393
Views
5
CrossRef citations to date
0
Altmetric
Research Article

FSH/AMH Ratio and Adipocyte Size are Linked to Ovarian Dysfunction

ORCID Icon, , ORCID Icon &
Pages 174-189 | Received 06 May 2019, Accepted 16 Jan 2020, Published online: 27 Jan 2020

ABSTRACT

Background and Aim

While the relationship between obesity and reproductive dysfunction is well known, the physiological mechanism behind obesity-related infertility remains unclear. Previous work suggests that follicle development prior to ovulation is disrupted in obese individuals. Follicle-stimulating hormone (FSH) and anti-Mullerian hormone (AMH) are two key regulators of follicle development, and the poorest reproductive outcomes have been recorded when these hormones are imbalanced. In order to understand how obesity impacts the reproductive axis, the present study induces reproductive dysfunction in female rats using a high-fat, high-sugar diet (HFHS). Results: In our study, several animals on the HFHS diet displayed abnormal estrous cycles. The HFHS diet also resulted in an increased prevalence of ovarian cysts and decreased formation of corpora lutea. Across all groups, the FSH/AMH ratio displayed a strong negative correlation with pre-antral, antral, and total follicle counts. Moreover, rats on the HFHS diet displayed larger adipocytes and produced higher levels of leptin than controls. When combined with average adipocyte size in multiple regression, the FSH/AMH ratio was strongly associated with cyst formation in the ovary. Conclusions: These findings provide strong evidence for the potential relevance of a combined FSH/AMH ratio as a marker of ovarian health and follicular status. Therefore, this ratio reflects a complex interaction between the reproductive and metabolic systems.

Introduction

Obesity has reached epidemic levels in recent years affecting as many as 2.1 billion people worldwide, with greater prevalence in women than men.Citation1 Links between obesity and impaired reproduction in women have been long-established, particularly observations of oligomenorrhea in obese women.Citation2 Women who are obese fail to ovulate at the normal rate and may be more likely to display polycystic ovary syndrome (PCOS) than their normal-weight counterparts.Citation3,Citation4 As such, obesity corresponds closely with infertility, yet the mechanism remains unclear.

The distribution of adipose tissue plays a prominent role in inducing infertility, with evidence suggesting that central deposition of fat tissue in obese women drives infertility.Citation5 This may be a result of a shift in adipose tissue of obese individuals toward hypertrophic adipocytes which secrete elevated levels of leptin and lower amounts of adiponectin, in addition to greater inflammatory cytokines.Citation6 Hyperleptinemia and resulting leptin resistance have been further correlated with diminished reproductive function.Citation7

One mechanism for leptin’s effects on reproduction results from its action on the pituitary gland, where it stimulates the secretion of both luteinizing hormone (LH) and follicle-stimulating hormone (FSH).Citation8 FSH is a key component of ovarian follicle development, known to promote continued growth leading to ovulation.Citation9 Elevated levels of FSH have previously been associated with reproductive aging due to reduced negative feedback from ovarian hormones, leading to overstimulation of the ovary and rapid depletion of the ovarian reserve.Citation10,Citation11

Another key marker of ovarian function is anti-Müllerian hormone (AMH), which is produced predominately by pre-antral and small antral follicles.Citation12 Evidence from cell culture experiments and AMH knockout mice has demonstrated that more primordial follicles begin growth in the absence of AMH, suggesting a potential role for AMH in regulating initial follicle recruiment.Citation13,Citation14 The possibility that AMH is a regulator of follicle development is further evidenced by rapid depletion of the follicle pool in AMH-knockout mice.Citation15 Importantly, AMH may play a key role in modulating the follicular response to gonadotropins, as AMH has been found to inhibit steroidogenesis by attenuating granulosa cell response to FSH.Citation16,Citation17 In opposition to rising FSH, a fall in AMH corresponds with the onset of reproductive decline.Citation18 Though not widely considered responsive to gonadotropins, AMH has displayed intracycle variation in a number of studies, suggesting that whether AMH represents a cycle-independent measure as previously thought remains a point of contention and may be individual-specific.Citation19Citation22 Interestingly, leptin has been demonstrated to downregulate AMH mRNA, and it has been observed that obese women display lower levels of AMH.Citation23,Citation24

While both FSH and AMH have been used as biomarkers of ovarian follicular health, AMH has recently been considered to more closely relate to ovarian status.Citation25,Citation26 However, the poorest ovarian outcomes have been recorded when both FSH and AMH levels are abnormal, particularly when FSH levels are high and AMH is low.Citation27Citation29 These findings suggest that a combined FSH/AMH ratio could provide a useful measure of ovarian health, yet previous studies have not related this ratio to ovarian structures.

Consequently, this study sought to explore the relationship between the FSH/AMH ratio and follicular structures in the rat ovary, as well as examining the effect that diet-induced obesity has on this relationship. Our previous work demonstrated the ability of diet-induced obesity to promote cyst formation in the rat ovaryCitation30,Citation31, and the current study aimed to understand potential mechanisms underlying the diet-ovary connection. We hypothesized that rats with an elevated FSH/AMH ratio would display a reduced follicular pool, as well as greater frequency of cyst formation indicative of ovarian dysfunction. We further expected that cyst formation would be exacerbated in rats displaying greater adipocyte size.

