Fetal sex and the development of gestational diabetes mellitus in polycystic ovarian syndrome gravidae

Abstract

BACKGROUND:

Polycystic ovarian syndrome is characterized by elevated androgens and is a well-known risk factor for the occurrence of gestational diabetes mellitus. Androgens (particularly dehydroepiandrosterone-sulfate) are crucial for the development and characteristics of the male reproductive tract during fetal life, and fetal dehydroepiandrosterone-sulfate enters the placenta where it is metabolized and functions as an estrogen substrate. Given this unique sex-specific relationship with androgens and the association of serum dehydroepiandrosterone-sulfate concentration with insulin resistance, we hypothesized that metabolic comorbidities in pregnancy might differ by fetal sex in gravidae with polycystic ovarian syndrome, notably in those with infertility.

OBJECTIVE:

This study aimed to evaluate the data in a large population-based database to explore if fetal sex was significantly associated with gestational diabetes mellitus in gravidae with infertility and polycystic ovarian syndrome after controlling for confounders.

STUDY DESIGN:

This study was designed to evaluate the risk for the occurrence and rates of gestational diabetes mellitus among gravidae with infertility and a history of polycystic ovarian syndrome. We used a 2-hospital, single academic institution database comprising more than 30,000 subjects enrolled from September 2011 to June 2021 to identify all gravidae with diagnoses of infertility and polycystic ovarian syndrome at the time of delivery and to compare them with gravidae who lacked these comorbidities. Data on covariates, including but not limited to maternal age, body mass index, fetal sex, race, ethnicity, presence or absence of hypertensive disease, and presence or absence of gestational diabetes were identified. Unadjusted and adjusted odds rations were calculated.

RESULTS:

We found a statistically significant association between fetal female sex and the development of gestational diabetes mellitus in gravidae with polycystic ovarian syndrome (odds ratio for female vs male, 2.13; 95% confidence interval, 1.06–4.32; P=.03). After adjusting for potential confounders identified in our univariate analyses, there continued to be a statistically significant association between female fetuses and the development of gestational diabetes mellitus (adjusted odds ratio for female vs male, 2.10; 95% confidence interval, 1.04 −4.41; P=.04). In contrast, there was no significant association between fetal sex and the development of gestational diabetes mellitus in our similar analysis of gravidae without infertility and polycystic ovarian syndrome (P=.99).

CONCLUSION:

Although the origin of gestational diabetes mellitus is multifactorial, we found that female fetal sex is associated with gestational diabetes mellitus in gravidae with infertility and polycystic ovarian syndrome but not in their comparative controls. Further research on the molecular mechanisms driving the association between female fetuses and the development of gestational diabetes mellitus in the context of maternal polycystic ovarian syndrome is warranted.

Introduction

Polycystic ovarian syndrome (PCOS) is thought to be the most prevalent endocrine disorder found in females, affecting 6% to 12% (as many as 5 million) of American reproductive-aged females.¹ Persons with PCOS are often insulin resistant and have excess levels of systemic androgens, which can lead to anovulation and oligomenorrhea.² Consequently, PCOS is a common cause of female infertility.² When persons with PCOS do become pregnant (gravidae), they are at a higher risk for complications during pregnancy, including miscarriage and gestational diabetes mellitus (GDM).³ Gravidae with both PCOS and GDM can have altered metabolic function that persists well into the postpartum period and the interpregnancy interval.³ Patients with PCOS and GDM can have a 3.5-fold higher risk for continued impaired glucose metabolism at 18 months postpartum than controls without PCOS but with GDM.⁴ In addition, GDM can adversely affect the developing fetus, leading to macrosomia and adiposity in newborns,⁵ as well as to impaired glucose tolerance and obesity in childhood.⁶ Thus, it is important to understand the relationship between PCOS and GDM given the potential long-term maternal and offspring effects.

As evidence for a direct link between PCOS and the development GDM becomes stronger, the basis of this relationship and the role that infertility plays must be examined. Pregnant persons with both a history of PCOS and infertility necessitating treatment with assisted reproductive technologies have a 2-fold increased risk for developing GDM irrespective of body mass index (BMI).⁷ A history of GDM is associated with a 2.4-fold increased risk for developing metabolic syndrome—a constellation of metabolic disorders including abdominal obesity, glucose intolerance or insulin resistance, dyslipidemia, and hypertension⁸ that is estimated to affect one-quarter of infertile persons with PCOS in some studies.⁹ Thus, a better understanding of the associations among PCOS, GDM, and infertility is necessary to inform shared decision-making with patients and to offer timely interventions with the potential to improve maternal-fetal outcomes.

