A longitudinal analysis of hypothalamic–pituitary–adrenal axis and metabolic energy interactions pre- and post-menarche within and among Mayan Indigenous adolescent girls

Abstract

Adolescent girls face significant energetic trade-offs resulting from the concurrent demands of somatic growth and reproductive maturation. The hypothalamic pituitary-adrenal axis (HPAA) is a mediator of energy allocation. However, its longitudinal role during this critical transition is unclear. Hypothesis: HPAA activity should be positively linked with energy mobilization and inversely linked with energy storage, both pre- and post-menarche. We quantified first morning urine biomarkers of HPAA activity (cortisol), potential energy uptake (c-peptide), and energy storage (adiponectin) collected pre-to post-menarche from 46 Mayan girls. We used mixed effects models to characterize HPAA activity and metabolic energy status across menarche stages. Adiponectin and c-peptide were higher pre- versus post-menarche (p < 0.001) after adjusting for cortisol and time relative to menarche. Cortisol was positively associated with both c-peptide and adiponectin after adjusting for time relative to menarche (p < 0.001). Consistent with our hypothesis, HPAA activity was positively linked to energy mobilization across the adolescent transition. Energy uptake was higher pre-menarche, a possible response to energetic demands imposed by somatic and reproductive maturation. Energy storage was higher post-menarche, which may help meet future reproductive energy needs. Understanding changes in energy metabolism during the adolescent transition is paramount to optimizing women’s development and health across their life span.

Introduction

Adolescence is a critical developmental transition during which young human females (hereafter referred to as ‘girls’) face key energetic trade-offs between somatic growth and reproductive maturation. Girls typically begin reproductive maturation between the ages of 8 and 13 years^1,2. Menarche, defined as the first occurrence of menstrual bleeding, reflects the onset of ovarian cyclicity. This first menstrual bleed typically occurs once other growth and maturational processes have commenced (e.g., pelvic widening, development of secondary sexual traits) or milestones (e.g., peak height velocity) have been achieved^3–5. Initially, ovulatory frequencies tend to be low as adolescent girls’ hypothalamic-pituitary–gonadal axes (HPGAs) are not yet fully mature^3,6.

As both somatic growth and reproductive maturation are energetically demanding, energy metabolism is hypothesized to play a key role in the adolescent transition. Mechanistically, metabolic energy availability, i.e., the energy readily available for metabolic processes, increases following an insulin-resistant period that characterizes early adolescence. Insulin resistance appears to trigger a compensatory increase in insulin secretion from pancreatic beta cells⁷. Insulin facilitates glucose uptake into cells, either for immediate use or long-term storage in adipose tissue^8,9. These changes in insulin secretion patterns are paired with increases in adiposity levels that occur as girls mature¹⁰. These processes reflect critical changes in metabolic energy allocation, which have been proposed to be required for HPGA maturation^7,10–12.

Changes in energy metabolism are, in part, regulated by the hypothalamic–pituitary–adrenal axis (HPAA). For example, when individuals face urgent challenges, energy is diverted from lower priority tasks, such as reproductive function, to meet said challenges^4,13. Thus, the HPAA has been postulated to be involved in energetic trade-offs between physiological processes that support growth and reproductive transitions¹⁴. The role of the HPAA in regulating reproductive function in adult women has been previously explored^13,15,16. Yet, less is known about the specific mechanisms regulating the interactions between the HPAA, growth, and reproductive maturation in adolescent girls. Indeed, while studies have evaluated changes in HPAA and energy availability with maturation age¹⁷, to our knowledge no previous studies have longitudinally examined effects of the HPAA on metabolic energy, to evaluate the HPAA’s role in regulating metabolic energy across girls’ reproductive maturation transition. To evaluate the hypothesis that the HPAA is involved in the regulation of metabolic energy during the transition from pre- to post-menarche, we test three predictions. After controlling for time relative to menarche to adjust for variation in maturation status, we expect to observe within girl differences in energy uptake (evaluated via c-peptide, a marker of insulin and rough proxy for potential uptake of glucose into cells) and energy storage levels (evaluated via adiponectin, an inverse marker of visceral adipose fat mass) pre-menarche and post-menarche. We also expect variation in HPAA activity (evaluated via cortisol) to be associated with variation in metabolic energy markers (c-peptide and adiponectin) both pre- and post-menarche (see variable definitions in Methods—Table 1).

Table 1.

Description of variables.

