Hormonal Stimulation of Human Ovarian Xenografts in Mice: Studying Folliculogenesis, Activation, and Oocyte Maturation

Abstract

Ovarian tissue cryopreservation and banking provides a fertility preservation option for patients who cannot undergo oocyte retrieval; it is quickly becoming a critical component of assisted reproductive technology programs across the world. While the transplantation of cryopreserved ovarian tissue has resulted in over 130 live births, the field has ample room for technological improvements. Specifically, the functional timeline of grafted tissue and each patient’s probability of achieving pregnancy is largely unpredictable due to patient-to-patient variability in ovarian reserve, lack of a reliable method for quantifying follicle numbers within tissue fragments, potential risk of reintroduction of cancer cells harbored in ovarian tissues, and an inability to control follicle activation rates. This review focuses on one of the most common physiological techniques used to study human ovarian tissue transplantation, xenotransplantation of human ovarian tissue to mice and endeavors to inform future studies by discussing the elements of the xenotransplantation model, challenges unique to the use of human ovarian tissue, and novel tissue engineering techniques currently under investigation.

Early in 2020 the American Society for Reproductive Medicine (ASRM) published a committee opinion classifying ovarian tissue cryopreservation (OTC) and banking as an acceptable fertility preservation technique that is no longer considered experimental (1). This is an important breakthrough since OTC is the sole fertility preservation option available to prepubertal girls undergoing anticancer treatments and patients who cannot delay anticancer treatments to undergo ovarian stimulation and oocyte retrieval (2, 3). Although an exact success rate is unknown, large scale studies from fertility centers conducting OTC and transplantation have reported up to 33% of women achieving pregnancy (4). As of 2017, autotransplantation of cryopreserved ovarian tissue has led to more than 130 live births (5-17). These numbers are especially encouraging considering the growing number of childhood cancer survivors (18), who may desire to have children in the future (19).

Yet, even as OTC becomes a more widely implemented practice in fertility centers across the world, autotransplantation remains contraindicated for patients with blood-borne cancers, such as leukemia, due to the potential risk of cancer cell reintroduction (1). Furthermore, the likelihood of success following ovarian tissue transplantation becomes increasingly variable the older a woman is at the time of OTC (20, 21). Beyond a general understanding that younger patients at the time of OTC are more likely to achieve pregnancy after transplantation (22), no reliable methods exist to determine if a specific patient will have successful restoration of ovarian function following grafting. Without improved methods to evaluate the number and quality of follicles transplanted, the question of whether a woman will achieve pregnancy or restoration of ovarian function following transplantation remains much of a mystery. As the field advances, it is critical to provide patients with as many options as possible to allow each woman to select the option(s) that best fit her health status and lifestyle and that maximize her chances to have children.

Ongoing research aims to address many of the challenges in ovarian tissue transplantation by developing methods to mitigate the risk of cancer reintroduction, identifying markers to predict each patient’s likelihood of achieving pregnancy, and optimizing protocols for implementation in assisted reproductive technology programs. Xenotransplantation of ovarian tissue to mice is one of the most common in vivo models used to study OTC and transplantation and has been conducted using ovarian tissue from numerous species including goat (23), bovine (24), swine (25), and humans. Xenotransplantation of human ovarian tissue (OTX) in mice is one of the most advanced and clinically relevant models for conducting this research and is the focus of this mini-review. A comprehensive literature search was conducted using the PubMed database and the search terms “xenotransplantation ovarian tissue” and “xenotransplantation ovarian follicle.” These terms returned 529 and 169 results, respectively, and the abstracts of these results were screened for relevance. Articles were excluded if transplanted tissue was from fetuses or transgender patients who had already begun hormone therapy, if no information was included about the source of donor tissue, if tissue was cultured in vitro prior to implantation, or if a study’s focus was anticancer therapies. After removing irrelevant articles, the remaining 65 articles were reviewed in full and then 3 additional articles were identified by screening citations.

OTX allows for the study of human folliculogenesis in a mammal with a functionally similar to human hypothalamus-pituitary-gonadal axis, while posing a considerably lower ethical and financial cost compared with nonhuman primate models. Recent demonstration that complete folliculogenesis from primordial stages to mature oocytes can be achieved in vitro with human follicles (26) paves the way for focused research in this area to optimize the approach for therapeutic benefit and achieve consistent outcomes (27). This leaves OTX as the only reliable method to study human follicles at all developmental stages from their resting primordial state to ovulation and corpus luteum formation. Although various models of OTX are available in many mammalian species (23-25, 28-34) the focus of this review is on the various models for human OTX in mice with specific emphasis on how recent findings can be capitalized upon to direct future research into fertility preservation for cancer survivors.

The Xenotransplantation Model

Selecting an appropriate host

Survival and function of xenografts requires suppression of the host immune system, either with pharmacological agents or via mutations that lead to immune cell depletion. Immunosuppressive agents, such as cyclosporine, may have significant side effects and require frequent administration and dose titration. Conversely, immunodeficient mice, such as NU/J (“nude”) or severe combined immunodeficient mice (SCID), serve as a more convenient and robust xenograft model.