Materials and Methods

Animals

Female Sprague Dawley rats (n = 30; P17) were obtained from Charles River (Raleigh, NC) and were housed with their mothers until weaning at day 23. At weaning, the rats were randomly assigned to either the HFHS diet (n = 16) or a control diet (n = 14) group. The control group had ad libitum access to water and standard rat chow (3.1 kcal/g, 17% calories from fat; LM-485; Envigo, Indianapolis, IN). The HFHS diet group had ad libitum access to water, a 32% sucrose solution, and high-fat chow, which we have previously validated as a model of diet-induced obesity (5.24 kcal/g, 60% of calories derived from fat; D12492; Research Diets, New Brunswick, NJ).Citation30 The rats were housed in pairs in a climate-controlled environment with a 12/12 light/dark cycle (lights on at 8 AM). The rats were weighed every other day throughout the study. The Washington and Lee University Institutional Animal Care and Use Committee approved all animal procedures (protocol no. NT0717). From animal birth to perfusion of the final animal, the total length of the study was 19 weeks. However, because perfusion took place on diestrus and timing of sacrifices was thus individually specific, perfusion occurred between weeks 17 and 19 of the study.

Estrous Cycle Monitoring

Estrous cycles were monitored by daily vaginal lavage during weeks 8 to 14 of the diet. Estrous cycle stages were identified via cytological examination, as previously described.Citation32 In brief, if leukocytes were the dominant cell type, the sample was identified as diestrus. If nucleated cells were abundant, the sample was considered proestrus. Samples with mostly cornified cells were classified as estrus. Large non-nucleated cells were an indicator of persistent estrus. The duration of each estrous cycle was determined as the number of days before the appearance of each diestrus phase. The rats were identified as demonstrating persistent vaginal cornification (PVC), or persistent estrus, if the smears revealed ≥5 consecutive days spent in the estrus phase of the cycle. The identification of PVC has been characterized in previous rodent models of PCOS and in aging animals with reproductive impairment.Citation33 To calculate the total time spent in each phase of the cycle, we divided the total number of days spent in each specific phase (proestrus, estrus, and diestrus) by the total duration of the sampling time. Comparisons of the difference in cycle distribution of each stage of the cycle between the two diet groups were evaluated using a chi-square test. Differences were considered statistically significant at p < .05.

Fasting Insulin and Glucose Monitoring

Insulin and glucose testing occurred between weeks 16 and 17 of the study. Prior to testing, rats were fasted overnight to deplete glycogen stores and reduce baseline variability between subjects. Blood samples were obtained via an indwelling jugular catheter, as previously described.Citation31 Fasting blood glucose levels (mg/dL) were measured using an Accu-Chek Compact Plus whole-blood glucose monitor. To examine the fasting insulin levels, blood was collected at the same time, allowed to clot, and centrifuged to obtain serum. Serum samples were then stored at −20°C until the insulin levels were determined using a rat insulin enzyme-linked immunoassay kit as per the manufacturer’s instructions (model 90010; Crystal Chem). In order to calculate HOMA-IR, a clinical measure of insulin resistance, fasting insulin (µU/L) was multiplied by fasting glucose (nmol/L), and this product was divided by 22.5.

Perfusion and Tissue Collection

On the morning of diestrus I, the rats were deeply anesthetized with isoflurane, blood was collected through cardiac puncture, and the rats were transcardially perfused with saline and 4% paraformaldehyde. Sacrifices occurred when rats were between the age of 121 and 133 days. Visceral fat pads were collected and weighed immediately after perfusion. We separated visceral fat pads by retroperitoneal fat and parametrial fat to represent the upper abdominal and gonadal fat, respectively. Adipose tissue and ovaries were collected and fixed in 4% paraformaldehyde at 4°C overnight before being transferred to a 30% sucrose solution in phosphate-buffered saline. The ovaries were then cleaned by removing the oviduct and any surrounding adipose tissue, weighed, and embedded in optimal cutting temperature matrix (23-730-571; Fisher) for storage at −80°C until cryostat sectioning.

Hormone Measurements

Blood samples collected at the time of perfusion were used for hormone quantification. Reproductive hormone assays were performed by the University of Virginia Ligand Core Facility. Serum follicle-stimulating hormone (FSH) was measured using a rat multiplex assay (reportable range 2.4 to 300.0 ng/mL). Anti- Müllerian hormone (AMH) was measured using an enzyme-linked immunosorbent assay according to the manufacturer’s instructions (model CSB-E11162r; CUSABIO). Adiponectin was also measured using enzyme-linked immunosorbent assay (CrystalChem-80570), as was leptin (CrystalChem-90040).

Ovarian Histologic and Follicular Assessment

Follicular development was assessed using one ovary from each rat. The ovaries were serially sectioned into 10-µm-thick sections, with every fifth section collected on gelatinized glass slides. Histological examination of tissue sections was conducted after hematoxylin and eosin staining, a well-documented method for histological examination.Citation34 While serial sectioning may result in the same morphological feature being counted in multiple cross-sections, it has been previously demonstrated that this method does not alter the validity of relative values obtained from the procedure so long as it is applied uniformly across all tissue samples.Citation35,Citation36 Images were captured at ten-times magnification with an Olympus BX-41 microscope using an Olympus DP80 camera. The total numbers of cysts and corpora lutea (CL) for each rat were identified according to previously described criteria.Citation36,Citation37 In brief, cysts were identified as those follicles displaying a large antral space surrounded by an enlarged and densely stained thecal cell layer and lacking an ovum. Cysts were distinguished from antral and other atretic follicles based on the characteristics of the granulosa cell layers, with cysts displaying few to no visible granulosa cells within the antral space. Corpora lutea were identified according to tissue density and the presence of a grainy luteinized cell appearance. Corpora lutea were counted only when they exhibited more than a 60% area with a higher tissue density.