Variable name	Variable definition
Time relative to menarche	Number of months before and after a participant reached menarche relative to when she was in her pre-menarche and post-menarche stage, calculated as the difference between age at menarche and age at which the participant provided her pre-menarche and post-menarche samples/data
Cortisol (ng/mL) pre-menarche	Cortisol is a marker of HPAA activity; data collected every other day for 3 weeks during pre-menarche
C-Peptide (ng/mL) pre-menarche	C-peptide is a proxy of insulin—a marker of potential energy uptake activity; data collected every other day for 3 weeks during pre-menarche
Adiponectin (ng/mL) pre-menarche	Adiponectin is a proxy for visceral adipose fat—a marker of energy storage; data collected every other day for 3 weeks during pre-menarche
Cortisol (ng/mL) post-menarche	Data collected every other day for 4 months post-menarche
C-Peptide (ng/mL) post-menarche	Data collected every other day for 4 months post-menarche
Adiponectin (ng/mL) post-menarche	Data collected every other day for 4 months post-menarche

Results

Baseline characteristics

The average age at menarche for the 46 participants included in this analysis was 13.6 years of age (Table 2). The average age of the same participants during the pre-menarche data collection period was 10.9 years (range 8–16), and their average age during the post-menarche data collection period was 15.5 years (range 12–21 years). By their pre-menarche stage, girls had achieved on average 89.8% of their final height, and had an average waist to hip ratio of 0.897. By the time we conducted post-menarche data collection, girls had achieved 96.3% of their final height and had an average waist to hip ratio of 0.833. During the pre-menarche data collection period, their median cortisol was 31.7 ng/mL (25th percentile (Q1), 75th percentile (Q3): 17.8, 55.6), median c-peptide ng/mL was 770.0 (Q1, Q3: 34.54, 5077.06), and their median adiponectin was 6.63 ng/mL (Q1, Q3: 0.047, 100.46). During the post-menarche data collection period, median cortisol was 22.0 ng/mL (Q1, Q3: 13.2, 45.0), median c-peptide was 393.46 (Q1, Q3: 24.26, 3359.71), and median adiponectin was 0.344 (Q1, Q3: 0.012, 6.39). Time relative to menarche when girls were pre- and post-menarche is described in Fig. 1.

Table 2.

Summary statistics for hormone and anthropometry measures pre- and post-menarche.

Variable	Pre-menarche	Post-menarche
Age at Menarche
Mean (SD)	n/a	13.6 (1.50)
Median [Min, Max]	n/a	13.5 [11.0, 17.0]
Age at data collection
Mean (SD)	10.9 (1.67)	15.5 (2.43)
Median [Min, Max]	11.0 [8.00, 16.0]	15.0 [12.0, 21.0]
Cortisol (ng/mL)
Mean (SD)	41.6 (36.0)	38.9 (47.7)
Median [Q1, Q3]	31.7 [17.8, 55.6]	22.0 [13.2, 45.0]
C-Peptide (ng/mL)
Mean (SD)	11,155.0 (47,233.0)	4571.0 (12,168.0)
Median [Q1, Q3]	770.0 [34.54, 5077.06]	393.46 [24.26, 3359.71]
Adiponectin (ng/mL)
Mean (SD)	138.0 (393.0)	19.7 (78.3)
Median [Q1, Q3]	6.63 [0.047, 100.46]	0.344 [0.012, 6.39]
Proportion of final height
Mean (SD)	0.898 (0.0489)	0.963 (0.0353)
Median [Min, Max]	0.880 [0.826, 1.00]	0.975 [0.848, 1.00]
Waist to Hip Ratio
Mean (SD)	0.897 (0.378)	0.833 (0.0379)
Median [Min, Max]	0.840 [0.400, 2.99]	0.840 [0.760, 0.930]

Non-transformed values for cortisol, c-peptide, and adiponectin.

Figure 1.

Participant time relative to when they reached menarche during pre- and post-menarche data collection periods. Red points reflect data collected pre-menarche; blue points reflect data collected post-menarche.

