Skip to main content
NCBI home page
The Journal of Clinical Endocrinology and Metabolism logoLink to The Journal of Clinical Endocrinology and Metabolism
. 2022 Aug 15;107(10):e4187–e4196. doi: 10.1210/clinem/dgac484

A Shifting Relationship Between Sex Hormone-Binding Globulin and Total Testosterone Across Puberty in Boys

Zhijie Liao 1, Daniel E Vosberg 2,3, Zdenka Pausova 4,5,6, Tomas Paus 7,8,9,10,✉
PMCID: PMC9516180  PMID: 35965384

Abstract

Context

Sex hormone-binding globulin (SHBG) is associated with levels of total testosterone (total-T), and both total-T and SHBG are associated with obesity.

Objective

We aimed to clarify the nature of the relationship between testosterone and SHBG and improve our understanding of their relationships with obesity. We hypothesize that the hypothalamic-pituitary-gonadal axis contributes to the homeostasis of testosterone by increasing the production of gonadal testosterone through a feedback mechanism that might operate differently at different pubertal stages.

Methods

We investigated the dynamics of the relationship between SHBG, total-T, and body mass index (BMI) throughout puberty (from age 9 to 17) using longitudinal data obtained in 507 males. The directionality of this relationship was explored using polygenic scores of SHBG and total-T, and a two-sample Mendelian Randomization (MR) in male adults.

Results

Consistent with our hypothesis, we found positive relationships between SHBG and total-T at age 15 and 17 but either no relationship or a negative relationship during the earlier time points. Such shifting relationships explained age-related changes in the association between total-T and BMI. Polygenic scores of SHBG and total-T in mediation analyses and the two-sample MR in male adults suggested an effect of SHBG on total-T but also a somewhat weaker effect of total-T on SHBG. Two-sample MR also showed an effect of BMI on SHBG but no effect of SHBG on BMI.

Conclusion

These results clarify the nature of the relationship between testosterone and SHBG during puberty and adulthood and shed new light on their possible relationship with obesity.

Keywords: body mass index, HPG axis, polygenic scores, Mendelian randomization, ALSPAC

Testosterone is a key regulator of reproductive organs and other tissues; its actions are associated with cardiovascular morbidity (1) and metabolic syndrome (2, 3). In males, a major part of the circulating testosterone is bound with high affinity to the sex hormone-binding globulin (SHBG). The remaining nonbound testosterone is termed as “bioavailable testosterone” (4). SHBG prevents bound testosterone from diffusing out of the bloodstream into the cell and binding to its intracellular (androgen) receptor (5), thus suppressing testosterone bioactivity (6). Thus, the bioactivity of testosterone is believed to be determined mainly by SHBG (7). SHBG is mainly produced in the liver, and its level in blood is influenced by hormonal factors as well as the metabolic status (6, 8). Protection from the excess of endogenous or exogenous sex steroids is thought to be a key physiological function of SHBG (6, 9-11). But there is continuing debate over whether — and how — SHBG regulates the levels of sex steroids (7, 12).

As total testosterone (total-T) is the sum of SHBG-bound testosterone and bioavailable testosterone, one can assume that a higher level of total-T should be accompanied by a higher level of SHBG to maintain a stable level of bioavailable testosterone. But the nature and directionality of the relationship between SHBG and total-T are unclear. On the one hand, very limited evidence suggests a positive effect of androgens on the SHBG (13). Other studies suggest the opposite: oxandrolone and testosterone treatment of adolescent males with a delay of growth and puberty appears to result in the reduction of SHBG levels (14, 15). Thus, it is commonly suggested that androgens suppress rather than upregulate the production of SHBG (6). While the above studies assume that testosterone impacts SHBG levels, others have argued for the opposite directionality, namely that higher levels of SHBG lead to higher total-T concentrations via the hypothalamic-pituitary-gonadal (HPG) feedback (12, 16). This hypothesis suggests that increased SHBG leads to a decrease in bioavailable testosterone that, in turn, relieves the negative feedback of testosterone on the hypothalamus and pituitary and, ultimately, stimulates the production of luteinizing hormone (LH) and gonadal (total) testosterone. This hypothetical effect of SHBG on total-T is supported by findings in both mice and human studies. Several human studies have found relationships between single-nucleotide polymorphism (SNP) in the gene encoding SHBG and the levels of total-T (10, 17). In adult males, carriers of genetic variants associated with high levels of SHBG also have high levels of total-T, while calculated free testosterone levels do not differ between the genotypes (10, 17). A mutation within SHBG, which results in SHBG deficiency, also leads to a low level of total-T (18). Furthermore, expressing a human SHBG transgene in male mice led to a marked rise in total-T levels while having no effects on the free testosterone (9). These findings provide strong evidence that high levels of SHBG can lead to high total-T concentrations.

Clarifying the nature of the relationship between testosterone and SHBG may also shed some light on our understanding of the relationships between testosterone, SHBG, and obesity. Both total-T and SHBG are associated with metabolic status but with different patterns in males and females (19, 20). Thus, in both male and female adults, low levels of SHBG are associated with obesity (21, 22). A higher risk of obesity is associated with low levels of total-T in males (23) while the opposite is seen in females (3, 22, 24). Underlying mechanisms of such sex-dependent association between testosterone and obesity remain unclear. According to the hypothesized role of the HPG feedback in regulating the SHBG-testosterone relationship, we speculate that the sex difference in the HPG feedback may partly contribute to sex differences in the relationship between total-T and obesity. Specifically, in male but not female adults, obesity-associated decreases in SHBG levels can lead to increases in bioavailable testosterone and in turn, decreases in total-T (via the HPG feedback mechanism). Thus, low levels of total-T might be related to a higher risk of obesity in males but not in females.

Puberty is a transitional period of physical development during which the HPG axis becomes (gradually) functional, and the production of testicular testosterone increases in males, while the level of SHBG declines (25). The reason for the decline in SHBG levels is not clear. Some authors suggest that it is partly due to the rising levels of testosterone, which is known to suppress the SHBG production (6, 14, 15). Overall, the relationship between SHBG and testosterone in human adolescents is unclear, as is the possible changes in this relationship throughout puberty. One study found a negative relationship between SHBG and total-T in boys aged 6 to 20 years (26), while another study found no relationship in boys with aged 5 to 8 years (27).

The present study describes the relationship between the levels of SHBG and total-T throughout puberty using a well-powered longitudinal dataset; it also investigates the directionality of this relationship using polygenic scores of SHBG and total-T. We also explore changes in the relationship between the sex hormones and body mass index (BMI) throughout puberty. According to the hypothesized role of the HPG feedback in regulating the SHBG-testosterone relationship, we expect no relationship between SHBG and total-T in early puberty and a positive relationship in late puberty when the HPG axis is fully functioning. Changes are expected in the relationship between total-T and BMI from the early to late puberty.