Although often overlooked, the placenta is an endocrine organ that may link PCOS and GDM. Recent evidence suggests that the complex feto-placental hormonal milieu is fetal sex dependent.¹⁰ Androgens, particularly dehydroepiandrosterone-sulfate (DHEAS), are crucial for the development and the characteristics of the male reproductive tract during fetal life. Fetal DHEAS undergoes placental metabolism to estrogen and also functions as an estrogen substrate. During pregnancy, the placenta is the primary site of estrogen synthesis, but placental estrogen production can only be achieved with the input of DHEAS from the fetal and maternal adrenal cortex.¹¹ The fetal adrenal gland—through its production of DHEAS and other androgens— and the placenta are important regulators of feto-maternal endocrine function and communication during pregnancy.¹¹ In fact, variations in maternal DHEAS levels have previously been associated with insulin resistance.¹² Thus, we propose that the fetus is potentially interacting with the maternal hormonal milieu through the placenta in a way that is fetal sex dependent and could be contributing to the development of GDM in some but not all gravidae with PCOS. Gravidae with PCOS have higher levels of circulating DHEAS than gestational age–matched controls.¹³ We hypothesized that the differences in the rates of GDM development in gravidae with PCOS and infertility may be modulated by fetal sex–driven differential placental metabolism of sex hormones.

Materials and Methods

Study design, subjects, and database characteristics

This retrospective cohort study, which was approved by the Baylor College of Medicine Institutional Review Board (H-26364), was designed to evaluate the rates of GDM in gravidae with infertility and PCOS according to fetal sex.

PeriBank is a comprehensive, 2-hospital institutional delivery database and biobank curated at the Baylor College of Medicine that focuses on detailed clinical data and accompanying specimens collected at the time of delivery. A detailed description of PeriBank has been published previously.¹⁴ At the time of admission, the PeriBank research personnel enrolled patients after written consent was obtained. Electronic medical records, prenatal records, and in-person interviews were used to obtain up to 4900 variables of clinical data on more than 39,000 deliveries during this period. The quality of the data was determined by regular verification of a subset of the inserted clinical data and by at least 1 board-certified maternal-fetal medicine physician scientist (K.M.A.) as previously published.¹⁴ No additional data were retrospectively abstracted from the patients’ electronic medical records for this study.

Inclusion criteria

Subjects included in this study were enrolled in the PeriBank database between September 2011 and June 2021 (n=39,324 entries). Each entry represented a delivery of a live born neonate. Each gravid person’s first-order pregnancy or first pregnancy with an infertility diagnosis was included in the analysis, and any subsequent higher order pregnancies were excluded so that each entry in our analysis represented a single individual gravid person. All deliveries of same-sex twins were counted as a single delivery or entry. Given our desire to analyze obstetrical outcomes on the basis of fetal sex, deliveries of fraternal twins (ie, male-female twin pairs) were excluded. All individuals with a history of fertility problems and PCOS were identified (Figure).

FIGURE. Exclusion criteria.

Flow chart depiction of the study exclusion criteria to achieve study sample in the nested cohort. Asterisk implies the index pregnancy and/or index pregnancy with infertility.

In total, 228 patients met these inclusion criteria. Of these 228 patients, 1 was excluded because data on the fetal sex were not available. Thus, 227 gravid persons were identified for analysis. The gravidae were divided into 2 groups based on the sex of the fetus (ie, male, n=110 or female, n=117). The primary outcome was presence or absence of GDM, and the comparator was the offspring sex. No data were available on GDM diagnosis for 1 male neonate and was excluded. Thus, our final study sample comprised 226 patients.

A separate group of patients without PCOS or infertility was used as controls for comparison with our PCOS group to see if there was an association between fetal sex and the development of GDM in patients without these comorbidities. Thus, all deliveries in the database from September 2011 to May 2014 without a history of fertility problems and without a history of PCOS were identified (n=7305) were grouped on the basis of fetal sex (ie, male, n=3728 or female, n=3577) and evaluated for the development of GDM.

Outcome measures and data analysis

The primary outcome of this study of gravidae with infertility and PCOS was the presence or absence of a GDM diagnosis in relation to the primary exposure of fetal sex. Our institutions used the Rotterdam criteria¹⁵ to diagnose PCOS; however, data on where the patient’s PCOS diagnosis was made and according to which criteria (ie, the National Institute of Health vs Rotterdam criteria) were not available for this analysis. Within our PeriBank database, infertility is a diagnosis that is coded for, and patients with an infertility diagnosis are easily aggregated. From this smaller subset of patients, we were able to review each individual patient’s medical problem list and identify those with a PCOS diagnosis.

