Fronto-Amygdala Resting State Functional Connectivity and Anxiety Symptoms in Adolescent Girls

Introduction

• Early adolescence marks the emergence of sex differences in anxiety, with girls reporting significantly higher anxiety symptoms than boys.
• High anxiety symptoms during this period increase the risk of anxiety disorders, depression, and suicide in adulthood, particularly for girls.
• The neurodevelopmental processes underlying this rise in anxiety symptoms in adolescent girls are not well understood.
• Changes in pubertal hormones may play a role, influencing brain development and increasing vulnerability to anxiety.
• Fronto-amygdala function, implicated in emotion regulation and anxiety, may be a plausible neurodevelopmental pathway for this increased vulnerability in girls.

Fronto-Amygdala Circuitry and Anxiety

• The amygdala is involved in emotional learning and attention to salient cues.
• The ventromedial prefrontal cortex (vmPFC) regulates emotions like fear and anxiety.
• The vmPFC includes medial structures of the orbital frontal cortex (OFC) (BA 10, 11, 12) and rostral anterior cingulate cortex (rACC) (BA 25, lower 24).
• Functional and anatomical subregions of the vmPFC may exhibit distinct developmental trajectories.
• There are direct and reciprocal projections between the vmPFC and the amygdala.
• Evidence supports the inhibitory role of the vmPFC on amygdala activity, including the regulation of physiological signals of arousal.
• The vmPFC and amygdala project to other regions involved in emotion processing and regulation, indicating their presence within several neural networks.
• The function and structure of the vmPFC and amygdala change considerably during late childhood and adolescence.
• Functional connectivity between these regions also undergoes age-related changes.
• When processing threat, vmPFC-amygdala functional connectivity shifts from more positive in childhood to more negative in adolescence.
• Altered regulatory control of the vmPFC on the amygdala is thought to contribute to anxiety symptoms.
• Youth with anxiety often show heightened amygdala activation and reduced fronto-amygdala connectivity to threat-related stimuli.
• Alterations in age-related vmPFC-amygdala resting state functional and structural connectivity have been associated with higher levels of anxiety and depressive symptoms.
• Puberty-related influences on fronto-amygdala connectivity could be a pathway by which girls become at increased risk for anxiety disorders during early adolescence.

Puberty, Fronto-Amygdala Function, and Anxiety

• Puberty involves the development of secondary sexual characteristics and attainment of reproductive capacity along adrenarcheal and gonadarcheal axes.
• Adrenarcheal changes begin around five to seven years of age, characterized by the release of adrenal hormones (DHEA and testosterone) and physical changes.
• Gonadarcheal changes typically follow adrenarche and are associated with maturation of primary and secondary sexual characteristics, driven by gonadal steroid hormones (testosterone and estradiol).
• Pubertal hormones have organizing effects on the brain, and pubertal timing could directly influence subsequent neurodevelopment.
• Individual variation in adrenarcheal and gonadarcheal processes are associated with anxiety symptoms.
• Earlier pubertal timing may be associated with higher anxiety symptoms, potentially mediated by hormonal effects on brain development.
• Early adrenarcheal timing may be linked to heightened anxiety via altered amygdala to inferior frontal cortex functional connectivity.
• The functioning of neural regions may be directly influenced by changes in hormone levels with pubertal maturation, given the high density of androgen and estrogen receptors in medial temporal regions.
• Testosterone's modulating effects on fronto-amygdala functional connectivity are evident from testosterone administration studies.
• Earlier pubertal timing could lead to a rise in anxiety symptoms, mediated by the effects of increases in pubertal hormones (i.e., testosterone) on vmPFC-amygdala resting state functional connectivity (i.e., reduced coupling).
• The rise in anxiety symptoms could also be linked to heightened sensitivity of the modulatory effects of pubertal hormones on the innervation of the amygdala to the prefrontal cortex.
• Individual variation in neurobehavioral sensitivity to changes in levels of pubertal hormones at the receptor level may contribute to hormone-related psychopathology.
• Some girls may be particularly sensitive to the modulatory effects of testosterone on fronto-amygdala function.
• Testosterone levels would play a moderating role in the association between fronto-amygdala functional connectivity and anxiety symptoms.
• Higher anxiety symptoms associated with earlier pubertal timing could also be linked to alterations in girls’ behaviors or the behaviors of those around them.

