Abstract
Recent scholarship has highlighted the importance of understanding relations between hormones, rather than studying hormones in isolation. Considering neuroendocrine coupling, or the coordination of hypothalamic-pituitary-adrenal (HPA) axis and hypothalamic-pituitary-gonadal (HPG) axis hormones over time, is one way to investigate how systems governing stress responsiveness and pubertal development covary during critical periods. To date, however, most work has considered hormone coupling cross-sectionally. The current study investigated neuroendocrine coupling in a longitudinal sample from the Northeastern United States. Youth (N=437, 53% male, 90% White) provided saliva samples for analysis of diurnal hormone activity at ages 9 (three samples per day across three days) and 12 (one sample per day on the same weekday for four weeks). At both timepoints, samples collected 30-minutes after waking were assayed for cortisol, dehydroepiandrosterone (DHEA), and testosterone. Multi-level growth modeling (MLM) was used to determine how levels of morning cortisol changed in tandem with DHEA and testosterone. Morning cortisol-DHEA coupling varied by child sex, as males’ cortisol-DHEA diminished over time, especially among pubertally-advanced males. Females, in contrast, demonstrated strengthening cortisol-DHEA coupling over time, especially more pubertally-advanced females. Morning cortisol-testosterone coupling did not vary by sex or pubertal status, demonstrating strengthening associations between ages 9 and 12. The current findings contribute to the literature on hormone coupling across development and expand this work into an earlier developmental phase than previously investigated.
Keywords: cortisol, DHEA, testosterone, hormone coupling, sex, hormones, adolescence
Adolescence is a period of marked biological, psychological, and social changes. These changes, however, do not occur independently. Rather, the psychological and social changes of adolescence are heavily influenced by the primary biological change of this life stage - puberty (Pfeifer & Allen, 2021). Research demonstrating age- and puberty-related changes in cognition, affect, and relationships has led to increased interest in the biological events of puberty and how their emergence may be related to the increase in behavioral and emotional difficulties often seen in adolescence (Barendse et al., 2021; Colich et al., 2020; Hamlat et al., 2019; Roberts & Lopez-Duran, 2019).
Puberty is initiated via a complex cascade of hormonal events (Grumbach & Styne, 2003). Gonadatropin releasing hormone (GnRH) is first released from the hypothalamus, which then stimulates the release of luteinizing hormone (LH) and follicle stimulating hormone (FSH) from the anterior pituitary. These hormones, in turn, are responsible for signaling sex organs and the adrenal cortex to release, among others, androgens such as dehydroepiandrosterone (DHEA) and testosterone, and estrogens such as estradiol. The pathway upon which this sequence of events unfolds is known as the hypothalamic-pituitary-gonadal (HPG) axis.
DHEA is the one of the earliest androgen hormones to be released; the events initiated by the release of DHEA are known as adrenarche. Features of adrenarche include increases in body hair, skin changes, and changes in body odor (Auchaus & Rainey, 2004). Although puberty is typically conceptualized as occurring during early adolescence, normative DHEA increases can be detected as early as age 6 (Wood, Lane, & Cheetham, 2019). Concurrent to adrenarche, surges of testosterone and estradiol facilitate the development of secondary sex characteristics, such as breast/gonad development, voice changes, and changes in body shape (Grumbach & Styne, 2003); these changes are known as gonadarche.
Importantly, while testosterone and estradiol dominate the hormonal milieu during later puberty, DHEA is still being produced. While it continues to play an important role in pubertal development, DHEA is also involved in stress responding and the functioning of the hypothalamic-pituitary-adrenal (HPA) axis (Hucklebridge, Hussain, Evans, & Clow, 2005). Indeed, while DHEA’s role as a pubertal hormone may be brief, it continues to play a critical part in the activities of the HPA axis throughout the lifespan.
The HPA axis is responsible for initiating and managing the physiological events that occur when humans are faced with stress (Dickerson & Kemeny, 2004). In response to acute threat, corticotropin-releasing hormone (CRH) is released from the hypothalamus, which triggers the subsequent release of adrenocorticotropic hormone (ACTH) from the pituitary and glucocorticoid release from the adrenal glands. Humans release glucocorticoids in the form of cortisol, which is released normatively throughout the day and surges in response to acute stressors.
The HPA axis also undergoes developmental changes. Prior to adrenarche, HPA axis reactivity is limited and primarily activated in response to stressors involving the child’s primary caregiver or home environment (Kamin & Kertes, 2017; Yim, Quas, & Rush, 2015). In middle childhood, however, HPA axis activity increases and is increasingly sensitive to psychological and physiological challenges (Foilb, Lui, & Romeo, 2011). Previous research, for example, has indicated that acute stress among youth at familial risk for major depressive disorder is associated with blunted cortisol activity to stressors prior to puberty, but increased cortisol release in reaction to stressors following puberty (Colich, Kircanski, Foland-Ross, & Gotlib, 2015; Hankin, Badanes, Abela, & Watamura, 2010).
The HPA and HPG axes have traditionally been considered in isolation. An emerging literature, however, is investigating how these axes work together, especially during developmentally sensitive periods such as puberty (for a review, see Zakreski et al., 2018). This work is especially relevant considering research suggesting, for example, that early life stress may influence the timing of pubertal development (Joos, Wodzinski, Wadsworth, & Dorn, 2018). Investigations of dual-axis hormone coupling, or covariance between hormones across axes, have typically measured relationships between cortisol and androgen hormones such as DHEA and testosterone (e.g., Dismukes et al., 2015; Johnson et al., 2014; Marceau et al., 2015; Ruttle et al., 2015).
Traditional views on dual-axis coupling suggested that the HPG and HPA axes should down-regulate one another as a way to both promote reproductive function during times of rest and inhibit it during times of stress (Gomez, Manalo, & Dallman, 2004; Viau, 2002). This would result in “negative” or inverse HPA-HPG axis coupling (i.e., when HPA axis hormones increase, HPG axis hormones decrease and vice versa). This system is likely advantageous during adulthood; mutual down-regulation of these axes during developmentally sensitive periods, however, may have long-lasting negative impacts on an individual’s reproductive development and subsequent stress responding (Shirtcliff & Ruttle, 2010). A developmentally sensitive system may result in either more “positive” coupling (i.e., as HPA hormones increases, HPG hormones increase as well), or even null coupling (i.e., increases in HPA hormones are not dependent upon activity of HPG hormones).