Methods

Participants

The sample of adolescents studied here was drawn from the Avon Longitudinal Study of Parents and Children (ALSPAC). This is a birth cohort that recruited pregnant women residing in Avon County, United Kingdom, with an expected delivery date between April 1, 1991, and December 31, 1992. Since then, the children and their parents have been studied extensively, and longitudinal data, including self-administered questionnaires and clinical examinations collected during study visits, are available (28, 29). The initial number of pregnancies enrolled was 14 541, resulting in 14 062 live births and 13 988 children who were alive at 1 year of age. Ethical approval for the study was obtained from the ALSPAC Law and Ethics Committee and the Local Research Ethics Committees. Please note that the study website contains details of all the data that are available through a fully searchable data dictionary and variable search tool (http://www.bristol.ac.uk/alspac/researchers/our-data/). We obtained and analyzed testosterone and SHBG levels from the blood samples of a subset of 507 male participants who had been recruited for further study using magnetic resonance imaging.

Quantification of Sex Hormones and BMI

Blood samples were taken at selected visits over the course of the longitudinal assessments, roughly at the following years of age: 9, 11, 13, 15, and 17 (Supplementary Table 1) (30). All participants in this study had a maximum of 5 and a minimum of 3 blood samples across these 5 visits (5 samples: n = 232; 4 samples: n = 208; 3 samples: n = 65). Measurements of height and weight were obtained to calculate BMI (kg/m2) for each of the 5 visits. Details of quantifying total-T and SHBG have been described in our previous report (31). Briefly, plasma levels of testosterone and SHBG were determined by conducting enzyme-linked immunosorbent assays using commercially available kits (testosterone: EIA-1559; DRG International; Springfield, New Jersey, USA. SHBG: IB79131; Immuno-Biological Laboratories, Inc.; Minnesota, MN, USA). The assays had lower limits of sensitivity of 0.28 nmol/L for testosterone and 0.77 nmol/L for SHBG, with average inter- and intra-assay coefficients of variation less than 10%. The assay sensitivity is calculated by subtracting 2 SD from the mean of 20 identical runs of the zero standards. The possible effects of circadian rhythm on testosterone levels were modeled (and adjusted for) in each visit separately (details in (31)). Outliers that were below or above 4 SD from the average levels of total-T or SHBG at each age were excluded from the analyses reported here.

Genetic Data and Polygenic Scores

There were 451 out of 507 participants who were genotyped using the Illumina HumanHap550 quad genome-wide SNP genotyping platform by 23andMe subcontracting the Wellcome Trust Sanger Institute, Cambridge, UK and the Laboratory Corporation of America, Burlington, NC, USA. Phasing and imputation were carried out with ShapeIt v2 and Impute2 using the 1000G Project as the reference set. The SNPs with minor allele frequencies less than 1%, call rate < 95%, or did not pass an exact test of Hardy-Weinberg equilibrium at P < 1 × 10-6 were removed. There were 7 054 621 SNPs remaining after the quality control (QC). All participants passed the QC, which included removing individuals with a mismatch between reported and genetic sex, minimal or excessive heterozygosity, high levels of genotype missingness (> 5%), and cryptic relatedness. We calculated polygenic score (PGS) of total-T and SHBG, respectively, using PRSice-2 (32). Specifically, PGSs were calculated by computing the sum of each participant’s genotypes, weighted by the corresponding genotype effect-size estimates derived from genome-wide association studies (GWAS) of total-T and SHBG using UK Biobank (UKBB) data obtained in male adults (see details below). We used summary statistics from adults (UKBB) rather than from adolescents because there are no published GWASs for these phenotypes available in adolescents; our sample (n = 500), we have insufficient statistical power for estimating reliably the weights for SNPs to calculate PGSs. The PGSs were computed from SNPs that were genome-wide significant (P < 5e-8) in the GWASs and independent based on the clumping carried out with the recommended parameters (i.e., kb = 250, r2 = 0.1). There were 475 and 286 SNPs used to derive PGS of SHBG and total-T, respectively (see details of these SNPs in Supplementary Table 2 (30)). Note that the weights of SNPs used to derive PGS in present study are highly correlated with the weights of SNPs from Ruth et al (SHBG: r2 = 0.996; total-T: r2 = 0.996 (20)). A greater PGS indicates a genetic predisposition for higher levels of total-T or SHBG.

To investigate the possible directionality of the relationship between SHBG and total-T, we fitted the PGSs of total-T (PGStotal-T) and SHBG (PGSSHBG) together with the phenotypic total-T and SHBG in mediation models. A PGS estimates an individual’s genetic liability to a phenotype, and it has an advantage in interpreting causal relationships over other types of predictors: correlations between polygenic scores and traits can only be interpreted in one direction causally, when not considering the possibility of horizontal pleiotropy. Thus, individuals with different PGS values are analogous to treatments in a randomized controlled trial, in that they differ in the exposure of interest, but should not differ in confounding factors. To help reduce the possibility of horizontal pleiotropy of the PGS (e.g., high PGSSHBG leads to increased total-T via other pathways rather than via SHBG), we also tested the mediation effect of phenotypic exposure in the relationship between the PGS and outcome. In other words, if there was a causal effect of the exposure on the outcome (for example SHBG influencing total-T), the relationship between the PGS of exposure (PGSSHBG) and the outcome (total-T) should be mediated by the phenotypic exposure (SHBG), and the direction of the total effect and the mediation effect in the mediation models should be the same.

UK Biobank and Genome-Wide Association Studies of SHBG and Total-T

The UKBB is a cohort study of approximately 500 000 adults aged 37 to 73 years at recruitment, comprising multimodal phenotyping and genotyping (33). The UKBB study received approval from the UK Biobank Resource Application 43688 and local ethics committees, and signed, informed consent was obtained from the participants. In this study, QC for GWAS was conducted among males. Participant exclusions were dropouts, heterozygosity, and missingness outliers (defined by the UKBB), genetic and reported sex mismatches, sex chromosomal aneuploidy, non-European ancestry, and kinship greater than third-degree. The last step was done by first removing participants with greater than 10 third-degree relatives, then by using the R package, “ukbtools” version 0.11.3 (34), and a KING kingship coefficient cutoff of 0.0884. The SNP exclusions were missingness > 5%, MAF < 0.01, significant deviation from HWE (P < 1e-10), and an INFO score of <0.8. Following participant and SNP QC, there remained 181 389 males and 8 644 321 SNPs.

