The relationships of sex hormone‐binding globulin, total testosterone, androstenedione and free testosterone with metabolic and reproductive features of polycystic ovary syndrome

Abstract Objective A recent Mendelian randomization study has suggested a causal role for sex hormone‐binding globulin (SHBG), total testosterone and free testosterone in the pathogenesis of polycystic ovary syndrome (PCOS). The aim of this study was to assess the relationships of SHBG, androstenedione, total and free testosterone with the individual metabolic and reproductive features of PCOS. Design Cross‐sectional data in PCOS patients (n=96) prospectively collected in a secondary/tertiary clinic for menstrual cycle disorders. Methods Multivariable regression analyses were conducted to study the associations between SHBG, androstenedione, total and free testosterone with metabolic (BMI, waist circumference, systolic and diastolic blood pressure, total cholesterol, LDL cholesterol, HDL cholesterol, triglycerides and homeostatic model assessment for insulin resistance [HOMA2‐IR]) and reproductive features (menstrual cycle length, antral follicle count, anti‐Müllerian hormone, luteinizing hormone, follicle‐stimulating hormone and Ferriman‐Gallwey score) of PCOS. Results Serum SHBG and free testosterone, but not total testosterone or androstenedione, were significantly associated with BMI, waist circumference, serum triglycerides, HDL cholesterol, LDL cholesterol and HOMA2‐IR. The strength of the associations with serum lipids was reduced after adjustment for BMI, but not for HOMA2‐IR. Total testosterone was significantly associated with antral follicle count. SHBG, total testosterone and androstenedione were significantly associated with serum AMH. Only the strength of the association for SHBG was reduced after adjustment for BMI. Conclusions Serum SHBG is associated with primarily metabolic features, whereas total testosterone and androstenedione are associated with reproductive features of PCOS. These results suggest a differential underlying pathophysiology for the metabolic and reproductive features of PCOS.


| INTRODUC TI ON
Polycystic ovary syndrome (PCOS) is the most common endocrine disorder among premenopausal women, with an estimated prevalence of 10%. 1 Many, but not all women with PCOS exhibit metabolic disturbances, including obesity, insulin resistance, hypertension and dyslipidaemia, and endocrine abnormalities such as an increased ratio of luteinizing hormone (LH) to follicle-stimulating hormone (FSH), and increased anti-Müllerian hormone (AMH) levels. 2 Androgen excess is a fundamental, diagnostic feature of PCOS that is present in approximately half to three quarters of PCOS patients. 3,4 A recent Mendelian randomization study has shown that genetically predicted SHBG, total testosterone and free testosterone levels are associated with the risk of PCOS. 5 This establishes a potential causal role of free testosterone, and its determinants, in the pathogenesis of PCOS. Nevertheless, there is an ongoing discussion on the role of free testosterone in the development of the individual features of PCOS.
The role of androgens in the pathophysiology of PCOS is likely a complex, multifactorial process, driven by genetics, hormonal imbalance and lifestyle factors. [5][6][7] Some have speculated that free testosterone plays a central role in the pathogenesis of all PCOS features, by actively contributing to the arrest of follicular development, theca cell dysfunction, ovarian stromal hyperplasia, abnormal gonadotrophin-releasing hormone (GnRH) secretion and insulin resistance. 2,[8][9][10] Others have argued that hyperandrogenism is merely a consequence of metabolic dysfunction or ovarian and endocrine changes, and does not, in itself, contribute to the pathophysiology of PCOS. 11-14 SHBG may reflect primarily metabolic changes, while total testosterone and androstenedione may reflect reproductive dysfunction. 11 The aim of the present study is, therefore, to study the associations of serum SHBG, androstenedione, total testosterone and free testosterone with the individual metabolic and reproductive features of PCOS. Irregular menstrual cycle was defined as a menstrual cycle length ≥35 days. Hyperandrogenism was defined as a free androgen index (total testosterone (nmol/L) * 100 / SHBG (nmol/L)) >4.5, 16 total testosterone >1.9 nmol/L, androstenedione >9.6 nmol/L, or a Ferriman-Gallwey score ≥4 for women of Caucasian, Black or Mixed ethnicity, and ≥6 for women of Middle-Eastern and Asian ethnicity. 17

| Clinical assessment
All patients filled out questionnaires regarding demographics (age and ethnicity), lifestyle (smoking status and alcohol consumption), self-reported history of acne and hirsutism (defined according to the aforementioned Ferriman-Gallwey score cut-off values) and gynaecological history (length and regularity of menstrual cycle). A regular menstrual cycle was defined as a menstrual cycle <35 days, oligomenorrhea as a menstrual cycle ≥35 days, amenorrhea as no menstrual period during the prior six months, and metrorrhagia as vaginal bleeding at irregular intervals.
Physical examination was performed to determine body mass index (BMI; calculated as body weight [kilograms] divided by length

