQ-BC, a 21-item self-report questionnaire. The PQ-BC asked children about specific PLEs severity which were endorsed with a binary response (i.e., yes or no). The score could range from 0 to 21 (M=5.07, SEM=0.1186); higher total scores indicate more PLEs endorsed.

2.2.2. Depression symptom severity.

The Childhood Behavioral Checklist (CBCL) was completed in a self-report, automated version that has been validated (Farahdel et al., 2021; Nelson et al., 2022; Steinberger & Barch, 2021; Thompson et al., 2019). This 113-item questionnaire measures aspects of the participants’ behavior across the past six months; each item was rated on a 3-point scale: not true, somewhat, or sometimes true, very or often true (Farahdel et al., 2021; Steinberger & Barch, 2021; Thompson et al., 2019). Responses were used to generate ABCD curated total depression symptoms score.

2.3. Menarche Assessment.

The Pubertal Development Scale collected menarche status as a self-report measure with a single item: “Have you begun to menstruate (started to have your period)?” This marker was chosen as there is a great deal of variability in an individual’s sensitivity to estrogens (Shirtcliff et al., 2000), and menarche is a marker that estradiol levels were sufficient to cause cascading biological effects (Cheng et al., 2021; Damme et al., 2020).

2.4. Hippocampal Volumes.

Participants completed a high-resolution T1-weighted structural MRI scan (1-mm isotropic voxels). Structural MRI data were processed using FreeSurfer version 7.0 (http://surfer.nmr.mgh.harvard.edu/) according to standard processing pipelines. Processing included removal of nonbrain tissue, segmentation of gray and white matter structures. Quality control for the structural images comprised visual inspection of T1 images and FreeSurfer outputs for quality (Casey et al., 2018; Hagler et al., 2019) conducted by the ABCD team. Subjects whose scans failed inspection or were identified as extreme outliers by Rosner’s test were excluded. Hippocampal volume was defined by the Desikan-Killiany atlas (Hagler et al., 2019). Hippocampal volumes were divided by the whole brain volume and are reported as a percent of brain volume to correct for differences in total brain volume.

2.5. Hippocampal Functional Network Connectivity.

Resting-state data were taken from the curated ABCD fMRI data for Gordon Network (Gordon et al., 2016). Grand averages were created across the left and right hemispheres to reduce the total number of connectivity networks examined. Functional networks were chosen based on previous literature highlighting a tri-network model of psychopathology (Menon, 2011) and from a whole brain analyses of menarche and individuals at clinical high risk for psychosis (Damme et al., 2020). This resulted in the following networks being included in the model: hippocampal-default mode (DMN), hippocampal-salience (SAL), hippocampal-executive control (ECN), (Menon, 2011) and hippocampal-visual networks (Damme et al., 2020). All analyses excluded individuals who did not pass quality assurance and were identified as extreme outliers in Rosner’s test.

2.6. Analytical Strategy.

Prior to analyses, one sibling was randomly selected from each family (n=4422 unique families/individuals). In a multilevel model, PLE symptoms severity was the outcome predicted by menarche status (Pre-Menarche, Post-Menarche) at each time point, hippocampal volume (as a percentage of total brain volume), and hippocampal connectivity to resting state networks (DMN, ECN, SAL, Visual), also accounting for between subject variables including income, education, age, and BMI where participants where time points were clustered within an individual as a fixed effect using lme4 in Rv.4.2.1, analytic code in Supplemental Material (Bates et al., 2015; Saragosa-Harris et al., 2022). All hippocampal parameters were entered into a single model to control for the specificity of each hippocampal effect; parameters were largely independent (Variable Inflation Factor < 1.22). Follow-up analyses examined whether hippocampal effects vary by menarche status (pre/post) as well as over time within individuals who were post-menarche at both time points, pre-menarche at both time points, or pre-menarche at baseline and post-menarche at follow-up to explore the features driving estradiol effects.

