Modeling delay of age at natural menopause with planned tissue cryopreservation and autologous transplantation.

Our analysis predicts that harvesting tissue earlier in life (e.g., by age 30) may allow women to delay menopause significantly, whereas the delay is reduced if harvest occurs near or after age 40, and particularly so for women with low ovarian reserve at birth. The PF survival rate post-cryopreservation and -transplant ischemia is a key variable, and our analysis predicts that survival may need to be near 100% for harvest after age 40 to be worthwhile for most women. Our analysis also predicts that fractionation of cryopreserved tissues into multiple transplants maximizes the delay in ANM. For tissue harvest at age 30, returning one third of the removed cortex in three separate surgeries delays menopause by at least 12 years for most women, even when only s = 40 % of follicles survive after transplantation. The biological rationale for this finding is that when all follicles are re-implanted at once, their loss due to growth activation and oocyte aging occur concurrently. However, when fractionated, ovarian aging within still-frozen tissue remains suspended until transplantation. Finally, our analysis predicts how effects of planned ovarian tissue harvesting and transplantation vary across a population of women. Such variability is to be expected, considering the variability in PF endowment at birth 3 that corresponds to ANM variability 26 . Indeed, women with large PF reserves may benefit greatly, but planned ovarian tissue harvesting and transplantation may not be advisable for women with small PF reserves. Though seemingly paradoxical, our model predicts that women with lower ovarian reserve should preserve smaller amounts of ovarian tissue. This is because removal of a large portion of cortex may immediately exhaust their ovarian reserve. Planning should be informed by the model provided here and also ovarian reserve assessment such as by serum measurements of AMH. Reserve assessment can place candidates into three categories of expected, below-expected and above-expected, allowing outcomes can be modeled. This may lead to the decision to forgo the treatment in some cases, while in other cases the procedure may be justified even in more advanced ages. No treatments are available to delay menopause and extend the natural fertility period in women, and ovarian tissue freezing may be the first successful approach. Our modeling of biological PF behavior provides a starting point for testing ovarian tissue cryopreservation and transplantation to delay menopause in healthy women. Combined with recent success of ovarian tissue cryopreservation in patients who preserved their tissues prior to chemotherapy, our work suggests that performing the procedure in healthy women would likely extend ovarian function and, potentially, the fertile lifespan. The provided model allows optimization of the amount and the timing of tissue harvest for cryopreservation as well as the transplantation strategy. Some health concerns have been cited with extended natural estrogen production. One of the commonly cited risks is breast cancer as the incidence is generally higher in women with late menopause 27 , 28 . However, not only is this risk small, but based on reanalysis of Women’s Health Initiative (WHI) data, the benefits of continued estrogenization (here, synthetic hormones) outweigh the clinical risks 29 . Natural menopause has been recorded as late as at age 62 30 . One can infer that delaying menopause to around the age of 60 should not be considered “unnatural”, and potentially, significant benefits including improved quality of life can result within this time frame. These can include reduced risk of cardiovascular disease, stroke, atherosclerosis, and osteoporosis 31 . Presented with an accurate risk assessment that includes family history and individual genetic information, women can consider the age that ovarian tissue transplantation is appropriate for them. For most women, menopause will occur > 10–15 years after cryopreservation. During this time, new information is likely to emerge that can further optimize transplantation timing. The amount of ovarian tissue transplanted can be based on patient risk perception and available risk/benefit information, to achieve an appropriate and safe delay in menopausal age. If extending menopause beyond 60 years of age is supported as safe, patients can return for repeated transplantation(s). Additional surgical procedures can be avoided by harvesting ovarian tissue during medically-indicated procedures such as cesarean sections, tubal ligations, endometriosis surgeries and others 14 , 17 . Since simple techniques of subcutaneous ovarian tissue transplantation have been developed by the senior author of this report 32 , 33 , transplants can be performed under local anesthesia in the