The phenotypes of interest were total-T (UKBB field 30850), SHBG (UKBB field 30830), and BMI (UKBB field 21001), with total-T and SHBG measured in nmol/L, using the immunoassay system (Beckman Coulter Unicel Dxl 800). The total-T and SHBG phenotypes were log10 transformed and outliers greater than 4 SD from the mean were excluded. Among the sample surviving QC, 170 830 males had values for total-T and 158 429 males had values for SHBG. Genome-wide association studies were conducted using PLINK2 (https://www.cog-genomics.org/plink/2.0/), estimating a general linear model for each SNP, adjusting for age and the first 10 principal components of genetic ancestry.

Mendelian Randomization

As our mediation models indicate a possible causal relationship between SHBG and total-T in late puberty, we conducted a 2-sample Mendelian randomization in male adults from the UK Biobank to corroborate results of the mediation-based analysis. To conduct the 2-sample Mendelian randomization analyses with nonoverlapping samples, the participants who passed QC in the above GWASs and had both total-T and SHBG values (n = 157 047) were randomly split into 2 subsamples, and GWASs were conducted again for each phenotype (n = 78 523 males each). There were no significant differences between the subsamples in age, the first 10 principal components of genetic ancestry, or recruitment site. Mendelian randomization was conducted using the R package “TwoSampleMR” version 0.5.6 (35), using the recommended clumping parameters (kb = 10 000, r2 = 0.001). The causal effect of SHBG on total-T and the causal effect of total-T on SHBG were assessed using independent genome-wide significant SNPs as exposures and the full summary statistics of the other phenotype as outcomes. Steiger filtering and heterogeneity filtering were applied, resulting in 67 exposure SNPs for SHBG and 11 exposure SNPs for testosterone (see Supplementary Table 7 for instrument strength of each SNP). MR Egger and inverse-variance weighted were used to test heterogeneity and did not identify significant evidence of heterogeneity (P value range, 0.43-0.637). In addition, no significant evidence of pleiotropy for the exposures was identified (MR Egger, P value range, 0.287-0.87). To avoid arbitrary selection of methods and improve power and minimize false discovery rates, the machine learning mixture of experts (MoE) technique, MR-MoE 1.0, was applied to estimate the reliability of Mendelian randomization (MR) testing (with “mr_moe” function in “TwoSampleMR”, 35). The MR-MoE method exploits characteristics of the summary data and provides a quantitative estimate of reliability of MR methods, with high MoE values suggesting high reliability (36, 37). We focused on the top 5 most reliable tests with Steiger and heterogeneity filtering. For both causal directions, these 5 methods were fixed effect inverse-variance weighted (FE IVW), random effect inverse-variance weighted (RE IVW), simple median, weighted median, and simple mean. In addition, the results using other methods (fixed effect and random effect MR Egger (38), Weighted median, Simple mode, Weighted mode, Penalized mode, Penalized median) are also reported in the Supplements. To compare the strength of the 2 causal effects (i.e., effect of SHBG on total-T and effect of total-T on SHBG), a Z-test was applied to compare beta values of the MR tests (39). Similar 2-sample Mendelian randomization analyses were applied to explore the causal relationship between BMI and SHBG.

Statistics

In the initial analysis, we used a simple linear regression model (total-T ~ SHBG) to describe the relationship between the levels of SHBG and testosterone in adolescents at each visit, and we investigated whether and how it changed throughout adolescence (with corrections for false discovery rate [FDR] for 5 visits). For the third visit (age 13), we observed a clear bimodal distribution of total-T concentrations with 2 peaks at around 3.5 nmol/L and 12 nmol/L, respectively. For this reason, we also divided the participants into low and high total-T subgroups and, for the third visit only, analyzed the relationship between SHBG and testosterone in each subgroup. The boundary (6.51 nmol/L) between low and high subgroups was determined based on the distribution of the levels of total-T during this visit, by finding the lowest point between 2 peaks using the “find_peaks” function from R package “ggpmisc” version 0.4.5.

To understand the longitudinal relationship between SHBG and total-T during puberty, a time-lagged model using path analysis was built to model total-T (SHBG) on both SHBG and total-T at the previous visit; this was done in a subsample of individual with all 5 samples available between 9 and 17 years of age (n = 232). The model also included within-time covariance terms between SHBG and total-T. The path analysis was implemented with R package “lavaan” version 0.6-10 (40), and all variables were standardized before the analysis. The comparative fit index (CFI), the root mean square error of approximation (RMSEA), and the standardized root mean square residual (SRMR) were used to assess model fit; RMSEA values less than 0.08 and CFI values greater than 0.90 indicate acceptable model fit (41, 42).

To investigate the possible directionality of the relationship between total-T and SHBG, PGS of total-T and SHBG together with the phenotypic total-T and SHBG were fitted into 2 mediation models for each of the 4 visits (age 9, 11, 15, and 17), and each subgroup at the third visit (age 13). The first model tested the effects of SHBG on total-T with PGSSHBG as the exposure, phenotypic SHBG as the mediator and phenotypic total-T as the outcome. In the opposite direction, the second model tested the effects of total-T on SHBG with PGStotal-T as the exposure, phenotypic total-T as the mediator and phenotypic SHBG as the outcome. The direction (negative/positive) of the total effects and the mediation effects were expected to be the same within the model if the levels of SHBG (total-T) had an influence on total-T (SHBG). All variables were standardized within each age before the analysis. The significance level for the mediation effect was determined with a bootstrapping procedure (n = 5000; FDR for 4 visits and 2 groups at age 13).

To explore the relationship between the sex hormones and BMI throughout puberty, we described the linear relationships between SHBG, total-T, and BMI in adolescents at each visit (with correction for FDR for 5 visits). As we speculated that the SHBG-testosterone relationship may contribute to the testosterone-BMI relationship, we investigated the relationship between total-T and BMI after controlling the SHBG at each visit with a linear regression model (BMI ~ total-T + SHBG).

All statistical analyses were conducted using R software (version 4.0.5). The significant threshold for P value is 0.05 (2-sided), with FDR corrections as noted above.