3. Results

3.1. Participants.

Our sample included ₄₄₂₂ participants; participant demographics and the distribution of study variables can be found in Table 1. Analyses below treated examined status at each timepoint, to better understand the demographics within the sample individuals are described in Table 2 by menarche status a group (post-menarche only, pre-menarche only, pre/post-menarche). There was a significant difference in age at baseline (F=182.1, p<.0001) post-menarche group was older than the pre-menarche (p_Tukey<.0001, mean difference 7.49 months) and pre/post-menarche group (p_Tukey=.0014, mean difference 2.2 months); the pre/post-menarche group was also older than the pre-menarche group (p_Tukey<.0001, mean difference 5.28 months), which holds at follow-up. The post-menarche group had a higher percentage of lower income households (χ²(4)=85.152, p<.0001) and a distinct racial distribution (χ²(8)=139.63, p<.0001) such that there was a lower percentage of White individuals (36.08%) and a higher percentage of Black individuals (45.36%). All models below account for these parameters and effects reported are beyond shared variance related to these factors.

Table 2.

Sample Demographics by Menarche Status Group

Demographic Parameter	Post-Menarche	Pre-Menarche	Pre/Post-Menarche
Race	%
Asian/American Indian/Alaska
Native/Native Hawaiian/Pacific	1.03%	2.36%	4.34%
Islander
Black	45.36%	11.82%	21.28%
Multiple	17.53%	12.18%	16.94%
White	36.08%	73.64%	57.44%
Household Income
<50K	58.46%	3.13%	5.99%
>=50K & <100K	23.08%	39.18%	44.89%
>=100K	18.46%	57.68%	49.13%
Age
Baseline	10.33 (.528)	9.75(0.572)	10.166(0.55)
Follow-up	12.5 (.41)	11.66(0.586)	12.166 (0.579)

3.2. Psychotic-Like Experiences Related to Menarche and Hippocampal Metrics.

All model statistics can be found in Table 3.There was a significant main effect of menarche status (t=4.64, p>0.0001, η²_partial=.004) in an unexpected direction such that post-menarche individuals experienced higher PLE severity (M=5.43; SEM=0.35) than pre-menarche (M=5.01; SEM=0.13; d=.045). This effect was driven by individuals who were post-menarche at each time point (menarche was prior to the first time point) who showed significantly higher PLE severity (M=8.53, SEM=0.99) than the individuals who were pre-menarche at baseline and post-menarche at follow-up (M=4.94, SEM=0.26, t= 3.51, p=0.0005) and individuals that were pre-menarche at both time points (M= 4.48, SEM=0.17, t= 3.51, p<0.0001), Figure 1.

Table 3.

Psychotic-Like Experience Severity Model Parameters With and Without Moderation

Parameter	Original Model			Moderation Model
	t	p	partial η2	t	p	partial η2
Menarche Status	4.64	>0.0001	0.0042	4.95	>0.0001	0.0048
Hippocampal-Visual Connectivity	2.00	0.0455	0.0006	0.98	0.3261	>0.0001
Menarche Status × Hippocampal-Visual Connectivity				1.99	0.0400	0.0008
Total Hippocampal Volume	0.48	0.6332	0.0001	0.48	0.6296	>0.0001
Hippocampal-Salience Network Connectivity	1.34	0.1804	0.0003	1.37	0.1707	0.0003
Hippocampal-Executive Control Network Connectivity	0.33	0.7422	>0.0001	0.31	0.7535	>0.0001
Hippocampal-Default Mode Network Connectivity	0.08	0.9365	>0.0001	0.03	0.9743	>0.0001
Age	−11.44	>0.0001	0.0300	−11.50	>0.0001	0.0300
Race			0.0095			0.0095
Race/Ethnicity Black vs. White	5.99	>0.0001		5.98	>0.0001
Race/Ethnicity Hispanic vs. White	2.75	0.0060		2.68	0.0073
Race/Ethnicity Asian vs. White	1.17	0.2435		−1.16	0.2469
Race/Ethnicity Other vs. White	1.42	0.1544		1.37	0.1695
Income			0.0078			0.0079
Income [>$50,000 vs. 50–100,000]	2.08	0.0379		−2.08	0.0375
Income [>50,000 vs. >100,000]	−5.40	>0.0001		−5.40	>0.0001
BMI	0.34	0.7343	>0.0001	0.34	0.7354	>0.0001

Figure 1.