office setting, with minimal discomfort and cost. While we can derive confidence intervals for the duration of ovarian aging so that “best” and “worst-case” scenarios can be considered (see also below, Strengths ), it is critical that we compare patient outcomes given varied timing, amount of tissue removed, etc. with model output. For now, we can provide first-of-their-kind expected outcomes, and the model can only be entirely validated moving forward. We anticipate that, given the relatively consistent performance of ovarian tissue transplants in cancer patients, the procedure should yield similar or better results, particularly because this healthy population is not confounded by malignancies or their treatments. A critical near-term research objective is ensuring that the greatest number of PFs survive post-transplantation, in order to maximize expected ANM delay and health benefits. Progress has been made in that area with approaches that greatly enhance post-transplantation follicle survival. These include neovascularizing agents 14 such as sphingosine-1-phosphate 34 , peri-operative pharmacological treatments, and robotic surgery 15 , 18 . The main strength of our approach is that we can consider the distribution of likely trajectories for ovarian aging after tissue removal and transplantation, and estimate patient variability. Because we account for acute PF loss following transplantation and simulate ovarian aging outcomes given differential acute loss, our model provides realistic predictions of “best” and “worst-case” scenarios. Because of variables that, for now, cannot be accounted for (Limitations, below), determining ranges of possible outcomes is critical. A limitation of this approach is its dependence upon a somewhat uniform spatial distribution of PFs within harvested tissue. Our focus on relatively young ages for the removal of tissue means that we expect a high density of PFs per unit volume of tissue 35 , 36 . Follicle distribution may be more variable in older patient specimens, and this could lead to variable performance post-transplantation. This will be difficult to address until rapid, indirect estimation of viable follicle numbers within tissue can be achieved. Given removal of at least 25% of the reserve is usually indicated for efficacious menopause delay, differences in follicle density within different volumes of ovarian cortex should balance out.

Study parameters were established as follows. Menopause delay (denoted D ) of planned ovarian tissue harvesting and transplantation is the increase in menopausal age that results from the intervention . Mathematically, D can be expressed as: D = menopause age with intervention − menopause age without intervention The concept is shown in Figure 1 . The gray curve depicts the declining PF reserve (total number of PFs across both ovaries) in a woman where no intervention takes place. In this example, menopause is reached at age 51, when the reserve depletes to 2 × 10 3 remaining PFs (gray curve crosses horizontal dashed line). A correspondence between 10 3 remaining PFs per ovary with the timing of the ANM onset was identified in a previous study 3 , and we used this menopausal threshold in our original model 19 . The blue curve depicts the same woman if p = 25 % of her ovarian cortex (50% of the cortex of one ovary) is removed at age 30, cryopreserved, and returned at age 49. Ovarian aging was modeled to continue normally from this point, and menopause is shown to be delayed until age 62 (blue curve crosses horizontal dashed line). Menopause delay D here is 62 − 51 = 11 years. We next probe how patient and technique variability impacts D and assess several key variables: patient age at tissue removal, fraction of total ovarian reserve removed (denoted p ), fraction of PFs that survive the procedure ( s ), and compare the performance of transplantation aof cryopreserved cortex at once vs. in three consecutive fractions. As an example in Figure 1 , we simulated interventions using “impending menopause (when the simulated ovarian reserve reached 2 × 10 3 remaining PFs)” as the time of tissue return to patients to delay menopause. While we cannot directly measure PF reserve in practice, the intervention can be timed with the change in serum ovarian reserve markers such as the anti-Müllerian hormone (AMH) and menstrual changes 25 . Figure 2 summarizes the dependence of menopause delay upon age at tissue removal. First (2A) conditions were established where p = 25% of the entire ovarian reserve was removed between ages 21 and 40, and s = 100 % of PFs survived . Expectedly, earlier tissue removal resulted in greater menopause delay. The dark blue curve shows delay for a woman born with the population median of 6.4 × 10 5 PFs 19 . The light blue zone surrounding the dark blue curve shows how delay varies between women according to the distribution of the number