Dehydroepiandrosterone supplementation and the impact of follicular fluid metabolome and cytokinome profiles in poor ovarian responders

Poor ovarian responders (POR) are a sub-group of infertile women that account for 9–26% of in vitro fertilization (IVF) indications [1, 2]. In patients designated as “poor responders,” so-called due to poor response to ovarian stimulation given during IVF workup, the limited number of obtained oocytes remains the major problem in optimizing the live birth rates [3]. As a result of a lower number of oocytes retrieved, there are fewer embryos to select and transfer, and subsequently these patients have lower pregnancy rates per transfer and lower cumulative pregnancy rates per started cycle compared with normal responders. In PORs the mechanism of ovarian insufficiency can be multifactorial with causes such as ovarian surgery especially in case of endometrioma [4, 5], uterine artery embolization for the treatment of uterine leiomyoma [6, 7], genetic defects, chemotherapy, radiotherapy, autoimmune disorders, single ovary, chronic smoking [8, 9], or linked to diseases such as diabetes mellitus Type I [10]. However, in most cases, follicular depletion plausibly reflecting premature ovarian aging [11], clinically translates into a reduction of implantation rates, an increase of early pregnancy loss, and disappointingly low IVF success [12, 13].

In each menstrual cycle, human ovaries produce a single dominant follicle. Growth of the dominant follicle encompasses enlargement of the oocyte, replication of follicular cells, and formation and expansion of a fluid-filled follicular antrum or cavity, providing a specialized microenvironment for the development of oocytes. Follicular fluid (FF) that fills the antrum cavity is derived from the surrounding theca capillaries, abundant and easily accessible during IVF procedures due to ample volume being produced during follicle maturation [14]. FF are rich in metabolites, notably hormones, amino acids and lipids that are critical for oocyte growth and development, which determines subsequent potential to achieve fertilization and embryo development [15]. As such, constituents of the FF surrounding the oocyte provides a unique biochemical window to the growth and differentiation of the oocyte [16]. To unravel the biochemical composition of human FF and its impact on oocyte development, metabolomic analyses using gas chromatography–mass spectrometric and proton nuclear magnetic resonance metabolomic analyses have been conducted [17‐22], as were proteomic analyses [23‐28]. These studies mainly report the FF profiles of IVF patients, whereas the FF metabolome of POR remains poorly characterized. Furthermore, the effect of dehydroepiandrosterone (DHEA) on the FF metabolome has not been previously studied.

DHEA is a steroid produced in the adrenal cortex and the ovarian theca cells in women that is converted into more active forms of androgens such as testosterone [5]. It has been suggested that DHEA supplementation may increase the number of available follicles in PORs through an increased serum level of insulin-like growth factor, increased follicular response to follicle stimulating hormone (FSH), shifts to aerobic metabolism [29] and improved quality of oocytes [30]. However, the efficacy of DHEA pre-treatment has been controversial, with partial to reasonable clinical evidence being observed [31‐37]. Based on these findings, we conducted a study to evaluate whether the FF metabolome differed in POR patients treated with DHEA or not, and whether the FF metabolome may be predictive of IVF outcome. Furthermore, DHEA has immunoregulatory functions [38], and a large-scale study of FF cytokines was conducted in parallel to reveal DHEA immunomodulatory targets.

In this study, we report the first, and largest metabolome and large-scale cytokine study to-date of the FF of POR/DOR patients with DHEA supplementation. A total of 118 metabolites that included lipids, fatty acids, glucocorticoids, hormones, bile acids, peptides, and 22 cytokines were identified in the FF of POR/DOR patients. Among these, four FF metabolites, namely glycerophosphocholine, linoleic acid, progesterone, and L-valine varied significantly between DHEA + and DHEA- control subjects. This LC–MS/MS metabolomics study extends the human FF metabolome space in terms of characterization of its constituents, providing new insights into the complexities of oocyte development especially in POR women [48], as well as with DHEA supplementation [17‐19, 49, 50].

