Ultrasonographic features of ovarian morphology capture nutritional and metabolic influences on the reproductive axis: implications for biomarker development in ovulatory disorders

Abstract

Ultrasonographic imaging of ovarian morphology is used widely to inform reproductive health status in women. Metabolic disturbances induced by a negative energy balance (e.g., undernutrition) or positive energy balance (e.g., overnutrition, obesity) are known to drive or exacerbate reproductive dysfunction. Whether the utility of ultrasonographic metrics of ovarian morphology could be extended as biomarkers that detect and monitor the integration of metabolic and reproductive dysfunction is an emerging research area, and recent evidence is discussed. We note that unique variations in ovarian morphology emerge across the adiposity spectrum and highlight the potential for reproductive and metabolic “tipping points” upon which such morphological variations may be detected on ultrasonography.

Graphical abstract

Introduction

Ovarian ultrasonography is a widely accessible non-invasive imaging modality used in the point-of-care assessment of gynecological health, particularly in the context of ovulatory dysfunction. Standard morphological evaluations include measurements of ovarian volume (OV) and an estimation of the number antral follicles present, expressed either as the number per single ovary (FNPO), sum of both ovaries (AFC), or those observed in a single cross-sectional image (FNPS). These metrics are used diagnostically and prognostically to inform on menstrual cycle stage, ovarian reserve, response to ovulation induction, and the presence of certain ovulatory disorders, such as polycystic ovary syndrome (PCOS). Organization of follicles within the ovary (peripherally versus heterogeneously distributed) and characteristics of the ovarian stroma (area and echogenicity) are more qualitative indicators that can reflect on abnormalities in tissue hypertrophy, folliculogenesis and/or steroidogenesis – albeit their performance as indicators of reproductive health outcomes are less defined.

Serial ovarian ultrasonography has enabled an understanding of antral folliculogenesis, which occurs in a wavelike process in women with regular ovulatory cycles [1]. Briefly, an ovulatory cycle consists of the recruitment (2-5mm stage) and synchronous growth of a cohort of follicles two to three times during an inter-ovulatory interval. A subset of these follicles undergo continued growth to the selectable stage (6-9mm) from which a dominant follicle (≥10mm) is selected for preferential growth and ovulation [1]. Cross-sectional analyses of ovarian morphology therefore enable a snapshot into the antral stages of folliculogenesis and any potential deviation stemming from hypothalamic-pituitary-ovarian (HPO) axis dysfunction. Because antral follicles produce hormones in a stage-specific manner and are under varying degrees of local versus systemic regulation that coordinate folliculogenesis, particular consideration of follicle size populations (abundance, depletion, relative proportions), could be used to infer the degree to which certain stages or events in folliculogenesis are affected.

The emerging hypothesis considered herein is that under- and overnutrition alongside associated metabolic derangements influence the trajectory of folliculogenesis in ways that manifest as distinct and ultrasonographically detectable alterations in ovarian morphology. Recent evidence pertaining to the utility of ovarian morphology on ultrasonography to serve as a biomarker for the detection and progression of metabolic and reproductive dysfunction is discussed. We posit that unique variations in ovarian morphology emerge across the adiposity spectrum and are suggestive of physiological “tipping points” wherein the integration of metabolic and reproductive dysfunction presents as differences in ovarian morphology on ultrasonography.

Undernutrition and Ovarian Morphology

Low energy availability, as in the case of undernutrition, inhibits gonadotropin production thereby suppressing the main driver of antral follicle development and steroidogenesis [2]. A spectrum of ovarian morphology has been described in underweight and low-weight women. At the extreme, women with severe manifestations of anorexia nervosa exhibit small amorphous ovaries with no ultrasonographically detectable follicles, consistent with global suppression of the HPO axis [3,4]. In these cases, progressive weight gain coincided with increases in OV, follicle number, and follicle diameter, which implies a progressive resolution of ovarian dysmorphology with restoration of energy balance and weight gain [3,4]. By contrast, the ovaries of underweight or normal weight women experiencing weight loss-related amenorrhea [a type of functional hypothalamic amenorrhea (FHA)] have been described as multi-follicular (MFO), hallmarked by a normal or slightly enlarged size with a modest accumulation of antral follicles (mostly 4-10mm) distributed throughout the ovary [3–5]. These early descriptions of MFO are reflected to some degree in recent studies of women with FHA [6,7], and are consistent with some degree of follicle arrest and failure to progress to ovulation. However, a more recent and limited focus to AFC [8] and OV [6,8] when evaluating FHA has generated concern of overlap in MFO with morphological and functional markers [i.e. anti-Miillerian hormone (AMH)] of polycystic ovarian morphology (PCOM) whose pathophysiology aligns more closely with obesity and related metabolic derangements (described below). In general, MFO are smaller than polycystic ovaries [5,8], and their stromal characteristics and relative proportions of small to medium size follicles would be expected to differ owing to the different pathophysiological mechanisms at play. Ultimately, there is currently a missed opportunity to use new and improved ultrasound imaging technology to reliably delineate between these two conditions and leverage morphological differences to stage disease progression or resolution in the context of undernutrition.