of PFs present at birth 3 , from women born with the top 10% of PFs (top of blue zone) down to women born with the bottom 10% (bottom of blue zone). Because it is unlikely that s = 100 % of PFs will survive after cryopreservation and transplantation, we next investigate how reducing PF survival (fraction s ) influences menopause delay. In ovarian xenografting studies about 40% of PFs survive after thawing and transplantation of human tissue 22 , 23 . While the survival rate in patients is not known, ovarian transplant longevity is presumed to be reduced compared to a non-transplanted ovary 18 . However, technological advances including revascularizing pharmacological approaches, and robotic surgery may improve follicle survival 15 , 18 . We therefore include conservative 40% survival and improved 80% survival in our models. Supplemental Figure S1 provides probability density curves for 80% survival and tissue removal at ages 25, 30, 35, or 40 years. Panel 2B shows the optimal tissue removal percentage assuming PF survival of s = 40 % (red) or s = 80 % (blue). Here, the optimal removal percentage is defined as the one that maximizes menopause delay D . As patient age at removal increases, the optimal amount of ovarian cortex to remove declines, and this again is greatly impacted by ovarian reserve size (blue and red shaded areas: middle 80% of ovarian reserve, dark curves: median ovarian reserve). While removing approximately 50% of all ovarian cortex maximizes D for women early in the age range, removing less tissue is preferable in later ages. Ovarian reserve size again influences the amount of tissue to remove, where women with larger numbers of PFs experience greater menopause delay when a larger percentage of cortex is removed. Conversely, women with lower numbers of PFs experience greater D when a smaller amount of cortex is removed. Panel 2C shows that when s = 40 % of PFs survive (red zone), removal and return of p = 25 % of tissue yields a D of approximately 12 years for a woman with median ovarian reserve (red line) when tissue is collected at age 25. D declines through collection age 40 where ANM is postponed by approximately 1.5 years. Analogous to Figure 2 , the “height” of the red zone represents the middle 80% of ovarian reserve and shows that the greater their reserve, the greater their menopause delay. If tissue is removed at age 40 and returned near menopause, a woman at the lowest end of ovarian reserve would experience no delay in ANM. Naturally, doubling PF survival to s = 80 % (blue zone, dark blue line is median ovarian reserve) increases menopause delay across the timespan. Predicted D for a woman born with median PF number when 25% of her ovarian reserve is removed at ages 25, 30, 35 and 40, given differential PF survival s is provided in the Table . Finally, we consider how D varies between the return of all cryopreserved cortex at once versus in three consecutive transplants ( R ). Figure 3 (see also Supplemental Figure S2 ) compares PF decay between R = 1 (green zone) and R = 3 (purple zone). Age range at removal was 21–40, the tissue was returned near menopause with PF survival s = 40 % (3A) or 80% (3B). Three surgical returns of one-third of removed cortex resulted in a greater menopause delay (more returns always increase the delay, though the marginal increase in delay decreases as the number of returns grows). Women at age 30 with three tissue returns are predicted to experience delayed ANM with a lower limit near 8 years, but an upper limit near 20 years. The interactive tool was used to interrogate six (and even more) tissue returns and given 80% PF survival, the upper limit of ANM delay can exceed four decades. Our analysis also predicts that returning an equal amount of cortex at each transplant results in larger menopause delay compared to returning unequal amounts.

Efficacy in this study is the induced menopausal delay (denoted D ), defined as the difference between the age of menopause with and without tissue harvesting and transplantation. The main mathematical result which we use to study this process is the following formula for menopause delay, (1) D = 1 λ ln 1 − p + R λ ln 1 + p s R e λ T 0 − t 0 . In formula (1), p is the percentage of cortex removed, s is the fraction of removed cortex which survives removal, cryopreservation, and transplantation, t 0 is the age at tissue removal, T 0 denotes ANM in the absence of planned ovarian tissue harvesting and transplantation, R is the number of surgical returns, and λ is the decay rate of PFs within the ovaries. Supplementary Information includes a detailed mathematical derivation of formula (1) and comparisons between this paper’s and historical models are provided in Supplementary Figure S3 . An interactive tool is available for public use at ( https://www.fertilitypreservation.org/contents/probability-calculator/nopauze-calculator ).

