The microdose GnRH agonist (GnRH-a) flare involves a 3-week course of oral contraceptives, followed by small doses of GnRH-a beginning 1–3 days after discontinuation of the oral contraceptives or the start of menses. Gonadotropins for ovarian stimulation are then begun on cycle day 5 or given concomitantly with GnRH-a. This protocol takes advantage of the burst of gonadotropin release immediately following onset of GnRH-a administration and the subsequent suppression of endogenous gonadotropins. Pretreatment with oral contraceptives or progesterone when using the microdose flare protocol optimizes the subsequent suppressive effect of GnRH-a by decreasing the likelihood of corpus luteum rescue and premature ovulation [16, 17]. In 1994, Scott and colleagues performed a prospective cohort study of 34 patients with a prior failed IVF cycle. They were treated with 20 mcg of leuprolide every 12 h, followed by initiation of gonadotropins on cycle day 5. As hypothesized, higher-peak estradiol levels, increased number of oocytes retrieved, and decreased gonadotropin requirement were observed in groups of patients compared to their previous cycles. However, they failed to show any difference in fertilization rates, and pregnancy data was not reported [18]. Another study by Schoolcraft and colleagues, also using patients as their own historic controls, involved administration of 40 μg of leuprolide acetate every 12 h beginning on cycle day 3 with initiation of gonadotropin on cycle day 5. They also defined poor responders as patients with a prior canceled cycle. They reported a 50 % ongoing pregnancy rate and decrease in cancelation rate from 100 % to 12.5 % [19]. Of note, patients in this study also received growth hormone, which is an important confounder that cannot be ignored. Despite Schoolcraft’s findings, other studies report continued low birth rates even in the presence of improvement of other cycle parameters [20, 21]. It is possible that these low live birth rates are also a reflection of the increased aneuploidy rate with increasing age, an egg quality issue which is unfortunately not improved by this protocol. Unfortunately, there is a paucity of data with regard to use of the microdose protocol in women with POI.

There has been concern that continued administration of GnRH agonist in the follicular phase following luteal-phase GnRH agonist administration has a detrimental effect on ovarian response. The stop protocol was introduced to resolve this quandary, by preserving the downregulatory effects of luteal-phase GnRH agonist while not incurring any of the detriments of follicular administration. GnRH agonist is administered in the mid-luteal phase and discontinued when menses begin. Gonadotropins are then initiated on cycle day 3. A randomized controlled trial evaluating this protocol observed a decrease in gonadotropin requirement and increase in number of mature follicles retrieved when compared to the long protocol [22]. However, both study arms showed similar cancelation rates, implantation, and pregnancy rates. Another randomized controlled trial showed higher cancelation rates and also failed to show any difference in stimulation protocols and clinical pregnancy rates [23]. The GnRH agonist stop protocol has not been shown to confer additional benefit and is thus not recommended for POI patients or other poor responders [22–24].

The GnRH antagonist protocol involves direct blockade of the pituitary gonadotrope receptors without any of the stimulatory effect which usually accompanies initiation of GnRH agonist. The GnRH antagonist is introduced a few days after gonadotropins have already been initiated and prevents undesirable LH surges and subsequent premature ovulation.

Craft and colleagues evaluated 18 poor responders with 23 prior failed cycles and observed a decrease in the cancelation rate from 57 to 29 % and increase in the average number of oocytes retrieved from 4.7 to 6.4 [25]. Unfortunately, they also observed low live birth rates at 11.8 % [25]. Similar findings were observed in a retrospective study by Dragisic in 2005 which studied 68 poor responders in 80 IVF cycles. Subjects underwent an estrogen priming as well as GnRH antagonist protocol. These patients showed a significant decrease in cycle cancelations and an increase in number of oocytes retrieved (6.4 vs. 8.3) when compared to prior cycles. They observed an ongoing pregnancy rate of 26.2 %; however, live birth rate was not reported [26]. Takahashi et al. in 2004 described 40 women treated with the GnRH antagonist protocol following failed cycles on the GnRH agonist long protocol. They observed no difference on ovarian response, but demonstrated an increased number of women with expanded blastocyst on day 5 and a markedly improved ongoing pregnancy rate of 42.1 % [27]. Data from a randomized trial comparing GnRH antagonist cycles to cycles with neither GnRH agonist nor antagonist showed a trend toward significance for lower pregnancy and implantation rates [28]. Overall, the data show improvement in cycle parameters, namely, decreased cancelation rates and a modest improvement in ongoing pregnancy in poor responders on GnRH antagonist protocols when compared to the GnRH agonist protocols.

The use of clomiphene citrate (Clomid) during IVF was first introduced to decrease the gonadotropin requirement, as well as physical and financial burden to patients undergoing in vitro fertilization. Also known as the mixed gonadotropin protocol, it involves administration of the tablets from cycle days 3 to 7 followed by administration of gonadotropins (usually FSH). Although clomiphene citrate has been thought to adversely impact uterine receptivity, the concomitant use of gonadotropins is believed to attenuate the negative effect. A randomized controlled trial by Ghont et al. in 1995 showed significantly lower pregnancy rates per cycle with this protocol when compared to the agonist protocol: 24.5 vs. 36.8 %; p < 0.02 [29]. Contrary to this, a prospective randomized trial by Weigert and colleagues demonstrated a nonsignificant increase in pregnancy rate per embryo transferred and per oocyte retrieved (42.95 % vs. 36.6 % and 42.2 % vs. 34.7 %, respectively) comparing clomiphene citrate-gonadotropin protocol to the long agonist protocol [30]. Weigert et al. also observed significantly lower FSH and LH requirements in women on the combined protocol; however, this was confounded by a significantly lower number of oocytes retrieved (7.7 ± 3.6 vs. 8.7 ± 5; p value = 0.02) [30]. A subsequent case–control study found that the clomiphene citrate-gonadotropin protocol, compared to the long GnRH agonist protocol, resulted in a fewer number of oocytes retrieved (3.7 ± 2.0 vs.13.1 ± 6.0, p value < 0.05) and this difference persisted in the women >35 years (3.9 ± 2.2 vs. 12.5 ± 4.5) [31]. However, the pregnancy rates per embryo transfer were similar.

