The Role of Androgen Supplementation in Women With Diminished Ovarian Reserve: Time to Randomize, Not Meta-Analyze

The management of patients with diminished ovarian reserve (DOR) remains one of the most challenging tasks in IVF clinical practice. Despite the promising results obtained from animal studies regarding the importance of androgens on folliculogenesis, the evidence obtained from clinical studies remains inconclusive. This is mainly due to the lack of an evidence-based methodology applied in the available trials and to the heterogeneity in the inclusion criteria and IVF treatment protocols. In this review, we analyze the available evidence obtained from animal studies and highlight the pitfalls from the clinical studies that prevent us from closing the chapter of this line of research.


INTRODUCTION
In women, testosterone and dihydrotestosterone (DHT), the bioactive androgens that bind directly to the androgen receptor (AR), are produced by peripheral conversion of androgen precursors (androstenedione, dehydroepiandrosterone and dehydroepiandrosterone sulfate) that are secreted from both the ovary and adrenal gland (1,2).
The AR is expressed at all levels of the female hypothalamic-pituitary-gonadal axis (2). In the ovary, the AR has been detected in several stages of oocyte development from the primary stage onwards, as well as in the ovarian stroma (3). The fact that hyperandrogenic women present an increased number of small antral follicles suggests a role for androgens in both follicular development and follicular arrest. Clinical examples of this effect include polycystic ovarian syndrome (PCOS) and congenital adrenal hyperplasia patients (4). On the other hand, although initial studies using histomorphologic criteria suggested that exposure to exogenous testosterone treatment in female-to-male transexual patients induced polycystic ovary morphology (5,6), more recent studies using both histologic and ultrasound criteria have not confirmed these findings (7)(8)(9).
Circulating androgen levels have been reported to decline with age, especially during the earlier reproductive years (10). Similarly, the reproductive aging process consists of a gradual reduction in oocyte quantity and quality, with a consequent age-related decrease in the reproductive potential (11,12). In the light of these findings, IVF centers have initiated androgen pretreatment in patients with diminished ovarian reserve, intending to improve their reproductive outcomes. In fact, a recent survey has shown that more than 40% of physicians in Europe and Australia are prescribing off-label androgens in this subgroup of patients (13). However, the evidence for including this approach in our clinical practice is scarce.
The aim of this review is to analyze the available evidence from animal studies regarding the impact of androgen supplementation on folliculogenesis, as well as the drawbacks from clinical studies that might preclude the obtention of definitive conclusions to guide an evidence-based approach for such a challenging population.

METHODS
The Cochrane Central Register of Controlled Trials (CENTRAL), MEDLINE via PubMed, the Web of Science and Scopus were screened with a combination of keywords related to ART, poor responders, diminished ovarian response, androgens, testosterone and DHEA in various combinations. The search period was from the date of inception of each database until 1 December 2020. Only full text papers published in English were included.

Primordial Follicle Initiation
Previous studies in primates have shown that androgens increase the numbers of small-and medium-sized follicles but not large preovulatory follicles (14). In particular, testosterone and DHT pretreatment increased the number of primary follicles. Also, they resulted in a significant increase in insulin growth factor I (IGF-I) and IGF-I receptor mRNAs in the oocytes of primordial follicles, suggesting that androgen-induced activation of oocyte IGF-I signaling may trigger primordial follicle growth (15). More recently, mouse studies have corroborated that testosterone promotes primordial follicle to primary follicle transition via an AR-mediated pathway rather than by transformation into estradiol (16).

Preantral to Antral Stage Transition
Besides the effect on primordial follicle initiation, androgens also seem to have a role in the preantral to antral stage transition. In vivo studies in ovine models have shown that DHEA exposure stimulates early follicular growth during the preantral and early antral follicular stages (17). Studies in mouse models have also shown that both DHT and testosterone stimulate granulosa cell (GC) proliferation and both secondary and preantral follicle growth (18). Moreover, androgens seem to support follicle development during the FSH-dependent preantral stage by increasing the expression of FSH receptor mRNA levels and, therefore, enhancing FSH action (19,20). GC-specific AR-null mice experiments have also shown that AR signaling in GCs is necessary for progression beyond the preantral stage (21). Androgens enhance antiapoptotic pathways, thereby contributing to follicle survival, and improve sensitivity to FSH-induced follicle growth and progression to the antral stage (22). On the other hand, when AR signaling is blocked, preantral follicles cannot progress to antral follicles and, instead, are subjected to an increased rate of atresia.

