Clustering of PCOS-like traits in naturally hyperandrogenic female rhesus monkeys

Abstract

STUDY QUESTION

Do naturally occurring, hyperandrogenic (≥1 SD of population mean testosterone, T) female rhesus monkeys exhibit traits typical of women with polycystic ovary syndrome (PCOS)?

SUMMARY ANSWER

Hyperandrogenic female monkeys exhibited significantly increased serum levels of androstenedione (A₄), 17-hydroxyprogesterone (17-OHP), estradiol (E₂), LH, antimullerian hormone (AMH), cortisol, 11-deoxycortisol and corticosterone, as well as increased uterine endometrial thickness and evidence of reduced fertility, all traits associated with PCOS.

WHAT IS KNOWN ALREADY

Progress in treating women with PCOS is limited by incomplete knowledge of its pathogenesis and the absence of naturally occurring PCOS in animal models. A female macaque monkey, however, with naturally occurring hyperandrogenism, anovulation and polyfollicular ovaries, accompanied by insulin resistance, increased adiposity and endometrial hyperplasia, suggests naturally occurring origins for PCOS in nonhuman primates.

STUDY DESIGN, SIZE, DURATION

As part of a larger study, circulating serum concentrations of selected pituitary, ovarian and adrenal hormones, together with fasted insulin and glucose levels, were determined in a single, morning blood sample obtained from 120 apparently healthy, ovary-intact, adult female rhesus monkeys (Macaca mulatta) while not pregnant or nursing. The monkeys were then sedated for somatometric and ultrasonographic measurements.

PARTICIPANTS/MATERIALS, SETTING, METHODS

Female monkeys were of prime reproductive age (7.2 ± 0.1 years, mean ± SEM) and represented a typical spectrum of adult body weight (7.4 ± 0.2 kg; maximum 12.5, minimum 4.6 kg). Females were defined as having normal (n = 99) or high T levels (n = 21; ≥1 SD above the overall mean, 0.31 ng/ml). Electronic health records provided menstrual and fecundity histories. Steroid hormones were determined by tandem LC–MS-MS; AMH was measured by enzymeimmunoassay; LH, FSH and insulin were determined by radioimmunoassay; and glucose was read by glucose meter. Most analyses were limited to 80 females (60 normal T, 20 high T) in the follicular phase of a menstrual cycle or anovulatory period (serum progesterone <1 ng/ml).

MAIN RESULTS AND THE ROLE OF CHANCE

Of 80 monkeys, 15% (n = 12) exhibited classifiable PCOS-like phenotypes. High T females demonstrated elevations in serum levels of LH (P < 0.036), AMH (P < 0.021), A₄ (P < 0.0001), 17-OHP (P < 0.008), E₂ (P < 0.023), glucocorticoids (P < 0.02–0.0001), the serum T/E₂ ratio (P < 0.03) and uterine endometrial thickness (P < 0.014) compared to normal T females. Within the high T group alone, anogenital distance, a biomarker for fetal T exposure, positively correlated (P < 0.015) with serum A₄ levels, while clitoral volume, a biomarker for prior T exposure, positively correlated (P < 0.002) with postnatal age. Only high T females demonstrated positive correlations between serum LH, and both T and A₄. Five of six (83%) high T females with serum T ≥2 SD above T mean (0.41 ng/ml) did not produce live offspring.

LARGE SCALE DATA

N/A.

LIMITATIONS, REASONS FOR CAUTION

This is an initial study of a single laboratory population in a single nonhuman primate species. While two biomarkers suggest lifelong hyperandrogenism, phenotypic expression during gestation, prepuberty, adolescence, mid-to-late reproductive years and postmenopause has yet to be determined.

WIDER IMPLICATIONS OF THE FINDINGS

Characterizing adult female monkeys with naturally occurring hyperandrogenism has identified individuals with high LH and AMH combined with infertility, suggesting developmental linkage among traits with endemic origins beyond humans. PCOS may thus be an ancient phenotype, as previously proposed, with a definable pathogenic mechanism(s).

STUDY FUNDING/COMPETING INTEREST(S)

Funded by competitive supplement to P51 OD011106 (PI: Mallick), by P50 HD028934 (PI: Marshall) and by P50 HD044405 (PI: Dunaif). The authors have no potential conflicts of interest.

Introduction

Progress toward finding a cure for polycystic ovary syndrome (PCOS) has been slowed by the lack of a defined pathogenic mechanism, and the absence of an animal model with naturally occurring PCOS. While prevalent (~15%) and highly familial, with adolescent onset, PCOS has no known cause (Dumesic et al., 2015). Women with PCOS exhibit at least two of the following three diagnostic criteria: clinical or biochemical hyperandrogenism, intermittent or absent menstrual cycles and/or polycystic ovaries (Fauser et al., 2012; Conway et al., 2014; Dumesic et al., 2015). Establishing a diagnosis for PCOS in adolescence, however, is contentious because of sufficient overlap between true PCOS and normal, transient adolescent androgen excess with intermittent menstrual cycles and metabolic fluctuations (Carmina et al., 2010; Witchel et al., 2015). Accompanying PCOS sequelae include hypersecretion of LH and antimullerian hormone (AMH), insulin resistance and obesity, together with increased risks for metabolic syndrome, Type 2 diabetes, gestational diabetes, endometrial hyperplasia and cancer (Dumesic et al., 2015). The pathophysiology of PCOS thus extends well beyond ovarian dysfunction to include hypothalamic dysregulation and, of more concern, increased risks for metabolic diseases and cancer associated with long-term morbidity and mortality (Wild et al., 2010; Dumesic and Lobo, 2013).

Contemporary understanding of PCOS considers its etiology to be polygenic, with developmental origins likely preceding puberty (Dumesic et al., 2015; Dunaif, 2016). At least 21 replicated candidate genes have been identified, regulating gonadotropin secretion and action, extracellular matrix development and a variety of common cellular functions (Azziz, 2016; Dunaif, 2016). Combined, however, these candidate genes account for <10% of the estimated 70% heritability of PCOS, implying a considerable epigenetic contribution to the phenotypic expression of PCOS (Hayes et al., 2015; Azziz, 2016). The most comprehensive epigenetic phenotypes that mimic PCOS arise from animal models that employ experimentally induced excess of fetal testosterone (T) to permanently induce PCOS-like reproductive and metabolic traits in female rodents (Foecking et al., 2008; Caldwell et al., 2015; Cimino et al., 2016; Walters, 2016), sheep (Padmanabhan and Viega-Lopez, 2014; Cardoso et al., 2015) and nonhuman primates (Abbott et al., 2013; Dumesic et al., 2014). The absence of a naturally occurring PCOS animal model, however, has hindered progress towards understanding pathogenic mechanisms that may bestow both genetic and epigenetic contributions to the etiology of PCOS.

