Novel Concepts for Inducing Final OocyteMaturation in In Vitro Fertilization Treatment

Infertility affects one in six of the population and increasingly couples require treatment with assisted reproductive techniques. In vitro fertilization (IVF) treatment is most commonly conducted using exogenous FSH to induce follicular growth and human chorionic gonadotropin (hCG) to induce final oocyte maturation. However, hCG may cause the potentially life-threatening iatrogenic complication “ovarian hyperstimulation syndrome” (OHSS), which can cause considerable morbidity and, rarely, even mortality in otherwise healthy women. The use of GnRH agonists (GnRHas) has been pioneered during the last two decades to provide a safer option to induce final oocyte maturation. More recently, the neuropeptide kisspeptin, a hypothalamic regulator of GnRH release, has been investigated as a novel inductor of oocyte maturation. The hormonal stimulus used to induce oocytematuration has amajor impact on the success (retrieval of oocytes and chance of implantation) and safety (risk of OHSS) of IVF treatment. This review aims to appraise experimental and clinical data of hormonal approaches used to induce final oocytematuration by hCG, GnRHa, both GnRHa and hCG administered in combination, recombinant LH, or kisspeptin.We also examine evidence for the timing of administration of the inductor of final oocyte maturation in relationship to parameters of follicular growth and the subsequent interval to oocyte retrieval. In summary, we review data on the efficacy and safety of the major hormonal approaches used to induce final oocyte maturation in clinical practice, as well as some novel approaches thatmay offer fresh alternatives in future. (Endocrine Reviews 39: 593 – 628, 2018) I nfertility affects one in six couples and is recognized by the World Health Organization (WHO) as the fifth most serious global disability (). This may appear a controversial statement in an overpopulated world; however, Mahmoud Fathalla, former director of the WHO Human Reproductive Program (HRP), explained the rationale for this: “If public health policies encourage couples to delay and plan pregnancies, [then it is] equally important that they are assisted in their attempts to conceive in the more limited time available” (). The number of in vitro fertilization (IVF) cycles carried out across the world is increasing each year, with .% of all children born in the United States in  being conceived through assisted reproductive technology (). IVF treatment is a supraphysiological process that simulates many of the physiological processes occurring during the normal human menstrual cycle, namely follicular development, oocyte maturation/ ovulation, fertilization, and implantation. During IVF treatment, a pharmacological dose of FSH is used to induce the growth of multiple of ovarian follicles. As follicles grow, an LH surge that could lead to premature ovulation is prevented either through the use of a GnRH antagonist (, ), or continuous administration of a GnRH agonist (GnRHa) to downregulate the GnRH receptor (). Once follicles reach the requisite size, LH exposure is provided to simulate the mid-cycle LH surge, which induces the processes of oocyte maturation and subsequent ovulation (). Oocyte retrieval is thus precisely timed following provision of LH exposure to retrieve oocytes following oocyte maturation, but prior to the occurrence of ovulation. LH exposure initiates the resumption of meiosis and the maturation of the oocyte from the immature “metaphase I” stage to the mature “metaphase II” stage of development (). During this process of oocyte maturation, the first polar body is extruded such that a diploid cell transitions toward a haploid gamete and attains competence for fertilization by a spermatozoon (). Following LH-like exposure, the remainder of the follicle forms the corpus luteum, which produces sex steroids, particularly progesterone, to prepare the endometrium for implantation of the embryo (). When LH-like exposure is excessive in duration, there is an increased chance of development of a dangerous complication of IVF treatment termed “ovarian hyperstimulation syndrome” (OHSS) (). OHSS is a predominantly iatrogenic condition that may ISSN Print: 0163-769X ISSN Online: 1945-7189

I nfertility affects one in six couples and is recognized by the World Health Organization (WHO) as the fifth most serious global disability (). This may appear a controversial statement in an overpopulated world; however, Mahmoud Fathalla, former director of the WHO Human Reproductive Program (HRP), explained the rationale for this: "If public health policies encourage couples to delay and plan pregnancies, [then it is] equally important that they are assisted in their attempts to conceive in the more limited time available" (). The number of in vitro fertilization (IVF) cycles carried out across the world is increasing each year, with .% of all children born in the United States in  being conceived through assisted reproductive technology ().
IVF treatment is a supraphysiological process that simulates many of the physiological processes occurring during the normal human menstrual cycle, namely follicular development, oocyte maturation/ ovulation, fertilization, and implantation. During IVF treatment, a pharmacological dose of FSH is used to induce the growth of multiple of ovarian follicles. As follicles grow, an LH surge that could lead to premature ovulation is prevented either through the use of a GnRH antagonist (, ), or continuous administration of a GnRH agonist (GnRHa) to downregulate the GnRH receptor (). Once follicles reach the requisite size, LH exposure is provided to simulate the mid-cycle LH surge, which induces the processes of oocyte maturation and subsequent ovulation (). Oocyte retrieval is thus precisely timed following provision of LH exposure to retrieve oocytes following oocyte maturation, but prior to the occurrence of ovulation. LH exposure initiates the resumption of meiosis and the maturation of the oocyte from the immature "metaphase I" stage to the mature "metaphase II" stage of development (). During this process of oocyte maturation, the first polar body is extruded such that a diploid cell transitions toward a haploid gamete and attains competence for fertilization by a spermatozoon (). Following LH-like exposure, the remainder of the follicle forms the corpus luteum, which produces sex steroids, particularly progesterone, to prepare the endometrium for implantation of the embryo (). When LH-like exposure is excessive in duration, there is an increased chance of development of a dangerous complication of IVF treatment termed "ovarian hyperstimulation syndrome" (OHSS) (). OHSS is a predominantly iatrogenic condition that may result in serious adverse consequences for otherwise healthy women undergoing fertility treatment ().
Thus, the LH-like exposure required to initiate the process of oocyte maturation is a critical step in the success of IVF protocols enabling the efficacious retrieval of mature oocytes, as well as affecting the chance of pregnancy and the safety of IVF treatment. In current IVF protocols, LH-like exposure is provided through either the use of human chorionic gonadotropin (hCG) or GnRHa, which are colloquially referred to as the "trigger" of oocyte maturation (). hCG has sufficient homology to LH to be able to activate the LH receptor and was the primary and remains the most commonly used trigger of oocyte maturation (). GnRHa induces endogenous gonadotropin (LH and FSH) release from the pituitary gland and is a safer option, particularly in women at high risk of OHSS (). Unfortunately, owing to the induction of a shorter duration of LH exposure, the luteal phase is more dysfunctional following GnRHa than hCG, and thus in recent years, there has been an interest in combining the better safety profile of GnRHa with a small dose of hCG to improve pregnancy rates, in socalled "double" or "dual" trigger protocols (, ).
Although not in common clinical use, recombinant LH (rLH) has also been trialed as a possible alternative to hCG for inducing oocyte maturation (). More recently, kisspeptin, a neuropeptide that stimulates endogenous GnRH release, has been used to safely mature oocytes during IVF treatment even in women at high risk of OHSS (-) (see Fig.  for diagram illustrating site of action of each agent used to induce oocyte maturation).
Two main stimulation protocols are used to grow follicles and provide the context in which a trigger is administered, namely either the "long" or GnRHa cotreated protocol, and the "short" or GnRH antagonist cotreated protocol. In the short protocol, the GnRH antagonist used is a competitive antagonist, and thus its inhibitory effect can be overcome by GnRH agonism. Thus, the short protocol allows for the use of GnRHa or kisspeptin to induce final oocyte maturation, whereas hCG or rLH can be used in either short or long protocols. The short protocol therefore enables greater flexibility for the hormone stimulus to induce oocyte maturation [see Table  (, , -) for summary of agents used to induce oocyte maturation].
This may be of particular value if the risk of OHSS only becomes apparent during follicular development. Although for some years there were concerns that pregnancy rates could be reduced by using the short protocol (, ), several meta-analyses and a large randomized controlled trial (RCT) have established equivalence of these two protocols (-). In the United States, there has been an increased use of the short protocol (rising from .% in  to .% in ; P , .), to enable more frequent use of GnRHa for oocyte maturation and mitigate the risk of OHSS ().
In this review, we aim to discuss the predominant hormonal stimuli used to induce final oocyte maturation, with a particular focus on the endocrine requirements for efficacy (retrieval of mature oocytes) and the impact on the luteal phase and safety (risk of OHSS) following the use of either hCG, GnRHa, both hCG and GnRHa in combination, rLH, or kisspeptin.

Methods
This review was undertaken using a comprehensive literature search of all available articles on PubMed from inception until  December  utilizing the search terms "oocyte maturation," "trigger," "human chorionic gonadotropin; hCG," "gonadotropin releasing hormone agonist; GnRH agonist," "luteinizing ESSENTIAL POINTS · IVF (in vitro fertilization) therapy utilizes supraphysiological treatments to simulate many of the physiological processes occurring in the natural human menstrual cycle · Oocyte maturation is a critical process to the success of IVF treatment, during which the oocyte gains competence for fertilization · Oocyte maturation is initiated by LH-like exposure that can be provided by human chorionic gonadotropin (hCG), GnRH agonist, recombinant LH, or kisspeptin · The mode by which oocyte maturation is induced has significant impact on the efficacy of oocyte retrieval, the chance of pregnancy, and the safety of IVF treatment · Oocyte maturation is part of a continuum with ovulation, and the size of follicles at time of administration and the interval to oocyte retrieval can impact the efficacy of agents of oocyte maturation · The risk of ovarian hyperstimulation syndrome, a potentially life-threatening complication of IVF treatment that can affect otherwise healthy women undergoing fertility treatment, is strongly related to the agent used to induce oocyte maturationUse of alternative agents of oocyte maturation to hCG can significantly improve the safety of IVF treatment hormone releasing hormone; LHRH," "recombinant luteinizing hormone," "luteal phase support," "in vitro maturation; IVM," and "kisspeptin." Relevant articles commenting on endocrine requirements for oocyte maturation were included in the review. Evidence from randomized clinical trials or meta-analyses was prioritized over retrospective studies where available.

The Endogenous Menstrual Cycle
Many of the processes in IVF protocols simulate the physiological processes occurring during the natural menstrual cycle, albeit in a supraphysiological manner. During the early follicular phase of the natural cycle, serum estradiol and progesterone levels are low and inhibin B is secreted by small antral follicles (). Thus, the early follicular phase is characterized by both reduced negative feedback from low sex steroid levels but increased negative feedback on FSH secretion by inhibin B levels (, ), overall resulting in an~% increase in serum FSH (-). This modest rise in FSH stimulates folliculogenesis and aromatase action to increase estradiol production by ovarian granulosa cells (). In IVF protocols, the modest threshold FSH level for monofollicular development is exceeded by a pharmacological dose of FSH for a duration sufficient to prevent atresia of nondominant follicles and thus induce multifollicular growth (). As estradiol levels continue to rise, there is a critical switch from negative to positive feedback on GnRH secretion (), which increases LH synthesis and lowers the GnRH concentration required for LH production ().
Kisspeptin is a hypothalamic neuropeptide that results in gonadotropin release in both men and women and is requisite for ovulation in women (, , ). The sensitivity to kisspeptin increases during the preovulatory phase when estradiol levels are highest (). Although estradiol is key in initiation of the mid-cycle LH surge (, ), levels of progesterone during the follicular phase are also influential. Administration of progesterone can advance the timing of the LH surge (, ); coadministration of progesterone with estradiol results in an LH surge of greater duration and amplitude than by estradiol alone (). However, multifollicular development during IVF treatment may alter the hormonal milieu from the natural cycle beyond differences in sex steroid levels alone. For example, gonadotropin surge-attenuating factor (GnSAF) is a molecule produced by ovarian follicles () that reduces pituitary sensitivity to GnRH () and may act to attenuate the amplitude of the LH surge (). Differences in GnSAF have been proposed to contribute to the differential sensitivity to GnRH antagonism observed between cycles with monofollicular and multifollicular growth, whereby hypersecretion of GnSAF in cycles with multifollicular development may reduce the degree of GnRH antagonism required to prevent a premature LH surge ().

Oocyte Maturation
Final oocyte maturation is the process by which the oocyte resumes meiosis to transition from the metaphase I to the metaphase II stage of development, at which stage it attains competence for fertilization by a spermatozoon (). The definition can be extended to include the capacity to support embryo development to the blastocyst stage and to live birth (). It is initiated by LH-like exposure that induces a fall in intraoocyte cAMP and is commonly assessed by the production of a polar body to signify a mature/ metaphase II oocyte ().
In humans, meiosis is initiated during embryogenesis (), but it is halted at prophase with the nucleus contained within an intact envelope and possessing condensed chromatin (). At this stage, the oocyte is surrounded by precursors to follicular somatic cells in a single squamous layer, forming the primordial follicle. Oocyte meiotic development remains arrested at this stage until antrum formation (). Pituitary release of gonadotropins following acquisition of reproductive maturity at puberty stimulates follicular and oocyte growth, resulting in the formation of primary and secondary follicles. Thus, whereas primordial follicle growth is a gonadotropin-independent continuous process (), secondary recruitment is gonadotropin-dependent. Crosstalk with cumulus cells play an important role in oocyte maturation, providing the oocyte with metabolic support and regulatory cues ().

Nuclear maturation
Although nuclear and cytoplasmic maturation are linked processes, cytoplasmic maturation can occur independently of full nuclear maturation () [see Fig.  () for exposition of nuclear and cytoplasmic oocyte maturation]. During the initial growth phase of the oocyte, nuclear chromatin decondenses and is transcriptionally active. As folliculogenesis progresses, the oocyte acquires meiotic competence, as identified by the condensing and nuclear association of chromatin, and the formation of microtubule organizing centers, necessary for spindle formation (, ). Yet, although the oocyte now possesses the ability to progress through meiosis, this only occurs if the oocyte is removed from the follicle, with follicular signals ensuring that oocyte development is arrested at prophase I (). This allows the oocyte to undergo further differentiation between the late antral and periovulatory follicular stages, affording the oocyte developmental competence to sustain embryo development (). Developmental competence requires a series of nuclear and cytoplasmic cellular events that take place alongside meiotic stages to enable fertilization, DNA replication, and zygote ploidy. The resumption of meiosis is signaled by germinal vesicle breakdown (GVBD). Oocytes then progress through metaphase I in which paired homologous chromosomes align in the middle of the forming meiotic spindle. Nuclear chromosomes then separate, with half the genetic material being extruded in the first polar body, resulting in the formation of a mature, haploid, metaphase II oocyte, with competence for fertilization (). Normal meiotic spindle morphology in metaphase II oocytes assessed by polarized light microscopy was more likely to result in an euploid embryo (). A meta-analysis of  studies determined that when the meiotic spindle was present, fertilization rates were significantly higher (P , .), as were cleavage rates (P , .) and the proportion of top-quality cleavage embryos (P = .) (). The interval to GVBD is difficult to assess but is estimated to occur at a median of . hours and the interval between LH receptor activation and the first stage of meiosis is thought to be~ hours (). In humans, spindle assembly typically occurs~ hours following GVBD, and~ to  hours are required between GVBD and polar body extrusion (). The total duration of nuclear maturation including the time to GVBD has been estimated to be~ to  hours (). The oocyte is then arrested at metaphase II until fertilization ().