Overnutrition and Ovarian Morphology

Obesity resulting from a persistent positive energy balance, or overnutrition, is now recognized as a leading cause of menstrual cycle irregularity, reduced fecundity and anovulatory infertility [9]. Obesity coincides with numerous metabolic derangements such as dyslipidemia, insulin resistance (IR), compensatory hyperinsulinemia, and altered adipokine production – all of which are implicated in altered gonadotropin and ovarian hormone production, disrupted folliculogenesis, and chronic anovulation [9,10]. Animal models in particular, have been essential in establishing proof-of-concept and identifying mechanistic pathways by which diet-induced obesity impairs reproductive function and drives changes in ovarian morphology [11–13]. However, in women, the effects of obesity and its concomitant metabolic defects on the HPO axis are complex and to some degree exhibit competing mechanisms. On one hand, obesity can be a state of hypogonadotropic hypogonadism as evidenced by altered hypothalamic or pituitary feedback regulation, as well as suppressed gonadotropin secretion and ovarian steroidogenesis [14–16]. Yet, in the presence of PCOS, obesity is well known to exacerbate the clinical presentation of this disorder which includes increased luteinizing hormone (LH) secretion and ovarian androgen production. Ultimately, the potential for competing signals at the level of HPO axis has resulted in conflicting reports pertaining to an impact of obesity on morphological characteristics of the ovary. We posit that consideration of reproductive and metabolic disturbance on a continuum using reproducible ultrasonographic methods may be necessary to delineate the extent to which the ovary can serve as a biomarker in the context of obesity.

Reproductive and Metabolic Tipping Points

Evidence of obesity-induced suppression of gonadotropin production and ovarian steroidogenesis has been most often described in women with obesity that retain some degree of regular menstrual cyclicity (e.g. eumenorrhea) and presumably normal androgen status [14–16]. In this context, morphological indicators of folliculogenesis, such as FNPO and AFC, do not differ compared to lean, eumenorrheic women [17–19], despite functional markers of small to medium-sized follicles (i.e. AMH and inhibin B) showing negative associations with body mass index (BMI) in most [19–21] but not all recent studies [18]. While we are aware of one study that reported lower unilateral FNPO in women with obesity and regular cycles [20], most of the recent ultrasonographic evidence aligns with the suppressive effects of obesity being limited to hormonal indicators of ovarian function. Taken together, current evidence supports that changes are mild in the presence of cycle regularity and by consequence do not manifest in ultrasonographically detectable changes in ovarian morphology on cross-sectional analysis.

Evidence in women with irregular cycles and normal androgens (termed normoandrogenic anovulation) suggests that cycle irregularity may enable a shift in the ovary to serve as a biomarker for reproductive and metabolic status on ultrasonography. We have found that women with normoandrogenic anovulation have higher FNPO, attributed to an increased number of small follicles (2-5mm), versus controls, albeit no difference in OV was detected (8.2±3.4 vs 6.7±2.5cm³). Interestingly, LH levels were 60% higher than controls (8.2±5.6 vs 4.9±2.5mlU/mL) suggesting a shift from a suppressive to activational state of the HPO axis which is more in line with disruptions seen in androgenic types of anovulation like PCOS (H Vanden Brink, PhD thesis, Cornell University, 2020). Although differences in metabolic features did not reach statistical significance, on average women with normoandrogenic anovulation were heavier (29.2±9.2 vs. 26.0±6.5kg/m²), more insulin resistant (HOMA-IR, 2.4±2.1 vs 2.Oil.6) with lower sex hormone binding globulin (SHBG) levels (50.2±45.5 vs 58.8±26.3nmol/L) (H Vanden Brink, PhD thesis, Cornell University, 2020), whose production is known to be suppressed by the hyperinsulinemia of IR [22]. Collectively, these emerging data point to the potential for a metabolic tipping point that coincides with loss of menstrual cyclicity (i.e. reproductive tipping point) and enables ultrasonographically detectable increases in follicle number.