The left panel includes the average across both time points of psychotic-like experiences (PLE) severity by menarche status cohorts. The right panel shows the slope of change in PLE severity across time points for each menarche status cohort.

There was also a main effect of hippocampal connectivity to the visual network (t=2.72, p=.0067, η²_partial=.001), such that less hippocampal–visual network connectivity related to greater PLE severity (r²=.003), Figure 2. Follow-up analyses showed a significant interactive effect of menarche status (t=1.99; p=0.04, η²_partial=.0008; Figure 2), such that increased hippocampal-visual connectivity was related to decreased symptoms in the pre-menarche group (r=0.06, p<0.0001) but not in the post-menarche group (r=0.03, p=0.98). This interactive term moderated, the effect of hippocampal–visual network connectivity which was no longer significant (t=0.98, p=0.33, η²_partial<.0001) but not the menarche status (t=4.95, p<.0001, η²_partial=.0048), Figure 3. No other hippocampal-network (DMN, SAL, ECN) connectivity contributed significantly to the current model (p’s>.18, Table 3).

Figure 2.

Psychotic-Like Experiences (PLE) Severity Related to Hippocampal-Visual Network Connectivity: Left panel displays the main effect of hippocampal-visual connectivity related to PLE symptoms severity. The right panel displays the moderating effect of menarche status on hippocampal-visual connectivity.

Figure 3.

Menarche Status Moderates the Relationship Between Hippocampal-Visual Connectivity and Psychotic-Like Severity

Although theoretical relationships are expected to be unidirectional the statistical model testing these relationships examined a shared bidirectional relationship as reflected in the figure above.

3.3. Depression Symptom Severity Related to Menarche and Hippocampal Metrics.

Menarche status did not predict depression symptom severity (t=.97, p=.33), Table 4. Hippocampal-Salience network connectivity did relate to depression symptom severity (t=2.08, p=.0350, η²_partial=.0007), such that increased hippocampal-salience network connectivity was related to increased depression symptom severity, Figure 4.

Table 4.

Depression Symptom Severity Model Parameters

Parameter	t	p	η2partial
Menarche Status	−0.80	0.4244	0.0001
Hippocampal-Visual Connectivity	0.57	0.5665	>0.0001
Total Hippocampal Volume	0.51	0.6129	>0.0001
Hippocampal-Salience Network Connectivity	2.13	0.0329	>0.0008
Hippocampal-Executive Control Network
Connectivity	−1.31	0.1898	0.0003
Hippocampal-Default Mode Network Connectivity	−0.03	0.9745	>0.0001
Age	0.51	0.6099	>0.0001
Race			0.0097
Race/Ethnicity Black v White	−5.85	>0.0001
Race/Ethnicity Hispanic v White	−2.15	0.03170
Race/Ethnicity Asian v White	−2.28	0.0224
Race/Ethnicity Other v White	0.67	0.5011
Income			0.0100
Income [>$50,000 vs. 50–100,000]	−3.29	0.001
Income [>50,000 vs >100,000]	−7.04	>0.0001
BMI	4.04	>0.0001	0.0030

Figure 4.