Follicle identification was conducted in accordance with the size threshold methodology put forth in OsmanCitation38 in conjunction with the morphological categories proposed by Pedersen and Peters.Citation39 In brief, size thresholds were established by identifying the largest and smallest follicle of each class based upon follicular morphology. ImageJ software was utilized for follicle identification and measurement as previously documented. Namely, pre-antral follicles were classified as oocytes with two or more layers of granulosa cells but with no visible space between cells and displayed diameters between 0.06 and 0.30 mm. Antral follicles were defined as follicles containing an antral cavity but were subdivided into subcategories of early and late antral. Early antral follicles were identified as follicles containing space between granulosa cells, indicating the early formation of an antral cavity, and displayed diameters between 0.30 mm and 0.60 mm. Late antral follicles were identified as follicles containing a segmented antral cavity with two or more compartments and displayed diameters between 0.60 mm and 1.25 mm. Lastly, pre-ovulatory follicles contained one continuous antral cavity and displayed a minimum diameter of 1.25 mm. Primordial follicles were not quantified in the present work. If follicles lacked an ovum, they were considered atretic and were not included for analysis. As has been noted previously, a follicle classificatory scheme based upon size, as quantified by diameter, and morphology strengthens analyses by recognizing follicle development as a progression through multiple stages as well as a growth continuum. Importantly, granulosa cell count has been correlated with follicular diameter, suggesting that a combination of size and morphology for follicle identification is a valid approach.Citation40

Adipocyte Histology and Quantification

Adipocyte size was determined using a sample of retroperitoneal fat from each rat. Frozen, fixed adipose tissue was serially sectioned into 25-µm-thick sections, with every section collected on gelatinized glass slides. Histological examination of sections was performed after hematoxylin and eosin staining. Six randomly selected sections of adipocytes from each subject were analyzed at ten-times magnification using Adiposoft software as previously validated. Distributions of adipocyte cell size were constructed using individual counts for each rat.

Statistical Analysis of Histological and Physiological Measurements

Rats were separated into groups of cycling and non-cycling subjects. Non-cyclers were defined as rats spending a greater fraction than 0.48 in the estrus stage throughout cycle monitoring. The non-cycling group included 6 HFHS rats and 1 control diet rat which failed to cycle. Statistical analysis was conducted in Python. For between group comparisons, results are expressed as the mean ± standard error of the mean. The comparison of the means between groups was conducted using the nonparametric Mann-Whitney U test. Differences were considered significant at p < .05. The statistical significance of all correlations was assessed using linear regression analysis in Python’s SciPy package. The statistical significances for differences between group distributions was assessed using the Kolmogorov-Smirnov test. Multiple linear regression was performed with Python’s Statsmodels package using the variables of FSH/AMH ratio and average adipocyte size as predictors for average cyst size. In all analyses, missing data was handled by pairwise deletion, meaning individual animals were excluded from analyses for which they displayed missing data for the variable of interest but included in analyses where their data was complete. Two individuals were excluded from at least one analysis due to poor adipose tissue quality, two due to poor ovarian tissue quality, and four due to insufficient blood volume for specific hormone measurements.

Results

Adiposity Positively Correlates with Irregular Estrous Cycles

At termination, rats on the HFHS diet weighed significantly more than rats on the control diet (399.6 ± 10.0 g vs. 316.3 ± 7.72 g; p < .001; M-W U Test). To explore the effect of diet on estrous cyclicity, vaginal cytology was performed daily during the experiment. After 9 cycles, rats on the HFHS diet began to spend a greater portion of time in the estrus stage compared to controls, suggestive of irregular cycling (0.44 ± 0.06 vs 0.28 ± 0.03, chi-square test; ). This finding persisted through the end of the study. To assess the relationship between adiposity and estrous cycling, adipocyte size was quantified through histological examination. Across groups, adipocyte size displayed a positive correlation with fraction of time spent in estrus (p < .02, r = 0.53; ). This suggests that increased adiposity is associated with irregular estrous cyclicity. Due to this correlation, adipocyte size was compared between experimental groups. On average, non-cyclers and HFHS cyclers displayed significantly greater adipocyte size than control cyclers (p = .001, , M-W U Test).

Figure 1. HFHS diet increases fraction of time spent in estrus over time. Data points indicate mean fraction of time spent in estrus for each diet group. Error bars indicate standard error of the mean. Red squares indicate the HFHS group (n = 16), and blue triangles mark the control group (n = 14). The first estrous cycle presented was recorded at an age of 79 days. * indicates significance at the p < .05 level according to a chi-square analysis.

Figure 1. HFHS diet increases fraction of time spent in estrus over time. Data points indicate mean fraction of time spent in estrus for each diet group. Error bars indicate standard error of the mean. Red squares indicate the HFHS group (n = 16), and blue triangles mark the control group (n = 14). The first estrous cycle presented was recorded at an age of 79 days. * indicates significance at the p < .05 level according to a chi-square analysis.