The identification of four differential metabolites, glycerophosphocholine, linoleic acid, progesterone, and valine in this study may alert us to the metabolic effects of exogenous DHEA supplementation and plausibly using them to achieve improved outcomes [36, 51, 52]. Choline and derivatives are an emerging class of metabolites critical in developmental competence of fertilized oocytes [19], and glycerophosphorylcholine was found increased in the DHEA + group. Glycerophosphorylcholine is formed from the breakdown of phosphatidylcholine, and is an organic osmolyte, plausibly affecting concentrations of other constitutes of FF [16], and regulation of the diffusion of compounds into FF necessary for folliculogenesis and oogenesis [53]. PORs are known to exhibit a low diffusion of exogenous gonadotropin into FF, which is correlated with poor IVF outcomes [54]. It is conceivable that DHEA induced the metabolism of phosphatidylcholine to glycerophosphorylcholine. Progesterone is one of the key hormones for the progress of the first meiotic division in oocyte maturation, but changes to progesterone levels with DHEA supplementation has been controversial [55]. Our metabolomics study revealed for the first time that DHEA supplementation led to a decrease in FF progesterone levels, but to what impact lower progesterone induced by DHEA supplementation has on PORs remains to be investigated. Valine degradation has been previously reported in a proteomics study comparing competent versus incompetent buffalo oocyte proteome [56]. In a metabolomics study, valine metabolism was also identified in bovine cumulus and cumulus-oocyte-complex-conditioned media that undergo oocyte maturation [57]; although in both studies, valine was not directly detected in the omics profiling. In humans, degenerate oocytes or germinal vesicles that failed meiotically to reach metaphase II deplete valine more than competent oocytes. In other words, lower valine levels in culture media which is consistent with our results, and suggest plausible biological roles of valine in oocyte maturation. Interestingly, we noted a segregation of DHEA + patients with low and high level of valine, with the high valine group approaching concentrations of the DHEA- control group. Together with valine’s high AUC value in DHEA + , and that valine negatively correlated with testosterone, it is tempting to speculate that valine can be used as biomarker for monitoring individual DHEA supplementation. Linoleic acid is the most abundant polyunsaturated fatty acid in bovine [58] and human FF (Fig. 2), and varying concentrations of linoleic acid have reportedly different effects on oocyte maturation. At a concentration of 100 µM, linoleic acid added to maturation media inhibits bovine oocyte maturation and subsequent blastocyst development through increasing prostaglandin E₂ concentration in the medium, decreasing intracellular cAMP, decreasing phosphorylation of the MAPK1 and AKT and inhibited germinal vesicle breakdown [58, 59]. Conversely, at concentrations at 50 µM or below, linoleic acid improved oocyte quality by increasing the content of neutral lipids stored in lipid droplets [59]. FF linoleic acid’s high AUC value and its correlation with estradiol suggest that can be another biomarker for titrating and monitoring individual DHEA supplementation.

The elevated DHEA-sulphate levels coupled with a lack of difference in FF testosterone with DHEA supplementation suggest the following possibilities in POR patients: (i) inadequate DHEA conversion to testosterone due to polymorphism in SULT2A1, CYP19A1 and FMR1 genes [60], or (ii) long CAG repeats in androgen receptor gene which is linked to its lower transcriptional activity at the promoters of genes involved in the metabolism of DHEA to testosterone [61]. The former is unlikely: in a case–control study involving 94 subjects, androgen secretion was not impaired in pre-ovulatory follicles of POR compared to normal responders, and similar levels of follicular testosterone levels was reported [62]. However, ethnicity and genetic predispositions might play a role as Chinese women are reported to have higher free androgens and African American women lower [63], which might explain their differences in pregnancy rates in association with IVF than those observed among other ethnic groups. Conversely, long CAG repeats is associated with risk of POR and oocyte insensitivity to androgenic stimulation [64], thus hinting a tenable rationale on the observed similar FF androgen levels between the DHEA- controls and DHEA + subjects in this study and others [36]. The abundance of cortisol in DHEA + subjects is interesting, in particular that DHEA reduces circulating cortisol [65], indicating follicular versus systemic difference in how DHEA affects cortisol levels. In vitro, it was noted that DHEA suppresses cortisol activity [66], including the antagonist effects of DHEA on the anti-inflammatory responses induced by cortisol via glucocorticoid receptor-mediated pathways [67]. It is noteworthy that high FF cortisol levels found in fertilized IVF individuals compared to unfertilized individuals led to the postulation that oocyte exposure to cortisol is required with oocyte maturation [68]. The higher levels of FF cortisol observed in DHEA + subjects therefore argues for a compensatory response to modulate the ratio of the two hormones in an attempt to maintain a favourable FF response to mature oocytes [67].