Cytoplasmic maturation
Cytoplasmic maturation prepares the oocyte for nuclear maturation with specific chromatin configurations indicating the likelihood of the oocyte to resume meiosis () (see Fig.  for diagram of nuclear and cytoplasmic oocyte maturation). Thus, nuclear maturation mainly comprises chromosomal segregation, whereas cytoplasmic maturation involves organelle redistribution, changes in cytoskeletal dynamics, Golgi apparatus, calcium releasing activity, storage of mRNAs, proteins and transcription factors (). Cytoskeletal changes in microtubules, actin filaments, and chromatin create cell asymmetry and enable polar body extrusion with minimal loss of cytoplasm (). Although nuclear maturation is apparent by the presence of the extruded first polar body, it is more challenging to assess cytoplasmic maturity in clinical practice (). Whereas the oocyte only provides half of the genetic material, it provides nearly all membranous and cytoplasmic determinants required for embryogenesis. At metaphase II, the endoplasmic reticulum is redistributed from a fine network into clusters throughout the oocyte, and more functionally competent mitochondria are found beneath the oolemma.

Maintenance of meiotic arrest
In mammals, prophase arrest is in part maintained by the oocyte itself (). Cyclin-dependent kinase  (CDK) is a protein expressed by the oocyte that triggers chromosomal condensation and nuclear laminar breakdown, and thus is necessary for the progression from prophase to metaphase I. As the oocyte increases in size, so too does expression of CDK; however, despite this, the oocyte remains arrested in prophase until it is removed from the follicle. This suggests that inhibitory factors from granulosa cells play an important role in preventing the resumption of meiosis (). In follicle-enclosed  Figure 2. Final oocyte maturation. The midcycle gonadotropin surge causes a decrease in intraoocyte cAMP. The oocyte is removed from meiotic arrest and undergoes a series of coordinated changes affecting both the nucleus and cytoplasm. During nuclear maturation, the haploid metaphase I oocyte extrudes half of its genetic material in a polar body and transitions toward a haploid metaphase II gamete. To achieve this, the germinal vesicle breaks down (GVBD) and chromosomes align along the spindle before separation of genetic material occurs and the polar body is extruded. During oocyte maturation, cytoplasmic and nuclear maturation both occur in related but independent processes. Cytoplasmic maturation prepares the oocyte to meet the metabolic demands of fertilization and embryo growth through changes in organelles. Prior to germinal vesicle breakdown, mitochondria surround the germinal vesicle (GV), and the Golgi apparatus remains intact. By the end of oocyte maturation, mitochondria are associated with smooth endoplasmic reticulum, the Golgi body has been fragmented, and the polar body is extruded. Adapted from Mao et al. 2014 (69).
oocytes, LH results in a decrease in cAMP () and cyclic guanosine monophosphate (cGMP) to mediate the resumption of meiosis (, ). cAMP maintains CDK kinase in its inactive form through the action of protein kinase A. Thus, a fall in cAMP allows the formation of active forms of CDK kinase to initiate a cascade of events culminating in the resumption of meiosis (). cGMP is produced by granulosa cells and subsequently diffuses into the oocyte via gap junctions (), where it competitively inhibits the action of cAMP phosphodiesterase (). By preventing cAMP hydrolysis, cGMP therefore helps to maintain meiotic arrest.
In summary, human oocytes are arrested at meiotic prophase I until the mid-cycle LH surge signals a series of intracellular changes, including changes in oocyte and granulosa cell cAMP/cGMP levels that result in the resumption of meiosis, and the development of a metaphase II oocyte.

Current Modes to Induce Final Oocyte Maturation
hCG Both hCG and LH are complex heterodimeric glycoproteins with high cystine content. hCG has structural similarity to LH, sharing the same a subunit and % of the amino acid structure of the b subunit (). This property affords hCG the ability to stimulate the LH receptor () and to induce luteinization of granulosa cells and the resumption of meiosis ().
Although hCG activates the LH receptor, it does not do so in an identical manner to LH. Roess et al.
() demonstrated differences in the binding of LH and hCG to the LH receptor by rotational diffusion. Receptors bound by hCG were immobile, whereas those bound by LH were rotationally mobile, potentially accounting for differences in receptor activation (). hCG is % carbohydrate by weight and has greater glycosylation than does LH, which may also account for differences in receptor binding (). Furthermore, intracellular signaling following activation of the LH receptor differs depending on the ligand bound (). hCG possesses higher affinity for the LH receptor than LH, and it is fivefold more potent in stimulating human granulosa cell cAMP activity than equimolar concentrations of LH (). However, extracellular signal-related kinase / and AKT (protein kinase B) activation is greater following LH than hCG ().
In summary, hCG has a greater effect on cAMP and steroidogenic action than does LH, whereas LH has a greater effect on extracellular signal-related kinase / and AKT signaling, which are antiapoptotic proliferative signals. This difference in action is hypothesized to relate to their physiological roles in the normal menstrual cycle and in early pregnancy, whereby LH plays a key role in inducing oocyte maturation and ovulation, whereas hCG supports the developing embryo and decidua through stimulating steroidogenesis. The translation of these in vitro findings is further complicated by the presence of a complex hormonal milieu in vivo that can alter these in vitro behaviors (). In conclusion, although both LH and hCG activate the LH receptor, they are not equivalent with regard to both their receptor binding kinetics and the intracellular signaling that they induce.

Formulation of hCG
For decades, the only formulation of hCG was derived from the urine of pregnant women (). However, urinary hCG (uhCG) may contain significant batchto-batch inconsistencies, and it has the potential for immunological reactions and impurities (). The advent of recombinant DNA technology has made it possible for recombinant hCG (rhCG) to be synthesized in Chinese hamster ovary cells without the need for any human resource, thereby limiting the above issues (, ). uhCG is usually administered intramuscularly, whereas rhCG can be administered subcutaneously. Equivalence between uhCG and rhCG was demonstrated in a phase III double-blinded, randomized controlled study by Driscoll et al. (). The authors compared subcutaneous administration of  mg of rhCG with intramuscular administration of  IU of uhCG in  women, and they found no significant differences in the number of oocytes retrieved (rhCG . vs uhCG .), the number of oocytes retrieved per follicle . mm on day of trigger (rhCG % vs uhCG %), the number of mature oocytes (rhCG nine vs uhCG eight), or the number of cleaved embryos (). A larger randomized controlled study of  patients confirmed equivalence, with a similar number of oocytes retrieved following , IU of uhCG,  mg of rhCG, or  mg of rhCG (). Although the higher dose of  mg of rhCG resulted in two more zygotes/cleaved embryos than did the lower dose of rhCG, this came at the expense of an increased rate of OHSS (% vs %) (). Thus,  mg of rhCG was recommended for clinical use, being more convenient to administer than uhCG and causing lower rates of OHSS than the higher dose of rhCG. In , Bagchus et al. compared  mg (~ IU) of rhCG subcutaneously with , IU of uhCG intramuscularly in Japanese and white women (). Maximal hCG concentrations occurred between  and  hours before declining during  days following administration (). Interestingly, the mean exposure and mean maximum concentration (C max ) following rhCG was~% lower in Japanese women than in white women (). In Japanese women, C max was higher following uhCG than rhCG ( vs  IU/L) and occurred sooner [time to maximum concentration (t max )  hours vs~ hours] (). The half-life was similar at~ hours following both preparations and in both ethnicities (). However, importantly white women chosen for this study were weight-matched to Japanese women with a mean weight of  kg, and thus they were not chosen to exemplify a representative white population with higher body weights.
To investigate the possibility that urinary hCG may contain additional factors, such as epidermal growth factor (EGF), that could negatively influence trophoblast function, Papanikolaou et al. () randomized  women to receive either  mg of rhCG or , IU of uhCG to assess pregnancy rates. Live birth rate per protocol was higher following rhCG compared with uhCG (.% vs .%) owing to an increased rate of early miscarriage in the uhCG group (.% uhCG vs .% rhCG, P = .) (). The authors hypothesized that rhCG may have beneficial effects on placentation compared with uhCG, or that an embryonic factor could account for the difference (). However, the superiority of rhCG over uhCG has yet to be demonstrated, and two recent Cochrane reviews have found no difference in the rates of oocyte maturation, pregnancy outcomes, or OHSS (, ).

Dose of hCG
Animal studies in nonhuman mammals provide valuable insight into the effect of hCG dosage on oocyte maturation. In , Zelinski-Wooten et al.
() investigated the effect of varying concentrations of hCG administration in the female rhesus monkey. Following injection of  IU,  IU, or  IU of rhCG, or  IU of urinary hCG, peak concentrations of bioactive hCG at  hours were dose-dependent and similar following uhCG and rhCG (). The duration of the hCG surge (at levels . ng/mL) was also dose-dependent ( hours for  IU,  hours for  IU, . hours for  IU) (). Fewer animals yielded fertilized oocytes ( out of  animals) at lower doses ( and  IU of rhCG) compared with  IU of rhCG or uhCG ( out of ). Furthermore, peak progesterone levels declined sooner after the lower doses relative to  IU of rhCG and uhCG (). Thus, lower doses were able to induce oocyte maturation and granulosa cell luteinization, but they were insufficient to ensure optimal cytoplasmic oocyte maturation for fertilization and corpora lutea function (). Hence, a higher dose of hCG influences both the amplitude of hCG level attained, as well as the duration at which hCG levels are maintained over a threshold value.
The terminal half-life of rhCG in humans is estimated to be  6 . hours (, ), compared with a half-life of~ minutes for endogenous LH (). In ,  patients received uhCG at either  IU (n = ),  IU (n = ), or , IU (n = ) (). Significantly fewer successful oocyte retrievals occurred following  IU of hCG (.%) when compared with either  IU (.%) or , IU (.%, P , .), suggesting  IU as the minimum effective dose of uhCG (). Lin et al. () randomized  patients with a body mass index (BMI) of~ kg/m  to either  IU or  IU of uhCG and found no difference in either the number of mature oocytes ( IU . vs  IU .), oocyte maturity rate ( IU % vs  IU %), fertilization rate ( IU .% vs  IU .%), calculated mature oocyte yield from follicles . mm on day of trigger ( IU % vs  IU %), or rates of moderate to severe OHSS ( IU .% vs  IU .%) (). Follicular fluid hCG correlated with serum hCG levels and was proportional to dose (serum levels:  IU . vs  IU . IU/L; follicular hCG level: . vs . IU/L) (). Interestingly, clinical pregnancy rates per transfer were higher following  IU of uhCG ( IU .% vs  IU .%; P = .) (). Other retrospective studies have concurred that doses of hCG . to  IU are unlikely to confer further benefit on oocyte maturation, and an increase in pregnancy rates was not confirmed with higher doses of hCG (, ). Thus,  IU of uhCG is likely to be sufficient for most patients; however, other factors such as body weight may need to be taken into account for individual patients.
The impact of serum hCG levels was assessed in  patients who received either  IU, , IU, or , IU of hCG intramuscularly based on serum estradiol levels (). Serum hCG levels measured the day following administration suggested a proportional dose response:  mIU/mL following  IU,  mIU/mL following , IU, and  mIU/mL following , IU (). The oocyte yield based on aggregated data (number of oocytes divided by number of follicles . mm) did not increase at doses . IU (% at  IU, % at , IU, and % at , IU) (). Lin et al. () categorized their cohort by BMI and found that the serum hCG level achieved was lower in those with higher BMI values. Serum hCG at oocyte retrieval was  IU/L in those with BMI ,kg/m  and  IU/L in those with BMI of  of  kg/m  in patients receiving  IU of uhCG (). Other studies have also reported reduced circulating hCG levels in patients with higher BMI (-). Shah et al. () undertook a prospective randomized crossover trial to investigate whether route of administration (intramuscularly or subcutaneously) or BMI would affect pharmacokinetic properties of hCG. Twenty-two women received either intramuscular uhCG (, IU) or subcutaneous rhCG ( mg) during the follicular phase of the menstrual cycle (). The mean concentration ( vs  mIU/mL), maximum concentration ( vs  mIU/mL), and the area under the curve during the first  hours ( vs  mIU/mL) were higher in patients receiving intramuscular uhCG than "Kisspeptin could be a promising future option, particularly in the woman at high risk of OHSS." subcutaneous rhCG (). The mean concentration ( vs  mIU/mL), maximum concentration ( vs  mIU/mL), and the area under the curve during the first  hours ( vs  mIU/mL) were lower in obese women (BMI of  to  kg/m  ) than in women with normal BMI ( to  kg/m  ) ().
In , Gunnala et al. () retrospectively reviewed , IVF/intracytoplasmic sperm injection (ICSI) cycles in which uhCG was administered at varying doses based on serum estradiol on the day of trigger (, IU when estradiol was , pg/mL;  IU when estradiol was  to pg/mL;  IU was estradiol  to pg/mL; and  IU or dual trigger leuprolide  mg with  IU of hCG when estradiol was . pg/mL). The number of mature oocytes retrieved, fertilization rate, and number of embryos transferred did not differ by dose of hCG administered (). However, estradiol levels on the day of trigger correlate with the number of follicles available to provide a mature oocyte, and thus higher doses of hCG could have been administered in patients with fewer follicles and fewer anticipated oocytes had the same dose been used. Oocyte maturation rate (proportion of oocytes retrieved that were mature) varied by serum level of hCG on the morning after administration (% when hCG was  to  IU/L, % when hCG was  to  IU/L, % when hCG was  to  IU/L, and % when hCG was . IU/L; P , .) (). The same group analyzed , patients with serum bhCG levels $ mU/mL and  patients with serum bhCG levels , mU/L on the day following hCG trigger and found that patients with a BMI $ kg/m  had a -fold increased risk of having low serum bhCG level , mU/L (). Those with a lower bhCG level on the day following hCG trigger had lower oocyte maturation rate (.% vs .%, P , .), lower fertilization rate (.% vs %; P , .), and lower adjusted OR for live birth (adjusted OR = .) (). Similarly, Matorras et al.
() investigated serum hCG levels at  hours following  mg of rhCG subcutaneously in  women, again demonstrating that serum hCG negatively correlates with BMI (serum hCG =  2 . 3 BMI; r  = .) (). Patients with no oocytes retrieved had a lower serum hCG ( mIU/mL) compared with those with at least one oocyte retrieved (. mIU/mL) (). The mean number of oocytes retrieved was similar by categories of serum hCG level (. oocytes recovered even in those , mIU/mL at  hours) and oocyte recovery rates were similar across hCG levels (). This suggests that although most will have effective oocyte maturation with a standard dose, some individuals with higher BMI could benefit from higher hCG doses to ensure efficacious triggering.