Macaque monkeys, including rhesus (Macaca mulatta) and cynomolgus (Macaca fascicularis), share over 90% of their genome with humans, and provide close parallels in terms of fetal, infant and juvenile development, reproduction, metabolism and aging (Abbott et al., 2013). Experimental induction of T excess in fetal female rhesus macaques during early-to-mid gestation provides the most reproducible emulation of PCOS in women (Abbott et al., 2016). Together, these findings in female rhesus macaques raise the possibility that naturally occurring gestational T excess, as an early developmental mechanism, induces permanent epigenetic modifications of the female macaque genome, thereby emulating both genetic and epigenetic components of PCOS in women. Apart from humans, moreover, female macaques and other Old World nonhuman primates are unique in sharing similar menstrual cycle attributes, such as a relatively lengthy follicular phase, making them particularly vulnerable to a PCOS-like phenotype (Barnett and Abbott, 2003).

This study, therefore, tests the hypothesis that adult females with high levels of T in a laboratory population of rhesus monkeys will exhibit PCOS-like traits. We initially focus on only one of three laboratory populations (initial focal site: Wisconsin National Primate Research Center, WNPRC; additional sites: Yerkes NPRC and Wake Forest Primate Center) that form part of a larger study of genetic and epigenetic underpinnings of naturally occurring PCOS-like traits in laboratory nonhuman primates. This initial report investigates the coexistence of several PCOS-like traits, namely reproductive and metabolic dysfunction in hyperandrogenic adult female rhesus monkeys at WNPRC, to characterize phenotypic similarities among nonhuman primates and apparently healthy, hyperandrogenic women who have insufficiently high T to qualify for a PCOS diagnosis. Identifying such a population of high T female nonhuman primates for the first time suggests that androgen-dependent, PCOS-like traits are endemic to primates other than humans and likely have ancient origins (Azziz et al., 2011; Corbett and Morin-Papeun, 2013; Fessler et al., 2016). Such nonhuman primates provide a novel avenue to explore the genetic and epigenetic bases for such PCOS-like traits, along with their selective advantages, within well-characterized breeding pedigrees.

Materials and Methods

Animals

Ethics statement

The Institutional Animal Care and Use Committee of the Graduate School of the University of Wisconsin-Madison approved all study procedures. Care and housing of the monkeys were in accordance with the recommendations of the Guide for the Care and Use of Laboratory Animals and Animal Welfare Act and its subsequent amendments.

Study subjects

The 120 adult female rhesus monkeys (M. mulatta) used in this study were maintained at the WNPRC under standard husbandry conditions (Nicol et al., 2014), and fed ~100 kcal/kg body weight of a low fat, high-fiber diet (Purina monkey chow 5038, Ralston Purina) with occasional supplementation of fresh fruit and vegetables (~5–10% of daily calories). The chow formulation provides 13% of calories as fat, 17% as protein and 70% as carbohydrate, with ~1.2 g fiber/kg body weight (~4 times recommended daily fiber consumption for women of reproductive age; Institute of Medicine, National Academy of Sciences). Female subjects were of prime reproductive age (7.2 ± 0.1 years, mean ± SEM), represented a typical spectrum of adult body weight (8.3 ± 0.3 kg; maximum 12.5, minimum 4.6 kg), were apparently healthy, ovary-intact adults that were not pregnant or nursing, and were usually pair-housed with another adult female, a social housing condition conducive to regular menstrual cyclicity (Weinbauer et al., 2008).

Study design

After an overnight fast, circulating serum concentrations of selected pituitary, ovarian and adrenal hormones, together with insulin and glucose levels, was determined in a single, morning blood sample obtained between 07:00 and 10:00 h. Serum collection was limited to September through May in order to avoid summer oligomenorrhea (Dumesic et al., 1997, 2002; Eisner et al., 2000). Immediately following blood sample collection, females were sedated with ketamine HCl (10 mg/kg, i.m.) for ultrasonographic examination of the lower abdomen and somatometric measurements.

Menstrual and parity records

Incidence of menses, birth of live offspring, miscarriage, stillbirth, C-section and number of days housed with adult males were obtained from the WNPRC electronic health record system. Menstrual intervals were averaged for each female from ≥5 years of age (onset of regular ovulatory menstrual cycles) (Resko et al., 1982) until they entered the study, excluding when they were pregnant or nursing an infant. Parity was determined by at least one live offspring birth. An estimate of the time taken to conceive was derived from the percentage number of days each female (≥4 years of age; age at first conception) (Coe and Shirtcliff, 2004; Maestripieri, 2005; Gagliardi et al., 2007) was housed with an adult male (≥5 years of age; age at male sexual maturity) (Stephens and Wallen, 2013; Ramaswamy et al., 2014) until the end of 2012 (end of sample collection). For example, two females at 8 and 6 years of age, respectively, at the time of this study, with identical durations of 100 days housing with males would translate into 6.8% and 13.7% of percentage days available for male housing, respectively. The 6-year old would thus have had relatively more frequent opportunities to conceive than the 8-year old. Housing with adult males, and thus opportunities to conceive, are determined by WNPRC Animal Services in order to sustain colony numbers, genetic diversity and to provide sufficient numbers of monkeys for research. Research project assignment, alone, can also necessitate no housing with males. Female parity and opportunities to conceive were, therefore, determined by factors outside our study design.

Ultrasonographic ovarian and uterine measurements

Ultrasonographic measures were made using a General Electric Medical Systems LOGIQ e ultrasound machine with 12 L-RS, wide band linear array probe (5–13 MHz imaging frequency). Digitized calipers were employed to obtain the diameter of each ovary from static images of each largest ovarian cross section and uterine endometrial thickness from the longitudinal image of the entire uterus providing the largest endometrial depth (between the endometrial and myometrial interfaces, combining each endometrial layer on either side of the uterine lumen).