HPAA and energy uptake across the transition

To evaluate changes in c-peptide levels from pre- to post-menarche, we compared the fit of two models—one that allows for changes in c-peptide variance across the transition and another that assumes the variance to be constant. By comparing these two models, we were able to determine not only if there is a change in average c-peptide levels but also if the variance of these levels changes pre- to post-menarche. We observed no significant differences in the fit of the two models (likelihood ratio test, p > 0.05). Using the simpler model with variance assumed to be constant, we found no statistical evidence of an interaction effect between cortisol and menarche stage (p > 0.05). Thus, we removed the interaction and re-fitted the following final model, where:

We assume that, conditional on , (which assumes constant variance of c-peptide). Then, let be the log(c-peptide) of the girl at time point . Let be a girl-specific random effect, and let the ’s be independent. Let be the log(cortisol) of the girl at time point . Let be 1 if (i.e., the girl is post-menarche) and be 0 otherwise (for pre-menarche). Let be the time relative to menarche of the girl at time point j. Finally, we assume that the Yij’s are independent conditional on the Ui’s.

Holding cortisol and time relative to menarche constant, c-peptide, on average, was significantly higher by 0.54 log-transformed units pre- compared to post-menarche (p < 0.001) (Table 3, Fig. 2). Holding time relative to menarche constant, there was a moderate, positive association between c-peptide and cortisol (p < 0.001). Namely, for every unit increase in cortisol, c-peptide increased by 0.21 on the log-scale (Table 3, Fig. 2).

Table 3.

Estimated effects of cortisol and menarche stage on c-peptide ¹.

	Effect estimate	95% CI
(Intercept)	2.30***	1.92, 2.67
Cortisol (ng/mL)	0.21***	0.15, 0.26
Post-menarche	− 0.54 ***	− 0.84, − 0.23
Time relative to menarche	− 0.00	− 0.01, 0.00
SD (Intercept)	0.82

***p < 0.001; **p < 0.01; *p < 0.05.

Figure 2.

The fitted regression lines representing the estimated relationship between cortisol and c-peptide, adjusting for mean time relative to menarche. The red points represent values corresponding to the pre-menarche stage, and the blue points represent values corresponding to the post-menarche stage.

HPAA and energy storage across the transition

We observed a statistically significant difference in the fit of the two models used to evaluate changes in adiponectin levels pre- to post-menarche—one model allowing for changes in adiponectin variance pre- to post-menarche and the other assuming variance to be constant. Because the model allowing for change in variance was significantly different from the model assuming constant variance (likelihood ratio test p < 0.0001), we assume that variance does change and as such report results from this model that allows for change in the variance of adiponectin pre- to post-menarche. As with c-peptide, there was no statistical evidence that the relationship between cortisol and adiponectin varied by menarche stage (p > 0.05). Thus, we removed the interaction term for cortisol and menarche stage and re-fit the following final model, where:

We assume that, conditional on , (which allows a change in variance of adiponectin across the transition). Then, let be the log(adiponectin) of the girl at time point . Let be a girl-specific random effect, and let the ’s be independent. Let be log(cortisol) of the girl at time point . Let be 1 if (i.e., the girl is post-menarche) and be 0 otherwise (for pre-menarche). Let be the time relative to menarche of the girl at time point j. Finally, we assume that the ’s are independent conditional on the ’s.

With all other variables held constant, the re-fitted model shows that adiponectin was, on average, statistically significantly higher by 4.94 log-transformed units in pre- compared to post-menarche (p < 0.001) (Table 4, Fig. 3). Holding time relative to menarche constant, cortisol was moderately and positively associated with adiponectin (p < 0.001)—for every unit increase in cortisol, we observed an increase in adiponectin by 0.29 on the log-scale (Table 4, Fig. 3).

Table 4.

Estimated effects of cortisol and menarche stage on adiponectin¹.

	Effect estimate	95% CI
(Intercept)	2.15***	1.26, 3.04
Cortisol (ng/mL)	0.29***	0.21, 0.37
Post-menarche	− 4.94***	− 5.50, 4.39
Time relative to menarche	0.06***	0.05, 0.07
SD (Intercept)	2.57

***p < 0.001; **p < 0.01; *p < 0.05.

Figure 3.

The fitted regression lines representing the estimated relationship between cortisol and adiponectin, adjusting for mean time relative to menarche. The red points represent values corresponding to the pre-menarche stage, and the blue points represent values corresponding to the post-menarche stage.

Discussion

Our within-girl longitudinal analyses of HPAA activity and metabolic energy across the adolescent transition suggests that adolescent girls take up and spend more energy pre-menarche and store more energy post-menarche. In this cohort, girls’ c-peptide levels, a marker of insulin secretion and thus a proxy for potential glucose uptake, were higher pre- than post-menarche, consistent with the proposition that potential energy uptake is higher during earlier stages of the transition. Likewise, adiponectin levels were higher pre- than post-menarche. As adiponectin is inversely proportional to visceral adipose fat, these results suggest that energy storage increases after the onset of ovarian cyclicity. Consistent with our prediction that the HPAA seems to be playing a role in regulating energy allocation across the adolescent transition in girls, HPAA activity was positively associated with metabolic energy potentially available for uptake and energy storage both pre- and post-menarche.

