Animal studies

Hyperandrogenism, either from intraovarian origin, such as polycystic ovarian syndrome, or extraovarian origin, such as congenital adrenal hyperplasia or androgen-producing tumors, results in an increased number of developing follicles and an enlarged ovarian stroma. These observations were the starting point for the initial in vivo studies in primates showing that T and DHT treatment stimulated primordial follicle initiation and increased the number of growing small preantral and antral follicles and overall follicle survival. ^, Since then, several animal studies have followed, confirming the importance of T in the transition from the primordial to the primary follicle stage and from the preantral to antral stage. ^, Moreover, the importance of androgen support for overall follicle survival by decreasing follicle atresia and GC apoptosis and stimulating GC proliferation and differentiation has also been established.

The interaction between the FSH receptor (FSHR) and AR might explain some of these actions. In fact, studies in primates have shown that FSHR and AR messenger RNA (mRNA) levels in GC are positively correlated. Furthermore, T treatment has been shown to increase FSHR expression at all levels of follicular development, whereas FSH treatment increased AR expression in primary follicles. ^, Likewise, treatment with DHT in gilts has been shown to increase FSHR mRNA in preovulatory follicles, and studies in mice revealed that androgens increase FSHR protein levels, increasing follicular sensitivity to FSH and promoting preantral follicle growth and progression to antral follicles ( Figure 3 ).

Different androgen effects depending on the follicle developmental stage

COX-2 , cyclo-oxygenase; CYP19 , aromatase (also called estrogen synthetase); E2 , estradiol; FSHR , follicle-stimulating hormone receptor; IGF1 , insulin-like growth factor 1; IGF1R , insulin-like growth factor 1 receptor; mRNA , messenger RNA; P450scc , cholesterol side-chain cleavage enzyme.

Neves. Androgens and diminished ovarian reserve. Am J Obstet Gynecol 2022 .

Androgen receptor knockout (ARKO) studies

Despite the valuable findings provided by these experiments, they preclude the assessment of androgen action exerted directly through the AR because of the indirect effect exerted through ER activation ( Figure 2 ). Thus, genetic models with female mice homozygous for an inactivated AR (ARKO) have been developed to determine the direct action of AR on ovarian function. Based on these experiments, global ARKO female mice have demonstrated defective folliculogenesis, with a lower number of preantral follicles, higher rates of GC apoptosis, premature ovarian failure, lower response to COS, a defective luteal phase, and reduced litter size. Moreover, controls present normal cycles and fertility when ARKO and control ovaries are cross-transplanted, whereas ARKO hosts present abnormal cycles and reduced fertility, further suggesting an extraovarian neuroendocrine AR-mediated effect. Cell-specific ARKO models have also demonstrated that AR signaling in GCs is necessary for progression beyond the preantral stage, and AR importance in the pituitary signaling for a proper regulation of LH/FSH expression and secretion.

Human studies

In women, the AR is expressed at all levels of the female hypothalamic-pituitary-gonadal axis. Within the ovaries, the AR is expressed in the oocyte, GC, and theca cells, and in the ovarian stroma, and it is present in most stages of follicular development, although with different patterns of expression at different developmental stages. AR expression has not been detected in primordial follicles, but from the primary follicle stage onwards, and correlates linearly with follicle growth. Interestingly, the finding that early follicles acquire AR before FSHR and that AR expression in preantral follicles is higher than that of FSHR also points toward a role for androgens in early folliculogenesis. In line with these findings, data obtained from small antral follicles revealed a positive correlation between FSHR gene expression and AR expression on GC, and between the expression of FSHR and intrafollicular androgen (ie, A4 and T) concentrations, suggesting that androgens might have a role in the promotion of gonadotropin responsiveness of developing follicles.

To summarize, the available evidence seems to point toward a role for androgens in early folliculogenesis, maintenance of follicle health, and priming of later stages of follicle development.

Which androgen?

Vendola et al conducted the initial studies in primates that drove the scientific community’s attention to the potential role of androgen supplementation in increasing folliculogenesis. Although these initial studies were performed with the bioactive androgens T and DHT, the clinical role of androgens in ART was first proposed by Casson et al, who suggested that DHEA might augment ovarian response in women with previous POR.