Goal and Hypotheses of the Study

• The study tested primary hypotheses using a moderated-mediation model in early-adolescent girls varying in temperamental risk for anxiety disorders.
• The moderated-mediation model tested both mediating and moderating relationships between testosterone/pubertal development, vmPFC-amygdala functional connectivity, and anxiety symptoms.
• Levels of testosterone would moderate the relation between vmPFC-amygdala resting state functional connectivity and anxiety symptoms.
• Reduced vmPFC-amygdala functional connectivity would be associated with higher anxiety symptoms, particularly in girls with higher levels of testosterone.
• Pubertal status would play a moderating role in the relation between vmPFC-amygdala connectivity and anxiety symptoms.
• Reduced functional connectivity would be associated with higher anxiety symptoms in girls more advanced in self-reported pubertal status, particularly in adrenarcheal processes related to the rise in testosterone in girls.
• vmPFC-amygdala connectivity would mediate the relations between testosterone/pubertal development and anxiety symptoms.
• Mediated moderation models included ‘age’ as a covariate to ensure that developmental effects were specific to puberty and not to effects of chronological age.
• A vmPFC parcellation was used to explore which subregions of the vmPFC would be related to testosterone/pubertal development and anxiety symptoms.

Methods

• A total of 70 participants were used in the reported analyses which were derived from a larger sample. The overall sample consisted of 129 early adolescent girls ages 11–13.
• Recruitment was based on parent’s report of their child’s sex assigned at birth; gender identity was not assessed at intake.
• The current study is part of a larger longitudinal study investigating how social threat sensitivity in early adolescence confers risk for social anxiety and depression in mid- to late-adolescence.
• Recruitment was based on oversampling for shy/fearful temperament, which is a risk factor for social anxiety disorder into adolescence and adulthood.
• Temperament was assessed using online screening form prior to participants’ first visit using the Early Adolescent Temperament Questionnaire-Revised (EATQ-R; Ellis and Rothbart, 2001).
  • The sample was stratified such that approximately 2/3 of participants (n = 85) scored > 0.75 SDs above the mean on the parent- or adolescent-rated fearfulness (3.12 for parent-report, 3.48 for adolescent-report) or shyness scales (2.99 for parent-report, 3.16 for adolescent-report).
  • All other participants (n = 44) scored below this cut-off and were within the normative range of shy/fearful temperament.
• Exclusion criteria included:
  • meeting DSM-5 criteria for a current or lifetime diagnosis of any anxiety disorder (except for specific phobia), obsessive-compulsive disorder, post-traumatic stress disorder, major depressive disorder, or any psychotic or autism spectrum disorder, as determined by the Kiddie-Schedule for Affective Disorders and Schizophrenia (KSADS-PL; Kaufman et al., 2016);
  • IQ lower than 70, as assessed using the Wechsler Abbreviated Scale of Intelligence (WASI; Wechsler, 1999);
  • a lifetime presence of a neurological or serious medical condition;
  • the presence of any MRI contraindications;
  • presence of head injury or congenital neurological anomalies (based on parent report);
  • acute suicidality; and taking medications that affect the central nervous system and hormonal function (e.g., selective serotonin reuptake inhibitors, steroid medication).
    • Stimulants were permitted if use was discontinued for 36 h prior to the scan.
• Data for the current analysis focused on the baseline assessment and included participants according to the following criteria:
  • good quality resting state fMRI data (see below for criteria),
  • salivary assays to index basal levels of testosterone, and
  • self-report measures of pubertal status anxiety symptoms per the Screen for Child Anxiety Related Emotional Disorders (Birmaher et al., 1997).
• From the 129 participants, 90 completed resting state scans at baseline. Of these 90 participants, 89 had Pubertal Development Scale (PDS) score and 83 had hormone data.
• Resting state scans of 18 participants were excluded due to excess motion (see below) and one was excluded due to an incidental finding.
• Thus, a total of 70 participants (42 high in shy/fearful temperament; age: M = 12.30, SD =0.79) were included in the analysis linking resting-state fronto-amygdala functional connectivity, anxiety symptoms, and at least one measure of puberty. Out of those 70 participants, 64 had hormone data.

