A Case-Control Study of Follicular Fluid Cytokine Profiles in Women with Diminished Ovarian Reserve

Abstract

Ovarian reserve is an important determinant of a woman’s reproductive potential, and women with diminished ovarian reserve (DOR) often seek in vitro fertilization (IVF). The underlying etiology of DOR is unknown, but follicular fluid cytokine concentrations likely play a role in follicular development and maturation. The present study seeks to investigate the expression of cytokines in follicular fluid (FF) of women with DOR undergoing IVF and explore correlated functional pathways. 194 women undergoing ovarian stimulation were recruited at the time of oocyte retrieval. Women were classified as having DOR if they met one or more of the following criteria: AMH <1 ng/ml, FSH >10 mIU/ml and/or AFC <10. Controls included women undergoing IVF for male factor, tubal factor due to tubal ligation, or planned oocyte cryopreservation (non-oncologic). The concentrations of 480 cytokines and related growth factors in follicular fluid were determined using a multiplex immunoassay. Fifty-nine (59) cytokines had significantly different concentrations (53 higher and 6 lower) in the DOR relative to the control group after adjusting for age and body mass index (BMI) (false discovery rate; FDR<0.1). Using the most informative 44 biomarkers as indicated by a random forest (RF) model, an area under the curve (AUC) of 0.78 was obtained. Thus, follicular microenvironment differs between women with DOR and normal ovarian reserve. The differentially expressed cytokines belong to diverse processes that are primarily involved in follicular maturation and ovulation. These changes may play an important role in treatment outcomes in women with DOR.

Introduction

Infertility is a common reproductive health problem that affects more than 7 million (12–13%) women in the United States [1]. One of the most important determinants of a woman’s reproductive potential is her ovarian reserve, which is influenced by age, genetics, and environmental factors [2]. Reproductive aging is a natural process that centers on the generally accepted principle that women are born with a finite number of oocytes that peak during fetal life and decline in number thereafter through follicular atresia and apoptosis [2, 3]. While a decline in oocyte number and quality is a normal physiologic process as women age, some women experience diminished ovarian reserve (DOR), which refers to a more rapid decline in the follicular pool and reduced age-specific fecundability and reproductive potential [4]. It is characterized by the presence of regular menstrual cycles but abnormal ovarian reserve tests; the diagnosis affects 10–15% of women seeking fertility treatment [3, 4]. Diminished reserve negatively impacts potential oocyte (and consequently embryo yield) with in vitro fertilization (IVF) [5, 6], but likely does not cause infertility in isolation [7].

Although DOR can be due to genetic conditions, iatrogenic injury (e.g., chemotherapy) or environmental exposures [8], the etiological and pathological mechanisms of DOR in the majority of women are poorly understood. In addition, it is unclear whether DOR is due to increased follicular atresia, being born with a smaller than average follicular pool, or a combination of both [3].

Many women with DOR seek in vitro fertilization (IVF) to treat their age-related subfertility, and, since ovarian reserve is a determinant of embryo yield, it creates a challenge for patients undergoing IVF and their treating clinicians [9]. DOR can also have other health consequences, including an increased risk of early menopause and associated comorbidities, including an increased risk of osteoporosis and cardiovascular disease [10, 11].

One potential strategy to increase our understanding of DOR is by examining the follicular fluid that surrounds the oocyte during maturation. Follicular fluid includes plasma components transferred through the blood-follicular barrier and secretions from granulosa and theca cells [12]. A cascade of cytokines, chemokines and growth factors act upon the oocyte throughout the maturation process, hence their concentration in the follicular fluid affects oocyte viability and developmental potential [12, 13]. There have been some preliminary studies indicating altered concentrations of certain cytokines (e.g. interleukin-15) and related growth factors (e.g., amphiregulin) in follicular fluid of women with DOR [12–17].