Existing literature on hormone coupling supports the notion of a more developmentally sensitive neuroendocrine system. Cross-sectional investigations have explored coupling, typically operationalized as within-person variability in hormone levels, in response to acute stressors (reactive coupling; Bobadilla et al., 2015; Marceau et al., 2014) as well as of basal/diurnal hormone levels (Black et al., 2018; Dismukes, Johnson et al., 2015; Dismukes, Shirtcliff et al., 2015; Harden et al., 2016; Johnson et al., 2014; Marceau et al., 2014; Ruttle et al., 2015). While reactive coupling indexes relations between hormones in the context of an activated stress response system, basal/diurnal coupling provides a clearer picture of normative day-to-day patterns of hormonal covariance and allows for investigation of how these patterns may change as the HPA and HPG axes mature. As reviewed in Zakreski et al. (2018), the majority of the studies investigating basal/diurnal and reactive coupling in adolescence have supported the notion of positive HPA-HPG coupling during adolescence, with a switch to more negative coupling patterns in early adulthood.
Despite the surge of research investigating dual axis coupling in the last decade, longitudinal studies of hormone coupling are largely absent from the literature. This is a critical gap as the existing research does not elucidate how patterns of hormone coupling may change over development broadly, and the pubertal transition specifically. One exception is Ruttle et al. (2015), who reported on longitudinal patterns of hormone coupling in an adolescent sample assessed from ages 11–15. Their results suggest that positive HPA-HPG coupling predominates in early adolescence, with a switch to more negative, “adult-like” coupling patterns as adolescents age.
Investigations of hormone coupling have typically been limited to adolescence and adulthood; researching coupling in earlier developmental periods may further our understanding of the developmental trajectory of hormone coupling throughout the lifespan. The current study builds on a previous cross-sectional investigation of diurnal hormone coupling in 9-year-old children from a large community sample (Black et al., 2018) by examining how patterns of coupling change as children enter early adolescence. For the purposes of the current study, “coupling” is operationalized as within-person variability of cortisol, DHEA, and testosterone levels sampled 30-minutes after waking. Cortisol, DHEA, and testosterone all demonstrate circadian patterns associated with higher morning values (Guyton & Hall, 1996), so all hormones are theoretically at their daily peak 30-minutes from waking. Therefore, the goals of utilizing samples collected 30 minutes after waking was three-fold: 1. To examine hormones at their peak values during the day, wherein there would be maximal variability in hormone values; 2. To compare basal HPA and HPG relations rather than HPA-HPG relations in response to stress; and 3. To compare HPA and HPG hormones from the exact same collection time (as matching samples were not available at other times of day).
Based on the existing literature, we hypothesized that cortisol-DHEA coupling would become more positive as children age (from ages 9 to 12). Similarly, we hypothesized that cortisol-testosterone coupling would become less positive as children move through puberty (age 12). We also explored sex and pubertal development as moderators of these relationships. Changes in hormone coupling result from maturation of the HPA and HPG axes; maturation of the HPG axis, in turn, is the primary driver of pubertal development (Havelock, Auchus, & Rainey, 2004). Therefore, girls (who advance through puberty earlier than boys; Dorn et al., 2006) and those who are more pubertally advanced compared to their peers are likely to demonstrate more developmentally advanced patterns of hormone coupling (i.e., more positive cortisol-DHEA coupling and less positive cortisol-testosterone coupling) at younger chronological ages (Black et al., 2018; Marceau et al., 2014; Marceau et al., 2013; Ruttle et al., 2015; Simmons et al., 2015). Therefore, we hypothesized that sex and pubertal status would be significant moderators of this relationship, with girls and more pubertally advanced youth demonstrating more mature patterns of hormone coupling than boys and less-developed youth, respectively.
Materials and Method
Participants
Participants were 437 children from a larger longitudinal study of temperament and risk for psychopathology (Klein & Finsaas, 2017). Information from two waves of data collection, when children were approximately 9- and 12-years-old, was used for the current study. The sample was recruited when children were 3 years old using commercial mailing lists of families within a 20-mile radius of a Northeastern university. 559 children entered the study, and another 50 families were recruited 3 years later to increase the representation of families of color in the sample. Age 9 (Wave 1) data were collected between September 2010-April 2013, and Age 12 (Wave 2) data were collected between May 2014-June 2016. Participant ages were staggered such that one third of the sample turned 9 (or 12) per year of the data collection (e.g., one third turned 9 during 2010–2011, one third turned 9 during 2011–2012, etc.). Participants were scheduled for their study visits as close as possible to their 9th (or 12th) birthdays. The age 9 assessment was completed by 490 of the 609 families (80.5%). The age 12 assessment was completed by 472 of the 609 families (77.5%), and 91.2% (n = 447) of families who had completed the age 9 assessment.
The analysis sample was drawn from the youth who completed saliva collection at either age 9 (n = 419; 85.5% of the age 9 sample) or age 12 (n = 402; 85.1% of the age 12 sample). After applying exclusion criteria (see below), the final sample consisted of 437 participants. The final sample was 9.27 years old (range = 8.26–11.72 years, SD = 0.43) at the age 9 assessment, and 12.68 years old (range = 11.52–14.70 years, SD = 0.44) at the age 12 assessment. Based on parent-report, children were 52.9% male (n = 231), 90.2% White, 7.1% Black, 2.5% Asian American, 0.2% Native American, and 11.4% Hispanic/Latinx. Around half of parents in the current sample had earned a 4-year college degree or higher (57.4% of mothers and 43.3% of fathers), and the median family income was between $100,000 and $120,000 per year. Excluded children (including those excluded due to lack of participation in the age 9 or 12 assessments, lack of saliva collection at either assessment, saliva exclusion for conditions listed below, or other missing data; n = 172) were significantly more pubertally developed than those who participated (M Pubertal Development Scale [PDS] score = 12.08 versus M PDS score = 11.07; t(466) = 2.28, p <.05; although Hedges’ g = .31, suggesting a small effect). Mothers of children included in the current analyses had significantly higher levels of education (x2 = 13.09, p <.05), but fathers had significant lower levels of education (x2 = 17.15, p <.05) than excluded children. Otherwise, non-included participants did not differ from the final sample of children included in the analyses with regard to age, sex, family income, or race/ethnicity.
Procedure
At both ages 9 and 12, children and their parents completed a laboratory visit during which demographic and pubertal development information was collected. Saliva was collected in the home at both assessments. Parents were instructed how to conduct the at-home saliva collection during an in-person study visit and were subsequently contacted by study staff to answer any questions and coordinate sample pick-up.
Measures
Age 9 hormone collection and assay.
At age 9, children’s saliva was collected via passive drool immediately upon waking, 30 minutes after waking, and 30 minutes before bedtime on three consecutive weekdays, resulting in nine saliva samples per participant. Cortisol was assayed from all nine samples; for the purposes of the current study, however, only cortisol levels collected 30 minutes after waking were utilized to allow for real-time comparison between cortisol and other hormone values. Testosterone and DHEA were assayed only from the samples taken 30 minutes after waking. Participants were instructed to freeze saliva samples immediately after collection until a member of the study staff retrieved the samples from the participants’ homes. The samples were then stored at −20°C until they were transported on dry ice to the Biochemistry Laboratory at the University of Trier in Trier, Germany for analysis.