Figure 2. Adipocyte size associates with fraction of time in estrus. (a) Pearson correlation of individual fraction of time spent in estrus versus average adipocyte size. (b) Violin plot indicating the average adipocyte; red, non-cyclers (n = 7); pink, HFHS cycling (n = 9); blue, control cycling (n = 12). * indicates significance at the p < .05 level according to a Mann-Whitney U Test. One HFHS cycler and one control cycler were excluded from this analysis due to inadequate adipose tissue quality.

Figure 2. Adipocyte size associates with fraction of time in estrus. (a) Pearson correlation of individual fraction of time spent in estrus versus average adipocyte size. (b) Violin plot indicating the average adipocyte; red, non-cyclers (n = 7); pink, HFHS cycling (n = 9); blue, control cycling (n = 12). * indicates significance at the p < .05 level according to a Mann-Whitney U Test. One HFHS cycler and one control cycler were excluded from this analysis due to inadequate adipose tissue quality.

Abdominal Adiposity and Insulin Resistance are Associated with Adipocyte Size

Due to the relationship between adiposity and irregular cycling, the difference between distributions of adipocyte size were analyzed. HFHS cyclers and non-cyclers displayed significantly different distributions than control cyclers (p < .05, Kolmogorov-Smirnov; ). Control cyclers displayed unimodal distributions, while HFHS and non-cyclers are bimodal. This shift in the distributions was toward larger adipocytes in HFHS cyclers and non-cyclers. Because central adiposity and adipocyte size have been used as a predictor for insulin resistance, we examined the relationship between cell size and insulin resistance.Citation41 Average adipocyte size displayed a logarithmic relationship with percent abdominal fat, as well as HOMA-IR (, ). Due to the known association between adiposity, elevated leptin, and reduced adiponectin, we also explored the plasma concentrations of these adipokines.Citation42,Citation43 Compared to control cyclers, both non-cyclers and HFHS cyclers displayed significantly higher levels of circulating leptin (p < .005, , M-W U Test). However, there was no difference in circulating adiponectin between groups.

Figure 3. Distribution of adipocyte size differs between groups. (a) Distribution of adipocyte size for control cyclers (n = 12). (b) Distribution of adipocyte size for HFHS cyclers (n = 9). (c) Distribution of adipocyte size for non-cyclers (n = 7). Presented distributions are normalized between groups, meaning probability densities are presented on the y-axis. Cell size frequencies were normalized on the y-axis through multiplication by a normalization factor to account for differences in raw counts of adipocytes. HFHS and non-cycler distributions are significantly different from the control distribution at the p < .05 level, according to a Kolmogorov-Smirnov Test. One HFHS cycler and one control cycler were excluded due to inadequate adipose tissue quality.

Figure 3. Distribution of adipocyte size differs between groups. (a) Distribution of adipocyte size for control cyclers (n = 12). (b) Distribution of adipocyte size for HFHS cyclers (n = 9). (c) Distribution of adipocyte size for non-cyclers (n = 7). Presented distributions are normalized between groups, meaning probability densities are presented on the y-axis. Cell size frequencies were normalized on the y-axis through multiplication by a normalization factor to account for differences in raw counts of adipocytes. HFHS and non-cycler distributions are significantly different from the control distribution at the p < .05 level, according to a Kolmogorov-Smirnov Test. One HFHS cycler and one control cycler were excluded due to inadequate adipose tissue quality.

Figure 4. Metabolic measurements correspond to adipocyte size. (a) Logarithmic correlation between average adipocyte size and HOMA-IR scores for HFHS cyclers (n = 9), control cyclers (n = 11), and non-cyclers (n = 6). One HFHS cycler and one control cycler were excluded due to inadequate adipose tissue quality, and one non-cycler and one control cycler were excluded due to insufficient blood volume (b) Logarithmic correlation between average adipocyte size and percent abdominal fat for HFHS cyclers (n = 9), control cyclers (n = 12), and non-cyclers (n = 7). One HFHS cycler and one control cycler were excluded due to inadequate tissue quality. Percent abdominal fat represents the mass of abdominal fat pads divided by total body mass (c) Leptin levels compared between groups. Error bars represent the standard error of the mean. * indicates significance at the p < .05 level according to a Mann-Whitney U Test.

Figure 4. Metabolic measurements correspond to adipocyte size. (a) Logarithmic correlation between average adipocyte size and HOMA-IR scores for HFHS cyclers (n = 9), control cyclers (n = 11), and non-cyclers (n = 6). One HFHS cycler and one control cycler were excluded due to inadequate adipose tissue quality, and one non-cycler and one control cycler were excluded due to insufficient blood volume (b) Logarithmic correlation between average adipocyte size and percent abdominal fat for HFHS cyclers (n = 9), control cyclers (n = 12), and non-cyclers (n = 7). One HFHS cycler and one control cycler were excluded due to inadequate tissue quality. Percent abdominal fat represents the mass of abdominal fat pads divided by total body mass (c) Leptin levels compared between groups. Error bars represent the standard error of the mean. * indicates significance at the p < .05 level according to a Mann-Whitney U Test.