Characterizing the cytokine and chemokine composition of follicular fluid and their molecular functions may not only provide a better understanding of intrafollicular processes and the regulation of ovarian reserve, but also help identify potential interventions that could modulate ovarian aging and IVF treatment outcome [14, 16]. The aim of our study is to investigate the expression of cytokines in the follicular fluid of women with DOR undergoing IVF and explore correlated functional pathways. Though it is a discovery-oriented design, we hypothesize that women with DOR will have different cytokine profiles relative to controls, and that DOR-associated cytokines will be in pathways related to oocyte maturation and ovulation. Preliminary results were presented at the American Society of Reproductive Medicine 2020 annual meeting [18].

Materials and Methods

Participants

Women undergoing ovarian stimulation and planned oocyte retrieval were recruited from the Emory Reproductive Center in metro Atlanta, Georgia USA. 194 subjects were enrolled between May 2018 and April 2019 in the parent study. A nested case-control study of enrolled participants was conducted. Medical record data including infertility factors, demographics, and cycle characteristics was abstracted by a qualified physician and used to determine eligibility for this study. Women were eligible for inclusion in this study if they underwent oocyte retrieval at the Emory Reproductive Center, were between 18–50 years of age, and had not been exposed to prior chemotherapy. For the main analysis, women with DOR were diagnosed if they met at least one of the following criteria: anti-Mullerian hormone (AMH) level <1 ng/mL, basal follicle stimulating hormone (FSH) on days 2–4 of the menstrual cycle >10 mIU/ml and/or a total antral follicle count (AFC) <10 at cycle start [3, 19]. Controls were defined as women undergoing IVF for male factor, tubal factor due to tubal ligation, or oocyte cryopreservation for planned fertility delay (non-oncologic). 61 women with endometriosis, tubal factor infertility secondary to hydrosalpinges, prior oophorectomy, major ovarian surgery, history of radiotherapy or chemotherapy, underlying malignancy, and chronic inflammatory conditions were excluded from the main analysis. To maximize power for this discovery-based approach, all subjects with a follicular fluid sample and clinical data were analyzed in either the main, sensitivity, or second-cycle analysis. All subjects provided written informed consent. This study was approved by the Emory University Institutional Review Board.

Biological Sample Collection and Processing

Before initiation of treatment, all participants had a vaginal ultrasound on day 2–3 of their cycle to ensure that no follicles ≥10 mm were present. A basal hormonal profile, including assessment of estradiol (E2) and progesterone (P4) levels, was also obtained. Participants underwent controlled ovarian stimulation (COS) using stimulation protocols based on the patient’s age, body mass index (BMI), and ovarian reserve parameters. The dose of gonadotropins was adjusted according to follicular development, which was monitored by ultrasound and serum E2 level until ovulation was triggered. Final oocyte maturation was achieved with human chorionic gonadotropin or a gonadotropin-releasing hormone (GnRH) agonist when at least 2 follicles achieved an average diameter size ≥ 18 mm. Ultrasound-guided oocyte retrieval was performed 35 hours later. At the time of oocyte retrieval, 70–100 mL of fluid was pooled from multiple follicles from the same participant. Generally this volume represents fluid from all follicles, but in some cases in which the patient had a very large number of follicles, volumes >100 mL were discarded. No additional inclusion criteria are imposed at the level of the sample. An average of 7.3 and 18.2 oocytes were retrieved from women with DOR and controls. Follicular fluid samples were processed by centrifugation at room temperature for 15 min at 1200 × g, after which the supernatant was stored at −80 °C until assayed. The follicular fluid samples were then assessed using multiplexed sandwich ELISA-based quantitative array platform.

Cytokine Detection using a Multiplex Immunoassay

The concentrations of 480 cytokines were measured in quadruplicate using quantitative, sandwich-based immunoassays as described previously [17]: Quantibody^® Human Cytokine Antibody Array Q440 (cat. No. QAH-CAA-440) and Quantibody Human Cytokine Antibody Array Q12 (catalog number QAH-CYT-12; RayBiotech Life, Inc.; Peachtree Corners, GA). Follicular fluid samples were diluted 2-fold with Quantibody Sample Diluent and incubated on the glass-based antibody arrays following the manufacturer’s protocol. After processing, the slides were scanned with the Innopsys 710AL scanner in XDR mode, and the fluorescent signals were extracted with Magpix software.