Somatometric measurements

Immediately following ultrasonographic examination, somatometric measurements were performed with the monkey supine or in right recumbency, as previously validated for rhesus monkeys (Eisner et al., 2003; Abbott et al., 2012). Each female was assessed for body weight, crown-rump length (for BMI), abdominal circumference at the umbilicus, total abdominal skinfolds, hip circumference, clitoral volume and anogenital distance. Abdominal (waist) and hip circumference, as well as anogenital distance, were measured with a cloth tape measure to the nearest 1 mm; abdominal skinfold thickness was determined using a Lange caliper at ~5 cm above, below, right and left of the umbilicus (totaling all values); and clitoral volume was assessed using the same caliper to obtain length, width and height of the clitoris protruding outside the skin. In order to measure anogenital distance, each female was moved into right lateral recumbency, and the distance between the centers of the anus and the urethra were recorded.

Hormone determinations and HOMA-IR determination

Serum concentrations of steroid hormones were measured at the Optimized Analytical Solutions Laboratories, LLC (Durham, NC) using HPLC/tandem mass spectrometry Agilent MassHunter Workstation Data Acquisition for Triple Quad B.03.01 (B2065) and Agilent MassHunter Quantitative Analysis for QQQ (B.04.00/Build 4.0.225.0), as previously described (Abi Salloum et al., 2015). The lower detection limits were: androstenedione (A₄), 0.1 ng/ml; cortisol and dehydroepiandrosterone (DHEA), 0.5 ng/ml; corticosterone, 11-deoxycortisol and deoxycorticosterone (DOC), 0.05 ng/ml; estradiol (E₂) and estrone, 5 pg/ml; 17-hydroxyprogesterone (17-OHP) and progesterone, 0.2 ng/ml and T, 0.02 ng/ml. The intra-assay coefficients of variation for these hormone were as follows: A₄, 3%; cortisol, 11%; DHEA, 8%; corticosterone, 16%; 11-deoxycortisol, 5%; DOC, 19%; E₂, 2%; estrone, 4%; 17-OHP, 2%; progesterone, 4% and T, 8%.

Serum concentrations of LH and FSH were determined by in-house radioimmunoassays (RIAs) (Dumesic et al., 2002); insulin was also determined by RIA (Linco Research Inc., St. Charles, MO) (Eisner et al., 2000), and glucose was measured by the glucose oxidase method (Kemnitz et al., 1994). Consistent with previous findings that human AMH EIA (enzymeimmunoassay) recognizes monkey AMH (Lee et al., 1994; Atkins et al., 2014), there were no problems of cross reactivity in using a human-reagent EIA (Ansh Labs, Webster, TX). All assays were validated for rhesus monkeys. Intra-assay coefficients of variation, respectively, were as follows: LH 7.0%, FSH 4.2%, AMH 5.0% and insulin 2.0%.

Homeostasis assessment model-insulin resistance (HOMA-IR = [insulin (mU/ml) × glucose (mmol/l)]/22.5) was used to determine insulin resistance, as previously employed for rhesus macaques (Lee et al., 2011; He et al., 2013; Qi et al., 2015).

PCOS-like phenotypes

The National Institutes of Health (NIH) Evidence-Based Methodology PCOS Workshop (National Institutes of Health, 2012) recommended that Rotterdam 2003 diagnostic criteria are used to classify PCOS phenotypes in women: Type A, HA (hirsutism or serum total T ≥2 SD above the local population mean for non-PCOS women) + OD (intermittent/absent ovulatory menstrual cycles) + PCOM (polycystic ovary morphology); Type B, HA + OD; Type C, HA + PCOM and Type D, OD + PCOM. Rhesus monkey equivalent criteria for PCOS-like phenotypes were similar to those previously established for prenatally androgenized PCOS-like female monkeys (Abbott et al., 2009): Type A, HA (serum T ≥1 SD above the mean T value derived from all 120 WNPRC rhesus monkeys studied) + OD (>34 days between menstrual cycles) + PCOM (serum AMH ≥10 ng/ml, mean AMH value for WNPRC high T female monkeys); Type B, HA + OD; Type C, HA + PCOM and Type D, OD + PCOM . Serum AMH of ≥10 ng/ml was used as a conservative surrogate measure of PCOM because the majority of women with such high AMH levels have PCOS (Tal., et al., 2014), there is a strong association between serum AMH and other criteria of PCOS in women, including OD and HA (Laven et al., 2004; Cassar et al., 2014), and there are positive correlations between AMH and antral follicle count in women with and without PCOS (DeWailly et al., 2011; Alebic et al., 2015) as well as in female macaque monkeys (Dumesic et al., 2009; Appt et al., 2010).

Statistical analysis

All results were expressed as mean ± SEM. Parameters not normally distributed were log transformed (continuous variables), arcsine transformed (% values limited to 0–100%) or square root transformed after addition of 1 (# days housed with males). Treatment group comparisons were conducted using Student's t-test or Fisher's Exact Test (parity variables). Regression analysis was used to examine associations between variables. Statistical significance was determined as P < 0.05.

Results

PCOS-like diagnostic traits

By establishing a criterion for female monkeys with high T levels as ≥1 SD above (0.31 ng/ml) a mean serum T level of 0.21 ng/ml for 120 adult female rhesus monkeys at WNPRC, we identified 21 females with high T and 99 with normal T levels (Fig. 1). Limiting subsequent analyses to those females in the pre-ovulatory, follicular phase or anovulatory period (serum progesterone < 1 ng/ml), thereby excluding post-ovulatory, luteal contributions to circulating T, reduced the numbers of females to 20 high T and 60 normal T subjects. This exclusion process is outlined in Fig. 2. The incidence of classifiable PCOS-like phenotypes among these 80 female monkeys (n = 12, 15%) is remarkably similar to that found in unselected populations of women with PCOS (Table I). Of high T female monkeys, 45% (n = 9) exhibited a PCOS-like phenotype (Types A–C) in comparison to 5% (n = 3; Type D only) of normal T females. In addition, five of six (83%) high T females with serum T ≥2 SD (0.41 ng/ml) above the overall mean did not produce live offspring (Fig. 1). Two of these five females exhibited a classifiable PCOS-like phenotype.