DHEA is a preandrogen with a low affinity for the AR, requiring further conversion into T and DHT to exert its effect in the AR. A recent RCT has shown that DHEA supplementation significantly increased serum T and DHEA-S levels starting from week 4 of treatment. However, the fact that no difference was reported in follicular T levels questions the potential effect of DHEA on follicle development. Conversely, T, which is 1 of the most potent androgens in women, binds directly to the AR. Transdermal T supplementation has been shown to result in an increased expression of the AR gene in cumulus cells, supporting a role for T pretreatment at the follicular level.

Dose and duration of treatment

Multiple studies on DHEA supplementation have been published reporting the use of oral micronized DHEA 50 to 90 mg daily, with treatment durations ranging from 4 weeks to 12 months. However, no dose-finding studies have been performed to determine the optimal dose and duration for DHEA pretreatment. Although previous RCTs have reported a higher antral follicle count and ovarian volume after 16 weeks of DHEA supplementation in women with premature ovarian insufficiency, and a higher intrafollicular DHEA-S concentration after 12 weeks of DHEA pretreatment, these studies were not designed to determine the optimal duration of treatment. ^,

Most studies regarding T treatment in DOR and POR patients have reported transdermal application, with dosages varying from a 2.5 mg per day patch to a 10 to 25 mg per day gel application, and a duration of treatment between 5 and 28 days before the beginning of COS. One exception was the study by Saharkhiz et al, in which transdermal T supplementation was started with COS and maintained until the day of trigger.

Most of the available studies used a hydroalcoholic gel containing 1% or 2% T. ^, After the first application, the serum T levels reach a steady state in 24 hours and remain so for the duration of gel application, with parallel increases in DHT and estradiol. T concentrations begin to decline 24 to 36 hours after gel application, justifying a once-daily dosage, and return to baseline levels within 4 days after stopping application. Other authors used a patch with a 2.5 mg per day nominal delivery rate of T, which was applied on the thigh at night and maintained for 12 hours. ^, This transdermal delivery system also maintains stable T levels with little intra- or interindividual variation.

Either with patch or gel application, doses of treatment have been based on Vendola’s studies on primates, reporting an effect on follicular development with 20 μg/Kg/day. ^, However, pharmacokinetic studies performed in postmenopausal women revealed that the administration of 4.4 to 5 mg T gel raised free T levels to the normal range for reproductive-aged women, whereas doses above 8.8 to 10 mg increased T levels above the physiological range, questioning the doses used in these previously mentioned studies. ^, In line with these findings, studies in female-to-male transgender patients have reported a significant suppression in anti-müllerian hormone serum levels after high-dose T pretreatment. Moreover, studies in mice have shown that supraphysiological T levels negatively affect meiosis resumption, further highlighting a potential detrimental effect of supraphysiological doses on ovarian reserve.

Regarding T treatment duration, some authors have adopted a 5-day pretreatment based on the previously mentioned primate studies. ^{, , ,} However, Kim et al have shown that T effects at the follicular level occurred after at least 3 weeks of T pretreatment. Therefore, most studies have reported a 21- to 28-day pretreatment. Considering that the time that elapses in progressing from a primary follicle to ovulation is approximately 85 days, the beneficial effect of longer pretreatment durations cannot be disregarded.

In summary, there is limited evidence regarding the dose and duration of T or DHEA required and the intraovarian androgen availability after systemic administration at nonvirilizing concentrations.

Side effects

A few side effects including hair loss, hirsutism, and voice deepening have been reported in women undergoing DHEA treatment, but they are negligible in most studies with 75 mg; acne and oily skin are the most frequently reported. ^, Wong et al reported 2 cases of asymptomatic transient polycythemia. Karp et al described a case report of a first generalized seizure episode in a 30-year-old patient with DOR and a previous brain lesion after a car accident, undergoing pretreatment with 75 mg of oral DHEA daily. The authors underlined that special attention should be given to women with risk factors for convulsive activity, taking into account that DHEA crosses the blood–brain barrier and affects central nervous system excitability.