Data normalization and calculation of cytokine concentrations based on a standard curve were performed using Analysis Tool software specific to the Quantibody^® arrays (RayBiotech Life, Inc.; Peachtree Corners, GA). First, outliers were removed. Outliers were defined as 35% above the median of the quadruplicate data points for each target since capture antibodies were printed in quadruplicate. Second, the average of the remaining replicates was used to calculate the cytokine concentration. Third, inter-array normalization was performed using positive control spots printed on each array. The positive control spots included two dilutions of biotinylated bovine IgG (POS1 and POS2) representing high and median signal intensities, respectively. Inter-array normalization was also performed using a mixture of standards with signals in the middle of the standard curves, which was run on all slides.

Statistical Analysis of Array Data

Statistical analysis of Quantibody^® array data was performed using the R statistical software. The demographic variables and IVF cycle characteristics were compared in the DOR and control groups using the Student’s t-test with results expressed as mean ± SD. Significance was assumed at a p-value (p) ≤ 0.0004, to adjust for multiple testing.

The raw data underwent quality control (QC) at both the cytokine and sample level before statistical analysis. At the cytokine level, only those with >50% of values within the confidence range were included. The confidence range for each cytokine was defined as three times the lower limit of detection (LOD) to the maximum limit of detection. At the sample level, outliers, which were identified via principal component analysis (PCA), were excluded from further analysis (Online Resource Figure 1).

Quantile regression was used to identify biomarkers significantly associated with DOR after adjusting for age and body mass index (BMI). A false discovery rate (FDR) below 10% was considered significant. To determine if a group of cytokines were more predictive of DOR status, random forest (RF) modelling was applied to the DOR-associated results from the quantile regression. Receiver operating characteristic (ROC) analysis was performed, and the area under the curve (AUC) was used as an estimate for the predictive accuracy of the panel of biomarkers.

To further investigate the generalizability of our results, we repeated the quantile regression model in a group of patients with other etiologies of infertility that did not meet inclusion criteria for the primary analysis. This included 7 DOR patients with at least one additional infertility factor and a non-DOR group comprised of 48 patients with other etiologies, including chronic anovulation and endometriosis. For the cytokines associated with DOR in our primary analysis, we repeated the quantile regression analysis in this independent group and compared the effect sizes from the primary analysis to those observed in patients with other etiologies using a Pearson correlation.

Finally, to validate the reproducibility of FF cytokine profiles at an individual level, we analyzed the FF from 19 participants who underwent a second IVF cycle. The most common reasons necessitating additional cycles included failure to achieve pregnancy after transfer of all available embryos, lack of euploid embryos after preimplantation genetic testing for aneuploidy (PGT-A), and cryopreservation of additional oocytes for planned fertility delay. Pearson correlations were applied to assess correlation between the FF cytokine concentrations of an individual’s first and second cycles.

RESULTS

194 subjects with cytokine data available were eligible for inclusion into the study. 8 subjects were identified as outliers and were removed. There was no significant difference in case-control status between outliers and subjects included in the main analysis. After quality control, 131 subjects were eligible for the main analysis with 56 women in the DOR group and 75 women in the control group. Demographic features of study participants and their IVF cycle characteristics are presented in Table 1. Participants in the control group were younger than participants with DOR (p<.004). BMI was not significantly different between the two groups. As expected, since plasma AMH, AFC and basal FSH were the criteria used to define DOR, the values of AMH and the AFC were lower (p<.004) in the DOR group compared to the control group. In addition, the number of follicles >14 mm on trigger day p<.004), oocytes retrieved (p<.004), metaphase II oocytes (MIIs; p<.004), and available embryos on day 3 (p<.004) and day 5 (p<.004) were lower in the DOR compared to the control group. An additional 55 samples were included in an analysis of patients with additional etiologies of infertility (Online Resource Table 1). There were no missing values for DOR criteria, age, BMI, or cytokine concentrations.

Table 1.