Nude mice contain a mutation in the Foxn1 gene, lack hair, and have a dysfunctional thymus (35). These mice have significantly reduced T-cell populations, but still produce functional B-cells and natural killer cells (35). Some OTX studies have reported adverse events due to the invasion of remaining immune cells, particularly lymphocytes (36). Interestingly, the natural killer cells of nude mice are more potent than natural killer cells in other common mouse strains (35). Although the nude mutation is available on a handful of genetic backgrounds with both inbred and outbred strains used in OTX research, over the past 10 years the field has been transitioning from nude mice to the use of more immunodeficient SCID models.

SCID mice lack both T-cells and B-cells due to a mutation in the Prkdc gene (35). The SCID mutation is available on a number of genetic backgrounds, but in the case of SCID mice the degree of immunodeficiency strongly correlates with the genetic background. For example, C57BL/6, Balb/c and CB-17 mice carrying the SCID mutation are susceptible to a phenomenon known as leakiness (35). Leaky SCID mice develop functional B-cells and T-cells as they age. This renders them no longer immunodeficient, leading to rejection of allo- and xenografts. However, leakiness is negligible or absent in nonobese diabetic (NOD) mice with the SCID mutation (35); for this reason, NOD-SCID mice are becoming an increasingly common model due to their stable immunodeficiency. While NOD-SCID mice are significantly more immunodeficient than nude models, they still carry the risk of immune response to allo- or xenografts due to unaffected production of healthy populations of natural killer cells, macrophages, and dendritic cells. The addition of a mutation in the IL2RG gene in NOD-SCID mice reduces or completely eliminates the response of these cell populations. These mice, called NOG and NSG mice, are considered some of the most immunodeficient mouse models available and several studies advocate for their use over NOD-SCID mice (37). In particular, Terada et al demonstrated higher rates of graft recovery and antral follicle formation in NOG mice compared with NOD-SCID mice (38).

While immune-deficient mice allow for the in vivo study of human ovarian tissue, it is known that the immune system plays a role in folliculogenesis and ovarian endocrine function (39-42). Both the peripheral blood and endometrium undergo characteristic changes in immune cell populations that correspond to various stages in the menstrual cycle (43). Furthermore, immune cells are known to play a role in the formation and maintenance of ovarian structures such as the corpora lutea (44), and dysregulated immune function is associated with some reproductive disorders such as endometriosis (45). It is difficult to hypothesize how immunodeficient hosts may affect the survival and function of ovarian grafts but it is important to consider that some processes may be dysregulated due to the absence or altered function of the host immune system.

Selecting an implant location

The grafting location plays a critical role in the survival, function, and longevity of implanted tissue. Fragments of human ovarian tissue and isolated follicles embedded in support matrices have been grafted in multiple locations in mice, including the ovarian bursa (OB) (46), ovarian fossa (47), subcutaneous space (SC) on the back, neck, or thigh (47-49), under the kidney capsule (KC) (50), in the intraperitoneal cavity (IP) (51), within the muscle (IM) (52), and under the muscle fascia (37). Each location presents a different set of benefits and limitations, and to date the field has not agreed upon a standard surgical site for OTX studies.

Fast, efficient, and stable revascularization after implantation of the ovarian xenograft is a key component in ensuring follicle survival. Van Eyck et al hypothesized that substantial ischemic follicle loss occurs before functional vasculature forms within the transplanted tissue (53). However, it is difficult to quantify the magnitude of this loss due to (i) the inability to noninvasively, nondestructively characterize follicle populations within tissue fragments; and (ii) the heterogeneity of follicle populations present in individual tissue fragments (54). By comparing revascularization of implanted cortical pieces between different grafting sites, Nisolle et al discovered that vascularization was superior in IP grafts over SC grafts (55). Additionally, IM grafts showed a greater number of blood vessels and superior integration with host tissue when compared with KC grafts (50). Finally, a qualitative analysis of vascularization reported small surface vessels and a reddish color in IP and IM grafts but found that ovarian tissue remained white and more separated from host tissue in SC or OB grafts (56).

The ability of the implantation site to spatially accommodate multiple growing follicles is another important consideration when selecting a grafting location. Human preovulatory follicles, which reach up to 2 centimeters in diameter, are nearly 45 times larger than the average preovulatory mouse follicle which only reaches around 400 microns in diameter. Additionally, multiple human follicles within implanted fragments can simultaneously grow, resulting in significant volumetric expansion of grafted tissue. Soleimani et al (50) reported significantly larger antral follicles formed in IM versus KC grafts. Terada et al (38) reported that significantly more antral follicles formed in fragments of ovarian cortex grafted in the OB compared with SC and KC grafts. Finally, Dath et al (56) observed antral follicles only in IP and OB grafts when they compared IP, OB, SC, and IM grafting sites. It is difficult to pinpoint a cause for these differences in antral follicle formation as these differences may be more closely tied to variation in surgical technique at each site rather than to physiological differences in the sites themselves. For example, Dath et al (56) reported greater contact area between graft and host tissue at the IM and IP sites, compared with OB and SC. Considering that both IM and IP grafts were fixed to host tissue with sutures, their presence and corresponding microinjury may be a contributing factor in promoting angiogenesis and host/graft integration. The methods of suturing and securing the grafts have not been specified in studies of OB and SC grafts.

Finally, it is important to know how the chosen grafting location impacts graft recovery and follicle activation. Two independent studies reported the highest rates of graft recovery at IP and IM sites when compared with KC, SC, and OB sites (50, 56). One study found grafts from the KC site had higher recovery rates than SC grafts (36). There were no differences in primordial follicle activation among the different grafting locations discussed.