GnRHa
Although the ability of GnRHas to trigger oocyte maturation has been recognized since the s (), their potential to induce oocyte maturation was fully realized with the advent of the competitive reversible GnRH antagonists in the s (). GnRHas displace the GnRH antagonist from the GnRH receptor, leading to receptor activation and gonadotropin release from the pituitary gland ().
Schally et al. (, ) first isolated GnRH (at the time more commonly referred to as "luteinizing hormone-releasing hormone") and synthesized analogs by substituting one or more of the  amino acids. Replacement in position  or  resulted in the formation of analogs that were both more potent than endogenous GnRH and had greater duration of action at the GnRH receptor, with examples including triptorelin, leuprolide, and buserelin (-). The half-life of endogenous GnRH is~ to  minutes; however, the half-life of GnRHa is extended (, ) according to amino acid replacement, for example, triptorelin half-life (t / )~ hours, nafarelin t /  to  hours, leuprolide t / . hours, and buserelin t / . hours (). The endogenous LH surge lasts~ hours and consists of three distinct phases (), whereas LH secretion following GnRHa is characterized by two phases, that is, a rapid ascending phase lasting  hours, and a longer descending phase () (see Fig.  for a diagram of hormonal profiles following agents used to induce oocyte maturation). Of relevance, GnRHa activates pituitary GnRH receptors to release both endogenous LH and FSH, whereas hCG possesses only LH-like activity (). Whereas the mid-cycle FSH surge is not critical for oocyte maturation to occur, FSH is known to increase LH receptor expression in granulosa cells and additionally may directly play a role in the expansion of cumulus-oocyte complexes and oocyte maturation (-).

GnRHa preparations and dosing regimes
A number of GnRHas have been used to trigger oocyte maturation, with the literature encompassing different agonists and dosages. Several studies have used buserelin at . mg (, ), whereas triptorelin is frequently prescribed at . mg (, ). A recent RCT demonstrated no difference in LH profiles, the number of mature oocytes, fertilization rates, or embryogenesis in oocyte donation cycles following doses of triptorelin between . mg, . mg, and . mg, suggesting that . mg is at the upper end of the dose-response range ().
Leuprolide acetate has been used at doses ranging from . mg () to  mg (). In , Pabuccu et al.
() performed an RCT, randomizing  women to receive either  mg or  mg of leuprolide acetate, and they found no significant difference in the number of oocytes retrieved, implantation, or clinical pregnancy rates. Similarly, Parneix et al. () compared  women undergoing ovulation induction with  of  different regimens for inducing ovulation, including triptorelin, buserelin (both intranasally and subcutaneously), leuprolide, naforelin, or hCG. The authors reported that all regimens resulted in ovulation with no evidence of superiority of one analog over another, and with similar pregnancy rates between groups and the control (hCG) group ().
In , Şükür et al. retrospectively compared patients who received triptorelin at . mg (n = ), leuprolide at  mg (n = ), or , IU of hCG when serum estradiol was , pg/mL (n = ) (). The efficacy of oocyte maturation appeared similar between the interventions; the number of mature oocytes divided by the number of follicles . mm on the day of oocyte retrieval (calculated on aggregated data) was % for triptorelin, % for leuprolide, and % for hCG (). Thus, at present, although dosing and type of GnRHa vary in the literature, there is insufficient evidence to support preferential use of any GnRHa over another ().

Efficacy of GnRHas compared with hCG
In , Oktay et al. (n = ) compared leuprolide acetate at  mg (n = ) and hCG at  to , IU to induce oocyte maturation in women undergoing fertility preservation treatment (). Although the total number of oocytes was similar between the two groups (GnRHa ., hCG .), a greater number of mature oocytes (GnRHa ., hCG .; P , .), higher fertilization rate (GnRHa .%, hCG .%; P = .), and more zygotes were observed following GnRHa (). A prospective randomized controlled study of  patients by Humaidan et al. () reported that although significantly more oocytes were retrieved following , IU of uhCG (.) than following buserelin at . mg (.), the oocyte maturation rate was higher following buserelin (% vs %), leading to slightly more mature oocytes. Nonetheless, luteal levels of progesterone and estradiol were lower in the GnRHa group, corresponding to reduced implantation (GnRHa  of , hCG  of ) and clinical pregnancy rates (GnRHa %, hCG %), with increased early pregnancy loss (GnRHa %, hCG %) (). In an early meta-analysis, including three studies (), GnRHa was determined to have similar efficacy to hCG with regard to the number of mature oocytes retrieved, oocyte maturation rate, fertilization rate, and embryo quality.
Thus, GnRHa induces a gonadotropin surge sufficient for oocyte maturation, but it induces a shorter duration of LH exposure than hCG. Whereas this affords an improved safety profile, the same property results in a smaller chance of functional corpora lutea and an increased emphasis on adequate luteal phase support to maintain pregnancy rates.

rLH
In the mid-s, rLH became available as a further therapeutic option for use in IVF treatment. Following intravenous administration, rLH has a similar pharmacokinetic profile to endogenous LH with a distribution half-life of~ hour and a terminal half-life of  to  hours (, , ). Peak serum LH levels were attained  to  hours following subcutaneous administration of , IU of rLH with a terminal half-life of~ hours (). Pierson et al. () investigated the use of rLH to trigger ovulation during  ovulation induction cycles trialing doses of , , , ,, or , IU of rLH, or uhCG at  IU. All  patients who received doses between  and , IU ovulated, but  of  patients in the  IU group and  of  patients in the , IU group failed to ovulate (). There was a trend toward a greater rate of ovulation per follicles $ mm with increasing rLH dose (). Sex steroid levels were increased at days  to  in a dose-dependent manner (progesterone,  to . nmol/L), but were still higher after  IU of uhCG (progesterone, . nmol/L) (). One patient who received , IU was diagnosed with moderate OHSS (). The authors concluded that the minimal effective dose of rLH to induce ovulation was  IU.
In , Manau et al. () randomized  women to receive either rLH at  IU subcutaneously or hCG at  IU intramuscularly to trigger oocyte maturation. Patients received additional doses of  IU,  IU, and  IU of hCG or rLH on the days of oocyte retrieval,  days later, and  days later as luteal phase support (). A similar number of mature oocytes were retrieved (. with hCG and . with rLH), with a similar mature oocyte yield (number of mature oocytes from follicles . mm) calculated from aggregated data of .% in the hCG group and .% in the rLH group (). Serum progesterone at  days after administration was higher in the hCG group ( ng/mL) than in the rLH group ( ng/mL) (). After~. embryos were transferred, the implantation rate was similar at % to .%; however, Serum LH or hCG (iU/L) Figure 3. Serum profiles of inductors of oocyte maturation. hCG has a long duration of action with peak serum levels at~18 hours following administration. GnRHa induces a peak serum LH at 4 hours following administration (26). Kisspeptin-54 (KP54) also induces a rise in serum LH at 4 to 6 hours following administration but to a lower amplitude than GnRHa (16)(17)(18). The profile of rLH is less certain and may peak higher and sooner (15). Serum LH at 24 hours following rLH was 20 IU/L following 5000 IU, 60 IU/L following 15,000 IU, and 90 IU/L following 30,000 IU of rLH (15).
 of  patients in the hCG group developed moderate OHSS compared with none in the rLH group (). Moreover, hemodynamic changes were less significant following rLH than following hCG, perhaps due to differential intracellular signaling following activation of the LH receptor ().
In , a multicenter trial across  centers in nine countries was conducted to investigate the efficacy of rLH to induce oocyte maturation in comparison with uhCG (). Two hundred fifty women treated with the long protocol had final oocyte maturation induced by either subcutaneous rLH at doses of  IU (n = ), , IU (n = ), , IU (n = ) (, IU plus , IU administered  days after the first injection; n = ), or intramuscular uhCG at  IU (n = ) (). Mean serum LH levels at  hours following rLH increased dose-dependently from . IU/L following  IU to  IU/L following , IU (see Table  for summary of clinical trial evaluating rLH for inducing oocyte maturation) ().
The number of oocytes and zygotes following rLH increased dose-dependently by approximately one per dosing category (). The estimated mature oocyte yield using aggregated data (number of mature oocytes/number of follicles . mm) was % following  IU of rLH, % following , IU of rLH, and % following , IU of rLH as compared with % to % following uhCG (). Two patients had no oocytes retrieved despite comparable serum LH levels (one in the , IU rLH group and another in the , IU rLH group). The oocyte maturation rate did not show a clear dose-response and was indeed worse at the highest dose of rLH (). Despite a mean of . to . embryos being transferred, the overall clinical pregnancy rate following a single bolus of rLH was disappointing at .% ( of ); however, this was improved in patients who received , IU of rLH followed by , IU  days later to % ( of ) to be equivalent to patients who received uhCG at % ( of ) (). Concomitant with this, serum progesterone levels in patients who received a single bolus of rLH were significantly lower than the uhCG group, but this was rescued in those who received a further bolus of rLH , IU  days later ( vs  nmol/L) (). However, the apparent increase in the number of functional corpora lutea as reflected by the increased estradiol and progesterone levels in those receiving a second bolus of rLH came at the expense of an increased incidence of moderate OHSS at %, as in the uhCG group (.%) (). The proportion of patients with any features of OHSS showed a dose-response of % after  IU of rLH and % following , IU plus , IU  days later of rLH (). Similarly, there was a relationship between dose of rLH and rise in plasma renin on day  after administration (). Given the half-life and assumed time of peak serum LH levels following rLH, it would be reasonable to speculate that serum LH levels following even the lowest dose of rLH (. IU/L at  hours after administration) were likely to have exceeded those found to be effective following GnRHa (~ IU/L at  hours after administration). Although one must be cautious in comparing data across different studies, this study suggested that  IU of rLH was not the top of the dose-response curve for rLH despite achieving such high serum LH levels. A lack of corresponding FSH response may account for some of the reduced efficacy of oocyte maturation in comparison with GnRHa at comparable LH levels. The exact profile of the LH levels achieved and in particular the rise in LH during the first  hours was not clear from the data collected, and thus the time to oocyte retrieval may not have been optimal. However, no clear advantage was observed from the use of rLH over hCG and very large doses were required to achieve efficacy. Thus, rLH is not currently in clinical use as an inductor of oocyte maturation. Oral LH agonists are also in development and have been investigated in ovulation induction cycles, but have yet to be evaluated during IVF cycles.

Kisspeptin
Kisspeptins are a group of hypothalamic argininephenylalanine amide peptides encoded for by the KISS gene on chromosome q (). Kisspeptin isoforms are derived from the proteolytic enzyme cleavage of the -amino acid gene product to yield kisspeptins of different amino acid lengths denoted by their suffix (e.g., kisspeptin- comprises of  amino acids) (). Their activity at the G protein-coupled kisspeptin receptor is conferred by a common C-terminal decapeptide sequence, equivalent to kisspeptin- (, ). Kisspeptin acts at the kisspeptin receptor on GnRH neurons in the hypothalamus to elicit endogenous GnRH release, sufficient to induce a subsequent rise in gonadotropin secretion across a range of mammalian species, including humans (-) (see The pivotal role of kisspeptin in control of the HPG axis became apparent in  when two seminal papers demonstrated that of loss-of-function mutations affecting kisspeptin signaling resulted in hypogonadotropic hypogonadism (, ). Moreover, a girl with an activating mutation in the kisspeptin receptor was reported to have precocious puberty (). These studies revealed the crucial role kisspeptin plays in regulating the function of the reproductive axis.
Furthermore, evidence from sheep () and rodent studies () have determined that kisspeptin signaling is necessary for ovulation () and that administration of kisspeptin can induce ovulation (, ). Kinoshita et al. () demonstrated that administration of a kisspeptin-neutralizing monoclonal antibody directly into the preoptic area of the hypothalamus of female rats during proestrus was sufficient to prevent ovulation. Matsui et al. () simulated an IVF protocol in prepubertal rats using pregnant mare serum gonadotropin to induce follicular growth and demonstrated that a subcutaneous bolus of kisspeptin ( nmol/kg or . nmol/rat) was able to induce ovulation to the same extent as hCG.
In , Dhillo et al. () conducted first in humans administration of kisspeptin; a -minute intravenous infusion of kisspeptin- led to a robust dose-dependent (. pmol/kg/min to  pmol/kg/min) rise in serum LH (~fivefold) in healthy men (). The intravenous half-life of kisspeptin- was determined to be . minutes displaying first-order kinetics (). Kisspeptin-, which has a shorter half-life (~ minutes) than does kisspeptin- (~ minutes), has also been shown to stimulate gonadotropin secretion, both when given as an intravenous bolus (-) and when given as a continuous infusion (, ).
Kisspeptin also stimulated gonadotropin release in healthy women; however, it was noted that there was variation in response to kisspeptin depending on the phase of the menstrual cycle (). Although a small subcutaneous dose of kisspeptin- (. nmol/kg) elicited a modest mean (6SEM) rise in serum LH from baseline in the follicular phase (. 6 . IU/L), the same dose elicited a much greater rise in the preovulatory phase (. 6 . IU/L; P , .) (). Thus, early rodent studies suggesting that kisspeptin was a key regulator of ovulation (, ) and data in women suggesting that kisspeptin could induce an ovulatory LH surge () led the group to investigate whether kisspeptin could be used to induce final oocyte maturation during an IVF protocol ().