Non-cyclers Display Ovarian Dysfunction

While our previous researchCitation30,Citation31 has found an increase in cyst formation and reduction in corpora lutea in rats on a HFHS diet, we sought to explore specifically how cycle status relates to these structures. Both HFHS cyclers and non-cyclers presented greater numbers of cysts than control cyclers (p < .005, , M-W U Test). While HFHS failed to display fewer corpora lutea than control cyclers (p = .056, M-W U Test), non-cyclers displayed significantly fewer corpora lutea than both groups (p < .01, , M-W U Test). Taken together, these measures suggest impairment of ovarian health in non-cycling rats. To elucidate how this might affect follicle development, follicle distributions were then analyzed (). Non-cyclers displayed fewer numbers of pre-antral follicles than HFHS (p = .038, M-W U Test) but not control cyclers (p = .10, M-W U Test). Non-cyclers did not display significantly lower numbers of early antral and late antral follicles compared to other groups, but high levels of variance were present within the group (CVearly antral = 335.5, CVlate antral = 340.5). Compared to control cyclers, neither HFHS cyclers nor non-cyclers contained fewer pre-ovulatory follicles (p < .08, M-W U Test). For total follicles, no significant difference between groups was identified (p < .09, M-W U Test, not shown).

Figure 5. Ovarian structures differ between HFHS cyclers (n = 10), control cyclers (n = 11), and non-cyclers (n = 7). (a) Bar chart indicating cysts per section for each cycle group. (b) Bar chart indicating corpora lutea per section by cycle group. (c) Follicle counts by cycle group, divided into categories of pre-antral, early antral, late antral, and pre-ovulatory. Error bars represent standard error of the mean. A difference in letters above bars indicates a significant difference between groups according to a Mann-Whitney U Test. Two control cyclers were excluded due to inadequate ovarian tissue quality.

Figure 5. Ovarian structures differ between HFHS cyclers (n = 10), control cyclers (n = 11), and non-cyclers (n = 7). (a) Bar chart indicating cysts per section for each cycle group. (b) Bar chart indicating corpora lutea per section by cycle group. (c) Follicle counts by cycle group, divided into categories of pre-antral, early antral, late antral, and pre-ovulatory. Error bars represent standard error of the mean. A difference in letters above bars indicates a significant difference between groups according to a Mann-Whitney U Test. Two control cyclers were excluded due to inadequate ovarian tissue quality.

AMH and FSH Concentrations Do Not Differ between Groups

Due to the altered ovarian histology, we sought to explore factors associated with follicle growth and maturation. AMH and FSH concentrations were analyzed because these hormones are major regulators of follicle development (). Contrary to expectations, no significant difference in AMH levels between groups was found (p = .23, M-W U Test). Surprisingly, despite the observed differences in follicle distributions, FSH levels did not differ significantly between groups. AMH and FSH alone were thus unable to explain the differences in follicle distributions.

Figure 6. Anti-Müllerian hormone and follicle-stimulating hormone are unaffected by cycle status between HFHS cyclers (n = 10), control cyclers (n = 11), and non-cyclers (n = 7). (a) Bar chart for mean serum AMH levels by cycle group. (b) Bar chart for mean plasma FSH levels by cycle group. Error bars represent standard error of the mean. Significance was calculated using a Mann-Whitney U Test. Two control cyclers were excluded due to insufficient blood volume for hormone measurement.

Figure 6. Anti-Müllerian hormone and follicle-stimulating hormone are unaffected by cycle status between HFHS cyclers (n = 10), control cyclers (n = 11), and non-cyclers (n = 7). (a) Bar chart for mean serum AMH levels by cycle group. (b) Bar chart for mean plasma FSH levels by cycle group. Error bars represent standard error of the mean. Significance was calculated using a Mann-Whitney U Test. Two control cyclers were excluded due to insufficient blood volume for hormone measurement.

Figure 7. Elevation of the FSH/AMH ratio correlates with decreased follicle counts for HFHS cyclers (n = 10), control cyclers (n = 9), and non-cyclers (n = 7). (a) Pearson correlation between FSH/AMH and total follicle counts. Total follicle counts include pre-antral, early and late antral, and pre-ovulatory follicles. (b) Pearson correlation between FSH/AMH and late antral follicle counts per section. The r value represents the Pearson correlation coefficient, indicating the magnitude and direction of the linear relationship between the variables. The p-value indicates the significance of the linear relationship. Two control cyclers were excluded from the analysis due to insufficient blood volume and two due to inadequate ovarian tissue quality.

Figure 7. Elevation of the FSH/AMH ratio correlates with decreased follicle counts for HFHS cyclers (n = 10), control cyclers (n = 9), and non-cyclers (n = 7). (a) Pearson correlation between FSH/AMH and total follicle counts. Total follicle counts include pre-antral, early and late antral, and pre-ovulatory follicles. (b) Pearson correlation between FSH/AMH and late antral follicle counts per section. The r value represents the Pearson correlation coefficient, indicating the magnitude and direction of the linear relationship between the variables. The p-value indicates the significance of the linear relationship. Two control cyclers were excluded from the analysis due to insufficient blood volume and two due to inadequate ovarian tissue quality.

Figure 8. FSH-AMH ratio and average adipocyte size are associated with cyst prevalence. (a) Pearson correlation between mean cysts per section and FSH/AMH for HFHS cyclers (n = 10), control cyclers (n = 9), and non-cyclers (n = 7). (b) Pearson correlation between mean cysts per section and average adipocyte size for HFHS cyclers (n = 10), control cyclers (n = 9), and non-cyclers (n = 7). (c) Multiple regression using the predictor variables of FSH/AMH and average adipocyte size and the outcome variable of mean cysts per section for HFHS cyclers (n = 9), control cyclers (n = 8), and non-cyclers (n = 7). The height of the lines indicates the value of the cysts per section, and the transparency of shading represents whether a value falls above or below the plane of the regression. Darker shading indicates a point above the plane, while more transparent shading represents a point beneath the plane. Five control cyclers were excluded due to inadequate blood volume or tissue quality, and one HFHS cycler was excluded due to inadequate adipose tissue quality.