As discussed above, none of the implant locations appear to have a consistent, significant advantage over the others. While IP, IM, and OB sites appear to be superior in terms of revascularization and graft retrieval, a shift in recent years has favored IM and SC sites for OTX studies. There is limited clinical translation of grafting to the OB, as this is a structure not present in humans; this outweighs the benefits of this choice and encourages further investigation of SC or IM sites instead, which more closely mimic the heterotopic implantation technique of placing cortical strips subcutaneously or the orthotopic technique of transplanting cortical strips to small pockets made in a women’s remaining ovary, both of which are commonly used in ovarian tissue transplantation (57). Additionally, the real-time data that may be obtained through noninvasive monitoring of follicle growth throughout a study via magnetic resonance imaging (MRI) or ultrasound justifies continuous search for minimally invasive grafting sites, such as the SC space. As the field advances and employs advanced tissue engineering techniques, it may become more challenging to discern the role a chosen grafting approach in terms of experimental outcomes. This relates to the great degree of variability originating from other factors, such as biomaterial support matrices, the exogenous cell populations to be added, and whether the technique involves implantation of isolated follicles or whole cortex fragments. Thus, prioritizing and selecting an implant location based on specific study goals and ensuring the use of proper experimental controls to account for the experimental variability is the key to obtaining optimal reproducible results in this complex landscape.

Ovariectomy

The potential effects of endogenous mouse hormones on survival and folliculogenesis of human follicles is an important consideration for OTX studies. Ovariectomy of the host’s ovaries prior to OTX of human ovarian tissues has not always been the standard in OTX models, but within the last 5 years has become increasingly common. Some data suggest that ovariectomized hosts lead to improved follicle survival and development. For instance, Maltaris et al reported higher follicle counts and formation of more antral follicles in ovariectomized mice stimulated with human menopausal gonadotropin (hMG) compared with intact mice undergoing the same stimulation regimen (58). Also of note, retrieval of MII oocytes was reported only from tissue grafted to ovariectomized mice. Arguments in favor of ovariectomy include the benefits of elevated gonadotropin levels during the neovascularization period in improving graft revascularization (58), minimizing follicle loss due to ischemia, and avoiding potential for competition and/or inhibition from the sex hormones and inhibitory factors secreted by the intact mouse ovary. An additional motivation for ovariectomizing host mice is an attempt to model the clinical scenario of primary ovarian insufficiency experienced by patients. In this case, ideally, the ovariectomies should be performed at least a few weeks before the OTX to allow endogenous hormones and other factors to wash out, which unfortunately, doubles the number of surgeries immunocompromised animals undergo, potentially posing an animal welfare concern.

Stimulation with exogenous hormones

Follicle development past the secondary stage depends on the presence of gonadotropins in both mice and humans (59). However, significant differences in the schedule and levels of circulating hormones between the 2 species (Fig. 1) have led many researchers to stimulate hosts with exogenous gonadotropins to promote the development of antral follicles in OTX models. Multiple groups have developed different schedules and dosages for administration of exogenous hormones. The most commonly used approaches are presented in Fig. 2 and Table 1. A major challenge in designing these stimulation schedules is an inability to directly compare circulating levels of follicle-stimulating hormone (FSH) between the 2 species. Values for human FSH are measured in mIU/mL, where mIU is a function of the biological activity of a given amount of FSH and is determined by the World Health Organization Expert committee on Biological Standardization (60). Conversely mouse FSH measurements are taken in ng/mL (61). This difference in units between species poses constraints in optimizing the pattern and concentrations of FSH and luteinizing hormone (LH) that the ovarian tissue will be exposed to after xenotransplantation to mimic the concentrations seen in women throughout their menstrual cycle. This restricts the ability to come to a universal conclusion on the necessity of stimulation and to assess to what degree endogenous mouse hormones affect human folliculogenesis in OTX models. Despite the lack of a universal stimulation protocol for human tissues implanted in mice, the majority of groups conducting long-term OTX studies have opted to stimulate host animals (20, 21, 36-38, 50, 58, 62-69).

Figure 1.

Significant differences exist in the schedule and levels of circulating follicle stimulating hormone (FSH) and estradiol between humans (top) and mice (bottom). These differences have led researchers to stimulate murine hosts with exogenous gonadotropins in OTX models to promote the growth of multiple follicles to antral stages. Of note, only the differences in cyclicity—not absolute concentration of circulating FSH—can be compared between the 2 species due to differences in the units used in the assays for each species. Human data from (70), mouse estradiol data from (71), mouse FSH data from (61).

Figure 2.

Common stimulation schedules used in ovarian tissue xenotransplantation models. A single bar with 2 colors represents the delivery of 2 drugs at a given timepoint. Bar frequency indicates dosage daily or every other day. Arrows indicate time of sacrifice. 0 is the time of ovarian tissue implantation. Letters correspond to outcomes listed in Table 1.

Table 1.