The use of kisspeptin to induce oocyte maturation
The first trial evaluating the use of kisspeptin- to induce oocyte maturation was undertaken in  using an adaptive design (). Single doses of kisspeptin- at . nmol/kg (n = ), . nmol/kg (n = ), . nmol/kg (n = ), and . nmol/kg (n = ) were administered  hours prior to oocyte retrieval following a standard short protocol (). Peak plasma levels of kisspeptin were observed at  hour following subcutaneous administration resulting in mean serum LH levels of . IU/L following . nmol/kg of kisspeptin, and . IU/L following . nmol/kg of kisspeptin at  to  hours following administration with serum LH levels tending toward baseline levels at  to  hours following administration (). Of the  patients in the study,  (%) had at least one mature oocyte retrieved, and % ( of ) had at least one embryo available for transfer (see Table  for summary of data from three trials using kisspeptin as a novel inductor of oocyte maturation in IVF treatment) (). The number of mature oocytes increased dose-dependently, although the oocyte maturation rate was similar across doses (). The mature oocyte yield (percentage of mature oocytes from follicles . mm on day of kisspeptin) increased dose-dependently: % to % at . to . nmol/kg, % at . nmol/kg, and % at . nmol/kg (). Standard luteal phase support was provided by Cyclogest (progesterone) at  mg twice daily per vaginal suppository and estradiol valerate  mg three times a day orally (). The live birth rate per protocol at all doses tested was  of  (%) and per transfer was  of  (.%) ().
Once "proof of concept" that kisspeptin could be used as a trigger of oocyte maturation had been established, a further trial was conducted in  to establish the safety and efficacy of kisspeptin in women at high risk of OHSS (). Women were identified as being at high risk of OHSS by having a total antral follicle count (AFC) $ or serum anti-Müllerian hormone (AMH) level $ pmol/L to confer an at least fourfold increase risk of OHSS (). Women were randomized to receive a single subcutaneous bolus of kisspeptin- at doses between . nmol/kg and . nmol/kg (). All women were routinely screened for the development of both early OHSS (assessed  to  days following oocyte retrieval) and late OHSS (assessed  days following embryo transfer) ().
The number of mature oocytes again increased dosedependently (see Table ) (). The mature oocyte yield (proportion of mature oocytes from follicles $ mm on the final ultrasound scan prior to kisspeptin-) was % at . nmol/kg, % at . nmol/kg, % at . nmol/kg, and % at . nmol/kg (). In this study, luteal phase support comprised of intramuscular progesterone (Gestone at  mg daily) in addition to oral estradiol valerate at  mg three times a day (). The live birth rate per transfer was more than doubled in comparison with the first trial at % following all doses tested (). Importantly, although three women (%) were diagnosed with mild early OHSS, no woman had moderate to severe OHSS ().
Thus, early results following a single dose of kisspeptin- were promising with an overall mean (6SD) oocyte yield of % 6 % and no clinically significant OHSS. A further study was designed to assess the variability in response encountered in some women. This third trial investigated whether prolonging the duration of the LH surge using a second dose of kisspeptin at  hours following the first could ensure efficacious oocyte maturation (). Sixty-two women at high risk of OHSS received kisspeptin- at . mol/kg  hours prior to oocyte retrieval (). Patients were then randomized to receive either saline placebo at  hours following the first kisspeptin injection (single group; n = ), or a second dose of kisspeptin- at . nmol/kg (double group; n = ) (). A second dose of kisspeptin improved the proportion of patients achieving an oocyte yield $% from % of patients in the single group to % of "Shorter durations of the LH surge are sufficient for oocyte maturation." patients in the double group (). It also eliminated the retrieval of fewer than four oocytes, but importantly not at the expense of an increased rate of ovarian overresponse (). A unique property of kisspeptin pharmacodynamics became apparent during the trial, whereby a variable rise in LH was observed following the second dose of kisspeptin (). Those who had a lesser LH response following the first dose had a greater subsequent rise following the second dose of kisspeptin (). Conversely, patients who already had a robust LH response following the first dose of kisspeptin had minimal further LH secretion following the second dose (). Thus, the second dose of kisspeptin provided an "individualized" LH response, whereby further LH exposure was only elicited in those patients requiring it (). This led to the second dose altering the distribution of the number of oocytes retrieved, whereby an increased proportion of patients had an intermediate ovarian response (). One patient in the single group was diagnosed with moderate early OHSS, as she was admitted for , hours for abdominal pain on the day following oocyte retrieval and her symptoms settled with conservative management (). The live birth rate per protocol was % ( of ) in the single group and % ( of ) in the double group ().
In summary, kisspeptin acts on the hypothalamus to stimulate the release of endogenous GnRH and subsequent gonadotropin release. To date, the trials using kisspeptin suggest that it could be a promising future option particularly in the woman at high risk of OHSS; however, further trials directly comparing kisspeptin to current modes of inducing oocyte maturation are required.

Endocrine Requirements for Oocyte Maturation and Ovulation
It is relevant to consider evidence from both animal and human studies when evaluating endocrine requirements for triggering final oocyte maturation.

Data from animal studies
The proportion of the LH surge required for oocyte maturation differs from that required for ovulation and the maintenance of functional corpora lutea. Peluso () perfused gonadotropin-stimulated rat ovaries with varying proportions of the gonadotropin surge for  hours to assess the minimum gonadotropin exposure required to induce oocyte maturation and ovulation. Whereas only % of the gonadotropin surge was required to induce oocyte maturation and maximal progesterone secretion, ovulation only occurred in those exposed to % of the gonadotropin surge (). Similarly, Ishikawa () observed that achieving a low level of LH for a longer duration was more able to induce ovulation in proestrus rats than a higher level of LH for a shorter duration. This suggests that a threshold value exists for LH to initiate the process of oocyte maturation/ovulation, and once this level is exceeded, the duration of exposure is more critical for inducing ovulation and supporting functional corpora lutea ().
The duration of LH exposure required for oocyte maturation and ovulation has been further explored in a series of elegant studies undertaken by the Stouffer group in female macaques. In , Zelinski-Wooten et al. () compared the following inductors of oocyte maturation in gonadotropin-stimulated female rhesus monkeys: a single intramuscular bolus of  IU of hCG, a single subcutaneous bolus of  mg of GnRH, three subcutaneous boluses of GnRH at -hour intervals, and two boluses of GnRH at  mg  hours apart. Serum hCG levels remained detectable at  days after administration, whereas a single injection of GnRH caused serum LH to peak at  hours, and return to baseline by  hours (). Three hourly GnRH injections elevated bioactive serum LH for  hours, whereas  hourly GnRH injections elevated bioactive serum LH for  hours (peak serum LH levels were similar between the groups) (). hCG induced a greater proportion of oocytes to be in metaphase I or II Peak levels of serum progesterone were achieved at  days following uhCG injection and resulted in a functional luteal phase of . days. Peak levels of progesterone in the luteal phase with two doses of pituitary hLH or rhLH were . and . ng/mL, respectively, and approached that of uhCG-treated monkeys (. ng/mL) (). However, a single dose of LH was insufficient to maintain functional corpora lutea (midluteal serum progesterone of . ng/mL) ().
In summary, although shorter durations of the LH surge are sufficient for oocyte maturation, a longer duration of at least  hours was required to maintain corpora luteal function in the macaques.

Empty follicle syndrome
In , Coulam et al. () described the "empty follicle syndrome" (EFS) in four cases from which no oocytes were retrieved following , IU of intramuscular hCG to induce oocyte maturation, despite apparently normally growing ovarian follicles. The purpose of the inductor of oocyte maturation is to provide sufficient LH-like exposure to initiate and thus synchronize initiation of the process of oocyte maturation over multiple follicles. This allows most oocytes to be mature and ready for retrieval at a defined time point following the trigger, but prior to ovulation. In the absence of sufficient LH-like exposure from the trigger, insufficient oocyte maturation will result, causing EFS. Immature oocytes are more often surrounded by dense unexpanded cumulus cells and are harder to retrieve than mature oocytes (). Thus, EFS represents a failure of effective triggering and provides a useful defined measure of the very minimum endocrine requirements requisite for oocyte maturation.
EFS is further subcategorized as either "false EFS" whereby an error in administration or reduced absorption of the trigger of oocyte maturation is responsible (two thirds of cases), or "genuine EFS" in which a hormonal response deemed to be sufficient for oocyte maturation is detected but oocyte maturation still does not occur (). The prevalence of all EFS is estimated to be .% to .% and of genuine EFS to be % to .% (). Some units will conduct a urine test for hCG (signifying a level of at least  to  IU/L) or a serum LH level following GnRHa to indicate whether EFS is genuine or false as a result from a problem with administration or absorption ().
To further complicate the diagnosis of genuine EFS, the threshold values for the hormonal response at which oocyte maturation should have occurred are not clearly delineated. The etiology of genuine EFS is thus  ) presented the case of a -year-old woman who suffered recurrent EFS following hCG, but had successful oocyte retrieval following the addition of the GnRHa triptorelin  hours prior to oocyte retrieval in combination with hCG at  mg  hours prior to oocyte retrieval. This case is instructive in suggesting that there may be some patients who may benefit from variation of standard triggering protocols. In , Haas et al. () trialed this protocol in eight additional women who had previously experienced ineffective oocyte yields following a single bolus of hCG and significantly improved the number of oocytes retrieved. Several authors have sought to study the endocrine profiles that predict successful oocyte retrieval to inform the endocrine requirements for oocyte maturation. In , Kummer et al. () analyzed data from  women at high risk of OHSS (. follicles measuring . mm on day of trigger) in whom oocyte maturation was induced by leuprolide at  mg  hours prior to oocyte retrieval and determined an incidence of EFS of .% ( of ). The mean serum LH at  to  hours was . IU/L and serum progesterone was . ng/mL, whereas in cases of EFS, serum LH level was , IU/L (). In patients with oocytes retrieved, the lowest serum LH was . IU/L and serum progesterone was . ng/mL (). Interestingly, BMI was negatively correlated with both LH rise (r = 2., P , .) and posttrigger progesterone (r = 2., P , .) (). The number of mature oocytes positively correlated with serum progesterone at  to  hours (b = .), serum LH at  to  hours (b = .), peak estradiol (b = .), and negatively correlated with age (b = 2.) and BMI (b = 2.) (). Hence, despite being the effector of oocyte maturation, serum LH was a poorer predictor of EFS than was the resultant serum progesterone rise. Nevertheless, a low serum LH , IU/L at  to  hours after leuprolide increased the risk of EFS.
Chang et al. () retrospectively analyzed  patients who received  to  mg leuprolide  hours prior to oocyte retrieval and analyzed hormone levels at~ to  hours. Median serum LH levels were . mIU/mL [interquartile range (IQR), . to .] and median serum progesterone levels were . ng/mL (IQR, . to . ng/mL) measured at a median of . hours (IQR, . to . h) following GnRHa (). BMI was again found to negatively influence LH rise; serum LH was . mIU/mL in women with BMI ,. kg/m  , but . mIU/mL in women with Serum LH prior to administration of GnRH correlates with the subsequent rise in LH (), and thus patients with low baseline serum LH have an increased likelihood of insufficient LH response to GnRHa.
Collectively, these data highlight that patient factors can lead to a variability in response that is important when counseling patients on an individual basis, even if most patients having the same protocol can be expected to have a positive outcome. Meyer et al. () observed that % of patients who received leuprolide at  mg had a serum LH value at  to  hours of , IU/L. Patients with a low serum LH (,. IU/L) on the day of GnRHa increased the risk of having a serum LH level , IU/L at  to  hours following GnRHa from .% to .%, and further to % when serum LH was ,. IU/L (). Thus, a low serum LH level prior to GnRHa increases the risk of a suboptimal rise in LH following GnRHa administration and consequently of EFS.
From  cycles triggered with kisspeptin using a variety of doses across three studies (-), three patients (.%) had no oocytes retrieved, all of whom had a serum LH ,. IU/L at  hours following administration (-). A further three patients had no mature oocytes retrieved and had serum LH values , IU/L at  hours following administration (-). However, many patients had mature oocytes retrieved following similar LH values at  hours, which could reflect a variation in a patient's LH requirement for effective triggering (-). It is noteworthy that oocyte maturation occurred at seemingly lower LH values following kisspeptin than following GnRHa and rLH. In addition to its hypothalamic role, kisspeptin is known to be present in the ovary (). Ovarian kisspeptin expression changes in a cyclical manner during the menstrual cycle, and although undetectable in immature oocytes, kisspeptin expression is increased at ovulation (). Kisspeptin enhances in vitro maturation (IVM) of ovine oocytes () and porcine oocytes (). Although, one can speculate that kisspeptin could enhance oocyte maturation in combination with gonadotropin exposure, it is not likely to do so in the absence of a gonadotropin rise (-).
In summary, an insufficient rise in LH (, IU/L) and progesterone (, ng/mL) following GnRHa increases the likelihood of EFS. However, there is crossover in the values obtained in patients with genuine EFS and in women having normal oocyte retrievals.