Depression Symptom Severity Related to Hippocampal-Salience Network Connectivity

4. Discussion

Estradiol availability moderated the effect of hippocampal-dysconnectivity on PLE severity. Prior to estradiol availability (pre-menarche) reduced hippocampal connectivity related to increased PLE symptoms severity; when estradiol was available (post-menarche) PLE severity was not related to hippocampal dysconnectivity. This pattern was specific to PLEs and no similar relationship to estradiol availability was observed in depression symptom severity. Estradiol availability (menarche status) also had a continued direct effect on PLE symptom severity after accounting for this interaction. This continued effect of estradiol status may suggest the persistence of direct and interactive mechanisms by which this hormone may impacts PLE severity. As expected, there are PLE symptoms before and after the availability of estradiol, which reflects the myriad of factors that contribute to early psychosis lability. Despite the persistence of PLEs, the underlying causes and mechanisms may change after the availability of estrogen. These findings emphasize the utility of women’s health lens to understand the mechanisms of psychosis and how the importance of risk factors may change across development (Culbert et al., 2022; Patel et al., 2021; Tamminga, 1997). Indeed, this is a critical initial study, but much more work is needed to examine the influence of estrogen and identify potential changes in mechanisms of psychosis lability during neurodevelopment.

Contrary to our hypothesis, estradiol availability (post-menarche) related to higher PLE severity. Elevated PLE severity was driven by individuals who were post-menarche at each time point whose menarche occurred before baseline (when participants were age 8–9). This discrepancy raises several possibilities. First, prior work suggested that earlier timing of menarche is related to more normative hippocampal connectivity in individuals at clinical high-risk for psychosis (Damme et al., 2020) and later onset of psychosis (Cohen et al., 1999). However, it is important to note that in that previous study sample the earliest age at menarche was 11 year of age (Cohen et al., 1999; Damme et al., 2020). It is notable that 11–12 years old is normative timing of menarche according to national studies of pubertal timing (Biro et al., 2018; Cabrera et al., 2014; Chumlea et al., 2003), but in the current sample the pre-menarche group had menarche prior to the baseline (age 8–9 years old) before that normative, beneficial timing. Taken together these findings suggest the possibility of an optimal developmental timing for menarche to impact psychosis lability. Second, early onset of menarche is associated with high levels of life stress (Holdsworth and Appleton, 2020), which may also increase psychosis risk (Vargas and Mittal, 2022). As a result, individuals with the highest levels of early-life stress may have both a higher psychosis risk burden and earlier menarche. Finally, PLE items may reflect some normative changes that are co-occurring with the early onset of menarche; items may unintentionally capture the normative changes of puberty with items such as ‘Did you feel that parts of your body had suddenly changed or worked differently than before?’ These possibilities emphasize the need for additional research in this area to parse these critical aspects of menarche and development on PLEs.

Hippocampus-visual network connectivity was related to PLE severity above and beyond the variance accounted for by the other hippocampal measures in the model, including hippocampal volume and hippocampal connectivity to other major resting-state networks associated with psychosis (Menon, 2011). Additionally, this relationship between hippocampal dysconnectivity and PLE severity was moderated by estradiol availability (menarche status). This finding is consistent with our hypothesis and prior studies (Damme et al., 2020) that suggest hippocampal-visual area connectivity was related to psychosis risk symptoms and normalized by menarche. The current findings build on this line of research in several critical ways, including examining individuals in the relevant developmental context and examining the same individuals over time. Hippocampal-visual network connectivity is critical for perceptual and cognitive tasks, including memory formation (Raganath et al., 2005) and visual processing (Chan, 2017). Future work should examine whether these changes in hippocampal-visual connectivity relate to changes in cognitive processes. Changes in hippocampal-visual network connectivity has also been associated with treatment response to antipsychotics in first episode and chronic psychosis patients (Nelson et al., 2023). As a result, the current findings add to a growing literature that suggests the importance of hippocampal-visual network connectivity in the critical cognitive features, early psychosis course, and the role of estradiol.

