As before noted, our center serves overall a very adversely selected patient population, mostly patients who have failed multiple IVF cycles, often at multiple IVF clinics in the U.S. and overseas, before presenting to The CHR. Such patients require very different treatment approaches because their responses to standard ovarian stimulation protocols often greatly differs from better-prognosis patients. Our center, therefore, on repeated occasions discovered that treatments proposed based on investigations in better-prognosis patients were not applicable to our center’s patient population. Good examples were closed incubation and imaging systems which we found in our patients to produce similar outcomes to standard embryology in third-party egg donor cycles but to adversely affect IVF outcomes in poor-prognosis patients, still pursuing treatments with autologous oocytes [20].
The rapidly gaining popularity of embryo banking has attracted our skepticism because it was primarily built on the hypothesis that ovarian hyperstimulation creates an unfavorable environment for embryo implantation that can be overcome by delaying the embryo transfer from the stimulation cycle into a future implantation cycle. Though skeptical of the hypothesis based on in-house data [6, 21], we have been especially concerned about adverse effects cryopreservation may have on the cumulative pregnancy chances of a cycle’s embryo cohort in poorer prognosis patients and about additional costs created by adding an additional thaw cycle. Those concern grew after analyzing the published literature and as noted in the introduction section, recognizing that practically all reports in the literature that claimed improved pregnancy and live birth rates after embryo banking, had been conducted in only good prognosis patients [1–3, 10, 12]. Where no such patient selection took place, delayed frozen-thawed transfers did not result in cycle improvements.
We then from these observations further concluded that in presence of good prognosis patients who, in a general infertile population, seemingly benefit from all-freeze cycles, such an unselected patient population must also contain a counterbalancing group of patients who experience detrimental effects from such a clinical approach. That such a counterbalancing patient population, likely, had to be made up of patients on the opposing prognostic extreme to good-prognosis patients, seemed obvious and that meant that this group must mostly represent older women and/or younger women with prematurely aging ovaries, the two patient groups that make up over 90% of our center’s patient population.
Though, because of potential treatment delays due to randomization to placebo, for ethical reasons our center’s patient population does not support many randomized placebo-controlled studies, the unique quality of our center’s patients allows for the retrospective analysis of highly homogenous patient groups in extremis, where pregnancy chances are very low and, therefore, smaller study sizes allow for statistically valid observations.
Such a study is presented here and confirms our current understanding of the published literature that has led us to the following conclusions regarding all-freeze IVF cycles with embryo banking and deferred embryo transfers: (i) Embryo banking, overall, does not improve IVF outcomes in unselected patient populations. (ii) In favorably selected patients, embryo banking may, with reference point embryo transfer, lead to improved pregnancy and live birth chances. (iii) While patient selection processes may benefit good-prognosis patients, they usually have compensatory detrimental effects on poorer prognosis patients. (iv) As this study demonstrates, this is also the case in utilizing embryo banking in poorer prognosis patients. We here demonstrated the latter point in a four-step study.
Third-party egg donation cycles: Fresh vs. frozen
We in a carefully case-controlled study of practically identical patients in fresh and frozen third-party donor egg cycles (Table 1), demonstrate no difference in clinical pregnancy rates (P = 0.1760), with adjustments for age and AMH not at all affecting this conclusion (P = 0.2487). Since oocyte donors are highly selected and since recipients in this study were practically identical in basic characteristics, this study can be viewed as a baseline control study, demonstrating no visible effect of embryo freezing on IVF outcomes in unselected patients. This study, thus, reaffirms prior prospectively randomized studies of unselected patient populations [11], and thereby, also reaffirms the format of our retroactive third-party donor analysis.
In Table 1, reported basically identical miscarriage rates in donor recipient cycles, whether fresh (26.1%) or frozen-thawed (25.0%), are, however, deserving of further commentary: Both rates are unexpectedly high, considering that oocytes in these cycles came from young third-party egg donors in their 20s. The 2016 CDC National ART Summary Report suggests in third-party egg donation cycles only an approximately 10.4% miscarriage rate (https://www.cdc.gov/art/pdf/2016-report/ART-2016-National-Summary-Report-pdf). Here observed more than double as high miscarriage rate in both study groups, therefore, must reflect the very advanced (also practically identical) ages of both recipient patient groups (45.6 ± 5.1 and 45.7 ± 5.9 years, respectively). While the literature extensively comments on implantation, pregnancy, and live birth rates in donor egg recipient cycles at different ages, surprisingly, we were unable to locate even a single study centered on miscarriage rates depending on recipient ages. Two publications commented peripherally on the subject, with one noting no differences [22] and the other noting small increases in pregnancy losses [23]. Here reported outcome data, therefore, for the first time offer evidence that recipient age does matter when it comes to miscarriages. The likely reasons are accumulating medical problems, unrelated to age of oocytes and, therefore, unrelated to chromosomal abnormalities.