Endocrine requirements for efficacious oocyte maturation
Although studying the endocrine profiles to prevent EFS provides an indication of the minimal endocrine requirements for oocyte maturation to occur, this in itself represents the very minimum standard required when assessing trigger efficacy. More usually, one would want to assess the endocrine requirements to provide "efficacious triggering." The number of mature oocytes can provide a valid indication of the efficacy of triggering during an appropriately powered prospective randomized study if an equal number of oocytes are expected in each group. However, the number of mature oocytes that can be expected will heavily depend on individual patient factors aside from the triggering agent studied, especially the number of follicles available to provide a mature oocyte if effective triggering is provided. Another frequently reported measure is the "oocyte maturation rate" (proportion of oocytes that are mature). However, immature oocytes are also more difficult to retrieve, and thus insufficient triggering can lead to fewer oocytes retrieved and thus a reduction in both the denominator as well as the numerator, making this a less reliable measure of trigger efficacy. A recommended approach for quantifying trigger efficacy is to report the "mature oocyte yield," whereby the number of mature oocytes retrieved is corrected for the number of follicles on the day of trigger, of a size from which one would expect a mature oocyte to be retrieved if effective triggering is provided (e.g., number of mature oocytes divided by the number of follicles of  to  mm on the day of trigger) (). The presence of nuclear oocyte maturation is easily assessed by the appearance of a polar body denoting a metaphase II oocyte; however, cytoplasmic oocyte maturation requires more detailed imaging to fully assess, which may not be readily available in many centers. The fertilization rate is used by some authors as a surrogate measure to indicate that cytoplasmic oocyte maturation has occurred.
Oocyte yield was assessed by Chen et al. () in  patients who received GnRHa triptorelin at . mg  to  hours prior to oocyte retrieval. Mean serum LH at  hours after GnRHa was . IU/L (range, . to . IU/L), but .% of patients had a serum LH , IU/L (). The oocyte yield (number of oocytes as a proportion of follicles . mm on day of GnRHa) was % and oocyte maturation rate was % in patients with a  hour serum LH , IU/L as compared with an oocyte yield of % to % and oocyte maturation rates of % to % in patients with higher LH values (). Serum LH on the day of GnRHa administration was . IU/L in patients with a  hour serum LH , IU/L as compared with . IU/L in remaining patients, again reinforcing the concept that a low endogenous serum LH may predict an inadequate rise in LH following GnRHa triggering (). No significant differences were observed in other outcomes, including oocyte maturation, fertilization rate, and clinical pregnancy rate (although cycles were supplemented with hCG for luteal phase support) (). Similarly, in , Shapiro et al. () found a modest reduction in oocyte yield (defined as proportion of oocytes from follicles $ mm on day of GnRHa) and mature oocyte yield (defined as ratio of mature oocytes to the number of follicles $ mm on day of GnRHa) when serum LH at  hours was , IU/L, but a more dramatic reduction when the serum LH was , IU/L. Oocyte yield was % when  hour LH was , IU/L, % when  hour LH was , IU/L, and % when  hour LH was . IU/L ().

Combination of hCG and GnRHa-a role of FSH in oocyte maturation?
HCG is an effective inductor of oocyte maturation but provides only LH-like exposure, suggesting that the mid-cycle FSH surge observed in the natural cycle is not requisite for successful oocyte maturation. A potential advantage of GnRHa and kisspeptin over hCG and rLH is the concomitant release of FSH in addition to LH-like activity. FSH can promote formation of LH receptors in luteinizing granulosa cells, nuclear maturation, and cumulus expansion (, , ). In , Zelinski-Wooten et al. () reported that a large bolus of  IU of recombinant FSH in isolation was able to induce oocyte maturation to a similar extent as hCG in the female rhesus monkey. Peak FSH concentrations were observed  to  hours following injection and had returned to baseline by  hours (). However, FSH alone was unable to sustain the luteal phase, suggesting that LH action is additionally required for maintaining corpus luteal function (). Consistent with this, Bianchi et al.
() reported the case of a -year-old woman with PCOS who was treated with a long IVF protocol, but administered  IU of recombinant FSH rather than hCG in error. The patient had nine oocytes retrieved, of which eight were mature, and treated oocytes underwent normal fertilization, although pregnancy did not ensue (). Furthermore, Rosen et al. () noted that intrafollicular FSH levels corrected for follicular size were higher in follicles that yielded an oocyte. In , Lamb et al. () conducted a randomized double-blind placebo controlled trial in  women to assess whether additional recombinant FSH could enhance oocyte maturation if given concomitantly with hCG at time of trigger. Women treated with the long protocol with serum estradiol levels , pg/mL were randomized to receive either hCG at , IU with  IU of recombinant FSH, or hCG alone (). Fertilization rate was significantly improved with supplementation of recombinant FSH (% vs %) and a greater likelihood of oocyte recovery was observed, defined as the rate of obtaining an oocyte from a single mature-sized follicle on each ovary (% vs %) (). Clinical pregnancy rate (.% vs .%) and ongoing/live birth rates (.% vs .%) were not significantly improved ().
Owing to the duration of the LH surge following GnRHa being insufficient to support functional corpora lutea and support implantation, there is increasing interest in using a combination of GnRHa with a small dose of hCG. Some investigators have given them simultaneously (termed "dual trigger"), whereas others have administered hCG later to rescue the luteal phase (termed "double trigger").
In , Haas et al. () conducted a pilot study assessing whether patients with a low oocyte yield (defined as the ratio of the number of oocytes retrieved divided by the number of follicles . mm on day of trigger of ,%) in response to hCG could be improved by the addition of a GnRHa and increasing the interval to oocyte retrieval. Eight patients with a low oocyte yield following hCG at  mg administered  hours prior to oocyte retrieval subsequently received . mg of triptorelin  hours prior and hCG at  mg  hours prior to oocyte retrieval (). The number of oocytes retrieved improved from . to , the number of oocytes from follicles .mm improved from % to %, and the number of oocytes from follicles . mm improved from % to %, with three of eight patients having ongoing pregnancies (). The same group showed a similar improvement in  patients with low oocyte maturation rates (,%) (). Again, improvements were seen in the number of oocytes with double trigger (. vs .), the number of mature oocytes (. vs .), the oocyte maturation rate (% vs %), and the ongoing pregnancy rate (% vs %) (). Since two interventions were investigated simultaneously, it is difficult to identify which intervention resulted in the improvements observed. Although there is biological plausibility to the addition of GnRHa providing additional LH and FSH exposure, and several other nonrandomized studies have reported that patients with poor oocyte maturation in a first cycle triggered with hCG can have improved outcomes in a subsequent cycle if supplemented with GnRHa (, , ), such a study design may be susceptible to "regression to the mean" if randomization to a control intervention is not also assessed.
In , Schachter et al. () conducted an RCT in  short protocol IVF cycles to compare hCG at  IU alone (n = ), with hCG and triptorelin at . mg "The need for additional GnRHa during triggering is yet to be clearly demonstrated and further data are required." given in combination (n = ) at  hours prior to oocyte retrieval. Participants had at least one previous failed IVF protocol, but not with EFS (). One of  patients (.%) in the hCG group and  of  patients (.%) in the GnRHa-supplemented group had no oocytes retrieved and a further  (.%) of the hCG group and  (.%) of the GnRHasupplemented group had no fertilization (). However, in contrast to these results, patients in the hCG group had . oocytes retrieved and the GnRHa-supplemented group had . oocytes retrieved (). Serum FSH on the day of oocyte retrieval was higher in the GnRHa-supplemented group (. vs . IU/L), as was serum LH (. vs . IU/L) (). The ongoing pregnancy rate per embryo transfer was % in the hCG group and % in the GnRHa-supplemented group ().
In , Decleer et al. () conducted an RCT comparing  patients who received  IU of hCG alone with  patients who concomitantly received GnRHa at . mg of triptorelin  hours prior to oocyte retrieval. A mean of . mature oocytes were retrieved in the GnRHa-supplemented group as compared with . mature oocytes in those receiving hCG alone (). There were . more zygotes formed in the GnRHa-supplemented group with % of patients having a high-quality embryo formed, as compared with % of the hCG alone group (). The ongoing pregnancy rate was % in the hCG group and % in the GnRHa-supplemented group, with no cases of OHSS reported in either group (). As expected, serum LH was~ IU/L  day one after receiving hCG with GnRHa, but interestingly serum FSH rose in both groups from~ IU/L on the day of trigger to~ IU/L  day after the day of hCG, and to  IU/L in the GnRHa-supplemented group (). Serum progesterone was highest on the day of oocyte retrieval in both groups at~ to  ng/L (). The analysis did not demonstrate any difference in the number of oocytes, mature oocytes, zygotes, or implantation rate, although it did find an increase in the pregnancy rate in the GnRHasupplemented group as compared with hCG alone (relative risk, .; % CI, . to .) (). In summary, additional FSH exposure is suggested to enhance oocyte maturation, although LH/hCG play a dominant role and the additional impact of FSH is likely to be small. Although several reports have suggested that GnRHa supplementation may improve oocyte maturation in patients with a history of poor oocyte maturation, the need for additional GnRHa during triggering is yet to be clearly demonstrated and further data are required. In the past, it had been suggested that GnRHa may directly impair implantation rates perhaps through a direct negative action on the endometrium, or through other effects beyond those due to a shorter duration of action than hCG; however, these data would suggest that such effects are unlikely to be clinically significant in practice.

Interval between hCG and oocyte retrieval
There is a continuum between the processes of oocyte maturation and ovulation if sufficient LH-like exposure is provided. Thus, it is critically important to schedule oocyte retrieval at a precise interval following administration of the agent of oocyte maturation, such that it takes place following oocyte maturation, but prior to ovulation. If the interval is too long, ovulation may have occurred prior to retrieval and oocytes will no longer be present within the follicles, whereas if the interval is too short, insufficient time may have been provided for optimal oocyte maturation and oocyte retrieval is more difficult.
In the natural cycle, the median interval from the rise in LH to ovulation is  hours (% CI,  to ) and following the peak in LH is . hours (% CI,  to ) (). Ovulation occurs in % of women between  and  hours after the first significant rise in LH and between 2 and  hours after the peak (). Andersen et al. () conducted a study in clomiphene-stimulated cycles to assess the time of first ovulation following intramuscular hCG. In % of cases, the largest follicle was the first to rupture and the mean time to ovulation following hCG was . hours, although the range was between  and  hours (). In , Nader and Berkowitz () found that ovulation occurred within  hours following intramuscular hCG administration in some women, and thus suggested that an interval , hours between hCG administration and oocyte retrieval be used. This was in keeping with evidence from clomiphenestimulated cycles that suggested that the interval between hCG administration and ovulation was , hours for most patients, but could extend to . hours in some patients (). These data highlight the variability in time to ovulation among different women; however, ovulation could be expected to occur sooner after ovulation induction cycles than in cycles with multifollicular development.
Nargund et al. () studied  long protocol cycles with an interval between intramuscular , IU of uhCG and oocyte retrieval ranging between  and  hours. No significant differences were observed in the ratio of oocytes retrieved divided by the number of follicles punctured among the interval groups ( to , hours, .%;  to , hours, .%;  to , hours, .%) (). Lengthening the interval up to  hours did not increase the proportion of cycles affected by premature ovulation, nor were there significant differences in the number of oocytes retrieved, fertilization rates, or clinical pregnancy rates (). Similarly, in , Bjercke et al. () found no difference in number of oocytes retrieved, oocytecumulus complex quality, embryo quality, implantation, or pregnancy rates when women were randomized to undergo oocyte retrieval at either  hours or  hours following hCG. Bosdou et al. () randomized  normo-ovulatory women treated with the short protocol to either an interval of  hours or  hours following  mg of rhCG and oocyte retrieval and found no difference in oocyte retrieval rate (number of cumulusoocyte complexes divided by number of follicles $ mm on day of hCG), maturation rate (% vs %), fertilization rate (% vs %), or number of zygotes. Importantly, no patient ovulated prematurely due to the extension of interval between hCG and oocyte retrieval (). Interestingly, a case report has suggested that mature oocytes can be retrieved following a prolonged interval of~ hours after hCG administration (). Successful fertilization and embryo transfer were performed, although no subsequent pregnancy ensued, suggesting that the interval between hCG and oocyte retrieval can be extended from the standard  hours in many patients without causing ovulation.
In , Wang et al. () conduced a metaanalysis of five trials investigating the interval between hCG and oocyte retrieval (, , -), including a total of  patients. Oocyte maturation rate was significantly higher when oocyte retrieval was performed . hours after hCG administration, compared with , hours after administration (risk ratio, .; % CI, . to .) (). There was no difference observed in fertilization rate, implantation rate, or clinical pregnancy rate (). However, the studies included were heterogeneous, with timing of the "short interval" of oocyte retrieval ranging from  to  hours, and the "long interval" ranging from  to  hours ().
Ghasemian et al. () investigated  cycles in which oocyte retrieval had been carried out at either , , , or  hours following , IU of uhCG and found that the oocyte maturation rate increased with interval to oocyte retrieval at  hours [.% (n = )],  hours [.% (n = )],  hours [.% (n = )], and  hours [.% (n = )]. Oocyte morphology (extracytoplasmic, referring to the perivitellin space and zona pellucida; cytoplasmic quality, referring to the presence of dark or granular cytoplasm suggestive of aggregates of smooth endoplasmic reticulum, and presence of vacuole) was most frequently normal in those with a -hour interval: retrieved oocytes at  hours .% (n = ), at  hours .% (n = ), at  hours .% (n = ), and  hours .% (n = ) (). Thus, a longer interval to oocyte retrieval may result in fewer good-quality oocytes and fewer highquality embryos (), although data supporting the value of oocyte morphology remain conflicting ().
Thus, overall these data suggest that there is likely to be significant interindividual variation in the interval between the LH surge and the time to ovulation in the natural cycle. Similarly, in stimulated cycles, there is evidence to suggest that extending the interval between hCG administration and oocyte retrieval beyond the standard  hours is unlikely to lead to the frequent occurrence of premature ovulation. Data from studies with the double trigger discussed below, in which a GnRHa is administered  hours prior to oocyte retrieval in combination with hCG  hours prior, also suggest that for some patients there may be a benefit to extending the interval to oocyte retrieval in those not responding well to standard protocols (). However, to date there is insufficient evidence to suggest an improvement in outcomes with longer intervals between trigger and oocyte retrieval for most patients, although further studies taking into account follicle size profiles on day of trigger and hormonal response following trigger, especially in patients with suboptimal responses to standard protocols, are required to fully resolve this issue.