Interdependence with Hyperandrogenlsm

Insulin resistance (IR) and compensatory hyperinsulinemia represent a pathophysiologic link between metabolic abnormalities and reproductive dysfunction given that reproductive tissues maintain insulin sensitivity despite peripheral IR [10]. Both direct and indirect roles for insulin signaling on the HPO axis have been described. Briefly, insulin can act directly as a co-gonadotropin on ovarian thecal cells to synergistically increase androgen production alongside LH and/or bind to granulosa cells to increase mitogenesis and pre-mature acquisition of LH receptors resulting in follicle arrest [23,24]. Insulin can also act centrally to alter gonadotropin-releasing hormone (GnRH) signaling and increase LH pulsatility whose tempo modulates ovarian steroidogenesis in favor of androgen production [25,26]. Indirectly, insulin suppresses hepatic production of SHBG [22], which leads to increased androgen bioavailability that can alter balances in regulatory feedback systems. Given the association between obesity and IR, obesity as a state of metabolically induced hyperandrogenism may be the physiological scenario in which the integration of metabolic and reproductive dysfunction results in alterations in ovarian morphology. To that end, we noted that in addition to increased FNPO, the women with normoandrogenic anovulation described above had significantly higher free testosterone concentrations versus controls (0.6±0.3 vs 0.4±0.2 ng/dL) – albeit levels on average remained within the upper limits of normal (H Vanden Brink, PhD thesis, Cornell University, 2020). Further, experimental evidence employing ovarian ultrasonography in non-human primates showed that consumption of a high fat, high calorie “western style diet” (WSD) with testosterone resulted in a higher AFC versus consumption of WSD alone. While a trend towards larger ovaries was noted, WSD alone had no significant impact on follicle counts and ovarian size despite dysglycemia [27,28]. This potential interdependence between IR and hyperandrogenism supports the notion that a progression from normoandrogenic to hyperandrogenic anovulation may reflect an integrative spectrum for metabolic and reproductive dysfunction that manifests in increasingly pronounced defects ovarian morphology on ultrasonography.

Clinical Evidence for Morphological Indicators of Hyperandrogenic Anovulation

Ovaries in hyperandrogenic anovulation have PCOM, characterized primarily by small antral follicle excess (2-5mm) and ovarian enlargement [24,29]. The notion that ovarian morphology can reflect an adverse impact of metabolic derangements on the clinical presentation of PCOS is supported by recent research evaluating the consequences of raising the threshold for follicle excess [12 to 20 follicles [30] or 12 to 25 follicles [31]]. Although women no longer meeting criteria for PCOM were judged to have some degree of metabolic risk, women that maintained their PCOS status owing to a higher FNPO had poorer metabolic profiles including increased lipids [30] and worse measures of glucoregulation, despite similar BMI [30,31]. Likewise, ovarian enlargement (ΟV >9 or >10cm³) was found to be more likely in women with PCOS that were insulin resistant (HOMA-IR >4) [32], and showed utility to distinguish between those with or without metabolic syndrome [33]. Although these recent reports [32,33] could not demonstrate predictive or diagnostic utility of follicle counts for metabolic indices, collectively the data point to ovarian dysmorphology, particularly ovarian size, as having some potential to reflect metabolic dysfunction.

Studies directly evaluating relationships between metabolic status and follicle size populations in PCOS reinforce the controversy over follicle excess to reflect the severity of metabolic disturbance. We have reported that in both women and adolescents with PCOS, BMI is negatively associated with FNPS [34] and AFC [35], with the latter observation being largely driven by an excess of small follicles. Although these findings were unexpected, we also noted inverse associations between follicles ≤5mm with measures of central adiposity, glucoregulatory status, hyperlipidemia and pro-inflammation [35]. In the case of medium-sized follicles (in the region of 5-9mm), associations were reversed with positive associations being apparent for BMI, waist-hips ratio, triglycerides and C-reactive protein [35]. Our findings are tempered by reports of a weak positive association between 2-5mm follicles and fasting insulin levels, and no metabolic factor accounting for the number of small (2-5mm) and medium (6-9mm) antral follicles by principal component analysis [36]. Discrepancies across studies may be attributed to methodological differences in counting and measuring follicles (real-time versus off-line) [37,38] and highlight a need for standardization of ultrasonographic assessments much like any other assay. With these caveats in mind, the potential for variable metabolic involvement in mechanisms driving follicle excess (accumulation of mostly 2–5mm follicles) or arrest (failure to transition to selection and dominance) in women with ovulatory disorders is intriguing and further investigation is needed to corroborate this hypothesis.