| Study population
Between March 2017 and February 2020, we retrospectively identified 111 women who fulfilled the diagnostic criteria for PCOS.
Fifteen individuals were excluded because they were pregnant  3.2 | Associations of serum SHBG, androstenedione, total testosterone and free testosterone with metabolic features of PCOS Figure 1 shows the associations of SHBG, androstenedione, total testosterone and free testosterone with nine metabolic features of PCOS. Serum SHBG and free testosterone, but not total testosterone or androstenedione, were associated with BMI ( Figure 1A).  Table 2). In contrast, additional adjustment for BMI reduced the strengths of all associations, whereas addition of HOMA2-IR to the regression models did not have a substantial effect ( Table 2). The strengths of associations did not substantially change when repeating the analyses in fasted individuals only (n = 89) ( Figure S1 and Table S1). Figure 2 shows the relationships of SHBG, androstenedione, total testosterone and free testosterone with six reproductive features of PCOS. None of these were associated with menstrual cycle length (Figure 2A). Total testosterone, but not SHBG, was statistically significantly associated with antral follicle count ( Figure 2B), which was not affected by adjustment for age and BMI ( Table 2). The strength of association was reduced and no longer statistically significant after further adjustment for HOMA2-IR (Table 2). Androstenedione and SHBG were significantly associated with serum AMH ( Figure 2C). The significant association between SHBG and serum AMH was lost after adjustment for age and BMI (Table 2). Total testosterone was significantly associated with serum AMH after adjustment for age and BMI (Table 2). No significant associations were observed for serum LH and FSH ( Figure 2D,E, respectively). Finally, although the direction of the associations of serum SHBG, androstenedione, total testosterone and free testosterone with the Ferriman-Gallwey score were as anticipated, that is inverse for SHBG and positive for androstenedione, total testosterone, and free testosterone, none of these associations were statistically significant ( Figure 2F). The strengths of associations did not substantially change when repeating the analyses in fasted individuals only (n = 89) ( Figure S2 and Table S1).

| DISCUSS ION
The aim of this study was to examine the associations of SHBG, an- was significantly associated with antral follicle count and serum AMH, while androstenedione was significantly associated with serum AMH. Adjustment for BMI substantially reduced the strength of association of free testosterone and SHBG with the metabolic features of PCOS, but hardly affected the associations of total testosterone or androstenedione with the reproductive features of PCOS.
The observed patterns of associations, that is SHBG mainly associates with metabolic features whereas total testosterone and androstenedione associate with reproductive abnormalities of PCOS, are in line with previous observational studies in PCOS. [11][12][13][14] A recent Mendelian randomization study showed that genetically predicted SHBG, total testosterone and free testosterone levels were associated with PCOS risk, 5 which is not surprising given the adoption of hyperandrogenism as a diagnostic criterion of PCOS. 15 However, PCOS is a complex disorder, which comprises several metabolic and ovarian sub-phenotypes. 2,25 The patterns of associations seen in this study, support that different features of PCOS could Analyses were conducted with Z-scores. Beta coefficients should therefore be interpreted as per standard deviation increase in serum SHBG, free testosterone, total testosterone or androstenedione. See methods section.
Abbreviations: AMH, anti-Müllerian hormone; BMI, body mass index; HDL, high-density lipoprotein; HOMA2-IR, homeostatic model assessment for insulin resistance; LDL, low-density lipoprotein; SHBG, sex hormone-binding globulin. Experimental studies have shown that hepatic de novo lipogenesis, which is increased in obesity and insulin resistance, 26 impairs SHBG synthesis in the liver. 27 We recently demonstrated that de novo lipogenesis, assessed by stable isotopes, is inversely associated with serum SHBG levels in women. 28 Hepatic de novo lipogenesis has also been associated with a disadvantageous lipid profile. 29 The reduction in the strength of the association between SHBG and serum lipids after adjustment for BMI in the current study is in line with these previous observations and suggests that BMI is an im-  39,40 yet we did not find a significant association between total testosterone or androstenedione with HOMA2-IR. The mechanism by which insulin resistance influences reproductive features of PCOS therefore deserves further investigation.
In the current study, there was no association between any of the androgen markers and the Ferriman-Gallwey score. Although this may be the result of insufficient statistical power, it also corroborates a recent meta-regression analysis in 6593 women with PCOS demonstrating that free testosterone levels were not associated with clinical hyperandrogenism. 41 Hirsutism is a phenotypic expression of several factors, including the androgen concentration, androgen receptor activity and 5α reductase activity at the pilosebaceous units. 41 Furthermore, the Ferriman-Gallwey score is a subjective measure with significant interobserver variability. 42 Both facets may contribute to the lack of an association between free testosterone levels and clinical hyperandrogenism.  (Figures 1 and 2).
In conclusion, the current observational study shows differential associations of SHBG, androstenedione, total testosterone and free testosterone levels with metabolic and reproductive features of PCOS. These differential associations highlight the heterogeneous nature of PCOS and suggest that the underlying pathways contributing to the features of PCOS are diverse. The combination of SHBG, total testosterone and androstenedione levels may provide information on the primary underlying pathophysiological pathway in women with PCOS.

CO N FLI C T O F I NTE R E S T
The authors declare that there is no conflict of interest.

DATA AVA I L A B I L I T Y
The data that support the findings of this study are available from the corresponding author upon reasonable request.

S U PP O RTI N G I N FO R M ATI O N
Additional supporting information may be found online in the Supporting Information section.