Participant Demographics and IVF Cycle Characteristics

Characteristics	DOR (N=56)	Control (N=75)
	Mean (SD)	Mean (SD)	T Statistic	P-value
Age (years)	37.7(3.8)	34.4(4.05)	4.7	6.36E-06
BMI (kg/m2)	26.2(5.8)	26.6(6.1)	−0.4	6.79E-01
AMH (ng/ml)	0.73(0.69)	3.4(2.0)	−10.9	< 2.2e-16
AFC, N	7.9(3.6)	21.2(8.7)	−11.9	< 2.2e-16
Basal FSH (mIU/ml)	9.37(3.2)	7.59(6.4)	1.6	1.18E-01
E2 (pg/ml)	46.4(21)	42.2(21.7)	0.9	3.53E-01
Follicles >14mm on day of trigger, N	5.7(3.4)	12.3(4.1)	−10.0	< 2.2e-16
Oocytes retrieved, N	7.3(4.4)	18.2(8.0)	−9.9	< 2.2e-16
Mature oocytes (MII), N	5.3(3.3)	13.5(7.1)	−8.7	3.96E-14
Cleavage embryos, N	3.5(2.2)	9.9(5.6)	−7.6	7.33E-11
Blastocyst embryos, N	1.59(1.3)	4.46(2.8)	−6.1	4.53E-08
Total gonadotropin dose (IU)	5278(2566)	4882(2004)	1.0	3.42E-01
E2 on day of trigger (pg/ml)	1524(796)	2997(1063)	−9.1	1.74E-15

Note: Continuous data are presented as mean ± standard deviation. Categorical data are presented as number (percentage). BMI= body mass index; AFC = antral follicle count; AMH = anti-Mullerian hormone; FSH = follicle-stimulating hormone; E2: Estradiol

Cytokines with significantly different FF concentrations in patients with DOR compared to controls, controlling for age and BMI are presented in Table 2. Unadjusted comparisons are presented in Online Resource Table 2. Of 289 cytokines sufficiently expressed in FF, 59 had significantly different concentrations between the two groups (FDR<0.1), with 90% having lower concentrations in DOR compared with controls. For example, cytokines with lower concentrations in FF of patients with DOR include several growth factors, including amphiregulin (AR; Figure 1A). We found lower concentrations of multiple inflammatory mediators essential for ovulation including macrophage inflammatory protein-1α (MIP1A), disintegrin and metalloproteinase domain-containing protein 9 (ADAM9) and tissue inhibitor of metalloproteases 4 (TIMP4). We also found higher concentrations of macrophage migration inhibitory factor (MIF; Figure 1B) and chemokine ligand 14 (CXCL14), a novel chemokine expressed by a variety of immune and non-immune cells, in the FF of patients with DOR.

Table 2.

Fifty-nine cytokines associated with DOR after controlling for age and BMI (FDR <0.1).