Follicle size at time of trigger administration
During controlled ovarian stimulation, follicles are stimulated to grow under the action of a pharmacological dose of FSH. It is widely accepted that follicles that are too small are less likely to yield mature oocytes following LH-like exposure (). Conversely, ovarian follicles that grow too large can become "postmature" and are also less likely to yield a mature oocyte (). Thus, most IVF centers will monitor follicular size and administer the trigger of oocyte maturation, once follicles are deemed to have grown to an appropriate size, typically judged as two to three lead follicles . to  mm in diameter.
Knowledge of the size of follicles at the time of trigger expected to yield an oocyte can also enable the more accurate quantification of trigger efficacy. In , Shapiro et al. () observed that GnRHa resulted in significantly more oocytes retrieved (.) than for hCG (.). However, those treated with GnRHa had more follicles on the day of trigger (GnRHa .; hCG .), making it difficult to accurately compare trigger efficacy between the two groups (). In view of this, Shapiro et al. () proposed the concept of "oocyte yield," whereby the number of oocytes is corrected for the number of follicles on the day of trigger administration. They reported a mature oocyte yield defined as proportion of mature oocytes from follicles $ mm on the day of trigger of % following GnRHa (). However, the low percentage encountered following an established dose of an effective trigger suggests that the follicle size denominator chosen empirically may have been too broad, perhaps encompassing follicles that were too small to yield an oocyte. Other denominators have been used, with some authors reporting both the "Some patients who received kisspeptin achieved mature oocyte yields over 100%." number of follicles $ mm and the number of follicles $ mm on the day of trigger to account for different estimations of oocyte yield (). Studies evaluating the efficacy of kisspeptin to induce oocyte maturation have used a denominator of follicles $ mm on day of trigger and achieved a reasonable dose response (-). Similarly, some patients who received kisspeptin achieved mature oocyte yields over %, suggesting that follicles , mm may have also contributed to the number of mature oocytes retrieved.
There are limited data to justify the categories of follicle size on day of trigger used to estimate oocyte yield in the current literature, and none of the thresholds includes an upper limit for follicle size at which postmature follicles may become more prevalent. There does, however, exist relevant data on the follicle sizes on the day of oocyte retrieval that are most likely to yield an oocyte.
Rosen et al. () observed that the odds of retrieving a mature oocyte from a follicle  to  mm on the day of oocyte retrieval in size is reduced by % compared with follicles . mm. Wittmaack et al.
() reported that follicles with a volume , mL (~. mm) or . mL (~. mm) on the day of oocyte retrieval had lower oocyte yields (%) when compared with those between  and mL (~% to %). Dubey et al. () determined that fertilization rates were increased in oocytes from larger follicles on the day of oocyte retrieval ( to  mm, .%;  to  mm, .%;  to  mm, .%). Ectors et al. () found that fertilization rates were greatest in follicles  to mm in size on the day of oocyte retrieval (%), compared with either those , mm (%) or those . mm (%). Oocyte maturation rates were ..% in those follicles . mm on the day of oocyte retrieval, as compared with .% in those , mm (). Overall, data suggest that follicles of  to  mm on the day of oocyte retrieval are most likely to yield an oocyte ().
In , Hu et al. () analyzed  IVF cycles treated with the short protocol and categorized patients by the proportion of follicles $ mm on the day of trigger that were also $ mm as low proportion (% of follicles $ mm were also $ mm), middle proportion (% to % of follicles $ mm were also $ mm), or high proportion (.% of follicles $ mm were also $ mm). The number of oocytes retrieved was greatest in patients with a low proportion of follicles $ mm (oocyte number: low, .; middle, .; high, .), suggesting that follicles . mm on day of trigger contribute less to the number of oocytes retrieved than do smaller follicles (). Oocyte maturation rate was low (%), middle (%), and high (%) and fertilization rate was low (%), middle (%), and high (%) ().
There also exist data investigating the impact of adjusting the day of trigger administration, although no clear consensus was apparent. Kolibianakis et al.
() randomized patients to receive either the trigger once three or more follicles had reached $ mm in diameter, or to delay administration of the trigger by  hours thereafter. Delayed triggering resulted in . fewer follicles of  to  mm and . more follicles of $ mm with an associated rise in progesterone of . ng/mL and detrimental effects on pregnancy potential, but a nonsignificant increase of . oocytes retrieved (). Similarly, Kyrou et al. () compared administration of hCG once three follicles were $ mm in diameter (early), or  hours later (late), and found that delaying triggering increased the number of mature oocytes retrieved (early ., late ., P = .) with an associated rise in serum progesterone levels by . ng/mL. Mochtar et al. () randomized women to receive hCG once the lead follicle was either  mm or  mm, and they observed that those with a lead follicle of  mm had a greater number of follicles of  to  mm on day of trigger (. vs .) and an increase of two oocytes retrieved. Vandekerckhove et al. () found that a -hour delay in trigger administration of patients with three or more follicles of $ mm (and % to % of follicles $ mm were also $ mm) increased the number of mature oocytes retrieved by ., but only when serum progesterone was # ng/mL. This could suggest that larger follicles with evidence of luteinization may less likely yield an oocyte than do larger follicles without evidence of luteinization. Conversely, Tan et al. randomized patients to receive hCG either once the lead follicle was mm with a further  follicles .mm, or  day later, or  days later, and observed no difference in the number of oocytes retrieved (). Similarly, Tremellen et al. () found that patients with "ideal" timing of the hCG trigger (defined as two or more follicles of $mm, with most follicles $ mm) had similar outcomes to patients triggered either a day earlier or later. In , Chen et al. () conducted a meta-analysis including seven RCTs and  IVF cycles comparing hCG administration as soon as three or more follicles were $ mm in size ("early") compared with either  or  hours later ("late"). Fertilization rates were higher in the  hours later group (P , .), although this result was predominantly attributable to the results of one study, and overall no significant benefit was observed ().
In summary, the size of follicles at the time of trigger can influence the likelihood that LH-like exposure can induce oocyte maturation. Most reproductive medicine centers administer hCG once two to three lead follicles are  to  mm in diameter. When follicles grow as a tight cohort behind the lead follicle, the lead follicle provides a reasonable representation of all of the follicles. However, when follicle sizes on day of hCG are more disparate, the lead follicle may perform less reliably as a representation of all follicles. Data on specific follicle sizes that are most likely to yield an oocyte have predominantly been generated on the day of oocyte retrieval, at which time follicles of  to  mm are thought to be most likely to yield oocytes (). Data from our own group suggest that follicles of  to  mm on the day of trigger contribute most to the number of mature oocytes retrieved (). Indeed, patients with a greater proportion of their follicles within this range had more mature oocytes retrieved (). These data also allow for a datadriven estimation of trigger efficacy. Thus, we recommend that "mature oocyte yield" defined as the proportion of mature oocytes retrieved from follicles of  to  mm on the day of trigger is used to more accurately assess trigger efficacy. Further prospective studies are required to identify whether administering the trigger by a measure other than lead follicle size can benefit outcomes, although such a benefit may only be apparent in patients with a wide distribution of follicle size during stimulation.
Intrafollicular changes following hCG, GnRHa, or kisspeptin Higher intrafollicular reproductive hormone levels have been associated with an improved chance of oocyte retrieval. Rosen et al. () observed that intrafollicular FSH levels were higher in follicles that yielded an oocyte. Lamb et al. () observed that oocytes fertilized by ICSI were % to % more likely to be retrieved from follicles with higher intrafollicular concentrations of estradiol and testosterone, whereas oocytes fertilized by IVF were % to % more likely to arise from follicles with higher estradiol or progesterone concentrations. Similarly, Itskovitz et al.
() found that intrafollicular estradiol and progesterone levels were higher in follicles containing a mature oocyte. Interestingly, intrafollicular kisspeptin levels are higher than corresponding serum kisspeptin levels, and they correlate with follicular fluid estradiol levels and the number of mature oocytes retrieved (). Haas et al. () assessed alterations in expression of genes related to steroidogenesis in granulosa cells of  women who received either GnRHa or hCG triggering. Expression of the enzymes CYPA (. vs ) and CYPA (. vs ), as well as b-hydroxysteroid-dehydrogenase (. vs ), vascular endothelial growth factor (VEGF; . vs ), and inhibin b B (. vs ), was significantly lower in the GnRHa group (). Expression of the FSH receptor was also significantly lower in the GnRHa group, but not expression of the LH receptor (). Amphiregulin and epiregulin are ligands of the EGF receptor on mural granulosa cells, and amphiregulin's expression was inversely related to fertilization rate (). These EGF ligands have been proposed to be paracrine mediators of the LH signal to stimulate oocyte maturation (). LH is known to stimulate upregulation of amphiregulin and epiregulin. Amphiregulin expression was .-fold higher in mural granulosa cells in the GnRHa group, although not in follicular fluid (). Expression of amphiregulin and epiregulin were both increased more than twofold in patients receiving both GnRHa and hCG in comparison with hCG alone (). Expression of pigment epithelium-derived factor (an antiangiogenic factor secreted from granulosa cells) was also increased .-fold, whereas cumulus cell conexin was reduced by % in the GnRHa-supplemented group ().
Owens et al. () investigated expression of genes involved in ovarian reproductive function, steroidogenesis, and OHSS in granulosa lutein cells following the use of hCG, GnRHa, or kisspeptin to induce oocyte maturation in  women undergoing IVF treatment. Kisspeptin- increased expression of genes involved in ovarian steroidogenesis, the FSH receptor, the LHCG receptor, steroid acute regulatory protein (STAR), aromatase, estrogen receptors a and b (ESR, ESR), b-hydroxysteroid dehydrogenase type  (BHSD), and inhibin A, when compared with either hCG or GnRHa (). Whereas in vitro treatment of granulosa lutein cells with hCG induced steroidogenic gene expression, kisspeptin- had no significant direct effects on either OHSS or steroidogenic genes ().
Although the increase in rates of OHSS with hCG have predominantly been ascribed to its longer duration of action, evidence for additional direct actions at the ovary may also be contributory. Neulen et al.
() observed that hCG dose dependently induced VEGF expression in luteinized granulosa cells. Kitajima et al. () reported that GnRHa caused involution of corpora lutea of superovulated rats and reduced expression of VEGF, VEGF receptor , and VEGF receptor  and reduced vascular permeability in the ovaries of hCG-treated hyperstimulated rats. Similarly, hCG has been shown to directly increase VEGF expression and VEGF levels in human granulosa cells (, , ), whereas GnRHa may act directly on ovarian GnRH receptors to induce luteolysis (, ). Furthermore, the kisspeptin receptor has been hypothesized to play a key role in the pathogenesis of OHSS (). Exogenous kisspeptin administration has been reported to reduce VEGF levels via a direct action on ovarian kisspeptin receptors to mitigate the risk of OHSS ().
Lessons from IVM IVM is the process by which immature cumulusoocyte complexes derived from antral follicles are matured in vitro (). IVM was originally described in the context of unstimulated cycles without gonadotropin priming () and thus has been proposed as a useful option for women with PCOS who may have large numbers of small antral follicles putting them at increased risk of OHSS (). There are typically three regimens used during IVM-the first is the original "The kisspeptin receptor has been hypothesized to play a key role in the pathogenesis of OHSS." unstimulated cycle, whereby antral follicles are collected once follicles reach  to  mm in size before follicle dominance is established (, ). Alternative protocols, including priming with either FSH (-) or hCG (, ), have also been introduced in an attempt to increase oocyte yield and maturation rates, although controversy remains regarding whether these should be strictly thought of as IVM given that some in vivo maturation may also occur (, ).
However, IVM provides a unique opportunity to gain lessons on the size of follicle from which mature oocytes can be retrieved (), as well as the optimal gonadotropin environment for oocyte maturation. Initial evidence from rodent studies suggested that meiosis was less likely to take place in oocytes retrieved from small follicles when cultured in vitro (), with % to % of oocytes retrieved from antral follicles ( to  mm in diameter) progressing to metaphase I or metaphase II, compared with only % of those from preantral follicles ( to  mm in diameter) (). However, in human studies, evidence of maturation potential has been observed in oocytes retrieved from follicles as small as  mm, and mature oocytes from follicles # mm following hCG priming had similar outcomes to those from larger follicles (). Furthermore, evidence from studies priming follicles with hCG prior to retrieval has revealed that follicles , mm may possess granulosa cells with hCG receptors and can resume meiosis despite their small size (). In their investigation of  hCGprimed IVM cycles in  patients with polycystic ovaries, Son et al. () reported that no significant difference in oocyte diameter, fertilization rate, or cleavage embryo quality was observed in oocytes obtained from follicles  to  mm, or those obtained from follicles , mm. Furthermore, .% of oocytes retrieved from follicles , mm underwent oocyte maturation, with a fertilization rate of .% ().
Typically IVM has been undertaken in women at high risk of OHSS, although more recently, increased use for either fertility preservation in women undergoing cancer treatment or as a "rescue" treatment of women with poor ovarian reserve has been investigated. A prospective study compared  patients with normal ovarian reserve to  patients with poor ovarian reserve, and retrieved hCG-primed immature oocytes (). For both normal responders, and those with low ovarian reserve, IVM increased the proportion of MII oocytes (). At  hours, significantly greater proportion of germinal vesicles from women with low ovarian reserve had reached the MII stage, compared with those with normal ovarian reserve () (.% vs .%; P = .). However, fertilization rates and cleavage rates were similar between both groups.
In summary, the size of follicle from which mature oocytes can be retrieved can additionally be gleaned from studies of IVM (). Follicles as small as  mm have been found to contain mature oocytes, and mature oocytes from follicles # mm following hCG priming resulted in similar outcomes to those retrieved from larger follicles (). However, the rate of in vivo-matured oocytes positively correlates with dominant follicle size (dominant follicle # mm, .%;  to  mm, .%; . mm, .%) (). Similarly, Triwitayakorn et al. () observed that oocyte recovery rate increased from % of follicles , mm to % of follicles  to  mm and further to % of follicles . mm on the day of oocyte retrieval.