Given our hypothesis that the development of GDM in gravidae with PCOS may be modulated by differential placental metabolism of sex hormones based on fetal sex, our groups of interest were gravidae with PCOS who delivered male neonates and those who delivered female neonates. These groups of interest were compared using chi-square or Fisher’s exact tests for categorical variables and Wilcoxon rank-sum tests for non-normally distributed continuous variables. Linear regression was performed, and the generalized linear models were used for the estimation of odds ratios (ORs), adjusted ORs (aOR), and the corresponding 95% confidence intervals for binomial outcome variables. An aOR was calculated after correcting for possible confounders identified on the univariate analysis. Variables with a P value <.1 and those known and important potential confounders were included in the linear regression. Adjustments were made for maternal age and BMI at first prenatal visit. A P value <.05 was considered statistically significant. Results are reported as mean±standard deviation unless otherwise specified. All analyses were performed in SAS (version 9.4) (SAS Institute Inc, Cary, NC).

Results

During the study period, complete data were captured for 39,324 deliveries in our PeriBank database. A total of 226 gravidae with a history of PCOS and fertility challenges were identified in our study sample. The maternal characteristics and pregnancy outcomes were similarly matched for deliveries of female neonates and male neonates in patients developed GDM, whereas 184 did not have an associated GDM diagnosis. Of the gravidae who developed GDM, 28 (66.7%) gave birth to female neonates and 14 (33.3%) gave birth to male neonates. Of the gravidae who did not develop GDM, 89 (48.4%) gave birth to female neonates, whereas 95 (51.6%) gave birth to male neonates. Thus, female fetal sex was identified as a statistically significant predisposition for the development of GDM (P=.032) in our study subjects (Table 2).

TABLE 2.

Gestational diabetes mellitus and fetal sex

Fetal sex	Gestational diabetes mellitus		P value .032a
	Yes (n=42)	No (n=184)
Female	28 (66.7)	89 (48.4)
Male	14 (33.3)	95 (51.6)

Data are presented as number (percentage).

^aIndicates a P value <.05.

Maternal age at delivery (33.7± 4.7 years vs 32.6±3.3 years for those with female and male neonates respectively; P=.054) and maternal BMI at delivery (34.7±0.61 kg/m² vs 33.2±0.72 kg/m² for female and male neonates, respectively; P=.059) did not reach statistical significance. Maternal BMI at first prenatal visit was identified as a potential confounder in the univariate analysis (30±0.63 vs 28.4±0.75; P=.038) and was adjusted for in the regression models. There was a statistically significant association between fetal female sex and the development of GDM in gravidae with infertility and PCOS (OR for female vs male neonate, 2.13; 95% CI, 1.06–4.32; P=.03). When adjusting for prenatal BMI, there continued to be a statistically significant association between female fetuses and the development of GDM (aOR for female vs male neonate, 2.10; 95% CI, 1.04–4.41; P=.04). The aOR for GDM development in female vs male neonates with adjustments for maternal age and prenatal BMI was found to be 1.89 (95% CI, 0.92–4.01; P=.08).

As a control for the comparison of risk interactions, we examined GDM rates stratified by fetal sex in a random cohort of 7305 gravidae without infertility and PCOS who developed GDM (male fetuses n=3728; female fetuses n=3577). Of note, 1656 of the 3577 (46.3%) gravidae who delivered female neonates developed GDM and 1726 of the 3728 (51.0%) gravidae who delivered male neonates developed GDM (P=.99). In this cohort, there was no statistically significant relationship between fetal sex and GDM.

Comment

Principal findings

Elucidating the relationships among infertility, PCOS, fetal sex, GDM, and the subsequent clinical outcomes are critical for our understanding of the metabolic disease processes in pregnancy. Many factors are likely involved. Our findings suggest that among these factors, female fetal sex may play a key role in the development of GDM in individuals with infertility and PCOS. Specifically, we found that gravidae with infertility with PCOS were approximately 2-fold more likely to develop GDM if they were carrying a female fetus than if they were carrying a male fetus. This association persisted even after adjusting for maternal BMI at the first prenatal visit. In contrast, there was no significant association between fetal sex and the development of GDM in the analysis of gravidae without PCOS and infertility.