All samples were assayed in duplicate. Cortisol was assayed using a time-resolved fluorescence immunoassay with flourometric end-point detection (DELFIA). Hormone outliers (i.e., greater than 3 standard deviations above the mean) were winsorized to normalize distributions and set to 3 SD above the mean. DHEA and testosterone were assayed from the same aliquots and at the same lab as cortisol, using commercially available enzyme immunoassays specifically designed for use with saliva according to the manufacturer’s recommended protocol (Salimetrics Laboratories, Irvine, CA). The intra- and inter-assay coefficients of variation for cortisol, DHEA and testosterone can be found in Table 1. Hormone outliers were winsorized to normalize distributions. A total of 12 cortisol samples from 12 participants, 26 DHEA samples from 20 participants, and 11 testosterone samples from 11 participants were winsorized. A natural log transformation was applied to all cortisol, DHEA and testosterone data to reduce skew, and all analyses used transformed hormone values.
Table 1.
Inter- and intra-assay coefficients of variability for hormones at ages 9 and 12 assessments.
Inter-assay coefficient | Intra-assay coefficient | |
---|---|---|
Age 9 assessment | ||
Cortisol | 9.8 | 5.4 |
DHEA | 7.1 | 2.5 |
Testosterone | 9.0 | 6.7 |
Age 12 assessment – Lab 1 | ||
Cortisol | 12.6 | 6.7 |
DHEA | 9.6 | 7.9 |
Testosterone | 6.2 | 4.0 |
Age 12 assessment – Lab 2 | ||
Cortisol | 11.6 | 5.1 |
DHEA | 13.7 | 7.1 |
Testosterone | 8.7 | 5.1 |
Age 12 hormone collection and assay.
At age 12, saliva collection occurred on the same weekday for three weeks (male participants) or four weeks (female participants) in a row. This collection change from age 9 was made to collect estradiol samples from females, as estradiol must be collected over the full menstrual cycle to derive a true baseline estimate (Choe et al., 1983).1 Saliva samples were collected 30 minutes after waking on all sample days, and cortisol, DHEA, and testosterone were assayed from all samples. As at the age 9 assessment, participants were instructed to freeze saliva samples immediately after collection until a member of the study staff retrieved the samples from the participants’ homes. The samples were then stored at −20°C until they were transported on dry ice to laboratories for analysis.
22.1% of samples (n = 89 participants) were assayed at Madison Biodiagnostics (Lab 1) in Madison, WI, while 77.9% of samples (n = 313 participants) were assayed at Salimetrics Laboratories (Lab 2) in Irvine, CA. Salimetrics kits were utilized for analyses at both labs. Approximately 9.4% of samples (collected from 38 participants) were assayed at both laboratories to verify comparable analyses between labs. Samples assayed at two laboratories were highly correlated with one another (r = 0.95 for cortisol levels, r = 0.82 for DHEA levels, and r = 0.86 for testosterone levels).2 Hormone outliers were winsorized to normalize distributions. A total of 14 cortisol samples from 11 participants, 18 DHEA samples from 15 participants, and 31 testosterone samples from 18 participants were winsorized. All cortisol, DHEA, and testosterone levels at age 12 were natural log transformed, and log transformed hormone values were used in the final analyses.
Sample exclusion.
First, samples that were not frozen in a home freezer or that melted in transit to the university were excluded, as sample accuracy may be compromised following a freeze-thaw cycle. Second, any saliva sample that was collected more than 15 (cortisol) or 30 (DHEA and testosterone) minutes after the intended time (30 minutes after waking; as indicated by a participant-completed diary) was excluded in order to account for the hormones’ diurnal patterns. Third, samples were excluded if the child was taking an oral or inhaled corticosteroid, antipsychotic, or methyphenidate – extended release (Concerta) during sample collection, as these medications have been shown to affect hormone levels in children and adolescents (Granger et al., 2012). Following sample exclusion, of the 419 children who completed home saliva collection at age 9, 387 (90.2%) had at least one useable cortisol sample, 368 (87.8%) had at least one useable DHEA sample, and 380 (90.6%) had at least one useable testosterone sample. For age 12 saliva collection, of the 402 children who completed home saliva collection, 372 (92.5%) had at least one useable cortisol sample, 375 (93.2%) had at least one useable DHEA sample, and 378 (94.0%) had at least one useable testosterone sample. There were no significant differences between those children with missing versus non-missing hormone values with regard to age, sex, race/ethnicity, or pubertal development.
Pubertal development.
The Pubertal Development Scale (PDS; Petersen et al., 1988) was administered to mothers at the age 9 and 12 assessments. The PDS assesses pubertal development using five items. Items for both boys and girls include growth of body hair, skin changes (especially pimples), and growth in height, while items for boys only include voice deepening and growth of facial hair, and items for girls only include breast development and menstruation. Mothers rated each item on a scale from 1 (not yet started) to 4 (seems complete) (with the exception of a question about menarche, which was rated on a binary response scale). A single sum score of the PDS items at both timepoints was used in the current analyses (PDS range at age 9: 5–14; PDS range at age 12: 5–18).
Body Mass Index (BMI).
Participants’ height and weight, collected at the age 9 and 12 assessments, were used to calculate BMI at each time point (Keys et al., 1972).
Data Analysis
First, descriptive analyses were conducted to characterize the sample, check for assumptions of normality, and assess bivariate associations between variables of interest. Descriptive statistics were analyzed in SPSS. Next, SAS 9.4 was used to estimate multilevel growth curve models (MLM) to account for longitudinal nesting patterns of all hormones. Considering all 6 repeated measures at wave 1 and 2, the ICC for cortisol was 0.61. Change in cortisol over the 6 repeated measures across the two measurement waves (Mage = 9.27 and 12.68, respectively) served as the outcome at level 1, while DHEA and testosterone served as repeated measure predictors at level 1 within-day. Additionally, age and time since waking were included as additional within-day predictors at level 1. Puberty and BMI were modeled as fixed effects within-waves but nested effects across waves, while sex was a fixed level 2 predictor.
Given that we were interested in developmental processes, hormone change over time was modeled on age (historical time, calculated as age in days) and puberty (biological time, measured by PDS at each wave). With such a structure, the age construct can account for differences in the repeated measures within and across waves, while the puberty construct accounts for the yearly structure of the data. The zero point for age was set at 9-years old and for puberty it was set at the lowest puberty value (5) on the PDS scale. Finally, sex was coded zero for females and one for males. All other constructs were grand mean centered. Nested chi-square tests were calculated to determine whether adding random effects for cortisol at 9 years old (intercept) and change in cortisol (slope) improved model fit. We also tested whether adding random effects for puberty and a quadric effect for age, puberty, BMI, and time since waking improved the model; however, neither the random effects for puberty and BMI nor the quadratic effects improved model fit and as such they were not considered further.