Figure 8. FSH-AMH ratio and average adipocyte size are associated with cyst prevalence. (a) Pearson correlation between mean cysts per section and FSH/AMH for HFHS cyclers (n = 10), control cyclers (n = 9), and non-cyclers (n = 7). (b) Pearson correlation between mean cysts per section and average adipocyte size for HFHS cyclers (n = 10), control cyclers (n = 9), and non-cyclers (n = 7). (c) Multiple regression using the predictor variables of FSH/AMH and average adipocyte size and the outcome variable of mean cysts per section for HFHS cyclers (n = 9), control cyclers (n = 8), and non-cyclers (n = 7). The height of the lines indicates the value of the cysts per section, and the transparency of shading represents whether a value falls above or below the plane of the regression. Darker shading indicates a point above the plane, while more transparent shading represents a point beneath the plane. Five control cyclers were excluded due to inadequate blood volume or tissue quality, and one HFHS cycler was excluded due to inadequate adipose tissue quality.

FSH/AMH Ratio Corresponds to Follicle Counts in Individual Rats

Due to the different stage-dependent roles FSH and AMH play in follicle pool regulation and maturation, this study sought to explore the combined effects of these hormones on follicular development ().Citation44 Between groups the FSH/AMH ratio did not differ significantly, due in part to large variance in each group (data not shown). However, the FSH/AMH ratio displayed a strong negative correlation with total follicles per section (p < .0005, r = −0.67; Fig 7A). Furthermore, the FSH/AMH ratio most strongly correlated negatively with late antral follicles per section (p < .00005, r = −0.65, Fig 7B). The FSH/AMH ratio also correlated negatively with the number of pre-antral follicles as well as early antral follicles, though no correlation with pre-ovulatory follicles was displayed (data not shown). Thus, the FSH/AMH ratio as a combined metric is closely associated with follicle structures in the ovary.

Together, FSH/AMH Ratio and Adipocyte Size are Associated with Cyst Formation

To determine if the FSH/AMH ratio was associated with ovarian health beyond follicular decline, we explored the correlation of the FSH/AMH ratio with cysts, a known marker of ovarian dysfunction (). While the FSH/AMH ratio correlated negatively with follicle counts in the ovary, the ratio displayed a strong positive correlation with cysts per section (p < .001, r = 0.71, Fig 8A). Given our earlier findings regarding the shift in adipocyte distribution toward larger cell size, we aimed to explore the relationship between adiposity and cyst formation to determine if adipose dysfunction is associated with ovarian dysfunction. Adipocyte size displayed a positive correlation with cysts per section (p < .01, r = 0.50, Fig 8B). While both the FSH/AMH ratio and adipocyte size were independently associated with cyst formation, these metrics were highly associated with cyst formation when combined in multiple regression (p < .00001, Fig 8C). In summary, the FSH/AMH ratio of individual rats in this study displayed strong associations with respect to follicle counts in the ovary and additionally corresponded to cyst formation in conjunction with adipocyte size. An interplay between adiposity and follicular hormone balance was thus essential to ovarian health along a spectrum of outcomes across experimental groups.

Discussion

Relevance of FSH and AMH to Follicle Development

In this study, we identify the relationship between an elevated FSH/AMH ratio and lower follicle counts in a rat model of diet-induced obesity. Furthermore, we establish that this measure, in conjunction with average adipocyte size, is associated with cyst formation in the rat ovary. These findings build a compelling argument for hormonal imbalance and metabolic dysfunction impairing reproductive function through convergent pathways.

The FSH/AMH ratio presents a possible clinical marker of ovarian health because the two hormones regulate follicle recruitment in tandem. AMH secretion begins after a follicle is selected from primordial to primary status, with maximal levels of expression in small antral follicles.Citation45 This process of initial recruitment drives the continuous selection of primordial follicles for subsequent growth and relies primarily on local factors within the ovary.Citation46 Interestingly, an absence of FSH achieved through hypophysectomy results in decreased initial recruitment.Citation47 However, it is important to note that follicles are not sensitive to gonadotropins until after initial recruitment has occurred.Citation48 Instead, AMH may indirectly inhibit the initial recruitment of primordial follicles and decreases the sensitivity of antral follicles to FSH-stimulated differentiation required for cyclic recruitment.Citation13

In contrast to the continuous process of initial recruitment, cyclic recruitment occurs only during the rise in FSH during the estrous cycle, allowing antral follicles to continue growth until ovulation.Citation44 In the absence of AMH, FSH will recruit a greater number of follicles from the antral follicle pool.Citation15 AMH induces a decrease in FSH sensitivity through suppressing FSH-induced adenylyl cyclase activity, and has been shown to have no effect onCitation16 or decrease FSH receptor content of granulosa cells.Citation49 AMH-KO mice have more growing follicles and fewer primordial follicles, suggesting that more follicles are recruited for growth in the absence of AMH. Furthermore, this appears to be a dose-dependent effect, as mice with some AMH displayed an intermediate phenotype.Citation14 Over time, AMH deficiency leads to a reduction in the number of growing follicles, since the primordial follicle pool has been depleted through over-recruitment.Citation50

Because a regular estrous cycle is required for successful cyclic recruitment, it was expected that acyclic rats in the present study would display abnormal follicle distributions. Interestingly, non-cycling rats only displayed fewer numbers of pre-antral follicles while displaying normal counts of antral follicles. Despite the altered follicle distributions, individual AMH and FSH levels did not differ between groups. This suggests that the combined FSH/AMH ratio may instead serve as a stronger measure of ovarian health. When using the FSH/AMH ratio, strong negative correlations were found with different follicle counts, confirming this possibility. While non-cyclers tended toward more elevated FSH/AMH ratios, the effect of the combined measure appeared independent of cycle status.