HPAA activation triggers the release of glucocorticoids, including cortisol, which is involved in energy allocation in response to energetic, immune, or psychosocial challenges^4,18. Thus, we hypothesized that HPAA activity would be associated with insulin secretion, implicated in energy uptake, and adiponectin, which is inversely related to the levels of energy stored as adipose visceral fat. Consistent with this hypothesis, we found cortisol to be positively associated with both c-peptide and adiponectin. These observations are in line with the proposition that HPAA activity modulates energy that could potentially be taken up by cells, and energy storage, across this reproductive transition (Table 2, Fig. 1; Table 3, Fig. 2). This first longitudinal evaluation of the relationships between cortisol and both c-peptide and adiponectin within girls pre- and post-menarche offers complementary information to previous studies. The positive association between cortisol and insulin that we observed, both during pre- and post-menarche stages, is consistent with Huybrechts and colleagues (2014) cross-sectional analysis of 426 post-menarche girls, in which cortisol was positively associated with insulin secretion¹⁹. These findings were, however, inconsistent with a study conducted by Glass and colleagues (2024) examining changes in cortisol and c-peptide across the pubertal transition among a cohort of Qom Indigenous adolescents from Argentina. The authors found c-peptide increased linearly with age, but no association was identified to suggest cortisol predicts c-peptide¹⁷. The discrepancies between findings may be due to nutritional differences between populations—adolescent girls in Guatemala often face nutritional and energetic challenges, while the group of Indigenous adolescents in Argentina have greater access to calories^13,17. Additionally, there may be differences in how maturation status was measured which could introduce a source for differences in how c-peptide changes with age. To account for those potential differences, we controlled for maturation status by adjusting for time relative to menarche.

The overall higher levels of c-peptide we observed pre compared to post-menarche suggest that the potential for energy uptake is higher before the onset of ovarian cyclicity. There is a characteristic insulin-resistant period before menarche²⁰, which has been proposed to trigger a compensatory increase in insulin secretion^5,7. This insulin resistant period precedes the increase in gonadotrophins—follicular-stimulating hormone (FSH) and luteinizing hormone (LH)—characteristic of early stages of reproductive maturation. Huybrechts and colleagues (2014) propose that this insulin resistance is linked to an increase in HPAA activity, which, in turn, triggers an increase in insulin secretion¹⁹. Our observations that HPAA activation is positively associated with insulin secretion, and that insulin secretion during this period is higher than at more advanced stages of maturation, are consistent with Huybrechts and colleagues’ (2014) findings. In sum, insulin resistance during early stages of reproductive maturation, together with increases in HPAA activity, and subsequent increases in insulin, may reflect a mechanism to meet the surge in energy demands imposed by somatic growth and reproductive maturation early during the adolescent transition. These associations may thus reflect one example through which HPAA activation acts as a mechanism to upregulate insulin, to increase energy availability. The compensatory increase in insulin secretion occurring in response to insulin resistance has been proposed to trigger increased energy uptake and stimulate HPGA maturation^5,7. Insulin enhances FSH- and LH-stimulated production of gonadal steroids. This process results in an increase in the production of estrogen and progesterone by granulosa and theca cells in the ovaries, which is part of HPGA maturation and the establishment of ovulatory cycles^21,22. In turn, increased estrogen stimulates breast budding, skeletal growth, and other secondary sexual traits, which develop simultaneously as the HPGA matures, prior to menarche^3,23. As girls advance through reproductive maturation, ovulatory cycles become more regular and frequent, which is associated with a return to previous insulin levels and regular insulin sensitivity⁷. Thus, this period of insulin compensation depends on timing of HPGA maturation.