Figure 1.

Individual serum T levels (ng/ml) in all normal T (<0.31 ng/ml, red-filled circles, n = 99) and high T (≥0.31 ng/ml, blue-filled circles, n = 21) adult female rhesus monkeys, illustrating the mean value (dotted line, 0.21 ng/ml) for all females combined (n = 120), 1 SD (0.31 ng/ml) above the mean (solid line), 2 SD (dashed line, 0.41 ng/ml) above the mean.

Figure 2.

Flow diagram depicting study selection of monkeys.

Table I.

Classification of PCOS-like phenotypes in adult female rhesus monkeys at WNPRC^a and in unselected human populations^b.

Female population	PCOS-like/PCOS phenotype (% of PCOS individuals)
	Type A	Type B	Type C	Type D
WNPRC female rhesus monkeys (n = 12)	25.0	8.3	41.7	25.0c
Unselected human populationsb	25.4	19.3	35.3	20.0

^aOverall incidence of PCOS-like monkeys with Rotterdam 2003 criteria (Types A + B + C + D combined) is 15%, and according to NIH 1992 criteria (Types A + B only) is 5%.

^bAll human data are derived from Lizneva et al. (2016). Worldwide range of incidence of PCOS among women in unselected human populations with Rotterdam 2003 criteria (Types A + B + C + D combined) is 4–21%, and according to NIH 1992 criteria (Types A + B only) is 4–7%.

^cOnly one of these three normal T female monkeys exhibited serum T (0.30 ng/ml) and LH (3.2 ng/ml) levels approaching the elevated levels found in high T females (T ≥0.31 ng/ml, LH mean value 3.8 ng/ml).

Traits associated with PCOS

Somatometric and reproductive-related parameters

High and normal T female monkeys were similar in age, body weight, BMI, abdominal circumference, waist-to-hip ratio and total abdominal skinfolds (Table II). Both female groups also exhibited equivalent durations of menstrual cycles, as well as incidence of intermittent cycles and parity (Table III). While parous females in both groups required an average of 36–40 days pairing with males to conceive each live offspring, non-parous high T females failed to do so after approximately twice that duration (Table III). In contrast, and mostly due to WNPRC colony management practices, non-parous, normal T females, spent the least number of days (P < 0.03) with adult males (~22 vs. 99–133 days; Table III), with 50% never encountering adult males. Non-parous high T females were not similarly deprived of conception opportunities, with 88% experiencing housing with adult males. Miscarriage (n = 4), stillbirth (n = 3) and C-section (n = 1) were rare and limited to normal T females.

Table II.

Selected somatic parameters (mean ± SEM) of adult female rhesus monkeys with normal (<0.31 ng/ml) and high (≥0.31 ng/ml) circulating T concentrations.

Characteristic	Normal T females (n = 60)	High T females (n = 20)	P-value
Somatic parameters
Age (years)	8.7 ± 0.4	7.6 ± 1.8	0.132
Body weight (kg)	7.4 ± 0.2	7.4 ± 0.2	0.837
BMI (kg/m2)	33.6 ± 0.9	34.2 ± 1.7	0.732
Abdominal circumference (cm)	36.3 ± 0.9	37.3 ± 1.3	0.582
Total abdominal skinfolds (mm)	22.6 ± 2.0	24.5 ± 3.7	0.650
Waist-to-hip ratio	1.0 ± 0.1	1.0 ± 0.1	0.190

Table III.

Selected reproductive parameters (mean ± SEM) of adult female rhesus monkeys with normal (<0.31 ng/ml) and high (≥0.31 ng/ml) circulating T concentrations.

Characteristic	Normal T females (n = 20)	High T females (n = 60)	P-value
Reproduction-related parameters
Menstrual cycle duration (days)	34.4 ± 1.8	31.5 ± 1.5	0.362
Intermittent menstrual cycles (%)	23%	20%	0.999
Largest ovarian diameter (mm)	7.3 ± 0.3	7.3 ± 0.7	0.985
Uterine endometrial thickness (mm)	5.4 ± 0.2	6.8 ± 1.0	0.014
Anogenital distance (mm)	18.3 ± 0.4	17.3 ± 0.7	0.173
Clitoral volume (mm3)	340 ± 36	409 ± 88	0.391
Parity parameters
Parous females* (%, n)	85% (40/47)	65% (11/17)	0.089
Parous females:	(n= 40)	(n= 11)
% Days available housed with male	3.7 ± 0.4	3.3 ± 0.5	0.156
# Days with male	133.1 ± 19	99.8 ± 22	0.256
# Live births per parous female	4.0 ± 0.4	3.4 ± 0.7	0.524
% Days with male per live offspring	1.4 ± 0.3	1.4 ± 0.3	0.642
# Days per live offspring	40.5 ± 6.6	36.2 ± 8.2	0.736
Non-parous females:	(n = 20)	(n = 9)
% Days available housed with male	1.1 ± 0.4	3.3 ± 1.3	0.037
# Days housed with male	22.6 ± 7.5	73.1 ± 30.0	0.030

*n for both normal and high T females = # females paired with male >20 days. Bold P-values are significant.

Ultrasonographic visualization of largest ovarian diameter yielded comparable measurements for both female groups (Table III). In contrast, uterine endometrial thickness was greater (P < 0.014) in high compared to normal T females (Table III). Ultrasonographic resolution was insufficient, however, to reliably quantify ovarian antral follicle numbers and thus could not be used to identify polycystic ovaries. Both female groups otherwise demonstrated comparable genital-related dimensions.

Hormonal parameters

Compared to normal T females, those with high T demonstrated elevated serum levels of LH, AMH, A₄, 17-OHP, E₂, cortisol, 11-deoxycortisol and corticosterone, but not of FSH, DHEA, progesterone or E₁ (Table IV). High T females also exhibited elevated serum hormone ratios for LH to FSH, T to E₂, 17-OHP to progesterone, 17-OHP to DHEA, A₄ to DHEA and 11-deoxycortisol to cortisol. The remaining ratios were comparable across the two female groups (Table IV).

Table IV.