It is worth noting that multiple gestations confer additional risks for adverse outcomes,¹⁷ and thus, gravidae with more than 1 fetus are inherently at greater risk for developing GDM regardless of preexisting metabolic conditions, such as PCOS.³ For the purposes of hypothesis testing in our study, we included only same-sex fetal twins in our analysis who were counted as a single delivery or entry. However, we also performed a post hoc analysis of all qualifying gravidae in our study sample of 42 gravidae that developed GDM and found that none had twin gestations. Mills et al³ have previously reported the rates of GDM among women with PCOS with twin pregnancies and similarly observed that PCOS and not multiples was a greater risk factor for GDM.³ The role of fetal sex in twin pregnancies and the development of GDM in patients with PCOS remain unknown.

Interestingly, in the regression model that included both maternal BMI at the first prenatal visit and maternal age as possible confounders, our results approached but did not reach statistical significance (aOR, 1.89; 95% CI, 0.92–4.01; P=.08). One explanation could be that our sample size was not powered to detect differences when multiple predictors were included in the regression model. Given that our study population had an infertility diagnosis, the mean age of the gravidae (33.7±4.7 years for the female neonate group and 32.6±3.3 years for male neonate group) may be slightly higher than that of a PCOS population without infertility. However, a recent analysis of 27,586 pregnant persons with PCOS identified the mean age at delivery to be 30.86±5.38 years.¹⁸ Thus, the mean age at delivery of our patient population (ie, older than 30 years) may be reflective of the PCOS population as a whole rather than a finding intrinsic to our study. Further studies are needed to see if our results are generalizable to younger PCOS patients.

Although we have identified an association between fetal sex and the development of GDM in our study cohort, the mechanism by which fetal sex imparts a risk for the development of GDM is unclear. Pregnancy itself is an insulin resistant state, and gravidae with GDM manifest greater insulin resistance than nondiabetic counterparts because their beta cells are unable to secrete enough insulin to compensate for the physiological insulin resistance of pregnancy.¹⁹ Previously, effects of maternal physiology on fetal metabolism have been recognized, but this relationship may be bidirectional²⁰ and may potentially be differentially affected by the sex of the fetus.

Results

The clinical effects of fetal sex on metabolic dysfunction have not been characterized well. Some studies have demonstrated that the presence of a male fetus is associated with increased odds of GDM,²⁰ a higher postprandial glycemia,²⁰ and a greater need for maternal insulin therapy.²¹ In contrast, the findings from other studies align with our findings, suggesting that gravidae with female fetuses are more susceptible to increased insulin resistance during pregnancy.^22,23 Levels of diabetogenic hormones of human placental lactogen, which is secreted by the placenta into maternal circulation and serves as a physiological antagonist to insulin, have been reported to be higher in pregnancies with female neonates than those with male neonates.^24,25 A possible fetal sex dimorphism in the maternal endocrine environment that allows for differential effects of certain GDM risk factors is plausible.

Clinical implications of our findings

It is important to be cognizant of this relationship for clinical management. Retnakaran et al²⁶ found that those with GDM who delivered a female neonate showed early progression of type 2 diabetes (T2DM). Long-term follow-up studies will be required to evaluate if there is a difference in the long-term health outcomes among gravidae with PCOS who developed GDM that are associated with fetal sex. Identification of these patients who may be at higher risk for cardiovascular and metabolic dysfunction later in life is crucial for providing optimal and timely screening and treatment.

Although our study was not designed to examine the relationship among fetal sex, the use of metformin—an insulin sensitizing agent commonly prescribed preconceptionally to PCOS patients— and GDM, it is a relevant consideration. We have previously shown that there was no difference in rates of GDM in a population of gravidae with PCOS based on metformin usage (P=.12).²⁷ However, metformin has been shown to reduce the levels of sex-hormone binding globulin (SHBG) in gravidae with PCOS who are carrying female but not male fetuses.²⁸ Lower preconception and early-pregnancy levels of SHBG have been associated with higher rates of GDM.⁵ Additional research is needed to assess if metformin could potentially have any sex-specific effects on the development of GDM.

Research implications of our findings

Future studies are needed to identify the mechanisms by which the maternal glucose metabolism is differentially affected by the sex of the fetus. The placenta, which serves as a maternal-fetal conduit, may serve as the regulatory vessel for enacting these sex-specific physiological changes. Recent work has suggested that placental androgen signaling is dependent on the fetal sex.²⁹ Placental villous tissue from female fetuses cultured in the presence of inflammation showed a reduction in the expression of insulin-like growth factor 1 receptor (IGF1R), an androgen-mediated downstream target.²⁹ These findings support the idea that sex-specific differences in placental androgen signaling could be driving clinically meaningful differences in intrauterine growth outcomes.²⁹ We argue that a better understanding of the sex-specific alterations to placental function and the feto-hormonal environment is a crucial piece in potentially explaining sex-specific fetal growth outcomes and the development of maternal metabolic disease.