Results

Correlations Between SHBG and Testosterone Across Adolescence

The levels of SHBG declined and total-T increased during puberty (Fig. 1A, Supplementary Table 3 (30)). The relationship between SHBG and total-T changed throughout adolescence (Fig. 1B). We found no significant relationship at age 9 (df = 426, r = −0.029, P = 0.552), a weak negative relationship at age 11 (df = 464, r = −0.151, P = 0.002), a strong negative relationship at age 13 (df = 407, r = −0.416, P < 0.001), a weak positive relationship at age 15 (df = 454, r = 0.138, P = 0.004), and a strong positive relationship at age 17 (df = 404, r = 0.397, P < 0.001). After dividing participants into low and high total-T subgroups at age 13, no relationship between SHBG and total-T was observed in either the low (df = 140, r = −0.161, P = 0.055) nor the high subgroup (df = 265, r = 0.037, P = 0.544). Similar results were found when only participants with 5 blood samples (n = 224) were included in the analysis (Supplementary Table 4 (30)). The time-lagged model has modestly good model fit (CFI, 0.923; RMSEA, 0.124 [90% CI, 0.101-0.149]; standardized root mean square residual, 0.081). One-year cross-lag correlations were found in the late puberty age categories (age 13 to age 15, and age 15 to age 17) but not in early puberty (Fig. 2).

Figure 1.

Figure 1.

Open in a new tab

The relationship between SHBG and total-T across adolescence. The levels of total-T and SHBG (A) and the relationship between SHBG and total-T (B) change across puberty. There were 2 types of participants according to the distribution of total-T at the third visit (age 13). Thus, we further divided the participants into low (gray triangles) and high (black dots) subgroups based on the levels of total-T at this age.

Figure 2.

Figure 2.

Open in a new tab

Time-lagged association between SHBG and total-T. Values represent standardized coefficients, the asterisk represents the level of significance (*P < 0.05, **P < 0.001).

Directionality of Relationships Between Testosterone and SHBG

According to our design, if there was a causal effect of the exposure on the outcome, a relationship between the PGS of exposure and the outcome (i.e., the total effect in mediation model) was expected, and such a relationship should be mediated by the phenotypic exposure, namely, a mediation effect with the same direction as the total effect in the mediation models would also be found. For the first model, which aimed to test the effect of SHBG on total-T, we found a positive total effect of PGSSHBG on total-T at age 13 (only the high total-T subgroup), 15, and 17, but not at age 9, 11, or the low total-T subgroup at age 13. The total effect was mediated by phenotypic SHBG at age 17, but not at age 15 or the high total-T subgroup at age 13 (Fig. 3, left side). For the second model, which aimed to test the effect of total-T on SHBG, a positive total effect of PGStotal-T on SHBG was found at all ages and in the 2 subgroups at age 13. The total effect was partially mediated by phenotypic total-T at age 17, but not the other ages (Fig. 3, right side), suggesting the relationships between PGStotal-T and SHBG at other ages could be due to genetic pleiotropy. Similar results were found when only participants with 5 blood samples (n = 224) were included in the analysis (Supplementary Table 5 (30)). Thus, these results are consistent with a bidirectional (causal) relationship between SHBG and total-T at 17 years of age.

Figure 3.

Figure 3.

Open in a new tab

Mediation models with path coefficients and levels of significance (*P < 0.05, **P < 0.001). The a and b indicate, respectively, correlations between PGS and SHBG (or Total-T) and correlations between SHBG (Total-T) and Total-T (SHBG) while controlling the PGS, the ab indicates the mediation effects. The c indicates the total effects of PGS on the outcome, and c’ indicates the direct effects of PGS on the outcome after taking into account possible mediating effects. The presence of a significant mediation effect (ab) indicates a causal effect of exposure on outcome (SHBG on Total-T in the left panel, Total-T on SHBG in the right panel).

To corroborate the above mediation-based analysis of directionality in the relationship between total-T and SHBG, we conducted a 2-sample Mendelian randomization in male adults (see “Methods”). These analyses provided causal evidence of a positive effect of SHBG on testosterone (beta range: 0.60-0.63; MoE range: 0.88-0.93; P values ≤ 2.34 × 10−48; Fig. 4A, Supplementary Table 6 (30)). There was also some evidence of a positive causal effect of testosterone on SHBG (beta range: 0.16-0.19; MoE range: 0.56-0.69; P values ≤ 2.53 × 10−3; Fig. 4B, Supplementary Table 6 (30)), albeit weaker than the effect of SHBG on testosterone (P values ≤ 5.76 × 10−14). Similar results were obtained with the other MR methods (Supplementary Table 6 (30)).

Figure 4.

Figure 4.

Open in a new tab

Mendelian randomization results of the effect of (A) SHBG on testosterone and (B) testosterone on SHBG. The techniques, selected as the most reliable according to MR-MoE, and with Steiger and heterogeneity filtering, were fixed effect (FE) IVW, random effect (RE) IVW, simple median, simple mean, and weighted mean.

Relationships Between Sex Hormones and BMI

We found a consistent negative relationship between SHBG and BMI throughout puberty (Table 1A). Additional Mendelian randomization in male adults provided causal evidence of a negative effect of BMI on SHBG (MoE range: 0.69-0.75; P values ≤ 3.17 × 10−3) rather than an effect of SHBG on BMI (MoE range: 0.74-0.78; P values ≥ 0.520; Supplementary Table 8 (30)). The relationship between total-T and BMI varied as a function of age: no relationship at ages 9 and 15, a positive relationship at ages 11 and 13, and a negative relationship at age 17 (Table 1B). With SHBG and total-T in the same model, the relationship between total-T and BMI disappeared, while the SHBG-BMI relationship remained (Table 1C).

Table 1.

Association of Total-T, SHBG, and BMI at 5 ages during puberty

Model Age Variable b (SE) t value P value P value (FDR) DF
A. BMI~ SHBG 9 SHBG -0.372 (0.044) -8.489 <0.001 <0.001 427
11 -0.449 (0.042) -10.773 <0.001 <0.001 466
13 -0.380 (0.045) -8.465 <0.001 <0.001 405
15 -0.408 (0.043) -9.535 <0.001 <0.001 445
17 -0.362 (0.046) -7.941 <0.001 <0.001 409
B. BMI~ total-T 9 total-T 0.060 (0.048) 1.259 0.209 0.261 418
11 0.103 (0.046) 2.215 0.027 0.045 458
13 0.148 (0.048) 3.067 0.002 0.006 404
15 -0.036 (0.047) -0.759 0.448 0.448 445
17 -0.177 (0.049) -3.583 <0.001 0.002 396
C. BMI~ total-T + SHBG 9 Total-T 0.050 (0.044) 1.12 0.263 0.703 417
SHBG -0.371 (0.044) -8.373 <0.001 <0.001 417
11 Total-T 0.037 (0.042) 0.875 0.382 0.703 455
SHBG -0.442 (0.042) -10.437 <0.001 <0.001 455
13 Total-T -0.019 (0.050) -0.382 0.703 0.703 401
SHBG -0.386 (0.050) -7.794 <0.001 <0.001 401
15 Total-T 0.018 (0.043) 0.406 0.685 0.703 443
SHBG -0.419 (0.043) -9.656 <0.001 <0.001 443
17 Total-T -0.023 (0.051) -0.456 0.648 0.703 394
SHBG -0.364 (0.052) -7.009 <0.001 <0.001 394
Open in a new tab