FSH/AMH Ratio Marks Ovarian Health

While the FSH/AMH ratio is a novel measure, previous work suggests that a balance of the two hormones is essential to reproductive success. Importantly, women who fail to cycle regularly are more likely to display abnormal levels of FSH and AMH.Citation29 However, conflicting results exist regarding whether FSH or AMH represents a better marker of ovarian health, largely because the two hormones represent different, yet interlocked pieces of the follicle development process. In women with high FSH and low AMH, for instance, the size of ovarian reserve measured by oocyte retrieval was the lowest.Citation27

In our study, an elevation in the FSH/AMH ratio was also associated with poorer ovarian health, as confirmed by lower follicle counts, greater cyst prevalence, and decreased corpora lutea formation. To our knowledge, this study is the first to examine the relationship between this ratio and specific follicle structures in the ovary through histological examination. The FSH/AMH ratio was strongly associated with follicle counts, correlating negatively with the number of pre-antral, antral, and total follicles. These findings suggest that the balance of FSH and AMH is essential to healthy follicle development and regulation of the ovarian reserve. Importantly, neither FSH nor AMH individually correlated with follicle counts as strongly, suggesting that the ratio could serve as an improved clinical marker in the determination of the ovarian reserve.

Obesity and Positive Energy Balance Disrupts FSH and AMH

Previous research suggests that obesity could alter the levels of FSH and AMH as a result of the extended presence of a positive energy balance. For instance, obesity induced by neonatal overfeeding has been found to increase the levels of FSH receptor mRNA in the ovary while also decreasing the expression of AMH, suggesting a possible mechanism by which obesity can alter the hormone ratio.Citation51 Obesity-driven reductions in AMH have also been observed in later developmental stages. Indeed, AMH levels are lower among obese women, independent of gonadotropin levels.Citation24 Additionally, leptin has been negatively associated with AMH in obese women, demonstrated by a leptin-induced reduction in AMH mRNA expression and AMH production through the JAK2/STAT3 pathway.Citation23,Citation52

Our study utilized a diet-induced obesity model in order to challenge the reproductive system. We believe our diet-induced model of reproductive dysfunction more accurately reflects the natural variability of the obese population. While obesity is often conceptualized as a unified condition, in actuality it presents with significant heterogeneity across a wide range of variables.Citation53 As opposed to pharmacological and knockout interventions which generally produce uniform findings, our ad libitum HFHS model can affect multiple obesity-related variables along a spectrum of intensities. In our study, negative correlations between the FSH/AMH ratio and follicle counts occurred despite the lack of a pinpoint manipulation, suggesting a greater robustness of results and more generalizable implications. The variability in our data and resulting absence of simple between-groups differences thus provide a potentially representative obese population. Consequently, direct impacts of a single obesity-related variable are unlikely to be observed; therefore, interpretation of our results requires a multi-faceted approach.

Leptin, for instance, plays a crucial role in signaling the availability of energy stores to the reproductive system. The ob/ob model has provided strong evidence for the importance of leptin in reproduction, due to the fact that leptin-deficient mice display infertility.Citation54 FSH has been demonstrated to increase secretion of leptin from adipocytes through an estradiol-mediated pathway, suggesting a positive feedback loop whereby leptin and FSH mutually increase one another in the absence of negative ovarian feedback.Citation55 Leptin stimulates GnRH release, leading to secretion of LH and FSH, as well as signaling the pituitary directly to release gonadotropins.Citation8,Citation56

In our study, the groups with the highest levels of leptin displayed the greatest reproductive abnormalities. Elevated leptin alone, however, did not correlate with any of the reproductive outcomes, suggesting an interaction with other processes to produce impairments. One potential mechanism by which this may occur is through alteration of the early follicle recruitment process, which could then further affect the levels of FSH and AMH. As such, while low doses of leptin have a stimulatory effect on the ovary to signal energy availabilityCitation57, it is possible that excessive doses of leptin may have hyperstimulatory effects on the reproductive system, resulting in rapid depletion of ovarian reserves.

Metabolism-Derived Endocrine Disruption of Follicle Development

Previous work suggests that pre-pubertal recruitment of primordial follicles may occur via a leptin-mediated pathway. For instance, use of a leptin antagonist during early life overfeeding was able to preserve primordial follicle numbers.Citation58 In female rats overfed from an early age, puberty occurs earlier than in controls.Citation59 Lower counts of primordial and preantral follicles have been observed in neonatally overfed rats, suggesting that obesity can affect the follicular reserve even prior to the onset of puberty.Citation58

Another mechanism by which obesity could induce early over-recruitment from the ovarian reserve could operate through insulin, which has been shown to promote the transition of follicles from primordial to pre-antral.Citation60 Since rats fed a HFHS diet in our study displayed hyperinsulinemia, insulin could have had an effect on follicle recruitment in these animals. Like leptin, insulin has also demonstrated a synergistic relationship with FSH to promote follicular development once follicles gain FSH sensitivity.Citation61 There are thus multiple endocrine pathways by which signals of positive-energy balance in obesity can affect follicle development.