Interestingly, the HPAA activity and insulin secretion patterns we observed during the adolescent transition appear to be parallel to those described in other reproductive life history transitions in women. During gestation, for example, increased cortisol secretion is associated with a reduction in insulin sensitivity, which, as in pre-menarche, leads to insulin resistance²⁴. During later stages of gestation, this insulin resistance triggers greater insulin secretion, resulting in an increase in glucose uptake²⁴. In the post-partum amenorrhea period that follows a birth, women continue to experience insulin resistance until the resumption of ovulatory cyclicity occurs²⁵. Initial increases in insulin help enhance FSH- and LH-stimulated estrogen and progesterone production by ovarian cells, markers of the re-establishment of ovarian function^21,22. The resumption of ovarian cyclicity, in turn, resolves the insulin resistant period during postpartum as increasing estrogen levels restore pre-pregnancy insulin sensitivity²⁵. These similar insulin secretion patterns during adolescence, gestation, and the postpartum period are consistent with the hypothesis that the HPAA plays a role in the mobilization of energy that supports the energetic demands of (re-)establishing ovarian function during reproductive transitions.

Insulin resistance is associated with a decrease in glucose oxidation (using glucose for energy), and greater fat oxidation (mobilizing adipose fat mass as a source of energy)²⁶. Fat oxidation yields more adenosine triphosphate (ATP) production per molecule compared to glucose oxidation, thus generating more energy per unit for use²⁶. Thus, the increase in compensatory insulin secretion as a response to insulin resistance during early adolescence may promote a shift towards an increase in use of fat storage to fuel earlier stages of maturation, prior to menarche. During this period, insulin continues to facilitate glucose uptake to promote more overall available energy either for immediate use or storage, but it could be that fat stores are prioritized as the main energy source for generating ATP for use. Our results suggest that when girls are pre-menarche, mechanisms may be in place to promote fat oxidation. Consistent with this proposition, in this sample of adolescent girls, HPAA activation was positively associated with elevated adiponectin pre-menarche, reflecting a decrease in visceral adipose fat. This finding may suggest that the mobilization of energy from fat stores is supporting growth and reproductive maturation during early stages of the adolescent transition¹⁸.

Several hypotheses have been put forward to explain changes in adiposity across the adolescent transition among girls, including the Rose Frisch hypothesis which posited that establishing a minimum weight for height and a minimum precent body fat is necessary to achieve menarche²⁷. However, the Frisch hypothesis has been heavily criticized, on the basis that there are many complex factors that contribute to reproductive maturation. Ellison (1990) suggests that it is not a case of achieving a specific weight or fat threshold, but rather it is achieving and maintaining a positive energy balance throughout the transition that promotes and supports reproductive maturation. In contexts of positive energy balance, excess energy can be allocated towards reproductive development and function²⁸. While we don’t have the data to test specifically when adiposity changes in relation to menarche timing, to contribute to this discussion, here we provide important evidence on some of the mechanisms that are involved in overall changes in adiposity across the menarche transition. Thus, taken together, our results provide support for the hypothesis that HPAA activation plays a role in the shift towards fat oxidation, providing the metabolic energy source during earlier maturation stages to simultaneously fuel both somatic growth and maturation of the reproductive axis.

As girls advance through maturation, both basal metabolic rate (BMR) and resting metabolic rate (RMR) have been observed to be higher compared to pre-transition adolescents (i.e., before thelarche, breast budding, the first stage of female maturation)²⁹. BMR and RMR reflect the energy used to maintain metabolic tasks when an organism is at rest and are associated with both greater energy use and storage, indicating greater energy needs. As higher BMR and RMR are associated with both greater energy use and greater visceral fat accumulation²⁹, these patterns are consistent with the observed higher levels of c-peptide among girls in this cohort leading up to menarche, suggesting greater energy needs for fuelling somatic growth and maturation of secondary sexual characteristics, and the shift towards greater energy storage needs post-menarche.

Consistent with the proposition that the adolescent insulin resistant period ends once ovarian cyclicity is established⁷, we observed that girls experienced lower c-peptide levels in post- compared to pre-menarche. Once the insulin resistance period is resolved post-menarche, glucose oxidation resumes as the predominant energy source (over fat oxidation)²⁶. A shift back towards prioritizing glucose oxidation, and away from fat oxidation, would allow for greater fat accumulation^26,30. This shift coincides with our observation that post-menarche girls had lower adiponectin levels, suggesting greater metabolic energy storage in the form of visceral fat. Our findings are consistent with those of other cross-sectional and longitudinal studies reporting that girls’ adiponectin levels decline slightly across the transition^8,10,26. Our results are also consistent with studies that observe an increase in other measures of energy storage, including weight, BMI, and fat percentage, as adolescent girls reach menarche and continue to mature^10,31. The increase in adiposity post-menarche may be critical to accumulate energy reserves in anticipation of potential energetic demands of future reproductive events such as pregnancy, lactation and maternal care, which are energetically costly³². Yet, the biological mechanisms driving these shifts in adiposity pre- to post-menarche may be specific to adolescent girls who are living in contexts of energetic challenges.