Interaction terms were used to test for moderation of cortisol-DHEA/testosterone coupling. Two-way interactions between DHEA/testosterone and potential moderators allowed us to examine how the within-person relation between cortisol and DHEA/testosterone across multiple days and timepoints was influenced by individual factors (including sex, pubertal development, and age; e.g., the DHEA × sex interaction tells us how cortisol-DHEA coupling patterns differ in males versus females). Similarly, three-way interactions between DHEA/testosterone and multiple potential moderators (e.g., DHEA × sex × age interactions) revealed whether the relations between cortisol and HPG hormones were conditional upon multiple individual factors (e.g., cortisol-DHEA coupling patterns in older males versus younger females, etc.). Log-likelihood ratio tests and Wald tests were used to ascertain the relative contribution of each interaction term to the final model. Briefly, log-likelihood ratio tests index goodness of fit of the model, while Wald tests speak to the individual contribution of an effect in the model. While they are used in tandem, Wald tests are not as reliable as the difference in log likelihoods across models when adding fixed effects. As such, if a log-likelihood difference test suggested a non-significant improvement in model fit when adding a predictor, it was assumed to be non-significant even if the Wald test indicated that the fixed effect was significant.
Results
Tables 2 and 3 present descriptive statistics by sex as well as overall associations between study variables. Briefly, females had higher levels of cortisol, DHEA, and testosterone than males at age 9, and higher cortisol and DHEA than males at age 12, while males demonstrated higher testosterone than females at age 12. Cortisol at age 9 was significantly positively associated with DHEA and testosterone at age 9, as well as cortisol at age 12. DHEA and testosterone were significantly positively associated within and across waves. Regarding pubertal development, girls were further along in pubertal development than boys at both the age 9 and 12 assessments and mean pubertal development scores increased for the sample between ages 9 and 12. This suggests that pubertal development had progressed typically for most participants in the period between the age 9 and 12 assessments.
Table 2.
Descriptive statistics for major study variables by sex.
Males (n=231) | Females (n=206) | Overall (N=437) | |
---|---|---|---|
Age 9 assessment | Mean (SD) | Mean (SD) | Mean (SD) |
Age (years) | 9.27 (.40) | 9.27 (.47) | 9.27 (.43) |
PDS | 6.60 (1.47) | 7.62 (1.84) | 7.08 (1.73) |
BMI | 18.05 (3.32) | 18.42 (3.75) | 18.23 (3.53) |
Cortisol (ug/dL) | |||
Day 1 | .34 (.18) | .36 (.21) | .35 (.19) |
Day 2 | .32 (.18) | .37 (.17) | .34 (.18) |
Day 3 | .34 (.20) | .37 (.21) | .36 (.20) |
DHEA (pg/mL) | |||
Day 1 | 98.57 (132.36) | 103.89 (138.60) | 101.12 (135.22) |
Day 2 | 82.16 (104.32) | 108.24 (129.02) | 94.57 (117.28) |
Day 3 | 81.62 (101.87) | 106.44 (149.42) | 93.08 (126.49) |
Testosterone (pg/mL) | |||
Day 1 | 26.65 (11.26) | 30.09 (13.20) | 28.29 (12.32) |
Day 2 | 26.42 (11.42) | 31.30 (13.71) | 28.72 (12.77) |
Day 3 | 25.65 (10.80) | 29.93 (13.71) | 27.63 (12.40) |
Age 12 assessment | |||
Age (years) | 12.71 (.42) | 12.65 (.45) | 12.68 (.44) |
PDS | 9.66 (3.06) | 12.67 (2.54) | 11.07 (3.20) |
BMI | 20.53 (4.64) | 21.21 (4.50) | 20.85 (4.58) |
Cortisol (ug/dL) | |||
Week 1 | .39 (.16) | .48 (.20) | .43 (.18) |
Week 2 | .41 (.19) | .47 (.22) | .44 (.20) |
Week 3 | .39 (.19) | .48 (.23) | .43 (.21) |
Week 4* | -- | .45 (.23) | -- |
DHEA (pg/mL) | |||
Week 1 | 146.44 (105.51) | 209.52 (136.63) | 176.01 (124.96) |
Week 2 | 143.18 (91.57) | 184.05 (112.58) | 162.15 (103.74) |
Week 3 | 132.09 (78.68) | 188.04 (119.52) | 158.29 (103.60) |
Week 4* | -- | 236.48 (189.77) | -- |
Testosterone (pg/mL) | |||
Week 1 | 87.53 (45.06) | 70.75 (27.96) | 79.72 (38.94) |
Week 2 | 89.10 (43.84) | 65.91 (24.73) | 78.43 (38.07) |
Week 3 | 89.94 (43.39) | 67.04 (24.83) | 79.33 (37.73) |
Week 4* | -- | 66.21 (24.58) | -- |
Note. PDS = Pubertal Development Scale; BMI = body mass index.
Week 4 saliva was only collected from females; therefore, only averages for the female portion of the sample are included.
Values above represent non-transformed hormone values.
Table 3.
Bivariate correlations between study variables.
Variable | 1 | 2 | 3 | 4 | 5 | 6 | 7 | 8 | 9 | 10 | 11 | 12 | 13 |
---|---|---|---|---|---|---|---|---|---|---|---|---|---|
1. Child sex | -- | ||||||||||||
2. Age 9 assessment age | −.01 | -- | |||||||||||
3. Age 9 PDS | -.30 | .01 | -- | ||||||||||
4. Age 9 BMI | −.05 | −.01 | .27 | -- | |||||||||
5. Age 9 cortisol | −.01 | −.05 | −.07 | −.07 | -- | ||||||||
6. Age 9 DHEA | −.10 | .09 | .20 | .19 | .24 | -- | |||||||
7. Age 9 testosterone | -.16 | .09 | .24 | .23 | .45 | .23 | -- | ||||||
8. Age 12 assessment age | .06 | .61 | −.04 | −.08 | −.13 | −.12 | −.07 | -- | |||||
9. Age 12 PDS | −.47 | .07 | .45 | .25 | −.03 | .19 | .26 | .04 | -- | ||||
10. Age 12 BMI | −.07 | −.07 | .25 | .86 | −.02 | .19 | .21 | −.01 | .28 | -- | |||
11. Age 12 cortisol | −.24 | −.02 | −.03 | −.07 | .14 | .02 | .03 | −.05 | .05 | −.10 | -- | ||
12. Age 12 DHEA | -.26 | .07 | .31 | .31 | .07 | .52 | .52 | .08 | .40 | .29 | .35 | -- | |
13. Age 12 testosterone | .27 | −.04 | .02 | .16 | .05 | .23 | .34 | .04 | .29 | .15 | .15 | .45 | -- |
Note. Hormone values represent the average value within a timepoint. Bolded values are statistically significant (p <.05).