Consequently, it is clear that no one factor can explain the ovarian impairments stemming from diet-induced obesity. In our study, rats fed a HFHS diet displayed lower numbers of pre-antral follicles, but this did not correlate directly with metabolic parameters. Similarly, between group differences in estradiol were not identified and did not correlate with metabolic parameters (data not shown). The metabolic state characterized by excess energy stores may thus interact with the reproductive system on a highly individual basis in order to produce a number of outcomes along a spectrum of abnormality.

Lipid-Induced Oxidative Stress and Ovarian Dysfunction

The adipocyte could play a key role in reproductive dysfunction through its joint endocrine and metabolic roles. In our study, both insulin and leptin were associated with the average size of adipocytes in logarithmic and linear relationships, respectively. Furthermore, non-cyclers and HFHS cyclers showed adipocyte distributions shifted toward larger cell size. Enlargement of adipocytes is problematic because as they become hypertrophic, they undergo oxidative stress and secrete high levels of inflammatory cytokines and lipids into the bloodstream.Citation6

A saturating relationship between abdominal fat and adipocyte size in our study suggests that as adipocytes become hypertrophic, lipids may begin to be stored in other tissues. One possible destination of these lipids could be the ovary. It has been demonstrated that the follicular fluid of obese women displays greater levels of these lipids and inflammatory cytokines which could impair ovarian function.Citation62 For instance, interleukin-1 (IL-1) knockout mice displayed markedly higher levels of AMH and greater numbers of antral follicles, suggesting a relationship between inflammation and follicle recruitment.Citation63 Such lipotoxic conditions in the ovaries of obese subjects can lead to oocyte apoptosis as a result of oxidative stress, which could dysregulate the follicle development process.Citation62

While the present study did not examine cytokine or lipid accumulation in the ovary, these lipotoxic effects could partially explain why adipocyte size in conjunction with the FSH/AMH ratio associates so strongly with cyst formation. The effects of obesity on the reproductive system may thus be two-fold, both directly acting on ovarian structures and altering the endocrine axis of reproduction. For this reason, direct effects may only rarely be observed, calling for more complex models of obesity and reproduction which draw on a wider range of contributing factors. The establishment of such heterogeneous models, particularly through examination of direct effects of lipotoxicity on ovarian structures, should serve as the subject of future research. Additional future work should examine the relationship of the FSH/AMH ratio to fertility measures, such as litter size and breeding success, to link its morphological effects to reproductive outcomes.

Our study is not without its limitations. First, our ovarian histological assessments were based upon serial tissue sectioning, which introduces inherent bias into our results. Such an approach precludes us from making judgments regarding absolute numbers of ovarian features since only every fifth section was analyzed. However, our analyses were focused on comparisons of relative values between groups, and it is thus essential that the bias we introduce through serial sectioning be consistent across groups. By applying the same methodology uniformly across the study, we feel the relative comparisons offered in this study are valid. Also, we acknowledge that the classificatory scheme in which we established size thresholds based on morphological features for follicular assessment comes with its limitations. While serial sectioning may not capture the largest cross-section of a given follicle, consistent application of this methodology should not alter the validity of between group comparisons made in this study. In addition, we establish the significance of an FSH/AMH ratio despite a paucity of between group differences across a number of outcomes measured in this study. For at least some parameters, this lack of between group difference may stem from relatively small sample sizes, which reduce our statistical power to detect significance. However, despite a lack of between group differences in this study, the FSH/AMH continuously displayed strong and consistent associations with follicle, cyst, and corpora lutea counts across experimental groups. This primary finding acknowledges that the complex interaction of diet-induced obesity and reproduction may not produce clear between group distinctions yet pushes individual animals along a spectrum of abnormality.

In summary, our results establish the potential relevance of the FSH/AMH ratio in association with follicle counts. An elevation of this ratio displayed strong negative correlations with pre-antral, antral, and total follicle counts, suggesting an impairment of ovarian health. Furthermore, when combined with average adipocyte size, the FSH/AMH ratio was strongly linked with the formation of ovarian cysts. The FSH/AMH ratio displayed significant variability between groups, suggesting it is a highly individual yet accurate indicator of follicle status. It is proposed that elevation of this ratio occurs through a complex interaction with the metabolic system, providing direction for further studies.

Disclosure of Interest

The authors report no conflicts of interest.

Acknowledgments

The authors would like to thank Veronika Pogrebna, Katrina Volk, and Jennifer Zachry for their contributions to the implementation of the experiment. In addition, the researchers would like to thank the Summer Research Scholars and other Washington and Lee students who participated in the completion of this work.

Additional information

Funding

This work was supported by the Washington and Lee University Summer Research Scholar’s Program, Livesay Fellowship, and the Advanced Research Cohort;Washington and Lee University [Livesay Research Fellowship,Summer Research Scholarship].

References

Reprints and Corporate Permissions

Please note: Selecting permissions does not provide access to the full text of the article, please see our help page How do I view content?

To request a reprint or corporate permissions for this article, please click on the relevant link below:

Academic Permissions

Please note: Selecting permissions does not provide access to the full text of the article, please see our help page How do I view content?

Obtain permissions instantly via Rightslink by clicking on the button below:

If you are unable to obtain permissions via Rightslink, please complete and submit this Permissions form. For more information, please visit our Permissions help page.