Regarding transdermal T, Singh et al reported few cases of acne, oily skin, increased hair growth, breast tenderness, vaginal spotting, and mild leg swelling. However, no local or systemic side effects related to the use of transdermal T patches or gel have been reported in most studies. ^{, , , ,}

Oocyte yield

The impact of androgen pretreatment on the number of oocytes retrieved was compared among 12 RCTs ( Figure 4 , A). No significant difference was found after DHEA pretreatment (MD, 0.76; 95% CI, −0.35 to 1.88; 5 RCTs; 538 patients; I ² =77%; random-effects model; moderate certainty). Conversely, a significantly higher number of oocytes was obtained after T pretreatment (MD, 0.94; 95% CI, 0.46–1.42; 7 RCTs; 603 patients; I ² =42%, random-effects model; moderate certainty).

DHEA and testosterone vs no treatment/placebo ovarian response to stimulation

CI , confidence interval; DHEA , dehydroepiandrosterone; IV , inverse variance; SD , standard deviation.

Neves. Androgens and diminished ovarian reserve. Am J Obstet Gynecol 2022 .

No significant effect on the number of MII oocytes retrieved was observed after DHEA pretreatment (MD, 0.25; 95% CI, −0.27 to 0.76; 3 RCTs; 372 patients; I ² =19%; random-effects model; moderate certainty) nor after T priming (MD, 0.62; 95% CI, 0.00–1.25; 6 RCTs; 553 patients; I ² =71%; random-effects model; moderate certainty) ( Figure 4 , B).

Embryo yield and embryo quality

Limited evidence exists regarding the effect of androgen priming on the number of embryos obtained. Only 2 RCTs evaluated the impact of DHEA supplementation on the total number of embryos. The authors found no statistically significant difference when comparing DHEA with the control group (respectively, median 2.5 [range 0–7] and median 2 [range 0–7], P =.297). Four RCTs evaluated the effect of T pretreatment on embryo yield. Although 3 did not report a significant effect, ^{, ,} Saharkhiz et al reported a significantly higher number of embryos after T supplementation. Taking into account the heterogeneity regarding the embryo stage (cleavage stage vs blastocyst), no meta-analysis was performed.

Similarly, although no meta-analysis could be performed regarding embryo quality because of the different grading systems, the impact of androgen pretreatment on embryo quality has also been analyzed in several RCTs.

The effect of DHEA priming on embryo quality has been evaluated in 5 RCTs. Although some reported improved embryo quality, others reported no difference ^{, ,} or even a lower embryo score after DHEA pretreatment. Notably, despite finding no statistically significant difference in the number of top-quality embryos after 12 weeks of DHEA pretreatment, Yeung et al reported significantly improved embryo quality in the subgroup of women who had higher intrafollicular DHEA-S levels. The authors hypothesized that higher intraovarian DHEA-S concentration, either natural or achieved through DHEA supplementation, might improve follicular maturation, leading to reduced aneuploidy and better-quality embryos.

The effect of T pretreatment on embryo quality has also been a matter of debate. Although some RCTs found no difference in the rate of top-quality embryos transferred, ^, others found a higher embryo score ^{, ,} after T priming.

Pregnancy rate and live-birth rate

The impact of androgen priming on reproductive outcomes is displayed in Figure 5 . Overall, pooled data showed a similar CPR after DHEA priming in DOR/POR patients undergoing IVF (RR, 1.17; 95% CI, 0.87–1.57; 8 RCTs; n=727; I ² = 0%; low certainty), and a similar LBR (RR, 0.97; 95% CI, 0.29–3.19; 3 RCTs; n=117; I ² = 0%; low certainty) compared with placebo or no treatment. On the contrary, T pretreatment showed a higher CPR (RR, 2.07; 95% CI, 1.33–3.93; 8 RCTs; n=653; I ² =0%; low certainty) and higher LBR (RR, 2.09; 95% CI, 1.11–3.95; 4 RCT; n=333; I ² =0%; low certainty).

DHEA and testosterone vs no treatment or placebo: clinical outcomes

CI , confidence interval; DHEA , dehydroepiandrosterone; IV , inverse variance; SD , standard deviation.

Neves. Androgens and diminished ovarian reserve. Am J Obstet Gynecol 2022 .

Miscarriage rate

The impact of androgen priming on the MR has been evaluated in 10 RCTs. No significant differences were found when DHEA was compared with controls (RR, 0.80; 95% CI, 0.29–2.22; 4 RCTs; n=322; I ² =0; moderate certainty). Likewise, similar MRs were yielded when T priming was compared with controls (RR, 1.83; 95% CI, 0.58–5.81; 6 RCTs; n=444; I ² =0; moderate certainty) ( Figure 5 , C).