Procedure

• The study was approved by the University of Pittsburgh Human Research Protection Office and occurred in three visits to the lab, including informed consent, screening procedures, and neuroimaging protocol.

Measures

• Risk status at recruitment was assessed using the EATQ-R (Ellis and Rothbart, 2001), considering shyness and fearfulness scales from both parent and self-reports.
• Demographic variables were collected, including age, sex, and socio-demographic status.
• Pubertal development was assessed using the Pubertal Development Scale (PDS) (Petersen et al., 1988) to derive overall, adrenarcheal, and gonadarcheal development scores.
• Testosterone levels were assessed using salivary assays, collecting samples immediately upon awakening on 4 separate days over a four-week period (one sample per week). Saliva samples were stored in an ultra-cold − 80◦C freezer. $CV= 10\%$
  • The averaged intra-assay CV was M= 1.20, SD= 0.23.
  • The average inter-assay CVs across plates was 17.26.
  • As expected, the distributions of hormones were positively skewed, with skewness of 7.16 (SE=1.57).
  • Few samples (n = 13) were below the level of detection of the assay and 3 were excluded for having low levels of testosterone for all four samples.
• Clinical assessments included the K-SADS-PL (Kaufman et al., 2016) to determine DSM-5 diagnoses and the SCARED (Birmaher et al., 1999) to assess child symptoms of anxiety.
• Neuroimaging data were acquired using a 3 Tesla Siemens Prisma magnet with a 64-channel phase array coil.
• Functional images were acquired using a multiband gradient-echo T2 * -weighted EPI sequence (multiband acceleration factor: 3, 60 slices, TR: 1500 ms, TE: 30 ms, flip angle: 55◦, echo spacing: 0.60 ms, field of view: 221 × 221 ×138 mm, voxel dimensions: 2.3 mm × 2.3 mm×2.3 mm).

fMRI Processing

• fMRI data were preprocessed using fMRIPrep (v1.5.3; Esteban et al., 2019), including steps for coregistration, susceptibility artifact correction, normalization, and motion parameter estimation.
• Additional processing steps included 36 parameter nuisance regression, high-pass spectral filtering at 0.008 mHz, and bandpass filtering.
• Framewise displacement (FD; Power et al., 2012) was used to measure motion contamination, removing volumes with FD greater than 0.3.

Regions-of-Interest (ROI) Mask and Extraction

• The set of ROIs used to characterize both the vmPFC and the bilateral amygdala was constructed by combining the 6 ROI vmPFC parcellation from Chase et al. (2020) (Chase et al., 2020) with the right and left amygdala ROIs of the AAL3 atlas (Rolls et al., 2020).
  Because of the amount of data loss, we did not examine the connectivity between these ROIs and the bilateral amygdala ROIs. We did include ROI 1 and ROI 6, when available, in the computation of the partial correlations for other ROIs.

Statistical Analyses

• Functional connectivity was operationalized as the lag-0 partial correlation.
• Moderated mediation models (Preacher et al., 2007) were used to examine the relationships between testosterone/pubertal development, vmPFC-amygdala resting state functional connectivity, and anxiety symptoms.

Results

• Two models contained a significant moderating effect for vmPFC ROI 3 – Right Amygdala: one for testosterone and one for adrenarcheal processes.
• One model for gonadarcheal processes showed a significant moderating effect on vmPFC ROI 2 – Left Amygdala.
• For models with testosterone levels and those with adrenarcheal development scores, high levels of testosterone or more advanced adrenarcheal development serve to potentiate the negative relation between vmPFC ROI 3 – Right Amygdala connectivity on anxiety, as evidenced by significant moderation effects:
  Testosterone: $\beta = -0.269$, SE = 0.117, p < .05
  Adrenarcheal Development: $\beta = − 0.256$, SE = 0.116,p < .05.
• We also found a significant moderated mediation model for gonadarcheal development where more advanced development moderated the relation between vmPFC R01 2 – Left Amygdala connectivity and anxiety symptoms ($\beta = − 0.249$, SE = 0.121,p < .05).