To justify a multilevel investigation, an unconditional means model was first fit to the data to determine whether significant variability existed in cortisol levels across participants and time. An initial model including cortisol as the outcome with no other predictors was fit to the data, and nested chi-square tests suggested that adding a random intercept (x2 (1) = 230.12, p < .001) and slope (x2 (1) = 189.14, p < .001) significantly improved model fit. The model with random components (BIC = 3957) suggested that there was significant variability in baseline cortisol at age 9 (ν = 0.32 [SE = 0.03]) and significant variability in change in cortisol across the repeated measures (ν = 0.02 [SE = 0.003]). Furthermore, there was a significant positive covariance between variability in cortisol at baseline and change in cortisol over time (σ = 0.02, SE = 0.003, z = 7.98, p <.0001) indicating that youth with higher levels of cortisol at 9 have steeper increases in cortisol over time.
Next, simplified models were fit to ascertain the main effects of age, puberty, sex, DHEA, and testosterone on cortisol; significant main effects of DHEA and testosterone indicated significant coupling. (Table 4 presents the results of a simplified MLM including only age, puberty, and hormone values, while Table 5 presents the results when sex is added to the simplified model.) Both simplified models indicated significant cortisol-DHEA and cortisol-testosterone coupling. Furthermore, age, puberty, and sex were negatively associated with cortisol over time.
Table 4.
Beta Coefficients and 95% confidence intervals from the baseline model.
b | Lower bound | Upper bound | |
---|---|---|---|
Intercept | −0.84 | −0.92 | −0.76 |
Age | -0.07 | −0.10 | −0.04 |
Puberty | -0.02 | −0.04 | −0.01 |
Testosterone | 0.46 | 0.39 | 0.54 |
DHEA | 0.14 | 0.11 | 0.18 |
Note. Cortisol levels collected 30-minutes after waking was the outcome of interest in this model. DHEA = dehydroepiandrosterone. Bolded values are statistically significant (p <.05).
Table 5.
Beta Coefficients and 95% confidence intervals from the baseline model including sex.
b | Lower bound | Upper bound | |
---|---|---|---|
Intercept | −0.63 | −0.73 | −0.53 |
Age | -0.05 | −0.08 | −0.03 |
Puberty | -0.04 | −0.06 | −0.02 |
Testosterone | 0.51 | 0.43 | 0.58 |
DHEA | 0.13 | 0.09 | 0.17 |
Sex | -0.30 | −0.38 | −0.22 |
Note. Cortisol levels collected 30-minutes after waking was the outcome of interest in this model. DHEA = dehydroepiandrosterone. Bolded values are statistically significant (p <.05).
Next, additional predictors and interaction terms were added to the final model individually to calculate R2 change for each predictor at each level of the model (Table 6; chi-square difference statistics above 3.8 are statistically significant). Table 7 shows beta coefficients and 95% confidence intervals (CIs) from the final model. The CIs suggest that most of the predictors and interactions (with the exception of the age × sex × testosterone, the age × sex × DHEA, and the puberty × sex interactions) had a reliable effect on cortisol. Pubertal development, BMI, testosterone, and DHEA all had reliable effects above 0, while age, sex, and time since waking did not. Puberty and BMI were negatively related to cortisol, suggesting that youth with higher BMIs and those who were more pubertally advanced had lower levels of cortisol.
Table 6.
Model Fit, Chi-Square Difference, and change in R2 as predictors were added to the final model.
−2 × log | BIC | Δx2 | ΔR2 L1 | ΔR2 L2 intercept | ΔR2 L2 slope | |
---|---|---|---|---|---|---|
Sex | 3727.2 | 3770.2 | 15.4 | 0.00 | 0.00 | 0.00 |
Time since waking | 3679.6 | 3728.6 | 47.6 | 0.00 | 0.00 | 0.00 |
Puberty | 3614.0 | 3669.1 | 65.6 | 0.00 | 0.01 | 0.05 |
BMI | 3578.1 | 3639.3 | 35.9 | 0.00 | 0.01 | 0.02 |
Testosterone | 3110.8 | 3177.8 | 467.3 | 0.11 | 0.14 | 0.02 |
DHEA | 2934.6 | 3007.5 | 176.2 | 0.02 | 0.00 | 0.14 |
Testosterone*age | 2924.9 | 3004.0 | 9.7 | 0.00 | 0.00 | 0.00 |
DHEA*age | 2856.7 | 2941.9 | 68.2 | 0.01 | 0.08 | 0.13 |
Testosterone*puberty | 2850.0 | 2941.2 | 6.7 | 0.00 | 0.01 | 0.00 |
DHEA*puberty | 2832.8 | 2930.0 | 17.2 | 0.00 | 0.03 | 0.01 |
Testosterone*sex | 2826.0 | 2929.3 | 6.8 | 0.00 | 0.02 | 0.04 |
DHEA*sex | 2817.2 | 2926.6 | 8.8 | 0.00 | 0.01 | 0.01 |
Age*sex | 2805.6 | 2921.2 | 11.6 | 0.00 | 0.01 | 0.00 |
Puberty*sex | 2802.5 | 2924.1 | 3.1 | 0.00 | 0.00 | 0.00 |
Age*sex*testosterone | 2799.8 | 2927.4 | 2.7 | 0.00 | 0.00 | 0.00 |
Age*sex*DHEA | 2798.6 | 2932.3 | 1.2 | 0.00 | 0.00 | 0.00 |
Puberty*sex*testosterone | 2788.1 | 2927.9 | 10.5 | 0.00 | 0.02 | 0.02 |
Puberty*sex*DHEA | 2782.7 | 2928.7 | 5.4 | 0.00 | 0.01 | 0.00 |
R2 | 0.63 | 0.20 | 0.38 |
Note. Cortisol levels collected 30-minutes after waking was the outcome of interest in this model. BIC = Bayesian information criteria; BMI = body mass index; DHEA = dehydroepiandrosterone. Bolded values are statistically significant (p <.05).
Table 7.