Limitations

Our biomarkers of metabolic energy provide novel longitudinal information about energy potentially available for uptake and energy storage within girls across the adolescent transition. Yet, they do not provide information as to the specific tissues that are taking up and using this energy, nor whether energy was actually taken up into cells, given c-peptide only tells us about the potential for energy uptake. Thus, we are unable to determine whether this energy, and how much, is allocated towards somatic growth or reproductive development and function. While first morning urine (FMU) cortisol, c-peptide, and adiponectin are valid measures of HPAA activity and metabolic energy, which is a strength of our study³³, we do not have data on other biomarkers such as corticosterone or growth hormone, which, together, would provide a more comprehensive characterization of mechanistic changes during puberty. Further, girls self-reported age at menarche, which may be introducing additional biases and may limit the accuracy of the timing of menarche. While we did evaluate reproductive hormone patterns to confirm whether girls were truly in their pre- or post-menarche stages for each data collection period, we should be cautious in our interpretations of metabolic energy changes relative to menarche timing.

While our study provides useful information regarding mechanisms that could be involved in the interactions between environmental exposures and growth and development, the generalizability of our results is limited by the socio-ecological context in which participants in this cohort live. The COPE study represents a very unique dataset, and findings from this study may be specific to the adolescent girls living in remote communities in Guatemala. As such, we are cautious about assuming that our interpretations of our results extend to universal, biological mechanistic processes that occur among all girls during their adolescent transition. Future studies should replicate our investigation in a variety of populations to ascertain whether these findings are consistent across different socio-ecological contexts.

Finally, the non-significant interaction term between cortisol and menarche stage that we included in our original proposed models may have been related to low power due to our small sample size. As well, we conducted inference using only our final models, which we selected after finding no evidence that the more complex models including interaction terms between cortisol and menarche stage we considered fit better. This process may have affected the accuracy of our resulting p-values and confidence intervals. A larger dataset would be required to conduct model selection and inference separately.

Significance and implication of findings

In sum, this is the first study to include longitudinal, within-individual analyses of the associations between HPAA activity and energy available for uptake and energy storage pre- and post-menarche. Our results show that girls present higher insulin (greater potential energy uptake) and adiponectin levels (lower energy stores) when they are pre-menarche compared to post-menarche. High insulin levels and low visceral fat (higher adiponectin) early in the transition may suggest that pre-menarche girls invest their available metabolic energy in somatic growth and reproductive maturation during this period. Once menarche is established and those processes are near completion, post-menarche girls begin to devote their available metabolic energy to sustain ovarian cyclicity and, importantly, to store energy in the form of visceral fat (lower adiponectin) in preparation for the high energy costs of gestation, lactation, and child rearing. Our observation that cortisol is positively associated with both energy markers, is consistent with the hypothesis that the HPAA plays a critical role as a mediator of energy allocation during this transition.

Future longitudinal studies should evaluate how metabolic energy allocation varies amongst girls with different physical phenotypes, variation in access to nutrition and resources, or different socio-ecological contexts, and their impact on the pace of reproductive maturation, including the establishment of ovulatory cycles after reaching menarche. Examining factors associated with the pace of the adolescent transition and variation in ovarian cycle quality and regularity will be important in informing on how socio-ecological context and specific challenges can lead to changes in critical reproductive milestones including, for example, coitarche (age at first sex) and age at first birth, which are important determinants of reproductive health outcomes throughout women’s lifespans. Understanding the factors that may affect said process is paramount to optimizing women’s health and critically important for public health policy and interventions.

Methods

Study population and study design

Our analyses were based on secondary data from the Consequences of Peri-conceptional Events (COPE) study^13,34, a longitudinal, naturalistic cohort study that began in 2000 and has followed the Indigenous Kaqchikel Mayan girls who provided the data used in this study since their conception. COPE’s main objective was to examine relationships between peri-conceptional maternal stress and early child development. COPE participants live in two rural, semi-subsistence communities in the highlands of Guatemala¹³. Most families own or lease land to produce corn, beans, coffee, fruit, and vegetables, which comprise the majority of their diet, in addition to consuming small amounts of chicken, meant, and freshwater fish (personal communications). Further, due to a market economy shift that has taken place in the last 20 years, access to processed and packaged foods in these communities has increased substantially, which has introduced nutritional challenges, including poor- or low-quality energy intake, among COPE participants. Daily stressors or environmental exposures that may also contribute to socio-ecological challenges include chronic, persistent exposure to infectious diseases or infections, strenuous or demanding physical activities, and exposure to psychosocial stressors such as financial stress or interpersonal conflicts with family and community members.