Are gonadotropins mandatory for maturation?
It was originally demonstrated that human oocytes may persist at the germinal vesicle stage in vitro for up to  hours after collection, but that beyond this, they could resume meiosis independently of gonadotropins (). However, given the variable maturation rates and cycle pregnancy rates in early IVM protocols (, ), the controlled addition of gonadotropins to culture medium was soon shown to improve the efficiency of IVM (). Gonadotropins are hypothesized to exert their effect on oocyte maturation indirectly via follicular cumulus cells; however, oocytes possess gonadotropin receptors and thus may also act directly ().
Studies have shown disparate results regarding the optimum ratio of FSH/LH required for IVM. Anderiesz et al. () found that the addition of recombinant FSH (rFSH) either alone or in combination with rLH in a ratio of : (to replicate gonadotropin concentrations during the endogenous mid-cycle LH surge) nonsignificantly increased oocyte maturation by % or by %, respectively (). Choi et al. () found that cumulus expansion increased in proportion to concentrations of FSH and LH in a bovine animal model, and was maximal at  ng/mL FSH and  mg/mL LH. Hreinsson et al. () compared culture media supplemented with either . IU/mL hCG or . IU/mL LH and observed no significant difference in the proportion of oocytes that underwent maturation in the different cultures (% hCG vs % LH). Although LH may not be a critical component of culture medium (), activation of the LH receptor mediates cellular effects contributing to oocyte stability, with EGF a key mediator in transmitting LH receptor activation signaling to the cumulus cells and oocyte (, ). Another factor, brain-derived neurotropic factor expressed by granulosa cells following LH/hCG signaling, also increases oocyte maturation ().
Several studies have suggested that kisspeptin may have additional direct effects at the ovary via ovarian kisspeptin receptors, beyond its predominant mode of action via endogenous GnRH release from the hypothalamus (-, ). Castellano et al. () observed that kisspeptin expression increased in a cyclical manner during the menstrual cycle of a rodent model; kisspeptin was predominantly localized to the theca layer of growing follicles and the corpora lutea. Ovarian kisspeptin expression increased at ovulation, but was undetectable in immature oocytes (). Kisspeptin has also been shown to increase IVM of ovine () and porcine immature oocytes and to increase blastocyst formation rate and blastocyst hatching (). Although yet to be directly compared, oocyte maturation rates following kisspeptin appear comparable to other triggers despite serum LH levels achieved following kisspeptin seemingly being lower than those observed following rLH or GnRHa. Furthermore, there is a suggestion that kisspeptin may mature oocytes from smaller follicles than current triggers (). However, kisspeptin has a short half-life, with circulating kisspeptin levels peaking at~ hour following subcutaneous injection and hence there is only a short duration of exposure to kisspeptin (). Although one can speculate that kisspeptin could enhance oocyte maturation in combination with gonadotropin exposure through its predominant mode of action at the hypothalamus, it is unlikely that in vivo administration can lead to oocyte maturation in the absence of a gonadotropin response (-). Further studies investigating whether IVM of immature oocytes can be enhanced when kisspeptin is added to the culture medium would be of interest.

Luteal Phase Characteristics Following Different Agents That Induce Oocyte Maturation
In the natural menstrual cycle, the luteal phase is defined as the period between ovulation and menstruation, or establishment of pregnancy (). The corpus luteum secretes estrogen and progesterone to support the endometrium for implantation and placentation (, ). Stimulation of the LH receptor is required to maintain survival of the corpus luteum (, ), and inhibition of pituitary LH by either GnRHa (, ) or GnRH antagonist () results in luteolysis, with regression observed after  hours without LH activity (, ).
All IVF cycles are characterized by luteal phase dysfunction, and thus hormonal supplementation with luteal phase support, especially progesterone, is required to maintain adequate pregnancy rates (, ). hCG has a longer duration of action () and is able to better maintain survival of the corpora lutea than shorter acting triggers such as GnRHa (, ). Increased survival of corpora lutea following hCG improves endogenous sex steroid production and better maintains pregnancy rates (, ), but this comes at the expense of an increased risk of OHSS (, ). Despite GnRHa demonstrating a better safety profile, early studies with GnRHa were associated with reduced pregnancy rates and increased early pregnancy losses (). For this reason, hCG is widely used as the preferred agent to induce oocyte maturation for most patients at low risk of OHSS, whereas GnRHa is predominantly reserved for patients at high risk of OHSS, although more intensive strategies to support the luteal phase are required.
The luteal phase is deficient following all agents used to induce oocyte maturation The luteal phase of all stimulated IVF cycles is dysfunctional (, ). When  women were randomized to receive either  mg of rhCG,  mg of rLH, or . mg of triptorelin to induce oocyte maturation without luteal phase supplementation, the luteal phase following all three was observed to be deficient (). Although the oocyte maturation rate (proportion of oocytes that are mature) was comparable between the three groups (rhCG %, rLH %, and GnRHa %) (), median serum LH on the day of oocyte retrieval significantly differed: rHCG . IU/L, rLH . IU/L, GnRHa . IU/L (). The length of the luteal phase was best maintained by hCG: peak progesterone occurred on day  after rhCG, day  after rLH, and day  after GnRHa (P , .), and the day of progesterone decrease was day  for rhCG, day  for rLH, and day  for GnRHa (). The study was prematurely terminated due to low pregnancy rates (% to %) in all three groups (). Thus, even though the luteal phase is better preserved following hCG than other triggers, luteal phase supplementation is a mandatory component of all IVF cycles (). The importance of progesterone for maintenance of pregnancy is long established, with early studies revealing that despite lutectomy, pregnancies could be supported by exogenous progesterone (). Fanchin et al. () observed that increasing progesterone exposure was associated with reduced uterine contractility and increased pregnancy rates.
In , Emperaire et al. () suggested that patients with a poor luteal phase in one ovulation induction cycle using GnRHa are likely to respond similarly in a subsequent cycle if GnRHa is used again. This remained the case even when the GnRHa dose was increased (. mg) or given over three boluses of .mg; however, luteal phase support with  IU of hCG brought the luteal phase closer to normal (). The authors therefore suggested that some women could have a tendency toward a dysfunctional luteal phase regardless of the trigger used ().

Routes of progesterone administration for luteal phase support
Vaginal progesterone results in lower circulating levels of progesterone when compared with intramuscular progesterone; however, local endometrial levels are much higher (). As intramuscular progesterone can "Some women could have a tendency toward a dysfunctional luteal phase regardless of the trigger used." be uncomfortable, vaginal progesterone is more often used, especially in hCG-triggered patients (). As progesterone produced by the ovary normally reaches the endometrium via the peripheral circulation, parenteral progesterone has been suggested as being more similar to physiological pathways. Oral progesterone was initially suggested as luteal phase support during the s, but early studies demonstrated a lack of endometrial secretory changes when compared with those receiving intramuscular or vaginal progesterone due to significant first-pass metabolism (, ). Recently, dydrogesterone, a metabolite of progesterone possessing biological activity and good oral bioavailability, has been shown to have similar efficacy to vaginal progesterone in hCG-triggered patients ().
Luteal phase support following different agents of oocyte maturation In , Humaidan et al. () randomized  women to receive either GnRHa or hCG to induce oocyte maturation. All received luteal phase support in the form of vaginal progesterone at  mg daily, and estradiol at  mg daily orally (). Mean serum hormonal levels at  days after oocyte retrieval for GnRHa-and hCG-treated patients were as follows: LH (. IU/L vs . IU/L), FSH (. IU/L vs . IU/L), estradiol (. nmol/L vs . nmol/L), and progesterone ( nmol/L vs  nmol/L) (). Clinical pregnancy rates per cycle were significantly reduced following GnRHa compared with hCG (% vs %) (). Thus, there was a recognition that the luteal phase required more intensive support following GnRHa-triggered cycles than hCG-triggered cycles.
One approach pioneered by Engmann et al. () was to use high-dose sex steroids with intensive intramuscular progesterone and estradiol supplementation. In an RCT comparing GnRHa and hCG to induce oocyte maturation in  women at high risk of OHSS, luteal phase support was provided intramuscularly by  mg of progesterone titrated up to  mg intramuscularly to maintain serum progesterone levels .  ng/mL (. nmol/l) and  3 .-mg transdermal estrogen patches on alternate days titrated up to four patches and  mg of oral estrogen twice daily to maintain serum estradiol level . pg/mL (. pmol/L) (). Serum progesterone and estradiol levels were both lower on the day of embryo transfer following GnRHa (serum estradiol,  pg/mL vs  pg/mL; serum progesterone,  ng/mL vs  ng/mL) (). However, the implantation rate (% vs %), clinical pregnancy rate (.% vs .%), and ongoing pregnancy rate (.% vs .%) per transfer following GnRHa and hCG were similar (). Unfortunately, other investigators were not able to replicate the same excellent pregnancy rates with this intensive luteal phase support regimen (, ). A retrospective cohort study compared  women at high risk of OHSS ($ follicles $ mm on day of trigger) triggered with hCG and  women triggered with GnRHa triptorelin at . mg with intensive luteal phase support (intramuscular progesterone at  mg daily and vaginal progesterone at  mg twice daily and  mg of estradiol valerate) (). Live-birth rates were similar (GnRHa .% vs hCG .%) between the groups, but although one late-onset severe OHSS case was observed in the GnRHa group (.%),  (%) were observed after hCG (). If luteal phase support strategies can be shown to reliably maintain pregnancy rates following GnRHa to the same extent as hCG, then the preferable safety profile of GnRHa could encourage its use as a first-line agent more widely.
Use of hCG for luteal phase support Given the reduction in pregnancy rates with GnRHa, there is great interest in supporting the luteal phase using a small dose of hCG either given at the same time as GnRHa or at an interval during the early luteal phase to stimulate endogenous progesterone production.
Four oocyte donors underwent four oocyte donor cycles within  year to assess the luteal phase characteristics following different regimens (). Following the short protocol, women received one of the following regimens: () hCG at , IU to induce oocyte maturation, followed by standard LPS ( mg of vaginal progesterone three times daily and  mg of estradiol valerate daily from the day after oocyte retrieval), () GnRHa (. mg of triptorelin) and  IU of hCG  hours thereafter with standard LPS, () GnRHa (. mg of triptorelin), or () GnRHa without luteal phase support (). Estradiol and progesterone levels were higher at  days following oocyte retrieval in women who either received hCG either to induce oocyte maturation or as part of LPS (). Estradiol on day  following oocyte retrieval was  ng/L (, IU of hCG plus LPS),  ng/L (GnRHa at . mg of triptorelin plus  IU of hCG),  ng/L (GnRHa plus LPS), and  ng/L (GnRHa) (). Progesterone on day  following oocyte retrieval was also lower at  mg/L (, IU hCG plus LPS),  mg/L (GnRHa at . mg of triptorelin plus  IU of hCG), . mg/L (GnRHa plus LPS), and . mg/L (GnRHa) (). Thus, a small dose of hCG given  hours following GnRHa was able to simulate the luteal phase characteristics of hCG-triggered cycles.
In , Humaidan et al. () reported that administration of  IU of hCG at the time of oocyte retrieval was sufficient to support the luteal phase in those receiving GnRHa to a similar extent as those receiving , IU of hCG to induce oocyte maturation. In , Humaidan et al. () randomized women at high risk of OHSS ( to  follicles $mm) to receive either () GnRHa buserelin at . mg followed by a single bolus of  IU of hCG for luteal phase support (n = ), or () to  IU of hCG (n = ). Women assessed as not being at high risk of OHSS (, follicles $ mm) were randomized to receive either () GnRHa buserelin at . mg followed by hCG at  IU on the day of oocyte retrieval and a further dose  days thereafter (n = ), or () hCG at  IU (n = ) (). All women also received micronized progesterone at  mg twice daily (). There was no significant difference in pregnancy rates between groups (ongoing pregnancy rate per randomization of % to %) (). Two cases of moderately late OHSS occurred in both group , who received hCG, and a further two cases in group , who received two small doses of hCG for luteal phase support (). However, Seyhan et al. () reported high rates of severe OHSS ( of ) when hCG supplementation was used in patients at increased risk of OHSS, and thus this approach is not recommended in patients with a very large number of follicles on the day of trigger ().
Kol et al. () investigated whether luteal phase support could be provided by hCG supplementation alone in the absence of progesterone supplementation. Fifteen patients were triggered with GnRHa (triptorelin at . mg) and received  IU of hCG following oocyte retrieval and again  days later, achieving an ongoing clinical pregnancy rate of . Similarly, Andersen et al. () conducted a proof-of-concept study in  women, demonstrating that low-dose hCG ( to  IU daily) could be used to generate endogenous progesterone production and support pregnancy without the need for exogenous progesterone.
Thus, although the luteal phase may be insufficient following GnRHa trigger in many patients, this may not be universal. A recent study sought to investigate this concept, termed "luteal coasting," whereby the luteal phase is monitored and a rescue dose of hCG is administered only when progesterone levels drop (). Three women at high risk of OHSS received a short protocol with . mg of triptorelin to induce oocyte maturation (). Serum progesterone was measured  hours after oocyte retrieval, and supplemental hCG was administered at varying doses when serum progesterone dropped to , ng/mL (). In two out of three patients, this approach was sufficient to support the luteal phase (). In a further observational study by Lawrenz et al. (),  women at risk for OHSS received GnRHa (. mg of triptorelin) to induce final oocyte maturation, and vaginal progesterone supplementation was started from the night of oocyte retrieval and continued at  mg three times daily thereafter. Serum progesterone was measured  hours after oocyte retrieval and used to assess whether participants required additional luteal support in the form of hCG supplementation (). Mean serum progesterone  hours after administration of GnRHa was . ng/mL (range,  to  ng/mL) (). Thus, luteal phase deficit can be variable, and in the future tailored supplementation regimens may be developed ().
Use of GnRH agonist for luteal phase support GnRH receptors are present throughout the endometrium in stromal and epithelial cells, and their expression is increased during the secretory phase (, ), as are LH receptors (, ). Hypotheses for the reduced pregnancy rates observed following GnRHa have included those related to factors aside from the duration of the LH surge, such as a direct endometrial action to prevent implantation, or a direct induction of luteolysis via GnRH receptors. However, there is evidence to suggest that GnRHa can be used as luteal phase support. A prospective placebo-controlled study by Tesarik et al. () investigated the effects of GnRHa administration at the time of embryo transfer in oocyte donor cycles, using either . mg of triptorelin  days following embryo transfer or placebo. The implantation rate was significantly higher (.% vs .%, P , .) in women receiving GnRHa on the day of embryo transfer (). Interestingly, intravenous administration of GnRH at  mg during pregnancy can stimulate production of hCG from the placenta (). In , Pirard et al. () found that similar numbers of patients achieved clinical pregnancy following hCG at , IU with  mg of micronized progesterone compared with those receiving buserelin at  mg and then  mg three times daily (two of five vs three of five, respectively). Despite the small sample size, this study suggested that low-dose GnRHa could be used to support the luteal phase ().
A randomized prospective study investigated the effect of mid-luteal administration of GnRHa in both  short and  long protocols, where hCG had been used to induce oocyte maturation (). Women were randomized to receive either GnRHa or placebo  days after oocyte retrieval (). All women received  mg of estradiol daily and  mg of vaginal micronized progesterone daily from the day of oocyte retrieval for  days, and additionally received  mg of rhCG on the day of embryo transfer (). Both estradiol and progesterone levels were greater at  days following oocyte retrieval in the GnRHa-treated patients (estradiol, GnRHa at  pg/mL, placebo at  pg/mL; progesterone, GnRHa at  ng/mL, placebo at  ng/mL) (). Implantation rates were .% luteal phase GnRHa vs .% placebo (P , .) and live birth rates per intention to treat were .% luteal phase GnRHa vs .% placebo (P , .) (). However, to date the use of GnRHa for luteal phase support is not widely used in practice.
In summary, all agents of oocyte maturation can induce luteal phase defect and require luteal phase support (). However, luteal phase deficit is more pronounced following GnRHa than hCG (), and thus more intensive luteal phase support is required. Intensive luteal phase support with high-dose sex steroid supplementation is an attractive option, as this strategy will not increase the risk of OHSS ().
"Improving luteal phase support regimens following GnRHa to achieve reliable pregnancy rates can extend the use of these agents." Conversely, care must be taken when using even a small dose of hCG for luteal phase support in women at very increased risk of OHSS to maintain the benefit for safety in avoiding hCG triggering (). Improving luteal phase support regimens following GnRHa to achieve reliable pregnancy rates can extend the use of these agents more widely in place of hCG.