Outcomes Obtained From Stimulation Schedules Displayed in Fig. 2

	Outcomes of Note	Ref
a	Secondary follicles present	(54, 64)
b	Antral follicle present	(37)
c	Antral follicles present, MII oocyte retrieved	(70)
d	Antral follicles present, up to 6mm diameter	(59)
e	Antral follicles present, up to 18mm diameter	(38)
f	Early antral follicle present	(36)
g	Reduced antral follicle counts	(63)
h	Significant reduction in follicle numbers	(63)
I	Antral follicles present, MII oocyte retrieved	(70)
j	Secondary follicles present	(71)
k	Antral follicles present, up to 12 mm in diameter	(46)
l	Antral follicles present, up to 6mm in diameter, MII oocytes retrieved	(21, 61)
m	Antral follicles present, up to 5mm in diameter	(72)
n	Antral follicles present, up to 6mm in diameter, MII oocytes retrieved	(67)
o	Antral follicles present, up to 2 mm in diameter, formation of corpora lutea	(20)
p	Antral follicles present, up to 2 mm in diameter, formation of corpora lutea	(20)

One of the most commonly administered hormones for stimulating follicular development is hMG, which is a combination of equal concentrations of FSH and LH purified from the urine of postmenopausal women (73). The FSH component stimulates the development of multiple follicles, while the LH activity is thought to promote ovulation (37, 74). Administration of supraphysiologic levels of hMG promotes development and ovulation of multiple follicles. Although there is considerable variability in hMG stimulation protocols used by various investigators, successful recruitment and growth of functionally competent large antral follicles have been achieved in most cases with human grafts implanted in mice. For example, some administer hMG daily while others administer every other day, with doses ranging from 1 to 5 IU (36, 69, 75). There is also no standard established for how soon after grafting hMG administration should begin. Some studies begin stimulation immediately following ovarian tissue grafting while others begin stimulation as late as 15 weeks after grafting (37, 62). Arguments in favor of beginning stimulation immediately include: (i) the potential for elevated gonadotropin levels to enhance neovascularization; (ii) the possibility of follicles capable of responding to gonadotropins being present in implanted tissue; and (iii) secretion of growth factors and anti-Müllerian hormone (AMH) from implanted tissues which could further enhance vascularization and prevent the premature activation of primordial follicles respectively. However, most groups wait to begin stimulation until at least 14 days following implantation. Since the vast majority of follicles that survive cryopreservation are primordial, it is likely to take time for follicles to activate and reach the size of gonadotropin dependence. While the start time may vary, once hMG administration is initiated it is always continued for the remainder of the study. In most cases, hMG was effective at promoting follicle development to the antral stage. Of particular interest are the 2 studies that reported impressively large antral follicles of 10 and 18 mm in diameter (37, 38). Man et al achieved a 10 mm follicle from a strip of ovarian cortex co-encapsulated with exogenous endothelial cells in a plasma clot and stimulated with 2 IU hMG daily for 8 weeks beginning 14 weeks after grafting (37). Importantly, grafts with endothelial cells developed larger and more antral follicles than grafts without (37). Terada et al (76) achieved development of an 18 mm antral follicle from grafted fresh ovarian cortex following daily stimulation with 5 IU hMG beginning 10 weeks after grafting and continuing for 14 days. In this case, it is important to note that the ovarian biopsies were transported to the laboratory in a solution containing 5 IU/L of FSH. No studies have directly investigated the impact of transporting ovarian tissue in solutions containing FSH, but it is hypothesized that inclusion of FSH in transport solutions may reduce ischemia-reperfusion injury by stimulating mitosis and inhibiting apoptosis in granulosa cells (63). Furthermore, addition of 8 IU/L FSH to the holding medium during transportation followed by daily stimulation with 4 IU hMG for 14 days or 5 IU/L FSH during transportation and daily stimulation with 5 IU hMG for 14 days resulted in the formation of antral follicles reaching 6 mm and 18 mm, respectively (38, 63). The jury is still out regarding the benefits of adding FSH to holding/transportation medium and additional studies are required to increase the strength of evidence.

Stimulation with FSH alone is another approach used to stimulate the development of antral follicles. Two different forms of FSH are commonly used, recombinant follicle-stimulating hormone (rFSH) and human urine-derived follicle-stimulating hormone (hFSH), although, there is no consensus in the scientific community over which form of FSH is most advantageous. While hFSH, purified from human urine, has long been the standard of care for ovarian stimulation, the batch-to-batch variability in purity (due to residual LH and other urinary proteins) presents challenges, especially in an experimental and more sensitive mouse model (77). This uncertainty in the ability to obtain absolute purity caused many to adopt the use of rFSH, which is engineered using an ovarian cell line, resulting in high purity and little batch-to-batch variability. Furthermore, rFSH has a longer half-life in vivo due to its composition of more acidic FSH isoforms than hFSH (77), which might have contributed to its preferential use. A recent meta-analysis concluded there were no significant differences in the amount of rFSH versus hFSH needed to achieve pregnancy or live birth in women undergoing ovarian stimulation (77) and neither form of FSH appeared to consistently lead to the formation of significantly more or larger antral follicles in the OTX studies (21, 50, 67, 78, 79). FSH is most commonly administered every other day, at a concentration of 1IU beginning 1 week following implantation. With this regimen, large antral follicles up to 12 mm in diameter have been observed (50). Other studies have also administered FSH only in the last 14 days of a study. In these cases, 1.5 IU (36) or 7.5 IU (79-82) were administered every other day for the last 14 days until sacrifice. One study used a step-up protocol wherein 1 IU FSH was administered every other day until a follicle ≥6 mm was observed via MRI, at which point the dose of FSH was increased to 5 IU every other day until the end of the study (50). With this protocol and the use of MRI to monitor follicle size, 7 MI and 24 MII oocytes were retrieved (50). Reliable outcomes with stimulation protocols may require designer recombinant FSH preparations that would allow initial stimulation with long-acting acidic mix of FSH followed by short-acting less acidic mix of FSH, reflecting the type of FSH isoform mix that prevail during different phases of the menstrual cycle (83-85).