Target ID	Entrez ID	Slope	Standard Error	T Statistic	P-value
Fas	355	−86.5	18.6	−4.7	7.98E-06
HAI2	10653	−57.6	13.4	−4.3	3.64E-05
TNF-R-I	7132	−1114.6	261.5	−4.3	3.92E-05
MIP1A	6348	−2	0.5	−4.2	5.78E-05
LIMPII	950	−39.6	9.6	−4.1	6.51E-05
HGF	3082	−1361.5	330.9	−4.1	6.91E-05
DR6	27242	−545.7	136.1	−4	0.0001
TGFBR3	7049	−1989.3	503.1	−4	0.0001
NKp44	9436	−146.3	37.3	−3.9	0.0001
B7-2	942	−74.6	19.5	−3.8	0.0002
TNF-R-II	7133	−899.8	243.4	−3.7	0.0003
AMICA	120425	−556.6	151.6	−3.7	0.0004
ADAM9	8754	−54.6	15.1	−3.6	0.0004
TIMP4	7079	−426.2	119.4	−3.6	0.0005
F7	2155	−1489.7	419	−3.6	0.0005
GDF15	9518	−105.9	31.1	−3.4	0.0009
IL12B	3593	−9.2	2.7	−3.4	0.0009
Shh.N	6469	−19.9	6	−3.3	0.0011
ESAM	90952	183.2	55.1	3.3	0.0011
Legumain	5641	−749.2	227.2	−3.3	0.0013
TLR2	7097	−98	30.7	−3.2	0.0018
MIP-1-beta	6351	−1.8	0.6	−3.1	0.0021
LYVE 1	10894	−159.9	51.3	−3.1	0.0023
SELE	6401	−873.5	280.7	−3.1	0.0023
TNFRSF13B	23495	−715.7	230.8	−3.1	0.0024
Cadherin 3	1001	−3240.6	1058.1	−3.1	0.0027
BCMA	608	−275.2	90.1	−3.1	0.0027
TYRO3	7301	−329.3	108.6	−3	0.0029
DPPIV	1803	1218.5	415	2.9	0.0039
AXL	558	−94.9	32.4	−2.9	0.0041
SDC 1	6382	−208.6	73.8	−2.8	0.0055
PROK1	84432	−224.7	80.4	−2.8	0.0060
Galectin 1	3956	−260.3	93.9	−2.8	0.0064
FGF 5	2250	−92.4	33.4	−2.8	0.0064
CDH 13	1012	−432.9	156.3	−2.8	0.0065
MIF	4282	26.4	9.6	2.8	0.0066
CXCL14	9547	162.4	59.3	2.7	0.0071
CD163	9332	−3310.5	1211.2	−2.7	0.0072
LGALS7	3963	240.5	88.3	2.7	0.0074
AGRP	181	−65.5	24.3	−2.7	0.0080
GPRASP1	9737	−198.6	74.5	−2.7	0.0087
IGFBP4	3487	−19137.5	7253.2	−2.6	0.0094
CD6	923	850.6	323.7	2.6	0.0097
CD23	2208	−34.6	13.2	−2.6	0.0097
FLT4	2324	−74.8	28.5	−2.6	0.0099
AR	374	−299.5	115.2	−2.6	0.0104
ANGPTL4	51129	−516.6	199.6	−2.6	0.0108
IL1R2	7850	−72.4	28.2	−2.6	0.0113
Prolactin	5617	−2639.9	1043.7	−2.5	0.0126
Endocan	11082	−50.8	20.2	−2.5	0.0131
G-CSF	1440	−6	2.4	−2.5	0.0137
L1CAM2	10752	−2248	900.2	−2.5	0.0138
IL21R	50615	−119.9	48.5	−2.5	0.0148
CEACAM1	634	−156.3	63.9	−2.4	0.0158
Granulysin	10578	−488.7	200.1	−2.4	0.0159
TIM1	26762	−7.5	3.1	−2.4	0.0162
Angiostatin	5340	−15323.1	6319.4	−2.4	0.0167
4-1BB	3604	−5.8	2.4	−2.4	0.0185
FLT3L	2323	−4.4	1.8	−2.4	0.0186

Figure 1.

Box plots show two of the 59 cytokines with different FF concentrations in individuals with DOR compared to controls (FDR<0.1). Individuals with DOR had significantly lower concentrations of AR and significantly higher concentrations of MIF. AR = Amphiregulin, MIF = Macrophage inhibitory factor.

The predictive value of cytokines with significantly different concentrations in FF of patients with DOR compared with controls for DOR state was evaluated in an RF model to determine the degree to which multiple cytokines may be more predictive than individual ones. We identified a set of 44 cytokines that generated an AUC of 0.78, suggesting a high predictive capacity. The was no significant improvement of AUC when all significantly different cytokines were used.

In order to determine the degree to which the results were generalizable, we compared the effect sizes for each cytokine associated with DOR in our primary analyses to those observed in patients with additional etiologies of infertility. 55 women were eligible for inclusion; 7 women were in the DOR group and 48 women were in the control group. We found a strong correlation (R=0.94, p<2.2e-16) in the magnitude and direction of effect (Figure 2), which further supports our findings that indicate an altered follicular microenvironment in patients with DOR compared to those with normal ovarian reserve even in the presence of additional infertility factors.

Figure 2.

Histogram of the correlation coefficients for the FF cytokine profiles of individuals that underwent two consecutive IVF cycles. The X-axis is the correlation coefficient of the FF cytokine concentrations from cycle 1 and cycle 2 per individual. The Y-axis represents the number of individuals within the corresponding correlation coefficient.