Discussion

According to the hypothesis that high SHBG leads to high total-T through the hypothalamic-pituitary feedback, the relationship between SHBG and total-T in males should only exist when the HPG axis is active. Consistent with this hypothesis, we found a positive relationship between SHBG and total-T, and a negative relationship between total-T and BMI only in the late but not early puberty. The shifting SHBG-testosterone relationship explained age-related changes in the association between total-T and BMI. Further analyses with genetic tools provided corroborating evidence supporting a causal effect of SHBG on total-T. Nonetheless, our results also indicate an additional effect of total-T on SHBG in late puberty. Thus, it appears that the 2 hormones interact in a bidirectional manner.

The positive correlation between SHBG and total-T emerged at 15 years of age and grew stronger at age 17. Such age-related changes in the relationship between SHBG and total-T support the hypothesis that a high SHBG level is counteracted by high total-T via HPG feedback, thus maintaining constant levels of bioavailable testosterone. After a peak of activity during infancy, the HPG axis enters a long period of relative quiescence until late childhood when pubertal maturation begins. The onset of puberty typically occurs between ages 9 and 14 for males (43). During puberty, the increased secretion of gonadotropin-releasing hormone (GnRH) regulates the production of LH in the pituitary gland, which in turn stimulates the maturation of the gonads, leading to an increased production of testosterone from 10 to 15 years of age (31, 44). The levels of testosterone reach the adult level at around 16 to 17 years of age (31). The timing of puberty can be quite varied (45), and most individuals may not have a fully active HPG axis until 17 years of age. Thus, only at late puberty, when the HPG axis is sensitive to testosterone feedback, SHBG-induced low levels of bioavailable testosterone will result in an increased production of LH and, in turn, an increase in the production of testosterone by the gonads until a new equilibrium is reached. Our results of the mediation models and 2-sample Mendelian randomization, which has been widely used to assess whether an exposure has a causal effect on an outcome, further support the effects of SHBG on total-T in males with a functional HPG axis (i.e., during late puberty and in adulthood). Note that the PGSSHBG for the adolescents was based on a GWAS in middle-aged male adults and, therefore, may be suboptimal in estimating the genetic liability of the levels of SHBG in adolescents. Despite this limitation, we found comparable prediction power of the PGSSHBG on the SHBG concentration at different ages in our sample of male adolescents (with r of 0.23-0.36).

The underlying mechanism of the shifting SHBG-testosterone relationship can also explain the changes in the relationship between total-T and BMI throughout puberty. A consistent negative correlation between SHBG and BMI was found throughout puberty, while a negative correlation between total-T and BMI only emerged at age 17 when a strong SHBG-testosterone relationship emerged. In line with previous studies in human and mice (46, 47), our two-sample Mendelian randomization suggests a causal effect of BMI on SHBG. Changes in pro- or anti-inflammatory cytokines and alteration in the transcription factors (such as HNF-4α and PPAR-γ) of SHBG gene that occur in obesity could account for the reduction in plasma levels of SHBG (47, 48). We suggest that the emergence of the association between total-T and BMI is contributed by the HPG axis–mediated SHBG-testosterone relationship. Specifically, with a fully active HPG axis, a high BMI-associated decrease in SHBG levels can lead to an increase in bioavailable testosterone which, in turn, suppresses the production of gonadal (total) testosterone via the negative feedback of testosterone on the HPG axis. Thus, a high BMI appears to associate with a low total-T level, which is likely due to the low SHBG levels induced by high adiposity. This is also supported by a disappearance of the testosterone-BMI relationship after controlling for SHBG levels. Furthermore, these findings are, at least in part, consistent with the observed sex differences in the relationship between total-T and the risk of obesity. A lower total-T is associated with a higher metabolic risk in male adult, but not in female adults (3, 19). In addition to other possible mechanisms, such as the hypogonadal–obesity–adipocytokine hypothesis (23), the lack of negative feedback of the HPG axis in testosterone production and the absent relationship between SHBG and total-T in females (16) may also contribute to the sex-dependent association between testosterone and obesity.

In addition to the influence of SHBG on total-T, our results of mediation models and Mendelian randomization also indicated a possible—albeit somewhat weaker—positive effect of total-T on SHBG in late puberty and adulthood. This appears contrary to the classical postulation that testosterone inhibits SHBG production, which was mainly supported by studies of androgen treatments in males and in human liver cells (6, 14, 15, 49). Considering that estradiol is postulated to stimulate the production of SHBG (48), it is possible that high levels of total-T lead to high levels of estradiol derived from testosterone via the enzyme aromatase (5), which, ultimately, results in high SHBG concentration. Alternatively, since androgen response elements have not been found in the human SHBG promoter (6), testosterone may not regulate expression levels of SHBG directly, involving instead indirect pathways. For example, it is known that the production of SHBG can be regulated by metabolic factors and certain pro- or anti-inflammatory factors, such as adiponectin (which upregulates), tumor necrosis factor alpha (TNF-α, downregulates), and interleukin 1 beta (IL-1β, downregulates) (48). Testosterone exerts a significant inhibitory effect on the formation of adipose tissue, and the expression of various adipocytokines, such as TNF-α and IL-1β, and is positively correlated with adiponectin level (1, 50). A high level of testosterone may lead to low body fat and low levels of fat tissue–released inflammatory factors, which, in turn, leads to increased production of SHBG. None of these possible mechanisms could explain, however, the absence of the relationship between total-T and SHBG at early puberty. Considering that it is probably an indirect effect of testosterone on SHBG production, and could be affected by other mediators, the influence of testosterone on SHBG appears to be minimal during early puberty. Note that the PGStotal-T for the adolescents was based on a GWAS in middle-aged male adults and, therefore, may be suboptimal in estimating the genetic liability of the levels of total-T in adolescents. Despite this limitation, we found comparable percentage of variance in total-T explained by the PGStotal-T between older adolescents (at 15 years: 5.89%, 17 years: 8.19%) and adults (UKBB: 5.48%). We also note that plasma levels of total-T and SHBG in males are very similar between the late adolescence and mid-adulthood (25).