Ovarian, adrenal and pituitary serum hormone concentrations and ratios (mean ± SEM), as well as selected glucoregulatory serum parameters in adult female rhesus monkeys with normal (<0.31 ng/ml) and high (≥0.31 ng/ml) circulating T concentrations.

Hormonal parameter	Normal T females (n = 60)	High T females (n = 20)	P-value
Ovarian and adrenal hormones
AMH (ng/ml)	7.2 ± 0.7	10.0 ± 1.1	0.043
T (ng/ml)	0.2 ± 0.1	0.4 ± 0.1	0.0001
A4 (ng/ml)	0.4 ± 0.1	0.7 ± 0.1	0.0001
DHEA (ng/ml)	19.8 ± 1.8	23.9 ± 1.8	0.240
17-OHP (ng/ml)	1.0 ± 0.2	2.8 ± 0.8	0.008
Progesterone (ng/ml)	0.2 ± 0.1	0.2 ± 0.1	0.937
E2 (pg/ml)	96 ± 10	153 ± 28	0.023
Estrone (pg/ml)	20 ± 2	28 ± 5	0.103
Cortisol (ng/ml)	337 ± 21	453 ± 34	0.007
11-deoxycortisol (ng/ml)	1.1 ± 0.1	3.2 ± 0.7	0.0001
Corticosterone (ng/ml)	0.6.4 ± 0.6	9.8 ± 1.6	0.020
Steroid hormone ratios
T/E2	2.7 ± 0.3	4.2 ± 0.8	0.022
A4/DHEA	0.03 ± 0.01	0.04 ± 0.01	0.005
A4/T	2.0 ± 0.1	1.8 ± 0.1	0.082
A4/estrone	26 ± 3	37 ± 5	0.062
Estrone/E2	0.25 ± 0.02	0.23 ± 0.05	0.680
17-OHP/progesterone	6.8 ± 1.2	17.3 ± 5.9	0.010
17-OHP/DHEA	0.06 ± 0.02	0.14 ± 0.06	0.042
17-OHP/A4	2.3 ± 0.5	3.4 ± 1.0	0.332
17-OHP/T	4.5 ± 1.1	6.0 ± 1.7	0.513
17-OHP/11-deoxycortisol	1.0 ± 0.2	1.2 ± 0.5	0.654
Progesterone/corticosterone	0.04 ± 0.01	0.03 ± 0.01	0.587
11-dexoxycortisol/cortisol	0.0034 ± 0.0004	0.0065 ± 0.001	0.002
Corticosterone/cortisol	0.02 ± 0.01	0.02 ± 0.01	0.404
Pituitary hormones
LH (ng/ml)	2.9 ± 3.1	3.8 ± 0.8	0.036
FSH (ng/ml)	1.9 ± 0.2	1.7 ± 0.2	0.511
LH:FSH ratio	1.7 ± 0.1	2.2 ± 0.2	0.046
Glucoregulatory factors
Insulin (uU/ml)	11.2 ± 1.0	13.1 ± 2.9	0.445
Glucose (mg/dl)	86.4 ± 2.5	88.5 ± 7.9	0.741
HOMA-IR	2.5 ± 0.2	3.0 ± 0.7	0.378

AMH, antimullerian hormone; DHEA, dehydroepiandrosterone; 17-hydroxyprogesterone, HOMA-IR, Homeostasis assessment model-insulin resistance. Bold P-values are significant.

Serum T positively correlated with A4 and E₂ in both female groups, but additional correlations were highly group specific (Table V). In normal T females, alone, positive correlations were found between T and DHEA, and between T and the glucocorticoids cortisol, 11-deoxycortisol and corticosterone. In high T females, in contrast, positive correlations were found between T and progesterone, LH and FSH, and between A4 and LH (Table V). Excluding an outlying T value (P < 0.01) from a high T female, the only androgen-related correlations remaining in the high T group were between T and A₄ (r = 0.46, P < 0.049) and T with LH (r = 0.49, P < 0.033). Such outlier T levels (>0.6 ng/ml), nevertheless, have since been found in additional female rhesus monkeys (unpublished results) now designated as ‘high T females’. There were no correlations between T and insulin, nor between AMH and T, E₂, LH, FSH and insulin in either female group (Table V).

Table V.

Associations between T, AMH and selected hormonal parameters in adult female rhesus monkeys with normal (<0.31 ng/ml) and high (≥0.31 ng/ml) circulating T concentrations.

Hormone associations	Normal T females (n = 60)	High T females (n = 20)
Association between T and:
AMH	r = 0.21, P = 0.102	r = 0.15,P = 0.527
A4	r = 0.73, P < 0.0001	r = 0.73, P < 0.0004
E2	r = 0.34, P< 0.009	r = 0.49, P < 0.030*
DHEA	r = 0.33, P < 0.011	r = 0.15, P = 0.517
17-OHP	r = 0.10, P = 0.426	r = 0.33, P= 0.158
Progesterone	r = 0.07, P = 0.612	r= 0.51, P < 0.021*
Cortisol	r = 0.52, P < 0.0001	r = 0.23, P = 0.329
11-deoxycortisol	r = 0.34, P< 0.008	r = 0.05, P = 0.824
Corticosterone	r = 0.44, P < 0.0005	r = 0.32, P = 0.173
LH	r = 0.18, P = 0.175	r = 0.89, P < 0.0001
FSH	r = 0.12, P = 0.356	r = 0.68, P < 0.0009*
Insulin	r = 0.10, P = 0.435	r = 0.02, P = 0.933
Association between A4 and:
LH	r = 0.06, P = 0.658	r = 0.71, P < 0.0006*
FSH	r = 0.08, P = 0.523	r = 0.40, P = 0.085
Insulin	r = 0.12, P= 0.380	r = 0.28, P = 0.237
Association between AMH and:
E2	r = 0.07, P = 0.588	r = 0.07, P = 0.782
LH	r = 0.11, P = 0.394	r = 0.03, P = 0.913
FSH	r = 0.15, P = 0.258	r = 0.07, P = 0.763
Insulin	r = 0.03, P = 0.801	r = 0.40, P = 0.081

*Not significant when highest T value omitted from high T group. Bold P-values are significant.

Fasting serum insulin and glucose levels, as well as HOMA-IR, were comparable in high and normal T females (Table IV).