OHSS Following Different Agents That Induce Oocyte Maturation
OHSS is one of the most common complications of IVF treatment () and is predominantly related to the use of hCG to induce final oocyte maturation (). The prolonged duration of action of hCG results in overstimulation of the ovaries and the release of vasoactive substances from the ovary, particularly VEGF-A, which causes leakage of fluid from the vascular space into the third spaces of the body (). Thus, OHSS is a potentially life-threatening iatrogenic condition that can result in massive ovarian enlargement, ascites, hydrothorax, renal failure, acute respiratory distress syndrome, and rarely even death (estimated at  per ,) ().
The most commonly used diagnostic criteria for OHSS are those of Golan et al. () from  with the updated categorization by Navot et al. () in . Mild OHSS is reported to occur in one third of cycles, moderate OHSS in a tenth, and severe OHSS in % of IVF cycles using hCG to induce oocyte maturation (). Mild OHSS predominantly consists of symptoms alone and is likely to resolve with conservative management. Hence, mild OHSS is not regarded as clinically significant by some practitioners and is often not reported (). Moderate OHSS is characterized by the additional presence of ascites on ultrasound, and severe OHSS as additionally having evidence of hemoconcentration, renal impairment, or respiratory distress ().
A further subcategorization of OHSS is used to reflect a difference in pathophysiology by the time of onset following oocyte retrieval: "early OHSS' occurs within  days of oocyte retrieval, whereas "late OHSS" occurs  days or more following oocyte retrieval (-). Early OHSS relates to the use of hCG to induce final oocyte maturation (or for luteal phase support), whereas late OHSS relates to endogenous hCG production from a developing pregnancy and thus can be further exacerbated by multiple pregnancy (). Consequently, early OHSS can be prevented through the use of alternate triggers of oocyte maturation than hCG (, ), whereas late OHSS can be mitigated by segmentation (cryopreservation of all embryos with embryo transfer in a subsequent cycle) and avoidance of multiple transfers (). However, even the use of GnRHa for inducing oocyte maturation and in combination with segmentation does not completely eliminate the risk of severe OHSS (-).
Late OHSS is often more severe and harder to manage than early OHSS as the stimulus for hCG production is ongoing (pregnancy). Whereas late OHSS is often considered a separate entity to early OHSS, it is noteworthy that late OHSS almost never occurs in the context of frozen embryo transfers where ovarian stimulation has not recently been carried out, even in high-risk patients, suggesting that late OHSS represents an exacerbation of subclinical early OHSS by subsequent pregnancy-related hCG production. Consequently, the use of alternative triggers to hCG can be expected to reduce the risk of late OHSS as well as early OHSS. Increased use of segmentation can reduce the occurrence of late OHSS; however, the "risk of OHSS" remains one of the most frequent reasons for cycle cancellation prior to embryo transfer across the world. In Europe, .% of the , IVF cycles started in  were cancelled prior to oocyte retrieval and .% of cycles commenced did not have a fresh embryo transfer (). Similarly in the United States, .% of , IVF cycles commenced in  were cancelled prior to oocyte retrieval, of which % were due to "ovarian overresponse" and .% of IVF cycles were segmented (). In the United Kingdom, .% of cycles were cancelled prior to oocyte retrieval due to risk of OHSS, and this was also the most common reason for cycle cancellation between oocyte retrieval and embryo transfer (% of cycle cancellations at this stage) ().
Rates of OHSS from retrospective studies relying on patient-initiated presentation for assessment could lead to underreporting of OHSS rates in comparison with studies where routine assessments are made (). In a well-conducted prospective clinical trial, severe OHSS occurred in .% to .% of patients and moderate OHSS in a further .% to .% of patients depending on whether a short or long protocol was used (). Despite these high rates of clinically significant OHSS (up to .%), these were rates of OHSS in an unselected population not at increased risk of OHSS (). In the % of patients with "irregular cycles" (implying the presence of polycystic ovarian syndrome), the rate of severe OHSS was further increased to .% ().
In , Shapiro et al. () retrospectively analyzed  IVF cycles in which an hCG dose between  IU and , IU was used depending on each patient's weight and OHSS risk to evaluate whether hCG level could predict OHSS risk. Serum hCG levels ranged between  and  IU/L at  to  hours following hCG (). Of  cycles,  were diagnosed as OHSS (.%) and  required paracentesis (.%) (). Patients with OHSS had a mean serum hCG of  IU/L (range,  to  IU/L) (). The number of follicles on the day of hCG and the serum hCG levels were independent predictors of OHSS (). The risk of OHSS in patients with  follicles on the day of hCG increased from % in those with serum hCG at  IU/L, to % when serum hCG was  IU/L, to % when serum hCG was  IU/L, and to % when serum hCG was  IU/L (). The risk of OHSS in patients with  follicles on the day of hCG increased from % in those with serum hCG at  IU/L, to % when serum hCG was  IU/L, to % when serum hCG was  IU/L, and to % when serum hCG was  IU/L (). Patients with  follicles had an increase in risk from % to % with hCG level, whereas patients with  follicles on the day of hCG had minimally increased risk even with higher serum hCG levels (). Thus, both serum hCG level as well as number of follicles influenced the risk of OHSS.
Fábregues et al. () compared the characteristics of IVF cycles of  women who were diagnosed with severe OHSS in their first long protocol IVF cycle, but not in a subsequent cycle within  months. During  years the incidence of OHSS at the center was .% ( of  cycles) and % of patients included in this study had PCOS (). Patients received intramuscular hCG  to  hours prior to oocyte retrieval, coasting up to  days and transfer of up to  embryos (). Patients had more mature oocytes retrieved in the OHSS cycle (. vs .) and higher estradiol levels on day of hCG ( vs  pg/mL), but they had similar implantation rates (% to %) (). This study suggests that careful management of cycles in women at risk for OHSS can help to mitigate the risk in subsequent cycles ().
In , Mathur et al. () conducted a retrospective analysis of rates of OHSS in  cycles in  patients with serum estradiol levels ,, pmol/L and , follicles $ mm in diameter who received  IU of hCG  hours prior to oocyte retrieval. In patients with , oocytes retrieved or serum estradiol ,, pmol/L, hCG at  IU was administered on the day of embryo transfer (). Early OHSS occurred in  cycles (%) and late OHSS in a further .% at a median time of  days after oocyte retrieval (). Patients with OHSS had more oocytes retrieved (median  vs ) and were more likely to have PCOS (.% vs .%) (). The incidence of early OHSS increased with the number of oocytes retrieved from~% in those with  to  oocytes to~% in those with . oocytes (). Late OHSS rates also rose, but to a lesser extent (~% in those with . oocytes) (). The number of oocytes predicting early moderate to severe OHSS was nine (positive likelihood ratio ., negative likelihood ratio .) (). Serum estradiol level on day of hCG predicting early moderate to severe OHSS was  pmol/L (sensitivity %, specificity %, positive likelihood ratio ., negative likelihood ratio .) (). These and similar data help to inform the risk of OHSS following an hCG trigger. Although the number of follicles on day of trigger is the best predictor of subsequent OHSS, the cutoffs are not absolute and there remains uncertainty in the subsequent risk of OHSS. Other markers of increased risk of OHSS include serum AMH, AFC, estradiol levels, number of intermediately sized follicles on day of trigger, and number of oocytes retrieved (, -).
OHSS risk in high-risk populations Women with polycystic ovaries have an approximately fivefold increase in risk of OHSS (). MacDougall et al. () observed that polycystic ovaries on ultrasound were found in % of severe OHSS cases and in % of moderate OHSS cases compared with % of the general patient population. The use of GnRHa to induce final oocyte maturation can significantly reduce the incidence of OHSS in comparison with hCG; however, a number of case reports have suggested that severe OHSS may still occur in the high-risk patient even when triggered with a GnRHa and treated with segmentation (-). A retrospective analysis of SART database by Steward et al. () in  reported that retrieval of at least  oocytes was predictive of OHSS risk. Swanton et al. () reported that patients with PCO morphology or PCOS had between  and  oocytes retrieved, and reported severe OHSS rates of .% to .%. Similarly, Jacob et al. () recently reported a clinical trial in women with PCOS; despite a median of  to  oocytes being retrieved, the study reported moderate/severe OHSS rates of % to %, which the authors state may have been an underestimate due to a lack of routine screening (). Furthermore,  of  patients were cancelled due to risk of overresponse ().
In , Krishna et al. () conducted a randomized unblinded study of  women under the age of  years who met the Rotterdam criteria for PCOS. Patients with serum E , pg/mL received either GnRHa at . mg of triptorelin (n = ) or  mg of hCG (n = ) to induce oocyte maturation (). Approximately % of patients were oligomenorrheic, mean AFC was  to , mean AMH was . ng/mL in the GnRHa group, and . ng/mL in the hCG group (). There were . follicles . mm in the GnRHa group and . follicles . mm in the hCG group (). Only  patient (%) was diagnosed with mild OHSS in the GnRHa group whereas in the hCG group, only  patients (.%) were not diagnosed with OHSS: % had mild OHSS, % had moderate OHSS, and % had severe OHSS (). Patients in the GnRHa group had more oocytes retrieved (. vs .), more mature oocytes (. vs .), a higher oocyte maturation rate (% vs %), and a higher proportion of patients had a top quality cleavage embryo formed (% vs %) (). A calculated mature oocyte yield from aggregated data (number of mature oocytes divided by number of follicles . mm) was % in the GnRHa group and .% in the hCG group ().
"GnRHa is preferable to hCG in the patient at increased risk of OHSS." Thus, GnRHa is preferable to hCG in the patient at increased risk of OHSS.
To date there have been two clinical trials investigating the use of kisspeptin in populations at high risk of OHSS comprising  patients (, ). Women , years old and BMI , kg/m  were identified as being high risk for OHSS by serum AMH level $ pmol/L or AFC $ and received a single subcutaneous bolus of kisspeptin- at doses of . to . nmol/kg (, ). In the first trial of  women at increased risk of OHSS, % of patients had an AMH $ pmol/L, all patients had an AFC $, and % had an AFC $ (). Furthermore, % of women had . follicles and % of women had . follicles (). A quarter of the women (n = ) had previously had an IVF cycle using hCG to induce oocyte maturation, and % ( of ) of them had developed severe OHSS requiring admission to a hospital for medical intervention or intensive care support (). Despite the high risk of the cohort, only % were diagnosed with mild early OHSS and % with mild late OHSS, but no patient was diagnosed with moderate to severe OHSS (). The second trial using kisspeptin in a cohort of women at high risk of OHSS included  women with the same inclusion criteria (). Women received either one or two doses of kisspeptin  hours apart (). Despite a second dose of kisspeptin extending the LH surge, there was no increase in the rates of OHSS (). One woman was diagnosed with moderate early OHSS in single group (.%), and one mild late OHSS (.%) in the double group ().
A single center retrospective cohort study compared clinical parameters of OHSS in hCG (n = ), GnRHa (n = ), or kisspeptin- (n = ) in women at risk for OHSS identified by AFC . or total number of follicles on day of trigger . (). Women had a median of  antral follicles,  follicles $ mm on the day of trigger, and  oocytes retrieved (). Median ovarian volume at  to  days after oocyte retrieval was larger following hCG ( mL) than GnRHa ( mL; P , .), and in turn kisspeptin ( mL; P , .) (). Median ovarian volume remained enlarged -fold following hCG, -fold following GnRHa, and -fold following kisspeptin compared with prestimulation ovarian volumes (). Mean (6SD) ascitic volumes were lesser following GnRHa ( 6  mL) and kisspeptin ( 6  mL) than hCG ( 6  mL; P , .) (). Symptoms were most frequent following hCG and least frequent following kisspeptin (). Moderate to severe OHSS occurred in .% of patients following hCG, % following GnRHa, and no patient following kisspeptin (). The OR for OHSS was . (CI, . to .) following hCG and . (CI, . to .)following GnRHa, when compared with kisspeptin (). These data are consistent with a proposed role for kisspeptin in the pathogenesis of OHSS beyond that due to duration of action. Exogenous kisspeptin administration has been reported to reduce VEGF levels via a direct action on ovarian kisspeptin receptors to mitigate the risk of OHSS (). Nevertheless, the reduced rates of OHSS following kisspeptin observed during the trials so far require verification in prospective studies directly comparing kisspeptin to current triggers of oocyte maturation. Kisspeptin analogs are currently in development and may allow for a further novel triggering option in the future.

Summary
The mode by which oocyte maturation is induced has a significant impact on the ability to retrieve mature oocytes, the luteal phase characteristics predicating implantation, and the risk of OHSS. An appreciation of the endocrine and temporal requirements for oocyte maturation enables the optimization of current IVF protocols and the development of novel approaches to induce oocyte maturation to improve both the safety and efficacy of IVF treatment.