Our mathematical model, derived from biological data of human PF decay over time, indicates that ovarian tissue cryopreservation and transplantation can significantly delay the ANM in women aged < 40 years. Output suggests that even under circumstances where post-transplant follicle loss is at its worst, menopause could be delayed by many years given “early” tissue removal and perimenopausal replacement. Delay can be further extended by a fractionated return. As transplantation and revascularization enhancement methods continue to improve post-transplantation follicle survival, this strategy may become more feasible for older women, with less tissue needed for younger individuals 14 . While model validation will require a series of cases to be monitored over lengthy time periods, intervention efficacy post-transplant will be detectable earlier than the age(s) of expected menopause. For example, ongoing menstrual cyclicity and premenopausal AMH levels will indicate that ovarian function has been extended 37 . The tool provided here will allow clinicians, translational scientists, and patients to gauge the feasibility of planned ovarian tissue freezing to delay menopause.

Human ovarian cortex contains dormant primordial follicles (PFs) that are the reserve from which a limited number of ovulatory follicles bestow the individual fertility potential post-puberty. Because developing follicles engage in cyclical hormone production in women, PF exhaustion timing determines the age at natural menopause (ANM) 1 – 6 . Previous studies of human ovarian tissue provided quantitative information about how PF numbers decline over time 3 , 7 , 8 . While there can be more than 1,000,000 PFs at birth, only about 500 ovulate during the reproductive years. 99.9% of the ovarian reserve is lost to follicle atresia, and physiological reasons for this apparent “oversupply” have been proposed 9 . An initial characterization of the spatial distribution of follicles in the human ovary, including resting PFs, has been published recently 4 ; this provides a framework for our understanding of spatiotemporal follicle loss over time. Despite significant extension of human life in the last 50 years 10 , no proven interventions that delay ANM are available. Women still experience menopause at a mean age of 51.4 in North America 11 , 12 . As reported in the first successful case of ovarian cortex transplantation for a medical indication, the procedure can restore ovarian endocrine function, and the menopausal state can be reversed 13 , 14 . The procedure has evolved in the last two decades with increasing success in cancer patients, resulting in hundreds of livebirths worldwide 14 . As a result, in 2019, ovarian cortex cryopreservation and transplantation was removed from the experimental category for medical indications by the American Society of Reproductive Medicine, followed by other professional organizations globally 14 – 18 . However, whether planned ovarian cryopreservation can also be used to extend the duration of reproductive function and delay menopause is in question. Our prior work allows us to address issues surrounding resection, cryopreservation and return of ovarian cortex; we have established a mathematical model of PF behavior 19 , 20 that recapitulates patterns of ovarian follicle loss in individual women, and also produces the ANM population distribution when simulations are compiled. The model was based on our identification of fluctuating, stochastic signals that occur in PFs that impact whether they stay dormant or begin to grow 21 . After establishing patient-specific boundaries, including the known distribution of PF numbers during postnatal life, the mathematical model was applied to questions surrounding planned ovarian tissue harvest and transplantation. First, is there an optimal chronological age when ovarian cortex should be removed? Next, how much cortex should be removed from each patient (as a percentage of total ovarian cortex)? Post-transplant ischemia is associated with up to 60% of all follicle loss in cortex tissue 22 , 23 . With this in mind, how does follicle loss post-transplant impact ovarian function after tissue return? Last, would outcomes differ if all cryopreserved tissue was returned in single versus multiple procedures? We developed an interactive tool to investigate how these factors would influence ovarian aging post-planned ovarian tissue harvesting and transplantation, and address how outcomes could be optimized for (simulated) women. Feasibility and implications of the approach are considered.