Data used in our analyses were collected during three field seasons: in 2013 from 21 girls (aged 10–11 years old at the time) who were conceived as part of the COPE study in 2000–2001; in 2017 for the same participants (ages 14–15) as well as for 25 additional girls (ages 12–15), some of whom are sisters of the original COPE participants; and in 2023 for follow-up data from all girls that participated in the 2013 and/or 2017 field seasons (ages 18–21). Twenty-nine participants provided both pre- and post-menarche hormone data to evaluate within-individual changes across the transition. Additionally, 17 participants who participated only during their pre-menarche stage (i.e., initial data collection in either 2013 or 2017, but not follow-up data collection in either 2017 or 2023), or post-menarche stage (i.e., began participating in COPE only after having gone through menarche) were also included in this study to contribute additional information across the pre-post-menarche transition. Thus, data from a total sample of 46 girls was included in these analyses. Participants’ reproductive maturation status was evaluated in each follow-up field season to monitor their transition to post-menarche. Data collected in 2023 were included in this analysis only for girls who had not reached menarche by 2017 (n = 9).

Data and biospecimen collection

In 2013, 2017, and 2023, starting in January of each field season, participants self-collected first morning urine (FMU) samples in plastic, chemically inert, urine-collection containers. Collection frequency was daily for three weeks except Sundays in 2013 and every other day, except Sundays, for four months in 2017 and 2023. The 46 adolescent girls provided an average of 16 FMU samples when they were pre-menarche and 29 FMU samples when they were post-menarche. We processed aliquots (2 mL) of the FMU samples on the same day as FMU sample collection. Aliquots were stored in cryo-vials at -10 °C in the field laboratory. At the end of each field season, all samples were shipped on dry ice and stored at -80 °C in the Maternal and Child Health Laboratory at Simon Fraser University (SFU).

Demographic questionnaires, which included questions regarding reproductive status and age at menarche, were conducted in Kaqchikel or Spanish (the two local languages) by local, Indigenous research assistants, once for each participant during each field season.

Hormone assays

FMU samples provide information on biomarkers (hormone concentrations) of HPAA and HPGA activity and metabolic energy. FMU cortisol levels were used to evaluate participants’ basal HPAA activity (measured as urinary-free cortisol levels) throughout the previous night, since the last urinary void (Table 1). FMU cortisol is a validated measure of basal HPAA activity that is less likely to be affected by confounders such as exercise or food intake than samples collected at other times across the day, and makes between and within individual variation in basal HPAA activity easier to detect^13,33. FMU c-peptide levels provide information on metabolic energy that is potentially available for uptake. C-peptide was used as a proxy for insulin levels as it is a segment of proinsulin that is released in 1:1 equimolar amounts with insulin during the conversion of proinsulin to insulin⁹. Insulin levels reflect potential energy uptake, as insulin facilitates glucose uptake from the bloodstream into cells. FMU adiponectin was used as a marker of an individual’s stored energy reserves. Adiponectin, produced by white adipose tissue, is inversely related to the amount of accumulated visceral adipose fat mass³⁵. Higher adiponectin levels reflect lower levels of adipose fat storage, while lower levels of adiponectin reflect higher levels of adipose fat storage³⁵. Patterns of urinary beta sub-unit of follicle-stimulating hormone (FSH) and urinary conjugates of estradiol and progesterone (estrone-3-glucuronide (E1G) and pregnanediol glucuronide (PdG), respectively) were used to assess and confirm self-reported reproductive status, i.e., whether girls were pre- or post-menarche during each data collection period^36–38.

Hormone analyses were conducted at the Maternal and Child Health Laboratory at SFU. Urinary biomarkers of HPAA physiology (cortisol), energetic status (c-peptide and adiponectin), and reproductive status (FSH, E1G, and PdG) were measured using enzyme-linked immunosorbent assays (ELISAs) (Quansys Biosciences, Utah, USA), which our lab has previously validated (lower limits of detection for cortisol = 0.343 ng/mL; c-peptide = 0.090 ng/mL; and adiponectin = 0.023 ng/mL)³⁹. To adjust for variation in dilution levels for each urine sample due to daily variation in the hydration state of participants, all hormone concentrations were corrected for specific gravity using refractometry⁴⁰. To control for variation among assays, all samples from the same participant were run on the same plate. All intra-and inter-assay coefficients of variation were below 7% and 11%, respectively.