Interventions Aimed at Weight Loss and Glucoregulation

Interventions which target weight loss and/or improvements in glucoregulatory status point toward a potential concomitant amelioration in ovarian dysmorphology. Two recent randomized controlled trials in women with PCOS reported reductions in ΟV [39,40], FNPO [39] and/or stromal volume [40] following nutritional intervention [39] or glucagon-like peptide-1 (GLP-1) therapy [40]. Improvements in ovarian and metabolic features were noted across both studies including weight loss, glucoregulation, and lower free androgens [39–41]. By contrast, a randomized cross-over trial revealed no change in OV in response to two insulin sensitizing agents in women with PCOS [42]. In the case of myoinositol, no impact on reproductive or metabolic features was noted with treatment and aligns well with a lack of morphological changes in the ovary. While metformin administration was associated with improved insulin sensitivity, androgen status, and AMH, it is possible that the weight gain associated with treatment negated any potential impact on ovarian morphology. A recent chart review of women with and without PCOS also revealed no change in OV after bariatric surgery [43]. However, OV was found to be predictive of the magnitude of amelioration in HbAlC and triglycerides [43]. Together, recent intervention data suggests that improvements in metabolic status may be reflected in reductions in ovarian size or follicle number consistent with the ovary having a prognostic role for metabolic improvements.

Future Directions and Concluding Remarks

We have highlighted recent evidence that supports the hypothesis that ultrasonographic features of the ovary may reflect stages of nutritional and metabolic status as it integrates with reproductive dysfunction in women. We describe a spectrum of ovarian morphology across the adiposity continuum with the acknowledgement that the process may not be linear, nor limited to the involvement of IR, and must be viewed in light of genetic and environmental factors which contribute to the integration of reproductive and metabolic dysfunction [10,44]. Restricting ultrasonographic assessments to ovarian size and the total number of follicles present, as is the current standard for the diagnosis and evaluation of ovulatory disorders, is a missed opportunity to capture unique pathophysiologic mechanisms underlying disrupted folliculogenesis and ovarian dysmorphology. Future consideration of follicle size-populations and other anatomical features using reproducible ultrasonographic methods are necessary to clarify and enhance the diagnostic and prognostic role of ovarian morphology in women’s health. While our reflections have been focused to under- and over-nutrition, emerging reports related to other aspects of nutritional intake such as dietary patterns [45], diet quality [46,47], or specific nutrients [48] support a growing interest in delineating nutritional and metabolic influences on folliculogenesis and ovarian morphology. Because dietary intake represents a significant modifiable lifestyle factor that may propel, prevent, or ameliorate metabolic drivers of reproductive dysfunction, attention to the development of non-invasive and convenient biomarkers may reveal effective strategies to monitor integrative health status in women.

Table 1:

List of Abbreviations

AFC	Antral Follicle Count, sum of both ovaries
AMH	Anti-Müllerian Hormone
BMI	Body Mass Index
FHA	Functional Hypothalamic Amenorrhea
FNPO	Follicle Number Per Ovary, per single ovary or average of both ovaries
FNPS	Follicle Number Per Single Cross-Section, per single ovary or average of both ovaries
GLP-1	Glucagon-Like Peptide-1
GnRH	Gonadotropin-Releasing Hormone
HbA1C	Hemoglobin A1C
HOMA-IR	Homeostatic Model Assessment of Insulin Resistance
HPO	Hypothalamic-Pituitary-Ovarian (Axis)
IR	Insulin Resistance
LH	Luteinizing Hormone
MFO	Multi-Follicular Ovary
OV	Ovarian Volume
PCOM	Polycystic Ovarian Morphology
PCOS	Polycystic Ovary Syndrome
SHBG	Sex Hormone Binding Globulin
WSD	Western-Style Diet

Highlights:

• Unique ultrasonographic features of ovarian morphology emerge across the adiposity spectrum

• Ovarian size captures pronounced metabolic contributions of under- and over-nutrition

• Follicle size cohorts may better delineate pathogenic mechanisms in ovulatory disorders

• Interventions which target weight or insulin resistance coincide with improved ovarian features