In mouse models of polycystic ovary syndrome, treatment with DHEA resulted in increased production of cytokines such as serum TNFα, IL-6, IL12p70, and IFNγ [69, 70]. In this study, DHEA supplementation led to the reduction of FF IFNγ, LIF, MCP-1, and VEGF-D levels. It appears that DHEA modulates chemokines and growth factors in POR FF without a clear Th1 or Th2 immune response as proposed [52]. LIF or leukemia inhibitory factor is expressed in the ovary and controls follicular growth [71]. It was reported that LIF suppressed the growth of primary, secondary, and early antral follicles in cultured ovarian tissues [72]. The authors postulated that LIF produced in the late antral or graafian follicles is secreted to suppress the growth of the neighbouring primary, secondary, and early antral follicles as part of follicular growth [16]. Interestingly, when hCG is administered in rhesus macaques, at 12 h follicular LIF levels increase and induce follicle rupture and ovulation and decrease at 24 h [73]. In our study, the number of MII oocytes and oocytes trended higher in the DHEA + group, suggesting that the biological roles of LIF might have been achieved (follicular maturation and rupture) but inadequate to generate a clinically significant outcome. In vitro results suggested that follicles produce VEGF-A, with VEGF-A inducing the expanding vasculature to support the increased needs growing follicles [74]. The decrease in VEGF-A in DHEA + individuals is intriguing. Fisher et al., described that in cultured follicles, the rise in VEGF-A levels in faster-growing follicles are dependent on FSH dose and oxygen tension [75]. There have been reports that DHEA inhibits oxygen consumption in neurons [76], tempting the postulation that DHEA inhibited oxygen consumption in follicle that subsequently led to lower production of VEGF-A in DHEA + individuals. Further, the lack of correlation between the significant cytokines and metabolites suggests that DHEA converting to steroids which subsequently modulate cytokine production within the follicular microenvironment is more complex than originally thought.

We note various strengths of this study. Firstly, due to the highly confident identification based on MS/MS, and mass accuracy of LC–MS/MS-based metabolomics, we were able to distinguish progesterone from DHEA, an advantage over interference-prone immunoassays that face a cross-reactivity bioanalytical problem [55]. Similarly, LC–MS/MS-based determination of androgens was preferred over immunoassays due to strong interference from DHEA [77]. We did not detect E1 and E2; because for phenolic hydroxyl group of estrogens to act as proton donors, the signal would be more sensitive in the negative ion mode electrospray ionization [78] than in the positive ion mode which was used in this study. Aside from previously reported constituents of FF such as linoleic acid [17, 22], amino acids [18], and steroids including progesterone, testosterone [79], this study also captured metabolic products of ovarian steroidogenesis, cholesterols and glucocorticoids in the FF. Secondly, this study is the largest-to-date, providing a global view, specifically of the effect of DHEA on the FF metabolome and cytokine profile in POR patients. Thirdly, the recruitment of women of relatively advanced age, low mean number of recovered oocytes (4.9) and low clinical pregnancy rate (5/52 or 9.6%), which contrast with other studies [22, 33] suggest the appropriate inclusion of POR patients.

Many studies of POR, including this study, suffer from important limitations. Firstly, POR patients represent a heterogeneous group of patients of different prognosis and with a range of patient and biochemical characteristics. To-date, no definition has been able to correlate their presentation with pregnancy or live-birth prognosis [80, 81], contributing to the difficulties in designing studies and trials to evaluate therapeutic modalities. Secondly, with respect to the similar clinical pregnancy rate between DHEA + and DHEA- subjects, this might be construed as a limitation in terms of sample size. However, various studies, including randomized clinical trials have not led to a clear outcome in terms of an improvement in clinical pregnancy rate, the number of oocytes retrieved, and embryos formed [31‐37].

In conclusion, our study provided new insights to POR FF at the metabolome level, and as indicated from the FF metabolome analysis, exogenous DHEA to these patients altered the overall metabolome coverage and abundance to four metabolites. LC–MS global (untargeted) metabolomics analysis has provided the ability to reveal biologically relevant changes within a system, even at sensitive ranges before the precedence of gross morphological or phenotypical changes [82]. Hypotheses generated from this study included plausible mechanisms underlying DHEA metabolism, and the potential utility of glycerophosphocholine, linoleic acid, progesterone, and L-valine as markers to assess DHEA supplementation. Therefore, future directions include targeted quantitative LC–MS/MS approaches to be developed to detect and quantify four “responder” metabolites in approaches similar to those previously conducted on human peritoneal fluids and sera [83‐85] to design treatment based on metabolomics profiles. Steroid hormones including testosterone should also be quantified via LC–MS/MS to establish baseline levels before commencing DHEA supplementation. Further, comparing POR and normal responders will provide further insights to the alteration of the FF metabolome, and reach a deeper understanding of underpinning pathophysiology to PORDisclosure of conflict-of-interest statement.