In contrast to the positive relationship at late puberty, our initial analysis showed a negative correlation between SHBG and total-T at the age of 11 and 13. The changes of the levels of total-T and SHBG during puberty are opposite: SHBG declines while total-T increases (25). The mechanism for the decline in SHBG during puberty is unclear; it was suggested that decreases in SHBG are partly due to increases in testosterone (51). Our results, however, do not support such a suppression effect of testosterone on SHBG during puberty. The absence of the negative correlation in the 2 groups at age 13 indicated that the negative relationship between SHBG and total-T at early puberty could be due to the inter-individual variations on the timing of puberty. Thus, at 13 years of age, individuals at the earlier pubertal stage have higher SHBG but lower total-T than those at the later pubertal stage. Furthermore, our results of the mediation models suggest no causal relationships between SHBG and total-T at early puberty. But it should be noted that the PGStotal-T was calculated based on GWAS in male adults whose testosterone is largely produced in the testes, and thus may not be optimal in estimating the genetic liability of the levels of testosterone in these young boys. Further research is needed to understand the underlying mechanisms of the negative associations between total-T and SHBG in early puberty.

In summary, our study showed that, in male adolescents, the relationship between SHBG and total-T changes throughout puberty with a negative relationship in the early puberty and a shift into a positive relationship in late puberty. The dynamics in these relationships, together with our mediation modeling and Mendelian randomization, support the hypothesis that high SHBG leads to high total-T via the HPG axis. In addition, our results also indicate an additional—albeit weaker—effect of total-T on SHBG in late puberty, thus supporting the presence of a bidirectional interaction between the 2 hormones. By clarifying the SHBG-testosterone relationship, this study also provides insights into the relationship between sex hormones and risk of obesity.

Acknowledgments

We thank Ammar Khairullah, Laura Cousino Klein, Suzanne Ingle, and Margaret May for their help in quantifying sex hormones and statistical modeling. We thank all the families who took part in this study, and also thank the whole ALSPAC team, which includes midwives, interviewers, computer and laboratory technicians, clerical workers, research scientists, volunteers, managers, receptionists, and nurses.

Glossary

Abbreviations

ALSPAC

Avon Longitudinal Study of Parents and Children

BMI

body mass index

CFI

comparative fit index

FDR

false discovery rate

GWAS

genome-wide association study

HPG

hypothalamic-pituitary-gonadal

IVW

inverse-variance weighted

LH

luteinizing hormone

MoE

machine learning mixture of experts

PGS

polygenic score

QC

quality control

RMSEA

root mean square error of approximation

SHBG

sex hormone–binding globulin

SNP

single nucleotide polymorphism

total-T

total testosterone

Contributor Information

Zhijie Liao, Department of Psychology, University of Toronto, Toronto, Ontario, M5S 3G3, Canada.

Daniel E Vosberg, Centre Hospitalier Universitaire (CHU) Sainte-Justine, Montreal, Quebec, H3T 1C5, Canada; Departments of Psychiatry and Neuroscience, University of Montreal, Montreal, Quebec, H3T 1J4, Canada.

Zdenka Pausova, Research Institute of the Hospital for Sick Children, Toronto, Ontario, M5G 0A4, Canada; Institute of Medical Science, University of Toronto, Toronto, Ontario, M5S 1A8, Canada; Departments of Physiology and Nutritional Sciences, University of Toronto, Toronto, Ontario, M5S 1A8, Canada.

Tomas Paus, Department of Psychology, University of Toronto, Toronto, Ontario, M5S 3G3, Canada; Centre Hospitalier Universitaire (CHU) Sainte-Justine, Montreal, Quebec, H3T 1C5, Canada; Departments of Psychiatry and Neuroscience, University of Montreal, Montreal, Quebec, H3T 1J4, Canada; Department of Psychiatry, University of Toronto, Toronto, Ontario, M5T 1R8, Canada.

Funding Information

The UK Medical Research Council and Wellcome (Grant ref: 217065/Z/19/Z) and the University of Bristol provide core support for ALSPAC. ALSPAC genetics data was generated by Sample Logistics and Genotyping Facilities at Wellcome Sanger Institute and LabCorp (Laboratory Corporation of America) using support from 23andMe. A comprehensive list of grants funding is available on the ALSPAC website (http://www.bristol.ac.uk/alspac/external/documents/grant-acknowledgements.pdf). This research was specifically funded by Wellcome Trust and MRC (076467/Z/05/Z). This research was supported by a grant from the National Institutes of Health (R01MH085772 to T.P.). This publication is the work of the authors and T.P. will serve as guarantor for the content of this paper, which does not necessarily represent the official views of the National Institutes of Health. Z.L. is supported by the Chinese Scholarship Council (201806380177). D.V. is supported by the CHU Sainte-Justine Foundation scholarship.

Disclosures

The authors declare no conflict of interest.

Data Availability

Some or all datasets generated during and/or analyzed during the current study are not publicly available but are available from the corresponding author on reasonable request.