Finally, we sought to evaluate the degree to which follicular fluid cytokines are stable by examining the cytokine profile of 20 women who underwent more than one IVF cycle. Of those, one woman with a duplicate sample was excluded from the analysis due to a problem with her trigger shot injection. After quality control, 19 women (63% DOR) that underwent more than one IVF cycle remained. On average, the time between the two cycles was 2.68 months. Seven out of 19 (37%) patients underwent a different stimulation protocol for their second cycle, but, on average, the total gonadotropin dose was comparable between the first and the second cycle. As shown in Figure 2, cytokine levels were highly correlated between cycles for each woman (0.86<R<0.99). These findings suggest that the FF cytokine profile of an individual tends to remain stable under the influence of various stimulation protocols and over time.

DISCUSSION

The results of our study demonstrate that there are differences in the follicular microenvironment in women with DOR compared to individuals with normal ovarian reserve. Follicular fluid is a dynamic medium, and its biochemical properties, including its cytokine profile, play a critical role in determining oocyte quality and the subsequent potential to achieve fertilization and embryo development [16]. Although much is known about ovarian reserve testing and predicting a poor response to COS, our knowledge of differences in FF components, including the cytokine profile and their potential role in determining follicular fate, is limited. In our study, cytokines and growth factors with roles in various signaling pathways, including follicular growth, oocyte maturation, and inflammation, showed different concentrations between the two groups.

The results of this study suggest that there may be higher levels of intrafollicular inflammation and lower levels of growth factors that support oocyte maturity and maturation in women with DOR compared to those without. Increased levels of certain proinflammatory cytokines were indeed observed in the FF of individuals with DOR. We found higher concentrations of MIF, a proinflammatory cytokine that plays various roles in chronic inflammatory conditions including endometriosis [20–22], as well as CXCL14, a chemokine that is upregulated in response to inflammation and thus implicated in multiple chronic inflammatory conditions [23, 24]. However, we were not able to replicate associations with IL-15 concentrations found by Spanou and colleagues, though other studies have also not shown an association between DOR and IL-15 [12, 13].

We found lower concentrations of several growth factors in the FF of patients with DOR. Amphiregulin (AR) is an epidermal growth factor (EGF) like peptide that has been shown to be the most abundant and important epidermal growth factor receptor (EGFR) ligand in human follicular fluid [25, 26]. Multiple studies have shown that expression of AR and EGFR are rapidly induced by human chorionic gonadotropin (hCG) stimulation of granulosa cells, where it induces cumulus expansion and oocyte maturation [25, 27]. In support of these findings, Zamah and colleagues showed that immunodepletion of AR abolishes the ability of the FF to stimulate cumulus expansion and oocyte maturation in humans [27]. Lower concentrations of AR in FF of patients with DOR compared to those with normal ovarian reserve may result in lower oocyte maturation and fertilization potential, and ultimately lower pregnancy rates in this population.

Lower concentrations of granulocyte colony stimulating factor (G-CSF) in the FF of patients with DOR were measured in our study. Studies in patients undergoing IVF demonstrated higher levels of G-CSF in FF than in serum, pointing to a critical role in folliculogenesis [28]. In addition, G-CSF administration has been shown to improve the clinical pregnancy rate (CPR) in patients undergoing IVF [29]. Lower G-CSF concentrations in FF of women with DOR may also result in lower chemotactic activity of FF towards leukocytes which is an essential step in a successful ovulation [30].

Additional growth factors with lower concentrations in the FF of patients with DOR include hepatocyte growth factor (HGF) and its downstream component, toll like receptor 2 (TLR2). HGF is involved in follicular development and the oocytes’ developmental competence, including reduction of apoptosis of granulosa cells, organization of the extracellular matrix, tissue remodeling, neovascularization, and luteinization [31]. Kawano, et al. reported that the HGF level in FF is correlated with oocyte maturation and could serve as a marker of oocyte competence [32]. In a more recent study, Sahin and colleagues reported a significant correlation between HGF in the FF and fertilization rate. In addition, the authors reported that higher HGF levels were associated with higher-grade embryos [33].