Sensitivity analyses

• Refitting the models while controlling for mean FD showed the same pattern of significant findings with minimal change in either the effect estimates or standard errors, suggesting that in-scanner motion is not confounding our relations of interest.
• When ROI 1 and ROI 6 were not considered for any participant when calculating partial correlations the moderating effect of adrenarcheal development becomes marginally significant.
• Moderation only models were fit. Only the moderating effect of testosterone on ROI 3 - R Amygdala connectivity’s relation with anxiety was significant; however, the other two reported effects for adrenarcheal and gonadarcheal processes were marginally significant and of approximately the same magnitude as in the moderated-mediation models.
• AIC selects the moderated-mediation model, while BIC selects the moderation only model, albeit by less than 2 points.
• For our significant adrenarcheal development and gonadarcheal development models, both AIC and BIC selected the moderation only model, which corresponds to the lack of significant direct relations between the developmental measure and connectivity.

Discussion

• vmPFC-amygdala connectivity was associated with anxiety symptoms only at higher levels of testosterone and more advanced pubertal development.
• Specifically, we found that testosterone and adrenarcheal processes moderate the negative association between the right amygdala and vmPFC ROI 3 connectivity and anxiety symptoms.
• In addition, gonadarcheal processes moderate the negative association between left amygdala and vmPFC ROI 2 connectivity and anxiety symptoms.
• Our findings support our hypothesis regarding the moderating role of testosterone levels on the negative association between vmPFC-
  amygdala functional connectivity and anxiety symptoms.
• Findings were significant for both sexes in that study, the authors noted that the range of change in testosterone levels was smaller in female participants suggesting that girls may be more sensitive to the modu­latory effects of testosterone on this circuitry.
• We addressed this gap by using a vmPFC six-cluster parcellation (Chase et al., 2020) to explore puberty-related connectivity between the amygdala and subregions of the vmPFC.
• One mechanism could be via changes in the sensitivity of sex steroid hormone receptors in these vmPFC subregions or their connections with the amygdala.

Strengths and limitations

• Strengths of this study include
  • examining both basal levels of testosterone and self-reported pubertal development considering separate processes of adrenarche and gona­darche in a well-characterized sample of early adolescent girls varying
  • considering level of temperamental risk of anxiety disorder.
  • In addition, testosterone concentrations were derived using hierarchical linear modeling with four morning saliva assays collected over one month.
    *(Chase et al., 2020) provides an in-depth analysis of the connectivity of the amygdala with functional subregions of the vmPFC improving interpretation of findings through the mapping of these subregions to corresponding regions in rodent and non-human primate studies.
• limitations are important to consider for future research.
  • it remains unclear whether findings might extend to adolescent boys or whether they extend to non-cisgender youth (transgender, non-binary).
  • the moderate sample size within this age range may have limited our ability to detect variability in pubertal development as a function of age and examine the specific ef­fects of temperamental risk status.
  • the use of a cross-sectional design precludes drawing conclusions about the direction of the rela­tionship between pubertal development, fronto-amygdala functional connectivity, and anxiety symptoms
  • the lack of directionality of the functional connectivity findings precludes our
    ability to determine whether more advanced pubertal maturation is
    linked reduced inhibitory vmPFC connections to the amygdala or
    reduced amygdala input into the vmPFC.
  • the exploratory nature of
    the analyses with respect to the specific subregions of the vmPFC,
    combined with the moderate sample size, raises the possibility that these
    findings may include false positive effects.
    Exclusion of data from ROI 1 and 6
    impacted inference showed that upon removing ROI 1 and 6 from
    consideration entirely, the moderating effect of adrenarcheal develop­
    ment becomes marginally significant.

Conclusions

• Findings from the current study show that vmPFC-amygdala resting state functional connectivity is associated with anxiety symptoms only in early adolescent girls more advanced in adrenarcheal and/or gona­darcheal development and with higher levels of testosterone.
• The moderating role of perceived pubertal
  maturation and testosterone varies as a function of vmPFC subregions
  suggesting potentially different neurodevelopmental mechanisms in rise
  of anxiety in at-risk girls.
• Longitudinal studies
  examining puberty-related neurodevelopmental mechanisms underly­
  ing trajectories of anxiety symptoms in male and female youth are also
  needed to deepen our understanding of the sex differences in the
  development of internalizing disorders that emerge during adolescence.