The Peri-Ovulatory Stage
The effect of androgens in later stages of follicle development, namely in the pre-and peri-ovulatory stage, is controversial. Studies in primates have shown that testosterone treatment did not increase the number of preovulatory follicles (14). However, experiments in pigs have shown that androgens might have regulatory functions during late follicular development (23). In fact, DHT treatment resulted in an increase in the amount of FSH receptor mRNA in preovulatory follicles and increased ovulation rate (23). Similarly, experiments in mice have also shown that testosterone has a role in the maturation of oocytes arrested in prophase I of meiosis (24) and that DHT significantly increased the number of ovulated oocytes (22). On the other hand, Romero and Smitz reported that elevated levels of androstenedione and testosterone negatively affected meiotic resumption (25). These conflicting findings regarding the role of androgens in the late stages of follicular development suggest that further studies are needed to clarify the physiopathology behind such complex interactions. Figure 1 highlights the main androgen effects on folliculogenesis.

Genetic Studies
Finally, data from genetic models have also reaffirmed the role of AR-mediated activity in the regulation of ovarian function. Studies using female mouse models homozygous for an inactivated AR (ARKO) have revealed reduced fertility and a defective folliculogenesis (26)(27)(28), as well as a reduced litter size (27), increased follicular atresia and premature ovarian failure (21). Together, these data suggest the AR signaling pathway mediates both intra and extra-ovarian actions, with an essential role in maintaining normal ovarian function and fertility.

THE PITFALLS FROM CLINICAL STUDIES
All these promising data obtained from animal studies and the fact that both androgens and ovarian reserve decline steeply with age, led to the speculation that androgen replacement in women with DOR might delay these age-related effects. However, despite several lines of evidence supporting a role for androgens in folliculogenesis, the available data from clinical studies remains unconvincing. This might be related to the methodological inconsistencies observed in the available trials (Tables 1 and 2).

Dehydroepiandrosterone
A case series of five patients with unexplained infertility and previous poor response to ovarian stimulation was the first study to analyze the effect of DHEA pretreatment on ovarian response (51). In this study, 80 mg/day of oral micronized DHEA was given for 2 months, after which ovarian stimulation was started with recombinant follicle stimulating hormone (rFSH) for intrauterine insemination. The authors concluded that oral DHEA supplementation might improve ovarian response and reduce gonadotrophin consumption. Five years later, a case report of a 43-years old patient seeking embryo accumulation for preimplantation genetic screening draw the scientific community's attention to the role of androgens in ovarian response to stimulation (52). After her first stimulation cycle, the patient started self-administering 75 mg/day of oral micronized DHEA and initiated acupuncture treatment. In total, the patient performed 9 stimulation cycles with different stimulation protocols, and a significant increase in ovarian response was reported after four months of DHEA pretreatment. Since then, multiple observational and randomized controlled trials have followed, with varying DOR and poor ovarian reserve (POR) definitions, with DHEA doses ranging from 50 to 90 mg/day and a treatment duration ranging from 1 to 12 months, both before and during controlled ovarian stimulation (Tables 1 and 2). Importantly, no pharmacological studies have been performed to determine the optimal dose, duration or timing of DHEA supplementation in DOR patients.
Another key limitation regarding many studies on DHEA pre-treatment is the frequent use of patients as their own controls, comparing ovarian response after DHEA supplementation with a previous cycle. This study design does not take into account the importance of biological variability in the response to ovarian stimulation and the natural process of the regression to the mean, precluding definitive conclusions regarding the true effect of such treatment (77).
Also noteworthy is the fact that oral DHEA formulations are dietary supplements and therefore are not regulated by the US Food and Drug Administration (FDA) nor by the European Medicines Agency (EMA) and are exempt from pharmaceutical quality standards. Consequently, the true standardization of the formulations used cannot be guaranteed (78).