Estradiol availability moderated the relationship between the hippocampal dysconnectivity and PLE severity, such that before estradiol was available hippocampal dysconnectivity related to variance in PLE symptom severity; after estradiol became available PLE severity may be accounted for by other sources of psychosis lability. This finding is the first to demonstrate that menarche moderates the effect of hippocampal dysconnectivity on PLE severity within the neurodevelopmental context surrounding menarche. It is also noteworthy that menarche had a continued direct effect on PLE severity. This finding raises the possibility that menarche may have other mechanisms of impact that are related to PLE severity. This study reflects an initial investigation among a larger body of work that would be needed to elucidate the importance of estradiol events for psychosis.

Direct and mediated menarche effects were specific to PLE severity (Brockington, 2011; Culbert et al., 2022). Although estradiol has been related to increased onset and severity of depression symptoms, the current study found that there was no general impact of the availability of hormones along this symptom dimension. This finding is consistent with prior research that suggests that estradiol has a general benefit to psychosis but not depression symptoms (Bergemann et al., 2007). Indeed, a growing depression literature suggests that the impact of estradiol may be more related to acute changes during the menstrual cycle stage (Brockington, 2011; Culbert et al., 2022) rather than to hormonal events as in psychosis. It is notable, however, that depression symptom severity did relate to hippocampal-salience network connectivity above and beyond the other hippocampal metrics. Hippocampal-salience network connectivity has been associated with increased vigilance (Keefe et al., 2022) and depression treatment response (Philip et al., 2018). These distinct pattern of depression findings did not show a general or moderated benefits of menarche, suggesting a distinct relevance of estradiol to PLE severity.

These analyses highlight the importance of considering the developmental context surrounding emerging symptoms of severe mental illness, particularly PLE severity (Mittal and Wakschlag, 2017). However, there are notable limitations to the current study that require future research. First, the current sample is not enriched for psychosis. PLEs are a developmentally appropriate measure in the current sample age range (Patel et al., 2021; Vargas and Mittal, 2022), but these experiences are relatively common in childhood (Yung et al., 2009) and may not be a strong index of imminent risk for conversion to psychosis (Kelleher et al., 2012; Kelleher and Cannon, 2011). It is notable that PLEs have been related to risk for transition to psychosis, including predicting emerging prodromal syndromes (Cowan and Mittal, 2021). Second, menarche is a good general marker of early- or late-stage of pubertal development and the availability of peripheral estradiol (Cheng et al., 2021). However, menarche status cannot parse the acute from long-term, neurodevelopmental impacts of estradiol. As a result, more lifespan work is needed to understand the impacts of estrogen in terms of menstrual cycle fluctuations, developmental fluctuations, and the maturational/neural impacts of these hormones. Third, we believe that the hippocampus is a critical mechanistic region for these psychosis and menarche effects, but it is possible that other brain regions may also play a critical role. Future research should examine additional brain regions as potential sites for estradiol’s influence, especially the prefrontal cortex, striatum, and cerebellum. Fourth, there are several factors that may influence the timing of menarche including nutrition, social stress, trauma, and socioeconomic status, which may be important future targets to modulate the timing of menarche. Future studies should examine the mechanistic contribution of these factors to the timing of menarche. Finally, these estradiol effects may be sex-specific in their impact or relevance for symptoms. Indeed, hippocampal cells show sexual dimorphism (Tabatadze et al., 2015) that may result in sex-specificity of these mechanisms. Alternatively, mechanisms related to emerging symptoms may be general to both sexes though only targeted by menarche in biological females. In such a case, relevant mechanisms, i.e., hippocampal-visual network connectivity, may be a treatment target for neuromodulation that all individuals at risk for psychosis onset regardless of sex. Future studies should examine the sex specificity of these sites of estrogens action.

Highlights.

• Estrogen availability moderated the relationship between hippocampal connectivity and psychosis lability.

• Prior to menarche, hippocampal dysconnectivity accounted for more variance in psychotic-like experiences severity than after menarche, suggesting a change in the factors that contributed to symptoms.

• These results provide the first evidence of estrogen availability impacting psychotic-like experiences through hippocampal dysconnectivity in early adolescence.