The benefits and detriments of delivering LH alongside FSH versus FSH alone remain highly debated. This is because much of the investigation of the effects of LH administration centered on rates of pregnancy and live birth, failing to provide much insight as to how LH may affect folliculogenesis or ovulation in OTX models. The effects of delivering consistent, low levels of LH (such as those delivered in hMG) on folliculogenesis and oocyte maturation remain widely unknown. Likely because of insufficient favorable data regarding co-administration of LH alongside FSH and the more predictable quality of rFSH, the majority of OTX studies conducted thus far have chosen to stimulate with FSH alone.

Pregnant mare’s serum gonadotropin (PMSG) is the third hormone used in OTX studies to promote follicle growth. PMSG is isolated from the endometrial cups of pregnant mares and was used in the early days of assisted reproductive technology as part of a “two-step protocol.” The two-step protocol consisted of initial stimulation with 1 or 4 IU PMSG every other day for the last 2 weeks before sacrifice to promote follicle growth, followed by a single dose of human chorionic gonadotropin (hCG) administered to promote ovulation. This regimen has given rise to large antral follicles (2 and 6 mm in diameter) as well as MII oocytes and corpora lutea (20, 68). However, the use of PMSG has been abandoned and has not been used in OTX studies since 2005 because it can induce an unfavorable immune response in humans (86). While PMSG is not likely to invoke an immune response in immunocompromised hosts, the limited clinical translation of PMSG has led to abandonment of its use in OTX studies.

Lastly, since hCG interacts with the LH/choriogonadotropic hormone receptor of the ovary and acts as a surrogate for LH to induce ovulation (87), although not playing a role in promoting follicle growth, it is most commonly administered alongside hMG, FSH, or PMSG. Three different dosing regimens have been reported for hCG: (i) 20 IU 30 to 36 hours before sacrifice; (ii) 10 IU 36 hours before sacrifice; or (iii) 10 IU 24 hours before sacrifice. All studies which administered hCG alongside FSH reported collection of MII oocytes (21, 50, 78). Stimulation with both PMSG and hCG has led to the retrieval of MII oocytes and the formation of a greater number of corpora lutea when compared with PMSG alone (20, 68). However, recent data suggest that stimulation with hCG may lead to unphysiological conditions (particularly a stronger push for progesterone synthesis), which may be suboptimal for supporting pregnancy (87). While a host’s ability to support a viable pregnancy is not an outcome that can be examined in OTX studies, protocols that administer hCG should consider how increased progesterone synthesis due to hCG administration may affect the development of implanted tissue and how the technology would translate to humans as assisted reproductive technology programs move away from administering hCG (87).

In addition to promoting follicle growth and ovulation, there is also great interest in preservation of the primordial follicle pool and preventing premature activation. Two endocrine-mediated strategies for preserving quiescent primordial follicles involves delivery of AMH or gonadotropin-releasing hormone (GnRH) agonists. Delivery of AMH via osmotic pump beginning 1 week before transplant and continuing for 1 week until the end of the study failed to improve preservation of quiescent primordial follicles (52). On the contrary, inclusion of cells engineered to secrete AMH in ovarian grafts increased the percentage of primordial follicles remaining in ovarian grafts after 2 weeks (37). While the inclusion of AMH-secreting endothelial cells doubled the percentage of primordial follicles remaining in grafts compared with grafts containing unaltered endothelial cells (37), the inclusion of AMH-secreting mesenchymal stem cells increased the percentage of primordial follicles remaining in grafts 10-fold over grafts containing unaltered mesenchymal stem cells (37).

Studies that investigated the effects of GnRH agonists to preserve the primordial pool wherein 8 mg of triptorelin was administered every 4 weeks beginning either 2 weeks before ovarian tissue transplantation or the day of transplant (62, 64) have not provided conclusive evidence in support of use of GnRH agonists. Maltaris et al delivered 8 mg triptorelin via microcapsules every 4 weeks beginning 2 weeks before implantation, with or without daily 1 IU hMG beginning 2 weeks after implantation (64). While administration of triptorelin in the absence of hMG resulted in a significant decrease in follicle numbers at all stages (64), triptorelin delivery in the presence of hMG had no effect on the number of follicles at earlier stages albeit they reported significantly decreased antral follicles (64). Likewise, no differences in the number of primordial follicles were seen when triptorelin was administered once every 4 weeks in conjunction with 2 IU of hMG every other day (64).

In summary, in the majority of cases when ovarian tissue is cryopreserved and transplanted, stimulation is needed to obtain large antral follicles or MII oocytes and host animals should undergo hormonal stimulation. However, other outcomes, including follicle activation or angiogenesis, can be effectively investigated without the need for stimulation.