Autologous non-donor cycles: Fresh vs. frozen
In part two of here presented study, we switched to the investigation of the use of autologous oocytes in an overall highly unfavorable patient population and, as a first step, simply compared all fresh and all frozen-thawed cycles performed at the center during the study period (Table 2). Unsurprisingly, this comparison, in contrast to previously described third-party donor cycles, demonstrated two highly divergent patient populations; fresh cycles, not only were three -times as common but also represented significantly older women (P < 0001), with much lower AMH (P < 0.0001) higher FSH (P = 0.0003), though very similar number of transferred embryos. Describing both study groups in summary, one, therefore, can clearly state that women who underwent frozen-thawed cycles were not only significantly younger but also had much better functional ovarian reserve. That they achieved significantly better clinical pregnancy rates (P = 0.0014) with transfer of identical embryo numbers (P = 0.2536), therefore, cannot surprise and does not suggest that this improved outcome is consequential to delayed frozen-thawed embryo transfers.
This conclusion is confirmed by the observation that, once pregnancy outcomes were adjusted for age and AMH (as a representative of functional ovarian reserve), the significant difference in pregnancy rate disappeared (P = 0.2991). Adjusting for FSH instead of AMH made no differences in significance (P = 0.1564) and, since both FSH and AMH reflect FOR, we formally adjusted only for one (AMH). This observation, therefore, offers further evidence that seeming improvements in IVF cycle outcomes in frozen thawed over fresh cycles with use of autologous eggs are mostly due to underlying patient characteristics and not caused by embryo cryopreservation in place of fresh transfers.
Here, too, a comment on miscarriages is necessary: Though both patient groups in this study section are significantly younger than in above presented third party-donor section, fresh and frozen cycles, still, involved older women (41.2 ± 4.8 vs. 39.5 ± 5.9 years, respectively). Frozen cycles were, however, overall performed in younger women than fresh cycles (P < 0.0001). The fact that miscarriages were nominally lower in frozen cycles (28.1%) than fresh cycles (35.4%), therefore, has no practical meaning. What, however, is of interest, is the observation that miscarriages were uniformly higher in autologous than donor-recipient cycles, whether fresh or frozen, even though donor egg recipients were significantly older. Third-party egg donation in older women, therefore, clearly does appear to reduce miscarriage risk in comparison to autologous oocyte cycles; this advantage, however, shrinks with advancing recipient age, likely due to non-chromosomal maternal causes.
Autologous non-donor cycles: Fresh vs. frozen cycles in good-, intermediate- and poor-prognosis patients
Addressing the third and fourth steps of this study (Tables 3 and 4), we argued that every study population can be divided into better-, average-, and poorer-prognosis patients [13]. This must also have been the case in our 741 fresh cycles. Trying to identify within that total group of patients a best-prognosis sub-group must be possible by selecting patients based on number of embryos they produced in their IVF cycles since transferrable embryo numbers, after female age, are the second-most important predictor of pregnancy chances in IVF [13]. Moreover, since this study investigated an older patient population in which current IVF practice allows for transfer of multiple embryos, further patients selection should be achievable by selecting-out patients who produced more embryos than were immediately transferrable. In other words, among those 741 women, the 143 (19%) who ended up with cryopreserved embryos after undergoing a fresh transfer, must have been among best prognosis patients. We now compared these best-prognosis patients in fresh transfers to the same 217 frozen embryo transfer cycles from the previous analysis.
This change in patient selection resulted in highly significant outcome changes: Women undergoing fresh cycles now were suddenly significantly younger (P = 0.0001) from previously being significantly older in the complete autologous group (P < 0.0001), and significant differences in AMH and FSH to the benefit of frozen cycles completely disappeared, together with all prior outcome advantages in pregnancy rates for frozen cycles (P = 0.0086 before and P = 0.0451 after adjustment for age and AMH).
Moreover, once we compared the remaining 598 fresh cycles from women with moderate and poor prognosis with the 217 frozen cycles, patient demographics again reverted into similar ranges as had been seen previously in Table 2 for the complete autologous patient populations, with frozen transfer cycles seemingly outperforming pregnancy rates in fresh cycles. (P < 0.0001, P = 0.0028, Table 4).