Testosterone
Numerous observational and randomized controlled trials have also been published on the use of testosterone pre-treatment on POR and DOR patients ( Tables 1 and 2). Most studies report the use of transdermal testosterone, both in gel and patches, with doses of treatment based on Vendola's studies on primates (14,15). In these studies, an effect on follicular development was reported with transdermal testosterone 20 µg/Kg/day, obtained with a 12.5mg/day gel application or a 2.5mg/day patch. Importantly, however, pharmacokinetics studies performed in postmenopausal women revealed that the administration of 4.4-5 mg testosterone gel or cream raised free testosterone levels within the reference range for reproductive-aged women whereas higher doses increased testosterone levels above the physiological range (79,80). These findings question the potential clinical benefit (or harm) of using the high doses that have been reported so far.
The issue of the duration of treatment has also been another point of conflict in the published studies, ranging from 5 days, based on Vendola's studies (14,15), to 21-28 days, based on a RCT that reported that testosterone effects at the follicular level occurred after at least three weeks of testosterone pre-treatment (32).    This should come as no surprise, if we consider that the progression from a primordial follicle to a periovulatory follicle takes approximately 3 months (81).

Too Much Is Not Enough
The vast bulk of published original studies and meta-analysis on the use of androgens pre-treatment in DOR and POR patients is depicted in Figure 2. One of the limitations in analyzing the effect of these adjuvant strategies in DOR/POR patients is the definition of diminished and poor response itself. In this context, the Poseidon Group introduced the concept of 'low prognosis patients' and highlighted the need for tailored evidence-based clinical algorithms for each of the four proposed risk groups (82,83). Standardizing the inclusion criteria offuture studies based on these risk groups might be a further step in minimizing study heterogeneity. Despite the above-mentioned methodological limitations and the heterogeneity among the inclusion criteria and treatment protocols, original studies continue to be published in an attempt to optimize the clinical management of such a challenging population. With the same goal, a disproportionate number of meta-analysis has been published, especially when considering the number of original studies. Table 3 describes the meta-analysis published on the use of DHEA and testosterone supplementation in IVF and the study design of the included trials. If we consider the low level of evidence of some of the included study designs, the lack of evidence-based protocols for both DHEA and testosterone supplementation, the heterogeneity in the definition of POR and DOR and the diversity in the IVF protocols used in the different trials, the clinical impact of the conclusions drawn from these meta-analysis might be called into question. In this regard, an individual patient data approach could be of use in increasing the strength of the available evidence.
However, to break this vicious cycle, we are left with the need to write the story of androgens supplementation in patients with DOR/POR from the beginning. In order to do so, evidence from pharmacokinetics studies (79) as well as from the timespan of human folliculogenesis (97) must be taken into account in what concerns the optimal dose and duration of treatment. In this respect, the currently ongoing multicenter double-blind placebocontrolled randomized controlled trial T-TRANSPORT (NCT02418572, available at http://clinicaltrials.gov/ct2/show/ NCT02418572) might shed some light on this subject. With an intervention group undergoing 5.5 mg daily transdermal testosterone for two months prior to an IVF cycle and powered  with clinical pregnancy rate as the primary outcome measure, this trial is expected to clarify the role of androgens in IVF.

CONCLUSION
Despite the vast amount of available literature on the use of DHEA and testosterone in POR patients, the bulk of evidence is still limited to draw definite conclusions. More than reviewing the available data and publishing new studies based on the same pitfalls, we urge to restart this chapter with well-designed clinical trials.

AUTHOR CONTRIBUTIONS
AN designed the study, performed the literature review, contributed to the interpretation of the findings, wrote the manuscript and critically revised it. PM-B contributed to the interpretation of the findings and critically revised the manuscript. NP designed the study, supervised the writing of the manuscript, contributed to the interpretation of the findings and critically revised the manuscript. All authors contributed to the article and approved the submitted version.