Challenges Due to Variation in Human Tissue Samples

Despite consistency in strain selection, implant location, exogenous stimulation, and reliable host immunodeficiency, outcomes across different patients can be highly variable (20). Much of this variation can be attributed to differences in the number and quality of healthy follicles present in the human tissue samples at the time of grafting (21, 88).

Effects of tissue processing

When comparing outcomes (follicle size, follicle numbers, vascularization, steroidogenesis, etc.) across different OTX studies it is crucial also to consider differences in methods of tissue processing. Prior to implantation, ovarian tissue usually undergoes a great deal of laboratory processing, including transportation, tissue chopping, cryopreservation, thawing, and in some cases, tissue dissociation and/or follicle isolation (49, 89-92). Each of these steps carries a risk of follicle and/or stromal cell damage and differences in these processing steps have the potential to significantly affect study outcomes.

One of the most broadly investigated of the aforementioned processing steps is OTC. In most clinical cases ovarian tissue is cryopreserved for future use, when the patient is ready for ovarian function and fertility restoration. The current standard protocol for OTC is slow freezing (93), a process where ice crystals are introduced in a controlled manner in order to avoid tissue damage. Vitrification, a process where a sample is chilled into an amorphous glassy state, thereby avoiding the formation of damaging ice crystals altogether, is the standard process for egg and embryo freezing and is a current area of research for OTC. For more information on the state of cryopreservation technology for ovarian tissue the reader is directed to the published meta-analysis (91).

With the advent of bioengineering approaches, there has been an increase in OTX studies involving implantation of isolated follicles embedded in supportive matrices. The use of isolated follicles offers opportunities to reduce the risk of reintroducing cancer cells potentially harbored in the ovarian tissue removed prior to anticancer treatments, allows for precise calculation of follicle atresia/recovery rates, and opens the door for research into how intrafollicular communication and survival are impacted by follicle number or density. While data from studies implanting isolated follicles are limited in comparison to grafted tissue fragments, there is reason to hypothesize that isolation protocols increase rates of follicle activation (94). In one study that directly compared implanted tissue fragments with isolated preantral follicles embedded in plasma clots after 7 days of implantation, the follicle population recovered in plasma clots was composed of 13% primordial and 56% primary follicles while the follicle population in tissue fragments was 39% primordial and 24% primary (92). It is challenging to directly compare follicle recovery rates between tissue fragments and isolated follicles, because multiple factors affect follicle activation, survival, and growth. In the studies reviewed here, follicle recovery rate was below 25% after implanting isolated follicles. Some of the reasons for low follicle survival rate could be the lack of stromal cells co-implanted with the follicles, premature activation and/or damage to the follicles during isolation, and delayed vascularization.

Patient-to-patient variability in tissue quality

Perhaps of even greater importance is the impact of patient-to-patient variability in follicular reserve on graft function. Much of this variability can be attributed to patient age, but other factors such as use of oral contraceptives (95), underlying medical conditions (96), or previous gonadotoxic treatment may also play a role in determining how many healthy follicles are present in a patient’s ovarian tissue. The number of healthy follicles present at the time of transplantation strongly correlates with graft function and longevity (20, 21, 88). However, the field suffers from a lack of standardized approach as to how many follicles must be implanted to achieve a desired outcome. Without isolating individual follicles, it is challenging to nondestructively determine how many follicles reside in a piece of ovarian tissue. This, coupled with the paucity of ovarian tissue obtained from women of reproductive age, limits the ability to perform standardization studies. So, apart from a few studies that isolate and count follicles prior to implantation (49, 79, 92, 97), there are no conclusive data as to how the number of follicles transplanted affects graft outcomes.

Inferences can be drawn by examining the differential outcomes between tissue grafts from prepubertal and postpubertal individuals. When tissue from prepubertal girls was xenotransplanted, antral follicles formed in nearly 60% of implants after 21 weeks even without exogenous stimulation (98). In comparison, rates of antral formation varied from 0% to 37% following grafting of postpubertal tissue from patients ranging from 19 to 36 years of age in the absence of exogenous stimulation (20, 99). In contrast, stimulation with exogenous hormones significantly increased antral follicle formation in postpubertal tissue ranging between 12% up to 100% (20, 36, 79). Similar trends are seen when comparing the size of antral follicles formed from prepubertal versus postpubertal tissue. Using prepubertal tissue and no exogenous stimulation, 2 studies reported formation of antral follicles 7 mm in diameter and retrieval of 2 MII oocytes (88, 100). Compared with postpubertal tissue without stimulation, follicles of 7, 9, and 15 mm in diameter have been observed in transplanted ovarian tissue (99). Of note, male mice served as hosts for these studies. There have been no reports of large antral follicles developing from postpubertal tissue in female mice without stimulation. With stimulation of postpubertal tissue, several studies have reported formation of antral follicles greater than 5 mm with one study reporting an 18 mm antral follicle (38, 63, 65-68). Finally, corpora lutea and MI and MII oocytes have been observed in xenografted postpubertal tissue, but only when exogenous stimulation is used (20, 21, 66).