Beta Coefficients and 95% confidence intervals from the final model.
b | Lower bound | Upper bound | |
---|---|---|---|
Intercept | −0.75 | −0.88 | −0.62 |
Age | −0.03 | −0.07 | 0.02 |
Sex | 0.06 | −0.12 | 0.23 |
Time since waking | 0.00 | 0.00 | 0.00 |
Puberty | -0.04 | −0.06 | −0.02 |
BMI | -0.02 | −0.03 | −0.01 |
Testosterone | 0.89 | 0.67 | 1.11 |
DHEA | -0.14 | −0.23 | −0.04 |
Testosterone*age | −0.18 | −0.28 | −0.08 |
DHEA*age | 0.00 | −0.05 | 0.05 |
Testosterone*puberty | 0.02 | −0.03 | 0.07 |
DHEA*puberty | 0.05 | 0.03 | 0.08 |
Testosterone*sex | −0.06 | −0.33 | 0.21 |
DHEA*sex | 0.15 | 0.03 | 0.26 |
Age*sex | −0.08 | −0.14 | −0.02 |
Puberty*sex | −0.01 | −0.05 | 0.02 |
Age*sex*testosterone | 0.08 | −0.04 | 0.19 |
Age*sex*DHEA | 0.07 | 0.01 | 0.13 |
Puberty*sex*testosterone | −0.03 | −0.09 | 0.03 |
Puberty*sex*DHEA | −0.04 | −0.08 | −0.01 |
Note. Cortisol levels collected 30-minutes after waking was the outcome of interest in this model. BMI = body mass index; DHEA = dehydroepiandrosterone. Bolded values are statistically significant (p <.05).
Furthermore, the addition of covariates, predictors, and interactions to the final model influenced the pattern of hormone coupling seen in the baseline models. Specifically, while testosterone continued to have a positive association with cortisol, the main effect of DHEA (and therefore, cortisol-DHEA coupling) was negative in the final model. Coupling effects (i.e., main effects of DHEA/testosterone on cortisol), however, were qualified by several two- and three-way interactions with age, puberty, and sex. The age × testosterone, DHEA × puberty, DHEA × sex, age × sex × DHEA, and puberty × sex × DHEA interactions all had reliable effects on cortisol3. To assess how patterns of coupling changed as a function of age, puberty, or sex, simple slopes were calculated at the mean of age at wave 2 (12.6), the mean of pubertal development at wave 1 (M = 7.13) and wave 2 (M = 11.20), and one standard deviation above the mean on pubertal development at wave 2 (1 SD above the M = 14.42). Results can be found in Table 8 and suggest that cortisol-testosterone coupling at baseline became more positive as youth aged from 9 to 12. For females, the main effect of DHEA on cortisol was negative at age 9, and became non-significant at age 12, while male’s DHEA coupling became more negative as youth aged from 9 to 12. For females, DHEA coupling strengthened across pubertal development, such that the coefficient for DHEA was significant and negative at the lowest point on pubertal development but switched to significant and positive at the mean of pubertal development at wave 2 and at higher values of pubertal development. In contrast, males started off with a positive main effect of DHEA on cortisol at the lowest point in pubertal development which weakened at the highest point in pubertal development.
Table 8.
Tests of simple slopes for significant interactions predicting cortisol as outcome.
Moderator 1 | Moderator 2 | Hormone | Cortisol |
---|---|---|---|
Age 9 | - | Testosterone | 0.85 |
Age 12 | - | Testosterone | 1.32 |
Age 9 | Female | DHEA | -0.13 |
Age 9 | Male | DHEA | 0.15 |
Age 12 | Female | DHEA | −0.07 |
Age 12 | Male | DHEA | -0.47 |
Low puberty | Female | DHEA | -0.13 |
Low puberty | Male | DHEA | 0.15 |
Mean Puberty W1 | Female | DHEA | −0.00 |
Mean Puberty W1 | Male | DHEA | 0.06 |
Mean Puberty W2 | Female | DHEA | 0.24 |
Mean Puberty W2 | Male | DHEA | −0.14 |
High Puberty | Female | DHEA | 0.53 |
High Puberty | Male | DHEA | −0.36 |
Note. DHEA = dehydroepiandrosterone. Bolded values are statistically significant (*p<.05); italicized values are approaching significance (p<.10).
Discussion
Despite the increased interest in hormone coupling, longitudinal work on these processes have been lacking. Therefore, the current investigation aimed to add to the limited literature on the development of hormone coupling patterns from middle childhood to early adolescence.
Cortisol-DHEA coupling.
Based on previous literature, we hypothesized that cortisol-DHEA coupling would strengthen in magnitude between ages 9 and 12. We did not, however, find support for this hypothesis in the full sample. Cortisol-DHEA coupling, in fact, had weakened by age 12. However, this pattern varied by sex and pubertal development. Specifically, male cortisol-DHEA coupling became more negative over time, a pattern that was strongest among the more pubertally-advanced individuals. Female cortisol-DHEA coupling, in contrast, became more positive over time among more pubertally-advanced individuals.
To understand this pattern of results, it is important to consider that DHEA has roles in the functioning of both the HPA- and HPG-axes. As individuals mature, it is possible that there is a “peak” period during which DHEA is primarily functioning as an HPG hormone and playing a significant role in initiating pubertal development. Outside of this peak period, DHEA is likely to act more like an HPA-axis hormone, which would be expected to have positive associations with cortisol. Naturally, this peak period would vary for boys and girls, as pubertal development in girls typically precedes boys’ development by 1–2 years (Auchus & Rainey, 2004). Therefore, consistent with our results showing that cortisol-DHEA coupling is moderated by sex, boys would be expected to demonstrate weaker cortisol-DHEA coupling patterns further into adolescence than girls, while girls would begin demonstrating stronger associations between hormones at earlier ages.
Cortisol-testosterone coupling.
We expected that cortisol-testosterone coupling would become weaker between ages 9 and 12. We did not, however, find support for this hypothesis in the full sample. Rather than weakening, cortisol-testosterone coupling strengthened by age 12. In contrast to cortisol-DHEA coupling, cortisol-testosterone coupling patterns did not vary by sex in the current sample.
Although contrary to our hypotheses, these results are partially supported by the previous longitudinal coupling work. Specifically, Ruttle et al. (2015) found that cortisol-testosterone coupling was positive for the youngest adolescents in their sample (age 11), the age which is most similar to the participants in our study (9–12 years old). While it is curious that we observed the cortisol-testosterone coupling becoming stronger over time (in contrast to Ruttle et al.’s pattern of weaker coupling over time), the current study began investigating children earlier in pubertal development. It is possible, therefore, that the strengthened cortisol-testosterone coupling detected in the current sample is due to the steep increase in testosterone experienced by children between the ages of 9 and 12, which is concurrent with age-related increases in basal cortisol. Future studies would benefit by extending the duration of investigations to cover a broader window of pubertal development.
Strengths and limitations.