Although the mixed gonadotropin protocol confers the benefit of decreased gonadotropin requirement, there is no evidence to suggest improvement in cycle parameters, particularly pregnancy rate. On the contrary, studies have demonstrated a lower number of retrieved oocytes. Therefore, we do not recommend routine use of the mixed gonadotropin cycle in women with decreased ovarian reserve or primary ovarian insufficiency.

Administration of luteal-phase estrogen has been proposed to combat the early follicular recruitment in the luteal phase that is frequently seen in poor responders [26]. Treating patients with estrogen in the luteal phase has been shown to significantly decrease follicle size discrepancies, which in turn has led to an increase in the number of follicles > 16 mm in mean diameter, mature oocytes retrieved, and embryos available for transfer [26, 32]. Unfortunately, there are no prospective randomized trials to provide level 1 evidence for this protocol.

Shastri et al. performed a retrospective analysis of 186 young (<35 yo) patients with POI comparing the estrogen priming antagonist protocol to the microdose leuprolide protocol. Women in the estrogen priming arm had greater gonadotropin requirement, without a concurrent increase in number of oocytes retrieved and fertilized. However, they also showed a higher implantation rate in the estrogen priming arm, although this difference failed to achieve statistical significance (30.5 vs. 21.1; p value 0.09) [33]. Another retrospective chart review performed by Chang et al. reported lower cancelation rate and higher number of oocytes retrieved in the estrogen priming group. A higher number of good-quality embryos were also seen in the estrogen priming group, although this difference was not significant [34]. Another retrospective study done by Cakmak et al. evaluated the response of 30 patients who responded poorly and did not achieve pregnancy in a prior cycle and compared it to a subsequent estrogen priming cycle. They reported a higher number of dominant follicles in the estrogen priming arm (4.2 ± 2.7 vs. 2.4 ± 1.3), shorter length of ovarian stimulation, higher number of mature oocytes retrieved (4.9 ± 2.0 vs. 2.2 ± 1.1), and lower total gonadotropin requirement when compared to the patients’ prior cycle. The clinical pregnancy rate was 22.2 % compared to no pregnancy in the previous cycle [35].

Growth hormone (GH) is a polypeptide that is made, stored, and secreted into the bloodstream by somatotrophs in the anterior pituitary. The use of growth hormone to improve IVF outcomes in women is based on the premise that GH augments ovarian response to gonadotropins, as first described by Homburg in 1988 [36]. It has been shown to augment the action of FSH on granulosa cells by upregulating ovarian synthesis of insulin-like growth factor 1 (IGF-1) . IGF-1 in turn amplifies the action of gonadotropins on both the theca and granulosa cells [37]. Additionally, GH has been observed to upregulate DNA repair genes in liver cells, establishing the potential for a similar effect on the oocyte [38]. This may ameliorate some of the increased meiotic nondisjunction characteristic of oocyte aging [39].

A recent Cochrane review examined randomized control trials (RCTs) evaluating the role of growth hormone for in vitro fertilization. They identified eight RCTs that looked at the role of GH in women classified as poor responders, although the definition of poor responders differed depending on the study. They concluded that growth hormone significantly improved the live birth rates (OR 5.39, 95 % CI [1.89, 15.35]) and pregnancy rates (OR 3.28, 95 % CI [1.74, 6.20]) in women considered poor responders. However, given the varied definitions of poor responders, further study is needed to identify which subgroup of poor responders will benefit the most from GH. Given the findings from the Cochrane review meta-analysis, GH should be considered in women with POI [40].

Androgen receptors were first described in the pre-antral and antral follicles of rhesus monkeys, and androgen was noted to increase the expression of FSH receptors in the granulosa cells of these follicles [41–43]. In humans, androgen receptor mRNA and androgen levels in follicular fluid were found to correlate to FSH receptor mRNA [44]. This led to the suggestion that androgen action on pre-antral and antral follicles augments FSH receptor expression, which in turn leads to increased response to controlled ovarian hyperstimulation (COH). It has, therefore, been theorized that treatment with an androgen, such as dehydroepiandrosterone (DHEA), may confer a beneficial effect. Casson et al. first looked at this in 2000; a case series of five women who were poor responders were treated with 80 mg/day of micronized DHEA for 2 months and underwent COH while still on DHEA. They observed improvement in peak estradiol and peak estradiol/ampule of gonadotropin. Only one of the five women achieved pregnancy; this was a twin pregnancy [45]. Following this, multiple studies have been done to further evaluate the effect on DHEA on IVF parameters and outcomes, and the data remains inconclusive [46–48]. Given this, there is insufficient data to support the routine use of DHEA in IVF cycles for POI.