Associations among hormone levels, reproductive traits and somatic parameters

Anogenital distance, a biomarker for fetal T exposure during early-to-mid gestation, correlated positively with circulating levels of A₄ in high T females, alone (Table VI). Neither T nor A₄ demonstrated any additional correlations with reproductive or somatic parameters. Clitoral volume, a biomarker for T exposure during fetal and/or postnatal life, correlated positively with postnatal age in high T females, alone (Table VII).

Table VI.

Associations between selected androgens and reproductive and somatic parameters in adult female rhesus monkeys with normal (<0.31 ng/ml) and high (≥0.31 ng/ml) circulating T concentrations.

Association between	Normal T females (n = 60)	High T females (n = 20)
T and:
Anogenital distance	r = 0.08, P = 0.564	r = 0.40, P = 0.080
Clitoral volume	r = 0.11, P = 0.406	r = 0.12, P = 0.616
Ovarian diameter	r = 0.26, P = 0.068	r = 0,05, P = 0.857
Age	r = 0.07, P = 0.589	r = 0.09, P = 0.702
BMI	r = 0.10, P = 0.447	r = 0.37, P = 0.110
Waist-to-hip ratio	r = 0.04, P = 0.745	r = 0.13, P = 0.582
Skinfold thickness	r = 0.05, P = 0.716	r = 0.17, P = 0.466
Uterine thickness	r = 0.01, P = 0.999	r = 0.05, P = 0.830
A4 and:
Anogenital distance	r = 0.23, P = 0.079	r = 0.54, P < 0.015
Clitoral volume	r = 0.18, P = 0.174	r = 0.11, P = 0.655
Ovarian diameter	r = 0.13, P = 0.378	r = 0.16, P = 0.595

Bold P-values are significant.

Table VII.

Associations between non-androgen hormonal, reproductive and somatic parameters in adult female rhesus monkeys with normal (<0.31 ng/ml) and high (≥0.31 ng/ml) circulating T concentrations.

Association between	Normal T females (n = 60)	High T females (n = 20)
AMH and:
Ovarian diameter	r = 0.06, P = 0.655	r = 0.11, P = 0.718
Age	r = 0.42, P < 0.002	r = 0.01, P = 0.986
BMI	r = 0.30, P < 0.019	r = 0.04, P = 0.884
Waist-to-hip ratio	r = 0.21, P = 0.110	r = 0.21, P = 0.371
Skinfold thickness	r = 0.20, P = 0.136	r = 0.03, P = 0.903
Insulin and:
Age	r = 0.05, P = 0.733	r = 0.03, P = 0.903
BMI	r = 0.34, P < 0.0001	r = 0.26, P = 0.268
Waist-to-hip ratio	r = 0.23, P = 0.086	r = 0.28, P = 0.234
Skinfold thickness	r = 0.27, P < 0.044	r = 0.53, P < 0.016
Uterine thickness	r = 0.13, P = 0.324	r = 0.55, P < 0.013
Uterine thickness and:
Progesterone	r = 0.32, P < 0.012	r = 0.17, P = 0.468
E2	r = 0.16, P = 0.217	r = 0.03, P = 0.892
Age	r = 0.27, P < 0.040	r = 0.42, P = 0.062
BMI	r = 0.35, P < 0.006	r = 0.53, P < 0.017
Waist-to-hip ratio	r = 0.21, P = 0.107	r = 0.62, P < 0.004
HOMA-IR	r = 0.11, P = 0.419	r = 0.56, P < 0.010
BMI and:
Waist-to-hip ratio	r = 0.64, P < 0.0001	r = 0.71, P < 0.0006
Skinfold thickness	r = 0.63, P < 0.0001	r = 0.55, P < 0.013
Age and:
Anogenital distance	r = 0.32, P < 0.012	r = 0.13, P = 0.585
Clitoral volume	r = 0.14, P = 0.288	r = 0.67, P < 0.002

Bold P-values are significant.

In normal T females, alone, expected negative correlations were demonstrated between circulating levels of AMH and both age and BMI (Table VII). No associations were found in either female group between AMH and ovarian diameter, insulin or waist-to-hip ratio (Table VII).

While uterine endometrial thickness correlated positively with BMI in both female groups, other associations were distinct to each female type. In normal T females, positive correlations included circulating progesterone levels and age, whereas in high T females, positive correlations involved fasting insulin, HOMA-IR and waist-to-hip ratio (Table VII).

Fasting insulin levels demonstrated expected positive correlations with BMI and abdominal skinfold thickness in normal T females, but only with skinfold thickness in high T females (Table VII).

Discussion

In the present study of WNPRC adult female rhesus monkeys, we identified 15% with PCOS-like phenotypes using Rotterdam equivalent criteria and 5% by NIH. The incidence of PCOS-like phenotypes in WNPRC female monkeys closely align with worldwide PCOS incidence found in unselected human populations using Rotterdam (4–21%) or NIH (4–7%) criteria (Lizneva et al., 2016).

Hyperandrogenism occurs naturally in women, including apparently healthy women, female athletes and ballet dancers who do not exhibit diagnostic criteria for PCOS (Sjaarda et al., 2014; Allen, 2016; Lagowska and Kapczuk, 2016). Our study expands understanding of such naturally occurring female hyperandrogenism to nonhuman primates. Our findings add to a previous case report of a single, naturally occurring, hyperandrogenic, PCOS-like adult female cynomolgus macaque monkey with endometrial hyperplasia (Arifin et al., 2008). Our high T female monkeys exhibited several abnormalities commonly associated with PCOS, including elevated circulating levels of LH, AMH, A₄ and 17-OHP compared to normal T females (<1 SD of the population mean for T), as well as increased uterine endometrial thickness. Analogous endometrial development was found in the previous case report of a naturally occurring PCOS-like monkey (Arifin et al., 2008), as well as in PCOS women during anovulatory periods (Cheung, 2001), mid-follicular phase of menstrual cycles (Iatrakis et al., 2006) and immediately before hCG injection during ovarian stimulation for IVF (Amir et al., 2007). Most (83%) female monkeys with very high serum levels of T (≥2 SD above the population mean or ≥0.41 ng/ml), a Rotterdam 2003 criterion for a PCOS diagnostic T level (Rotterdam group, 2004; Fauser et al., 2012), demonstrated absent fertility or fecundity despite ample opportunity to conceive. Thus T-related impairment or delay to female fertility/fecundity occurs at the high end of a continuum of circulating T levels in adult female primates, including humans.