References

  • 1. Kelly DM, Jones TH. Testosterone: a metabolic hormone in health and disease. J Endocrinol. 2013;217(3):R25-R45. doi: 10.1530/JOE-12-0455 [DOI] [PubMed] [Google Scholar]
  • 2. Corona G, Monami M, Rastrelli G, et al. Testosterone and metabolic syndrome: a meta-analysis study. J Sex Med. 2011;8(1):272-283. doi: 10.1111/j.1743-6109.2010.01991.x [DOI] [PubMed] [Google Scholar]
  • 3. Vosberg DE, Parker N, Shin J, Pausova Z, Paus T. The genetics of testosterone contributes to “femaleness/maleness” of cardiometabolic traits and type 2 diabetes. Int J Obes (Lond). 2022;46(1):235-237. doi: 10.1038/s41366-021-00960-w [DOI] [PubMed] [Google Scholar]
  • 4. Larsen PR, Kronenberg HM, Melmed S, Polonsky KS.. Williams Textbook of Endocrinology. 10th ed. Elsevier Health Sciences; 2003. [Google Scholar]
  • 5. Fargo KN, Pak TR, Foecking EM, Jones KJ. Molecular biology of androgen action: perspectives on neuroprotective and neurotherapeutic effects. In: Pfaff DW, Arnold AP, Etgen AM, Fahrbach SE, Rubin RT, eds. Hormones, Brain and Behavior. 2nd ed. Academic Press; 2009:1219-1244. [Google Scholar]
  • 6. Hammond GL. Diverse roles for sex hormone-binding globulin in reproduction. Biol Reprod. 2011;85(3):431-441. doi: 10.1095/biolreprod.111.092593 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 7. Khosla S. Editorial: Sex hormone binding globulin: inhibitor or facilitator (or both) of sex steroid action? J Clin Endocrinol Metab. 2006;91(12):4764-4766. doi: 10.1210/jc.2006-1990 [DOI] [PubMed] [Google Scholar]
  • 8. Pugeat M, Nader N, Hogeveen K, Raverot G, Déchaud H, Grenot C. Sex hormone-binding globulin gene expression in the liver: drugs and the metabolic syndrome. Mol Cell Endocrinol. 2010;316(1):53-59. doi: 10.1016/j.mce.2009.09.020 [DOI] [PubMed] [Google Scholar]
  • 9. Laurent MR, Hammond GL, Blokland M, et al. Sex hormone-binding globulin regulation of androgen bioactivity in vivo: validation of the free hormone hypothesis. Sci Rep. 2016;6:35539. doi: 10.1038/srep35539 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 10. Ohlsson C, Wallaschofski H, Lunetta KL, et al. Genetic determinants of serum testosterone concentrations in men. PLoS Genet. 2011;7(10):e1002313. doi: 10.1371/journal.pgen.1002313 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 11. Hogeveen KN, Cousin P, Pugeat M, Dewailly D, Soudan B, Hammond GL. Human sex hormone-binding globulin variants associated with hyperandrogenism and ovarian dysfunction. J Clin Invest. 2002;109(7):973-981. doi: 10.1172/JCI14060 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 12. de Ronde W, van der Schouw YT, Pierik FH, et al. Serum levels of sex hormone-binding globulin (SHBG) are not associated with lower levels of non-SHBG-bound testosterone in male newborns and healthy adult men. Clin Endocrinol (Oxf). 2005;62(4):498-503. doi: 10.1111/j.1365-2265.2005.02252.x [DOI] [PubMed] [Google Scholar]
  • 13. Plymate SR, Hoop RC, Jones RE, Matej LA. Regulation of sex hormone-binding globulin production by growth factors. Metabolism. 1990;39(9):967-970. doi: 10.1016/0026-0495(90)90309-z [DOI] [PubMed] [Google Scholar]
  • 14. Crowne EC, Wallace WH, Moore C, et al. Effect of low dose oxandrolone and testosterone treatment on the pituitary-testicular and GH axes in boys with constitutional delay of growth and puberty. Clin Endocrinol (Oxf). 1997;46(2):209-216. doi: 10.1046/j.1365-2265.1997.t01-1-1150928.x. [DOI] [PubMed] [Google Scholar]
  • 15. Malhotra A, Poon E, Tse WY, Pringle PJ, Hindmarsh PC, Brook CG. The effects of oxandrolone on the growth hormone and gonadal axes in boys with constitutional delay of growth and puberty. Clin Endocrinol (Oxf). 1993;38(4):393-398. doi: 10.1111/j.1365-2265.1993.tb00520.x [DOI] [PubMed] [Google Scholar]
  • 16. Sinnott-Armstrong N, Naqvi S, Rivas M, Pritchard JK. GWAS of three molecular traits highlights core genes and pathways alongside a highly polygenic background. Elife. 2021;10:e58615. doi: 10.7554/eLife.58615 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 17. Svartberg J, Schirmer H, Wilsgaard T, et al. Single-nucleotide polymorphism, rs1799941 in the Sex Hormone-Binding Globulin (SHBG) gene, related to both serum testosterone and SHBG levels and the risk of myocardial infarction, type 2 diabetes, cancer and mortality in men: the Tromsø Study. Andrology. 2014;2(2):212-218. doi: 10.1111/j.2047-2927.2013.00174.x [DOI] [PubMed] [Google Scholar]
  • 18. Vos MJ, Mijnhout GS, Rondeel JMM, Baron W, Groeneveld PHP. Sex hormone binding globulin deficiency due to a homozygous missense mutation. J Clin Endocrinol Metab. 2014;99(9):E1798-E1802. doi: 10.1210/jc.2014-2055 [DOI] [PubMed] [Google Scholar]
  • 19. Brand JS, van der Tweel I, Grobbee DE, Emmelot-Vonk MH, van der Schouw YT. Testosterone, sex hormone-binding globulin and the metabolic syndrome: a systematic review and meta-analysis of observational studies. Int J Epidemiol. 2011;40(1):189-207. doi: 10.1093/ije/dyq158 [DOI] [PubMed] [Google Scholar]
  • 20. Ruth KS, Day FR, Tyrrell J, et al. Using human genetics to understand the disease impacts of testosterone in men and women. Nat Med. 2020;26(2):252-258. doi: 10.1038/s41591-020-0751-5 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 21. Laaksonen DE, Niskanen L, Punnonen K, et al. Testosterone and sex hormone–binding globulin predict the metabolic syndrome and diabetes in middle-aged men. Diabetes Care. 2004;27(5):1036-1041. doi: 10.2337/diacare.27.5.1036 [DOI] [PubMed] [Google Scholar]
  • 22. Wei S, Schmidt MD, Dwyer T, Norman RJ, Venn AJ. Obesity and menstrual irregularity: associations with SHBG, testosterone, and insulin. Obesity. 2009;17(5):1070-1076. doi: 10.1038/oby.2008.641 [DOI] [PubMed] [Google Scholar]
  • 23. Kelly DM, Jones TH. Testosterone and obesity. Obes Rev. 2015;16(7):581-606. doi: 10.1111/obr.12282 [DOI] [PubMed] [Google Scholar]
  • 24. Sowers MF, Beebe JL, McConnell D, Randolph J, Jannausch M. Testosterone Concentrations in women aged 25–50 years: associations with lifestyle, body composition, and ovarian status. Am J Epidemiol. 2001;153(3):256-264. doi: 10.1093/aje/153.3.256 [DOI] [PubMed] [Google Scholar]
  • 25. Handelsman DJ, Sikaris K, Ly LP. Estimating age-specific trends in circulating testosterone and sex hormone-binding globulin in males and females across the lifespan. Ann Clin Biochem. 2016;53(Pt 3):377-384. doi: 10.1177/0004563215610589 [DOI] [PubMed] [Google Scholar]