The current study had several strengths, notably the use of a large, community-based sample with multiple hormone collections over a particularly active period in pubertal development (between ages 9 and 12). Youth in the current sample, however, were primarily White, and from affluent families in the Northeastern United States. The results reported herein, therefore, may not be generalizable to samples that are more heterogeneous in terms of race, ethnicity, or socioeconomic status. Relatedly, other potentially significant moderators of hormone coupling (such as early life stress; Ruttle et al., 2015) were not investigated in the current analyses. Finally, saliva samples were collected on different schedules, and assayed at different labs, for the age 9 and 12 assessments. Assaying hormones at multiple labs introduces some error into the hormone measurements, although all labs reported acceptable inter- and intra-assay coefficients of variation (CVs). Furthermore, as the analyses were concerned primarily with within-subject, rather than between-subject comparisons, and all samples for a given individual were assayed at the same lab within timepoints, the use of different labs likely did not significantly influence the results.
Conclusion
Taken together, the current findings add to the burgeoning literature on longitudinal relations between hormones and contribute to our understanding of these relationships across earlier developmental periods than previous work has included. Following the current sample’s hormone and pubertal development into later stages of adolescence may clarify why some of the current study’s findings run counter to prior investigations. Relatedly, future work with a more heterogeneous sample (including more racial, ethnic, and economic diversity) would elucidate whether the current results may be generalizable to more diverse groups.
Acknowledgement:
This work was supported by National Institute of Mental Health Grant RO1 MH069942 to DNK.
Footnotes
Conflict of interest statement: The authors do not have any conflicts of interest to report.
Estradiol samples were not examined in the current study, as longitudinal investigation of cortisol-estradiol coupling was not possible due to lack of estradiol sampling at age 9.
Age 12 lab was included as a covariate in analyses but did not alter the pattern of results described below, so was excluded from the final model for parsimony.
Importantly, the age × sex × DHEA interaction did not have a significant likelihood ratio test but did have a significant Wald test for the fixed effect, while the puberty × sex × testosterone interaction had a significant likelihood ratio test and had a significant Wald test before the puberty × sex × DHEA interaction was added, but did not have a significant Wald test for the fixed effect in the final model. These patterns suggest that the age × sex × DHEA interaction is reliable but does not explain much variance (.3% at level 1, .2% of the intercept, and .1% of the cortisol slope), while the puberty × sex × testosterone and puberty × sex × DHEA interactions likely explain similar variance in cortisol given that puberty × sex × DHEA outperformed puberty × sex × testosterone in the final model.
Data availability statement:
The data from the current study may be made available upon request.
REFERENCES
- Auchus RJ, & Rainey WE (2004). Adrenarche–physiology, biochemistry and human disease. Clinical Endocrinology, 60(3), 288–296. 10.1046/j.1365-2265.2003.01858.x. [DOI] [PubMed] [Google Scholar]
- Barendse ME, Byrne ML, Flournoy JC, McNeilly EA, Guazzelli Williamson V, Barrett AMY, Chavez SJ, Shirtcliff EA, Allen NB, & Pfeifer JH (2021). Multimethod assessment of pubertal timing and associations with internalizing psychopathology in early adolescent girls. Journal of Abnormal Psychology. 10.1037/abn0000721 [DOI] [PMC free article] [PubMed] [Google Scholar]
- Black SR, Lerner MD, Shirtcliff EA, & Klein DN (2018). Patterns of neuroendocrine coupling in 9-year-old children: Effects of sex, body-mass index, and life stress. Biological Psychology, 132, 252–259. 10.1016/j.biopsycho.2017.11.004. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Bobadilla L, Asberg K, Johnson M, & Shirtcliff EA (2015). Experiences in the military may impact dual-axis neuroendocrine processes in veterans. Developmental Psychobiology, 57(6), 719–730. 10.1002/dev.21259. [DOI] [PubMed] [Google Scholar]
- Choe JK, Khan-Dawood FS, & Yusoff-Dawood M (1983). Progesterone and estradiol in the saliva and plasma during the menstrual cycle. American Journal of Obstetrics and Gynecology, 147(5), 557–562. 10.1016/0002-9378(83)90016-9. [DOI] [PubMed] [Google Scholar]
- Colich NL, Kircanski K, Foland-Ross LC, & Gotlib IH (2015). HPA-axis reactivity interacts with stage of pubertal development to predict the onset of depression. Psychoneuroendocrinology, 55, 94–101. 10.1016/j.psyneuen.2015.02.004. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Colich NL, Platt JM, Keyes KM, Sumner JA, Allen NB, & McLaughlin KA (2020). Earlier age at menarche as a transdiagnostic mechanism linking childhood trauma with multiple forms of psychopathology in adolescent girls. Psychological Medicine, 50(7), 1090–1098. 10.1017/S0033291719000953 [DOI] [PMC free article] [PubMed] [Google Scholar]
- Dickerson SS, & Kemeny ME (2004). Acute stressors and cortisol responses: A theoretical integration and synthesis of laboratory research. Psychological Bulletin, 130(3), 355–391. 10.1037/0033-2909.130.3.355. [DOI] [PubMed] [Google Scholar]
- Dismukes AR, Johnson MM, Vitacco MJ, Iturri F, & Shirtcliff EA (2015). Coupling of the HPA and HPG axes in the context of early life adversity in incarcerated male adolescents. Developmental Psychobiology, 57(6), 705–718. 10.1002/dev.21231. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Dismukes AR, Shirtcliff EA, Hanson JL, & Pollak SD (2015). Context influences the interplay of endocrine axes across the day. Developmental Psychobiology, 57(6), 731–741. 10.1002/dev.21331. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Dorn LD, Dahl RE, Woodward HR, & Biro F (2006). Defining the boundaries of early adolescence: A user’s guide to assessing pubertal status and pubertal timing in research with adolescents. Applied Developmental Science, 10(1), 30–56. 10.1207/s1532480xads1001_3. [DOI] [Google Scholar]
- Foilb AR, Lui P, & Romeo RD (2011). The transformation of hormonal stress responses throughout puberty and adolescence. Journal of Endocrinology, 210(3), 391–398. 10.1530/joe-11-0206. [DOI] [PubMed] [Google Scholar]
- Gomez F, Manalo S, & Dallman MF (2004). Androgen-sensitive changes in regulation of restraint-induced adrenocorticotropin secretion between early and late puberty in male rats. Endocrinology, 145(1), 59–70. 10.1210/en.2003-0565. [DOI] [PubMed] [Google Scholar]