Naturally occurring hyperandrogenism and co-occurring traits

Our selected cutoff (≥0.31 ng/ml) for high T female monkeys is remarkably similar to that of ≥0.28 ng/ml employed to identify apparently healthy women in the top quartile of normal serum T levels (Sjaarda et al., 2014), and both monkey as well as human studies utilized ‘gold standard’ LC–MS-MS methodology for steroid determinations. The women with top quartile T levels, however, did not exhibit sufficiently high T levels or experience infrequent menstrual cycles to achieve a PCOS diagnosis and, therefore, were considered part of the non-PCOS population (Sjaarda et al., 2014). In the human study, comparison of women in the top T quartile demonstrated increased serum ratio of LH to FSH, increased serum levels of AMH and E₂, as well as increases in intermittent menstrual cycles and diminished or delayed fecundity, compared to women in the lowest T quartile (Table VIII). Such naturally hyperandrogenic women thus not only closely emulate our female monkeys with high T regarding serum T levels, but also in terms of co-expressed PCOS-like phenotype (Tables I and VIII). Modestly higher cutoffs for serum T levels by LC–MS-MS have been used to identify PCOS women, ranging from 0.35 to 0.52 ng/ml (Barth et al., 2010; Legro et al., 2010; Gambineri et al., 2013; Salameh et al., 2014; Bui et al., 2015; Tosi et al., 2016). Concurrent elevations in circulating concentrations of androgens, particularly A₄, in women with PCOS (O'Reilly et al., 2014; Pasquali et al., 2016; Saito et al., 2016), combined with increased 17-OHP and glucocorticoids, implicate hyper-steroidogenesis accompanying such female hyperandrogenism. As our female monkeys with high T also demonstrate elevated serum ratios of T/E₂ and A₄/DHEA, a generalized androgenic bias may co-occur within such hyper-steroidogenic profiles that include contributions from the adrenal glands (Gourgari et al., 2016).

Table VIII.

Comparison of PCOS-like traits in apparently healthy, high T women and adult female rhesus monkeys.

Trait	High T womena ≥0.28 ng/ml (Quartile 4, upper quartile)	High T female monkeys ≥ 0.31 ng/ml (≥1 SD of mean)
LH or LH/FSH ratio	+b	+c
AMH	+b	+c
Intermittent menstrual cycles	+b	−
Diminished fecundity	+b	+c
E2	+b	+c
BMI	−	−
Fasting insulin and glucose	−	−

^aSjaarda et al. (2014).

^bP< 0.05 vs. Quartile 1, lowest T quartile for women.

^cP < 0.05 vs. normal T female monkeys.

Hyperandrogenism, particularly of ovarian origin, is the cornerstone of PCOS pathophysiology (Legro et al., 1998; Conway et al., 2014; Dumesic et al., 2015; Azziz et al., 2016; Rosenfield and Ehrmann, 2016), with coexisting adrenal hyperandrogenism in 25–50% of cases (Azziz et al., 2013; Luque-Ramírez and Escobar-Morreale, 2016). In PCOS, hyperandrogensim primarily arises from an intrinsic defect in ovarian theca cells (Nelson et al., 1999; McAllister et al., 2014; Rosenfield and Ehrmann, 2016), and excessive proliferation of theca cells contributes to enlarged ovarian stromal volume (Magoffin, 2005) and overall enlarged ovaries (Haney et al., 1986; Dewailly et al., 2014). Maximum ovarian diameter by sonographic determination in high T female monkeys, however, was comparable to that of normal T females, and typical for young adult females (Bishop et al., 2009). Increased circulating levels of 17-OHP, together with elevated serum ratios of 17-OHP/progesterone and 17-OHP/DHEA in high T versus normal T females, nevertheless suggest a relatively greater contribution of 17-OHP to circulating steroid levels in high T female monkeys and an ovarian source of hyperandrogenism (Rosenfield and Ehrmann, 2016).

Diminished fertility/fecundity in high T females

High T female monkeys with T ≥2 SD of the population mean demonstrated 17% fertility/fecundity compared to 85% fertility/fecundity in female monkeys with normal T, strongly implicating high T levels with subfertility, analogous to findings in comparably hyperandrogenic, apparently healthy women (Franks et al., 1985; Sjaarda et al., 2014). Prevalence of anovulatory cycles, despite normal menstrual regularity (Carmina and Lobo, 1999), or impaired uterine endometrial decidualization (Piltonen et al., 2015), two features of subfertility in women with PCOS, may contribute towards negative selection against female monkey reproduction at very high T levels. Since uterine endometrium expresses insulin receptors (Guidice, 2006) that can allow insulin excess to diminish uterine receptivity for embryo implantation (Schulte et al., 2015), monkeys with high T concentrations and increased endometrial thickness (depth) in proportion to fasting insulin levels, HOMA-IR, and waist-to-hip ratio may experience subfertility related to insulin resistance. In support of this, the insulin sensitizer, metformin, has been reported to reverse endometrial hyperplasia in a PCOS patient unresponsive to progesterone treatment (Session et al., 2003).

Impaired regulation of reproductive neuroendocrinology

High T in women with PCOS disrupts homeostatic E₂/progesterone-mediated negative feedback regulation of pituitary LH secretion (Pastor et al., 1998; Burt Solorzano et al., 2012), likely from enhanced hypothalamic GnRH release, causing LH hypersecretion in most cases (Franks, 1995; Conway et al., 2014; Dumesic et al., 2015). Administration of the androgen receptor antagonist, flutamide, to PCOS women, however, restores the sensitivity of circulating LH levels to negative feedback-mediated suppression from administration of E₂ (Daniels and Berga, 1997) with or without luteal phase progesterone levels (Eagleson et al., 2000), suggesting that extant adult hyperandrogensim may be a key pathogenic factor in inhibiting hypothalamic expression of progesterone receptor (Foecking et al., 2008). There may, however, be a component of LH hypersecretion that is independent of high circulating levels of T because administration of flutamide, alone, fails to diminish the high frequency of episodic release of LH in women with PCOS (Eagleson et al., 2000). While the molecular basis for LH hypersecretion has yet to be explored in our naturally hyperandrogenic female monkeys, positive associations in high T females alone between T and LH in an environment of elevated E₂, are consistent with hyperandrogenic disruption of LH negative feedback.