These results demonstrate that successful outcomes can be achieved with both prepubertal and postpubertal tissue, but these outcomes are less consistent and require greater intervention when postpubertal tissue is used. While many patients seeking OTC and transplantation are prepubertal it is important to develop technologies suitable for all women interested in fertility preservation. Furthermore, even within a single patient, the follicle density in individual tissue pieces can be highly variable (101). A meaningful step toward understanding the correlation between variation in patient tissue and the outcomes of ovarian transplantation will be development of a noninvasive, nondestructive method to characterize the unique follicle population present in each tissue fragment prior to grafting.

Tissue Engineering Approaches

Cryopreservation of ovarian tissue before exposure to toxic treatments, and subsequent auto-implantation after remission is the only fertility preservation option available to pediatric patients and women with hormone-responsive cancers. However, this technique presents a significant risk of reintroducing malignant cells harbored in the autologous transplant, particularly in the case of hematologic malignancies, which are common in children (102).To minimize the risk of reintroduction of cancer cells, primordial follicles could be isolated and purified from all the other somatic cells present in the tissue and transplanted back to the patient. As demonstrated through other tissue engineering approaches, transplanted cells survive poorly if implanted without a supportive scaffold (103). Thus, in recent years, there has been increased interest in engineering an optimal biomimetic scaffold, or an “artificial ovary,” which can reproducibly support the maturation of isolated primordial follicles and give rise to mature oocytes. Artificial ovarian tissue research focuses on developing and optimizing biomaterial-based scaffolds, which in combination with a tailored cell population and/or growth factor cocktail can reproducibly lead to live births and promote graft longevity (49, 51, 97).

Biomaterial selection and design

The selected biomaterial is the backbone of any artificial ovarian tissue technology. Isolated follicles require mechanical support to maintain their 3-dimensional structure (104) and properly develop. Hydrogels in particular are commonly used scaffolds for this purpose as they have the ability to retain secreted factors at efficient concentrations, can support growth and function of additional cell types such as endothelial or stromal cells, and in many cases can be tuned to optimize stiffness, diffusion, volume, or other properties to maximize follicle survival and meet clinical requirements (105). A range of biomaterial scaffolds have led to live births in mouse autograft models (106-109) and some of these materials are just beginning to be investigated as support matrices for human follicles and ovarian tissue (37, 46, 97, 110). To date, human tissue has been grafted to mice embedded in decellularized ovarian tissue (49), plasma clots derived from patients’ blood (37, 46, 79, 92), fibrin hydrogels derived from commercially available reagents (97), and collagen gels (111). Pors et al (49) encapsulated isolated human preantral follicles in decellularized human ovarian tissue bound together with Matrigel. Under these conditions, follicles developed to the secondary stage and the follicle recovery rate was approximately 25% (49). In comparison, isolated follicles embedded in plasma clots developed to antral stages (79). Finally, a recovery rate of 23% was reported for isolated follicles embedded in fibrin hydrogels, along with 50 000 isolated stromal cells (97). When hyaluronic acid was added to these gels, the follicle recovery rate lowered slightly to 20%, but a greater proportion of the primordial pool was preserved than in fibrin gels alone (97).

Tailoring stromal cell population

While transplantation of isolated ovarian follicles minimizes the risk of cancer cell transmission, the loss of supporting stromal cells negatively affects follicle survival, growth, and maturation. Ovarian stroma harbors a diverse cell population, including precursors for theca cells important for steroid production and folliculogenesis and endothelial cells crucial for revascularization (112). To investigate the impact of stromal cells on survival and development of isolated follicles, Paulini et al (97) encapsulated 30 to 50 isolated preantral follicles and 50 000 stromal cells isolated from ovarian biopsies in fibrin gels with or without hyaluronic acid. These gels were then implanted into a peritoneal pocket in nude mice for 7 days. Under these conditions, 20% to 23% of implanted follicles survived and the number of secondary follicles present in the graft increased, although this increase was not statistically significant. Dath et al specifically investigated the effects of the CD-34 positive endothelial cell population within the ovarian stroma and whether presence of these cells would improve revascularization of the grafts. In these studies, isolated follicles and stromal cells were encapsulated in plasma clots and implanted in mice (46). One group contained a normal population of stromal cells and the other group had all CD-34 positive cells removed (113). Dath et al also found that grafts containing CD-34 positive cells were larger and had significantly higher blood vessel density than grafts without CD-34 positive cells (46).

Use of pharmacological agents

An additional tissue engineering strategy employed by researchers is the delivery of pharmacological agents to transplanted cells or tissue. Some of the therapies investigated in OTX models include vascular endothelial growth factor-α (VEGF-A), basic fibroblast growth factor (bFGF), vitamin E, melatonin, and hyaluronic acid.

Abir et al investigated the effects of adding vitamin E to drinking water, incubating the graft in VEGF-A prior to implantation or doing both (114). Friedman et al (115) investigated the effects of adding melatonin to the drinking water, adding vitamin E to drinking water, incubation in a hyaluronic acid biological glue and incubation in VEGF-A in various combinations. In both studies, treatment of any kind caused a significant reduction in apoptosis when compared with no treatment at all. However, there were no significant differences in apoptosis, follicle survival, or follicle development between the different experimental groups. Tanaka et al (110) delivered bFGF via a gelatin hydrogel sheet and found that bFGF significantly increased the density of primordial and primary follicles preserved following cryopreservation.