We also found that the concentration of sonic hedgehog (Shh) was lower in the FF of individuals with DOR. The hedgehog family of proteins is essential for growth, differentiation, and morphogenesis of multiple organs during embryogenesis and postnatal life [34, 35]. While our knowledge about its role in regulating reproduction and follicular growth is rather limited, Shh may have a role in oocyte maturation. Nguyen, et al. demonstrated activation of the Shh signaling pathway in porcine cumulus oocyte complexes (COCs), with highest levels in small follicles as compared to medium and large size follicles. They also showed enhanced oocyte maturation after Shh supplementation to intact and denuded COCs, which was reversed by addition of cyclopamine, an Shh inhibitor [36]. Others have found in vitro treatment of mouse preantral follicles with Shh increased follicular growth, which was primarily associated with proliferation of granulosa cells [37]. Shh also increases proliferation of theca cells and consequently increases androstenedione production in cells from large and small follicles in mammals [38].

During ovulation in response to the luteinizing hormone (LH) surge, the entire follicle undergoes profound remodeling including proteolytic degradation, inflammation and angiogenesis. Ovulation is an inflammation-like reaction that requires active participation of leukocytes and their inflammatory mediators, with studies showing an influx of leukocytes and their presence in the ovary after the LH surge [38–42]. MIP-1A is a chemokine that is induced by cytokines (IL1, IFN -γ) and is the main chemoattractant for monocytes/macrophages, which are an essential leukocyte in ovulation [39]. MIP-1A potentiates monocyte/macrophage activation by IFN-γ [42]. In turn, higher levels of IFN-γ have been reported in the FF of embryos that cleave early, which has been described as a good prognostic factor [41]. Our results of lower concentrations of multiple inflammatory mediators, including MIP-1A, TIMP4, B7-2 and G-CSF suggest that downregulation of various processes required for ovulation occur in DOR patients.

Based on our findings, we could hypothesize that the higher concentrations of some inflammatory cytokines and/or lower concentrations of certain growth factors contributed to the diminished ovarian reserve, however, ultimately, we are not able to conclude whether the cytokine concentrations are one of the causative factors for the DOR pathology or are a downstream effect of having a diminished reserve. If, however, further future studies find continued evidence of a pro-inflammatory contribution to DOR, treatments aimed at anti-inflammation could be implemented in a clinical setting.

This study had a number of strengths and limitations that should be considered. First, compared to previously published reports, we were able to include a larger number of sample controls and women with DOR. This improves the reliability of our results with respect to differences identified between the two groups. A second strength of our study is the large number of cytokines examined, which enables a broad assessment of biological processes and molecular pathways involved in folliculogenesis, oocyte maturation, and ovulation. Other studies have investigated a limited number of pathways or a limited number of mediators within a certain pathway. Lastly, by optimizing the exclusion criteria, we believe our control group is an excellent representation of women with normal ovarian reserve without underlying conditions that would modify the intrafollicular environment.

Potential limitations of this study include the fact that women were assigned to different stimulation protocols for COS; this is consistent with other studies since the stimulation protocol and medication dose are primarily based on ovarian reserve parameters. Previous studies have shown that exposure to different gonadotropin concentrations may affect the follicular microenvironment, including endocrine and cytokine profiles as well as gene expression [40, 43]. To examine the extent to which this affects our findings, we analyzed the FF of patients who underwent two sequential cycles and found a strong correlation (R=0.92) despite changes in stimulation protocols in 7 of the 19 participants. Another limitation is that we used pooled FF samples and were not able to correlate each follicle with oocyte and embryo development. Bearing that in mind, we believe that analyzing the overall cytokine profile is a reliable and comprehensive method for detecting the differences in FF cytokine profile between the two groups. While we examine the expression of a large number of cytokines in the FF in this study, it is important to mention that cytokines cover a fraction of the many components involved in follicular growth, oocyte maturation, and function.

In summary, our results suggest downregulation of processes primarily involved in follicular growth and oocyte maturation as well as an altered immunologic microenvironment with increased intra-follicular inflammation in individuals with DOR. We also show downregulation of multiple signaling cascades involved in various aspects of ovulation, which may partly explain increased incidence of ovulatory dysfunction and abnormal luteinization in this population.

Availability of data and material:

Data is available from the authors upon reasonable request.