Ethics approval

The COPE study is part of a long-term, longitudinal study for which data collection began in 2000. Throughout the study’s history, ethics approval has been granted by the University of Michigan (IRB File Number: H03-00001759-R2), NIH (Protocol Number: 06-E-N145) and Simon Fraser University (Application/Approval Number: 2010s0557). Consent and assent forms, data and biospecimen collection protocols, and analyses for the COPE study were approved by Simon Fraser University’s Research Ethics Board (Application/Approval Numbers: 2012s0668 and 2016s0576). Informed consent and assent were obtained for the collection and analysis of all forms of personally identifiable data including biomedical, clinical, and biometric data. For participants younger than 18 years of age at enrollment, parents or legal guardians provided voluntary written informed consent, and participants provided voluntary written informed assent prior to participating. Participants that turned 18 during the study period were re-consented and provided their own voluntary written informed consent. The COPE study complied with all institutional guidelines and protocols for observational studies using human biospecimen samples for analyses, in accordance with the Declaration of Helsinki and the Tri-Council Policy Statement: Ethical Conduct for Research Involving Humans. We ensured that all study procedures followed these ethical regulations to minimize risk and protect the safety and rights of participants.

Statistical analysis

Cortisol, adiponectin, and c-peptide data distributions were right-skewed. We log 10-transformed their values so that the distributions of our response variables (adiponectin and c-peptide) were approximately normal, and to align these variables on a comparable scale⁴¹. To detect outliers, we identified datapoints well outside established average biological ranges and those outside the interquartile range for each variable, within subjects. We checked if extreme values in one hormone tended to occur with similarly extreme values of other hormones (suggesting an error in the sample). If so, those values were removed from our dataset (30 of 9660 data points).

To evaluate the associations between cortisol and metabolic energy pre- to post-menarche, we fit linear mixed-effects models using the lme() function in R⁴² (see final models in results).

Linear mixed-effects models take into account both fixed and random effects, the latter of which are included to account for the repeated measures collected from each participant. This approach also accounts for, and handles, unbalanced and missing data, which are a feature of our dataset⁴³. It also allows the variance of c-peptide and adiponectin to change from pre- to post-menarche. For both the c-peptide and adiponectin models, we conducted a likelihood ratio test to compare the fit of the model to that of the model that assumes that the variance of the response is constant. Both models were adjusted for time relative to menarche (defined as the difference between the time at which participants reached menarche and the time at which data was collected pre- and post-menarche), to adjust for variation in maturation status (time relative to menarche) of participants in each data collection period. Menarche status was included as a dummy variable to allow for longitudinal analyses, and effect modification, of hormone profiles within and among girls when they were pre-menarche compared to when they were post-menarche. Pre-menarche was the reference level (Menarche status = 0).

Author contributions

AR conducted the literature review, collected data and biospecimens in 2023, analyzed and interpreted the data, and wrote and revised drafts of the manuscript and figures. KGS contributed to the design of the COPE study, led the quantification of biomarkers, provided input on the interpretation of the data and results, and revised drafts of the manuscript. RMA provided input on the statistical analyses conducted and provided suggestions to improve the manuscript. CRV reviewed and revised drafts of the manuscript and provided input on the interpretation of the data, results and discussion. PAN designed the COPE study, collected data and bio-specimens, provided input on the analyses, data and results interpretation, and revised drafts of the manuscript.

Funding

The COPE study was funded through a Natural Sciences and Engineering Research Council (NSERC) Discovery Grant (Grant ID: RGPIN-2018-06750). AR was supported through an NSERC Vanier Canada Graduate Scholarship (Award ID: 434022).

Data availability

Data used for the purpose of this manuscript are part of the Consequences of Peri-conceptional Events (COPE) study, which was conducted by Dr. Pablo Nepomnaschy’s lab team. The corresponding author had full access to all the data in this study. To preserve the confidentiality of the participants who live in small communities, data will not be made publicly available.

Competing interests

The authors declare no competing interests.

Ethics approval

COPE’s data collection and use was approved by the Research Ethics Board (REB) at Simon Fraser University (SFU) (Application Numbers: 2012s0668 and 2016s0576).