- Granger DA, Fortunato CK, Beltzer EK, Virag M, Bright MA, & Out D (2012). Focus on methodology: Salivary bioscience and research on adolescence: An integrated perspective. Journal of Adolescence, 35(4), 1081–1095. 10.1016/j.adolescence.2012.01.005. [DOI] [PubMed] [Google Scholar]
- Grumbach MM, & Styne DM (2003). Puberty: Ontogeny, neuroendocrinology, physiology, and disorders. In Larsen PR, Kronenberg HM, Melmed S, & Polonsky KS, (Eds.), Williams textbook of endocrinology, 10th edition (pp. 1115–1286). New York: Elsevier. [Google Scholar]
- Hamlat EJ, Snyder HR, Young JF, & Hankin BL (2019). Pubertal timing as a transdiagnostic risk for psychopathology in youth. Clinical Psychological Science, 7(3), 411–429. 10.1177/2167702618810518. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Hankin BL, Badanes LS, Abela JR, & Watamura SE (2010). Hypothalamic–pituitary–adrenal axis dysregulation in dysphoric children and adolescents: Cortisol reactivity to psychosocial stress from preschool through middle adolescence. Biological Psychiatry, 68(5), 484–490. 10.1016/j.biopsych.2010.04.004. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Harden KP, Wrzus C, Luong G, Grotzinger A, Bajbouj M, Rauers A, Wagner GG, & Riediger M (2016). Diurnal coupling between testosterone and cortisol from adolescence to older adulthood. Psychoneuroendocrinology, 73, 79–90. 10.1016/j.psyneuen.2016.07.216. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Havelock JC, Auchus RJ, & Rainey WE (2004). The rise in adrenal androgen biosynthesis: Adrenarche. Seminars in Reproductive Medicine, 22(4), 337–347. 10.1055/s-2004-861550. [DOI] [PubMed] [Google Scholar]
- Hucklebridge F, Hussain T, Evans P, & Clow A (2005). The diurnal patterns of the adrenal steroids cortisol and dehydroepiandrosterone (DHEA) in relation to awakening. Psychoneuroendocrinology, 30(1), 51–57. 10.1016/j.psyneuen.2004.04.007. [DOI] [PubMed] [Google Scholar]
- Johnson MM, Dismukes AR, Vitacco MJ, Breiman C, Fleury D, & Shirtcliff EA (2014). Psychopathy’s influence on the coupling between hypothalamic–pituitary–adrenal and-gonadal axes among incarcerated adolescents. Developmental Psychobiology, 56(3), 448–458. 10.1002/dev.21111. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Joos CM, Wodzinski AM, Wadsworth ME, & Dorn LD (2018). Neither antecedent nor consequence: Developmental integration of chronic stress, pubertal timing, and conditionally adapted stress response. Developmental Review, 48, 1–23. 10.1016/j.dr.2018.05.001. [DOI] [Google Scholar]
- Kamin HS, & Kertes DA (2017). Cortisol and DHEA in development and psychopathology. Hormones and Behavior, 89, 69–85. 10.1016/j.yhbeh.2016.11.018. [DOI] [PubMed] [Google Scholar]
- Keys A, Fidanza F, Karvonen MJ, Kimura N, & Taylor HL (1972). Indices of relative weight and obesity. Journal of Chronic Diseases, 25(6–7), 329–343. 10.1016/0021-9681(72)90027-6. [DOI] [PubMed] [Google Scholar]
- Klein DN, & Finsaas MC (2017). The Stony Brook Temperament Study: Early antecedents and pathways to emotional disorders. Child Development Perspectives, 11(4), 257–263. 10.1111/cdep.12242. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Marceau K, Ruttle PL, Shirtcliff EA, Hastings PD, Klimes-Dougan B, & Zahn-Waxler C (2013). Within-person coupling of changes in cortisol, testosterone, and DHEA across the day in adolescents. Developmental Psychobiology, 57(6), 654–669. 10.1002/dev.21173 [DOI] [PMC free article] [PubMed] [Google Scholar]
- Marceau K, Shirtcliff EA, Hastings PD, Klimes-Dougan B, Zahn-Waxler C, Dorn LD, & Susman EJ (2014). Within-adolescent coupled changes in cortisol with DHEA and testosterone in response to three stressors during adolescence. Psychoneuroendocrinology, 41, 33–45. 10.1016/j.psyneuen.2013.12.002. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Petersen AC, Crockett L, Richards M, & Boxer A (1988). A self-report measure of pubertal status: Reliability, validity, and initial norms. Journal of Youth and Adolescence, 17(2), 117–133. 10.1007/BF01537962. [DOI] [PubMed] [Google Scholar]
- Pfeifer JH, & Allen NB (2021). Puberty initiates cascading relationships between neurodevelopmental, social, and internalizing processes across adolescence. Biological Psychiatry, 89(2), 99–108. 10.1016/j.biopsych.2020.09.002. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Roberts AG, & Lopez-Duran NL (2019). Developmental influences on stress response systems: Implications for psychopathology vulnerability in adolescence. Comprehensive Psychiatry, 88, 9–21. 10.1016/j.comppsych.2018.10.008. [DOI] [PubMed] [Google Scholar]
- Ruttle PL, Shirtcliff EA, Armstrong JM, Klein MH, & Essex MJ (2015). Neuroendocrine coupling across adolescence and the longitudinal influence of early life stress. Developmental Psychobiology, 57(6), 688–704. 10.1002/dev.21138. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Shirtcliff EA, & Ruttle PL (2010). Immunologische und neuroendokrine dysregulation in der folge früher deprivations und stresserfahrungen [Immunological and neuroendocrine dysregulation following early deprivation and stress] In: Heinz Brisch K, editor. Bindung und frühe Störungen der Entwicklung. Stuttgart, Germany: Klett-Cotta; 2010. pp. 167–202. [Google Scholar]
- Simmons JG, Byrne ML, Schwartz OS, Whittle SL, Sheeber L, Kaess M, Youssef GJ, & Allen NB (2015). Hormonal coupling in adolescence and the moderating influence of prior trauma and aversive maternal parenting. Developmental Psychobiology, 57(6), 670–687. 10.1002/dev.21275. [DOI] [PubMed] [Google Scholar]
- Viau V (2002). Functional cross-talk between the hypothalamic-pituitary-gonadal and-adrenal axes. Journal of Neuroendocrinology, 14(6), 506–513. 10.1046/j.1365-2826.2002.00798.x. [DOI] [PubMed] [Google Scholar]
- Wood CL, Lane LC, & Cheetham T (2019). Puberty: Normal physiology (brief overview). Best Practice & Research Clinical Endocrinology & Metabolism, 33(3), 101265. 10.1016/j.beem.2019.03.001. [DOI] [PubMed] [Google Scholar]
- Yim IS, Quas JA, Rush EB, Granger DA, & Skoluda N (2015). Experimental manipulation of the Trier Social Stress Test-Modified (TSST-M) to vary arousal across development. Psychoneuroendocrinology, 57, 61–71. 10.1016/j.psyneuen.2015.03.021. [DOI] [PubMed] [Google Scholar]
- Zakreski E, Dismukes AR, Tountas A, Phan JM, Moody SN, & Shirtcliff EA (2018). Developmental trajectories of HPA–HPG dual-axes coupling: Implications for social neuroendocrinology. In Routledge International Handbook of Social Neuroendocrinology (pp. 608–632). London: Routledge. [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Data Availability Statement
The data from the current study may be made available upon request.