Future Directions

The future of technologies to protect and restore ovarian endocrine function and fertility is promising. Numerous groups are investigating a myriad of approaches to overcome the deleterious effects of gonadotoxic anticancer therapy, including novel methods of cryopreservation, whole ovary cryopreservation (116), in vitro follicle maturation (117), ovarian suppression with GnRH agonists to protect the ovaries from gonadotoxic treatments (118), ovarian transposition (119), and even addressing the potential existence of oogonia stem cells (120). In the future, these approaches may provide additional options for women seeking fertility and/or endocrine preservation.

Techniques for OTC and transplantation have been greatly advanced by findings made possible by the OTX models. OTX using human ovarian tissue allows for investigation of clinically relevant samples, in a cost effective and ethical animal model. However, there are still numerous questions that cannot be answered using the currently available technology. The data that can be collected from OTX models is limited due to an inability to noninvasively monitor follicle growth and assess function. Improved techniques for assessing graft function through noninvasive imaging or hormone analysis would allow for longer term monitoring of grafts over multiple follicular and luteal cycles and help answer questions concerning long-term graft function and longevity. Similarly, the lack of nondestructive methods to characterize human tissue prior to implantation makes quantifying parameters such as follicle loss or activation difficult. There are several questions still to be answered regarding how follicle growth and function differs when implanted in an environment governed by the mouse endocrine system. Studies where exogenous hormones are administered in a manner to mimic the human menstrual cycle may help to elucidate some of these differences. Additionally, the majority of outcomes measured in OTX models includes follicle development and oocyte maturation, but oocyte quality is largely overlooked. Determining oocyte quality, and hence its potential for successful fertilization and embryonic development, is especially difficult when using human tissue because human oocytes cannot be fertilized for research purposes. However, outside of fertilization, the field has no conclusive markers to determine if any given egg has the potential to develop into a viable embryo following fertilization. This is an area in which information derived from xenotransplantation of tissue from other large mammals can be particularly useful. Mature oocytes collected from ovine, bovine, or even nonhuman primate tissue developed in immunocompromised mice can be fertilized in a laboratory setting to investigate the reproductive competence of these oocytes. Information derived from these studies could then be used to develop markers for oocyte competence that could be applied to human tissue. In parallel, efforts should be targeted toward developing recombinant FSH preparations that more closely mimic what exists in humans, with a more acidic mix for support of early follicular growth and less acidic mix for enhancing late follicular growth. Since the ultimate goal of artificial ovarian tissue is to restore fertility, a method to determine whether mature oocytes can in fact lead to a viable pregnancy would be a major breakthrough. Finally, there is great interest in understanding how to control follicle activation in artificial ovarian tissue. In a healthy human ovary, a small cohort of follicles activate each cycle and begin to grow (70). Furthermore, it is thought that activation is a significant source of follicle loss following transplantation (105, 121). This poses a challenge for artificial ovarian tissue design because if too many follicles become activated at once, or too quickly following grafting, the longevity of the graft can be greatly shortened. Current research in this area is investigating how both soluble factors and mechanical signaling affect follicle activation. Advances in understanding the exact mechanisms of follicle activation or how artificial tissue can be engineered to inhibit initial burst activation following implantation would make artificial ovarian tissue a more well-characterized and reliable technology for women seeking OTC and transplantation for fertility preservation.

As the field moves toward development of engineering artificial ovarian tissue, development of biomaterial scaffolds to support rapid neovascularization and human folliculogenesis becomes an increasingly important area of research. An ideal hydrogel should be robust enough to provide mechanical support to follicles to prevent follicle spreading and endure surgical implantation but degradable, to allow for remodeling as follicles grow and eventually become corpora lutea. Furthermore, these gels should retain and or sequester factors secreted by follicles in order to provide an ideal microenvironment. To date, very few biomaterials have been tested with human follicles.

Conclusion

Ovarian tissue cryopreservation and transplantation is a promising approach to fertility preservation for many women and girls who cannot undergo oocyte or embryo freezing. As this technology becomes more widely implemented it is more important than ever to make strides in engineering artificial ovarian tissue from cryopreserved banked ovarian tissue. The results from ovarian tissue xenotransplantation studies have demonstrated the feasibility of developing antral follicles and obtaining mature oocytes from cryopreserved ovarian tissues. Moving forward OTX models will continue to be a crucial research tool for developing artificial ovarian tissue and advancing the field of oncofertility.

Glossary

Abbreviations

AMH

anti-Müllerian hormone

bFGF

basic fibroblast growth factor

FSH

follicle-stimulating hormone

GnRH

gonadotropin-releasing hormone

hCG

human chorionic gonadotropin

hFSH

human urine-derived follicle-stimulating hormone

hMG

human menopausal gonadotropin

IM

intramuscular

IP

intraperitoneal

KC

kidney capsule

LH

luteinizing hormone

MRI

magnetic resonance imaging

NOD

nonobese diabetic

OB

ovarian bursa

OTC

ovarian tissue cryopreservation

OTX

human ovarian tissue xenotransplantation

PMSG

pregnant mare’s serum gonadotropin

rFSH

recombinant follicle-stimulating hormone

SC

subcutaneous

SCID

severe combined immunodeficiency

VEGF-A

vascular endothelial growth factor-α

Additional Information

Disclosure Summary: The authors state they have nothing to disclose.

Data Availability

Data sharing is not applicable to this article as no datasets were generated or analyzed during the current study.