  • 26. Sørensen K, Andersson AM, Skakkebaek NE, Juul A. Serum sex hormone-binding globulin levels in healthy children and girls with precocious puberty before and during gonadotropin-releasing hormone agonist treatment. J Clin Endocrinol Metab. 2007;92(8):3189-3196. doi: 10.1210/jc.2007-0231 [DOI] [PubMed] [Google Scholar]
  • 27. Abdelrahaman E, Raghavan S, Baker L, Weinrich M, Winters SJ. Racial difference in circulating sex hormone-binding globulin levels in prepubertal boys. Metabolism. 2005;54(1):91-96. doi: 10.1016/j.metabol.2004.07.018 [DOI] [PubMed] [Google Scholar]
  • 28. Boyd A, Golding J, Macleod J, et al. Cohort Profile: the ’children of the 90s’--the index offspring of the Avon Longitudinal Study of Parents and Children. Int J Epidemiol. 2013;42(1):111-127. doi: 10.1093/ije/dys064 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 29. Fraser A, Macdonald-Wallis C, Tilling K, et al. Cohort Profile: the Avon Longitudinal study of parents and children: ALSPAC mothers cohort. Int J Epidemiol. 2013;42(1):97-110. doi: 10.1093/ije/dys066 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 30. Liao Z, Vosberg DE, Pausova Z, Tomas P. A shifting relationship between sex hormone-binding globulin and total testosterone across puberty in boys. figshare. Posted June 15, 2022. 10.6084/m9.figshare.19358012.v2 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 31. Khairullah A, Klein LC, Ingle SM, et al. Testosterone trajectories and reference ranges in a large longitudinal sample of male adolescents. PLoS One. 2014;9(9):e108838. doi: 10.1371/journal.pone.0108838 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 32. Choi SW, O’Reilly PF. PRSice-2: Polygenic risk score software for biobank-scale data. GigaScience. 2019;8(7):giz082. doi: 10.1093/gigascience/giz082 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 33. Sudlow C, Gallacher J, Allen N, et al. UK biobank: an open access resource for identifying the causes of a wide range of complex diseases of middle and old age. PLoS Med. 2015;12(3):e1001779. doi: 10.1371/journal.pmed.1001779 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 34. Hanscombe KB, Coleman JRI, Traylor M, Lewis CM. ukbtools: an R package to manage and query UK Biobank data. PLoS One. 2019;14(5):e0214311. doi: 10.1371/journal.pone.0214311 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 35. Hemani G, Zheng J, Elsworth B, et al. The MR-Base platform supports systematic causal inference across the human phenome. Loos R, ed. eLife. 2018;7:e34408. doi: 10.7554/eLife.34408 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 36. Hemani G, Bowden J, Davey Smith G. Evaluating the potential role of pleiotropy in Mendelian randomization studies. Hum Mol Genet. 2018;27(R2):R195-R208. doi: 10.1093/hmg/ddy163 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 37. Li X, Meng X, He Y, et al. Genetically determined serum urate levels and cardiovascular and other diseases in UK Biobank cohort: a phenome-wide mendelian randomization study. PLoS Med. 2019;16(10):e1002937e1002937. doi: 10.1371/journal.pmed.1002937 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 38. Bowden J, Del Greco MF, Minelli C, Davey Smith G, Sheehan N, Thompson J. A framework for the investigation of pleiotropy in two-sample summary data Mendelian randomization. Stat Med. 2017;36(11):1783-1802. doi: 10.1002/sim.7221 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 39. Clogg CC, Petkova E, Haritou A. Statistical methods for comparing regression coefficients between models. AJS. 1995;100(5):1261-1293. doi: 10.1086/230638 [DOI] [Google Scholar]
  • 40. Rosseel Y. lavaan: an R package for structural equation modeling. J. Stat. Softw. 2012;48:1-36. doi: 10.18637/jss.v048.i02 [DOI] [Google Scholar]
  • 41. Hooper D, Coughlan JP, Mullen MR. Structural equation modelling: guidelines for determining model fit. Electron. J. Bus. Res. Methods. 2008;6:53-60. doi: 10.21427/D7CF7R [DOI]
  • 42. Hu LT, Bentler PM, Kano Y. Can test statistics in covariance structure analysis be trusted? Psychol Bull. 1992;112(2):351-362. doi: 10.1037/0033-2909.112.2.351 [DOI] [PubMed] [Google Scholar]
  • 43. Phillips DC. Encyclopedia of Educational Theory and Philosophy. Thousand Oaks, CA: Sage Publications; 2014; 18-19. [Google Scholar]
  • 44. Kaminski BA, Palmert MR. Chapter 65, Human puberty: physiology, progression, and genetic regulation of variation in onset. In: Pfaff DW, Arnold AP, Etgen AM, Fahrbach SE, Rubin RT, eds. Hormones, Brain and Behavior. 2nd ed. Academic Press; 2009:2113-2127. [Google Scholar]
  • 45. Palmert MR, Boepple PA. Variation in the timing of puberty: clinical spectrum and genetic investigation. J Clin Endocrinol Metab. 2001;86(6):2364-2368. doi: 10.1210/jcem.86.6.7603 [DOI] [PubMed] [Google Scholar]
  • 46. Goudswaard LJ, Bell JA, Hughes DA, et al. Effects of adiposity on the human plasma proteome: observational and Mendelian randomisation estimates. Int J Obes. 2021;45(10):2221-2229. doi: 10.1038/s41366-021-00896-1 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 47. Saéz-López C, Rivera-Giménez M, Hernández C, Simó R, Selva DM. SHBG-C57BL/ksJ-db/db: a new mouse model to study shbg expression and regulation during obesity development. Endocrinology. 2015;156(12):4571-4581. doi: 10.1210/en.2015-1677. [DOI] [PubMed] [Google Scholar]
  • 48. Simó R, Sáez-López C, Barbosa-Desongles A, Hernández C, Selva DM. Novel insights in SHBG regulation and clinical implications. Trends Endocrinol Metab. 2015;26(7):376-383. doi: 10.1016/j.tem.2015.05.001 [DOI] [PubMed] [Google Scholar]
  • 49. Edmunds SEJ, Stubbs AP, Santos AA, Wilkinson ML. Estrogen and androgen regulation of sex hormone binding globulin secretion by a human liver cell line. J Steroid Biochem Mol Biol. 1990;37(5):733-739. doi: 10.1016/0960-0760(90)90358-R [DOI] [PubMed] [Google Scholar]
  • 50. Bianchi VE. The anti-inflammatory effects of testosterone. J Endocr Soc. 2018;3(1):91-107. doi: 10.1210/js.2018-00186 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 51. Aydın B, Winters SJ. Sex hormone-binding globulin in children and adolescents. J Clin Res Pediatr Endocrinol. 2016;8(1):1-12. doi: 10.4274/jcrpe.2764 [DOI] [PMC free article] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Data Availability Statement

Some or all datasets generated during and/or analyzed during the current study are not publicly available but are available from the corresponding author on reasonable request.

Articles from The Journal of Clinical Endocrinology and Metabolism are provided here courtesy of The Endocrine Society

RESOURCES