Strengths and limitations to our study

The strengths of this study include the use of a large, population-based database with manually abstracted data instead of using diagnosis codes or patient-reported data with a risk of recall bias. Additional strengths include data collection performed by trained multilingual research staff and regular database audits to ensure the validity of the data obtained from patient interviews. In addition, we were able to distinguish between gestational and T2DM, assuring that we have minimized the risk for diagnostic misclassification. Although robust, our study does have limitations. This analysis is retrospective in nature and consequently, all limitations specific to such a research design are present. As an observational study, it is important to note that our findings cannot address causality and can only infer an association between outcomes. Most subjects delivered at a private hospital, and thus, our results may not be representative of all patient populations. In addition, our primary analysis focused on gravidae with infertility and PCOS. Information on the severity of infertility was not available for this analysis. It is possible that baseline metabolic derangements in our infertile PCOS cohort may play a role in reproductive failure that may additionally predispose them to GDM in pregnancy; these subjects would not be anticipated to be in our study cohort because subject data were limited to those with a viable (or potentially viable) pregnancy outcome. Future studies are needed to see if our findings are generalizable to gravidae with PCOS without an associated infertility diagnosis, which was outside the scope of this study. Finally, data on the patients baseline laboratory values for prepregnancy measurements of hemoglobin A1c, DHEAS, androstenedione, testosterone, SHBG, and anti-Müllerian hormone were not available in our delivery database. Future prospective studies are needed to evaluate how these levels may vary with GDM development in PCOS patients.

Conclusion

Our study is unique in that it specifically explores the relationship between GDM development and fetal sex in patients with infertility and PCOS specifically. This study demonstrated that fetal sex may be a previously unrecognized factor that impacts maternal glucose homeostasis in gravidae with PCOS. Further research on the physiology driving this association between female fetuses and the development of GDM is warranted.

TABLE 1.

Comparison of the baseline characteristics

Characteristics	Female neonate (n=117)	Male neonate (n=110)	P value
Maternal age at delivery (y)	33.7±4.7	32.6±3.3	.054
Gestational age (wk)	38.4±2.4	38.5±2.6	.3
Gravida	2.3±0.12	2.2±0.12	.3
Ethnicity
Hispanic	20 (25.6)	23 (20.9)	.4
Non-Hispanic	87 (74.4)	87 (29.1)
Hospital
County	6 (5.1)	3 (2.7)	.35
Private	111 (94.9)	107 (97.3)
Pregnancy type
Spontaneous	63 (54.8)	55 (50.9)	.56
ART	52 (45.2)	53 (49.1%)
Comorbidities
Hypothyroidism	25 (21.4)	18 (16.4)	.34
Hyperthyroidism	1 (0.8)	0	.99
Type 2 diabetes mellitus	4 (3.4)	5 (4.6)	.74
Gestational diabetes mellitusa	28 (23.9)	14 (12.8)	.032b
Hypertensive disease
None	91 (77.8)	89 (80.9)	.35
Gestational hypertension	11 (9.4)	13 (11.8)
Preeclampsia	15 (12.8)	8 (7.3)
BMI (kg/m2)
BMI at first prenatal visit	30.0±0.63	28.4±0.75	.038b
BMI at delivery	34.7±0.61	33.2±0.72	.059

Data are presented as mean±standard deviation or number (percentage) unless otherwise specified.

ART, assisted reproductive technology; BMI, body mass index.

^aDiagnosed based on ≥2 elevated values for the 3-hour 100 g oral glucose tolerance test based on Carpenter or Coustan values¹⁶

^bIndicates a P value <.05.

AJOG MFM at a Glance.

Why was this study conducted?

This retrospective cohort study was conducted to explore the significance of the relationship between fetal sex and the occurrence of gestational diabetes mellitus in gravidae with polycystic ovarian syndrome and a history of infertility.

Key findings

After adjusting for potential confounders, gravidae with polycystic ovarian syndrome and infertility were at a significantly higher risk for the development of gestational diabetes mellitus when carrying a female fetus than when carrying a male fetus.

What does this add to what is known?

Recent studies suggest that there is a potential for maternal-fetal interactions in the occurrence of dysregulated glucose metabolism (such as gestational diabetes), presumptively because of androgen differences between male and female fetuses. Although the precise mechanism is unknown and our study cannot directly address it, our findings suggest that future lines of investigation may benefit from considering sex-specific alterations in the placental function and feto-hormonal milieu.