PCOS-like ovarian parameters in female monkeys with high T

The combination of elevated serum levels of AMH with those of E₂ in female monkeys with high T suggests relatively increased ovarian follicle recruitment likely comprising increased numbers of small-to-medium sized antral follicles and pre-antral follicles (Stubbs et al., 2005; Pellatt et al., 2007). Granulosa cell-derived AMH levels in the circulation are a reliable biomarker for the numbers of growing, non-dominant follicles in the ovary (de Vet et al., 2002; Weenen et al., 2004; Pellatt et al., 2007, 2010). Increased levels of AMH indicate expanded numbers of pre-antral to medium-sized antral follicles in ovarian follicle populations and are thus commonly manifest in women with polycystic ovaries (Pigny et al., 2003; Rotterdam group, 2004; DeWailly et al., 2014). Normally, AMH acts to inhibit FSH action, aromatase and follicle selection (Jonard and Dewailly, 2004; Pellatt et al., 2011) to homeostatically regulate ovarian folliculogenesis (Dewailly et al., 2016). In a hyperandrogenic environment, however, ovarian AMH release from ovarian follicle granulosa cells is enhanced (Pellatt et al., 2007) and abnormally elevated (Pigny et al., 2003; Dewailly et al., 2016) potentially disrupting optimal folliculogenesis (Stubbs et al., 2005; Kissell et al., 2014; Dewailly et al., 2016).

Developmental programming of high T adult female monkeys

Our results are consistent with our pathogenic hypothesis of genetic- or epigenetic-determined androgen excess predisposing to the development of PCOS-like traits, including LH excess, elevations in AMH and infertility (Abbott et al., 2002, 2016; Dumesic et al., 2007). Gestational exposure of female monkeys to fetal male levels of T and A₄ (Resko et al., 1987; Abbott et al., 2008) induces PCOS-like reproductive and metabolic traits in adulthood (Abbott et al., 1998, 2005, 2013). Circulating A₄ levels, interestingly, may provide a more robust and reliable indicator of prevailing androgen excess than T levels alone (O'Reilly et al., 2014; Pasquali et al., 2016). Given this consideration, we found that circulating levels of A₄ in our adult female monkeys with high T positively correlate with anogenital distance, a reliable fetal biomarker for gestational exposure to androgens in primates (Goy et al., 1988; Wallen and Lloyd, 2011; Abbott et al., 2012), likely due to local 5alpha-reductase metabolism of androgens to dihydrotestosterone during the ‘in-utero programing window’ (Dean and Sharpe, 2013). In studies of young Spanish women, longer anogenital distance is associated with high adult levels of T (Mira-Escolano et al., 2014a), greater numbers of non-dominant antral follicles (Mendiola et al., 2012) and menstrual irregularity in their mothers before conception (Mira-Escolano et al., 2014b). More recently, in a North American population, anogenital distance of newborn daughters of women with PCOS exceeds that found in daughters of women without PCOS (Barrett and Hoeger, 2016). The androgen excess exhibited by our female monkeys with high T may thus have its origins during fetal life, possibly in the later part of the ‘in-utero programming window’ since there is no marked genital virilization accompanying the androgen–anogenital distance correlation.

In addition, clitoral volume correlates with greater postnatal age in female monkeys with high T, suggesting previous prolonged exposure to androgen excess either before (Goy and Phoenix, 1972) or after (Brown et al., 1999) birth. In a Turkish study, the clitoris was elongated in women with PCOS compared to those without PCOS (Kosus et al., 2016), but two earlier Italian studies failed to find such a difference (Battaglia et al., 2008; Morotti et al., 2013).

Limitations and strengths of the present study

This is an initial study in a single laboratory population of a specific nonhuman primate species. While numerically large for a study of nonhuman primates, the current investigation identified relatively few female monkeys with high T (21 out of 120). Although two biomarkers suggest lifelong hyperandrogenism, further demonstration of high T levels from mid-gestational fetal life through prepuberty and adolescence into mid-to-late reproductive to postmenopause years requires future longitudinal studies. Additional study limitations include the inability of abdominal sonography to accurately detect ovarian antral follicles of 0.5–2 mm in diameter, the absence of serial serum progesterone determinations to detect ovulation and the lack of LH pulsatility testing to assess hypothalamic contribution to high LH. In addition, greater in-depth metabolic assessment of high T females requires dynamic testing of insulin-glucose homeostasis, and dual-energy x-ray absorptiometry/magnetic resonance imaging quantification of visceral versus subcutaneous adiposity.

Strengths of our study include (i) the identification of a cluster of PCOS-like traits in female monkeys with high T levels, with (ii) use of ‘gold standard’ LC–MS-MS steroid determinations to identify such females by a monkey population-determined cutoff value for high T (0.31 ng/ml) resembling that in women and (iii) application of controlled and comparable development, diet, physical and social environment among female subjects. Identifying naturally occurring female hyperandrogenism in a defined nonhuman primate (NHP) population closely related to humans, therefore, holds promise for determining molecular genetic and/or epigenetic contributions linking high T to reliable co-expression of PCOS-like traits.

Authors’ roles

D.H.A., M.E.W., S.E.A., D.A.D., K.W. and J.E.L. contributed to study conception and design. A.K.E. and B.H.R. collected and processed samples and NHP measurements, and organized, summarized and performed preliminary analysis and interpretation of measurement and hormone data. K.C.L. conceived, designed and directed LC–MS-MS analyses and interpreted relevant data output and quality control. D.H.A., B.H.R., S.E.A., D.A.D., K.W. and J.E.L. contributed to data analysis and interpretation. D.H.A. wrote the manuscript, and D.A.D., J.E.L., M.E.W., K.W., B.H.R. and S.E.A. edited manuscript drafts. All authors approved the final manuscript.

Funding

A competitive supplement to P51 OD011106 (PI: Mallick), by P50 HD028934 (PI: Marshall) and by P50 HD044405 (PI: Dunaif).

Conflict of interest

None declared.