Cryopreserved oocyte versus fresh oocyte assisted reproductive technology cycles, United States, 2013.

Among patients using their own oocytes, those using cryopreserved oocytes were more likely to be ≥40 years old, have no prior births, and have one or more prior ART cycle(s) as compared with those using fresh oocytes ( Table 1 ). They also were less likely to have tubal factor and unexplained infertility but more likely to have ART treatment due to some other reason. They were more likely to attend a larger clinic (≥500 cycles), have ≥11 oocytes retrieved, use ICSI and assisted hatching, have zero embryos cryopreserved, and have only one embryo transferred. Cryopreserved donor oocyte cycles were more likely to be conducted among recipients ≥40 years old and using oocytes retrieved from women <30 years old than fresh donor oocyte cycles. They also were more likely to be conducted among recipients who attended a larger clinic, used ICSI and assisted hatching, and had zero embryos cryopreserved. The only factor significantly associated with live birth among autologous cryopreserved oocyte cycles was cryopreservation of at least one embryo ( Table 2 ). Autologous fresh oocyte cycles had higher rates of live birth among patients of younger age, and patients with no prior pregnancies, births, or ART cycles. Live birth rates varied among autologous fresh oocyte cycles by reason for ART use: rates were significantly higher when the reason for ART was endometriosis, ovulation disorder, male factor infertility, or unexplained infertility, and lower for diminished ovarian reserve, tubal factor infertility, uterine factor infertility, or “other” reason. Autologous fresh oocyte cycles also had higher rates of live birth when ≥11 oocytes were retrieved, at least one embryo from the cycle was cryopreserved, assisted hatching was not used, two or more embryos were transferred, or blastocyst embryos were transferred. Donor cryopreserved oocyte cycles had higher rates of live birth among recipients with no prior ART cycles or if at least one embryo was cryopreserved. Rates of live birth were higher among donor fresh oocyte cycles when recipients were 30–34 years of age or had an ovulatory disorder, when the woman providing the oocyte was aged <35 years, or when the ART treatment cycle involved the retrieval of ≥11 oocytes, the cryopreservation of at least one embryo, the transfer of two or more embryos, or the transfer of blastocyst embryos. Live birth rates were lower among donor fresh oocyte cycles when recipients had tubal factor infertility. When comparing outcomes of autologous cryopreserved oocyte cycles with autologous fresh oocyte cycles, after adjusting for significant covariates, cryopreserved oocyte cycles, transfers, and pregnancies showed no evidence of any difference in the rates of cancellation, implantation, pregnancy, miscarriage, or live birth ( Table 3 ). When comparing the outcomes of all donor cryopreserved oocyte cycles with all donor fresh oocyte cycles, cryopreserved oocyte cycles had lower cancellation rates before transfer (8.5% vs. 11.5%; adjusted risk ratio [aRR] 0.74, 95% confidence interval [CI] 0.57–0.96), lower pregnancy rates (51.1% vs. 58.5%; aRR 0.88, 95% CI 0.81–0.95), and lower live birth rates (43.0% vs. 49.4%; aRR 0.87, 95% CI 0.80–0.95) after adjusting for significant covariates. However, there was no evidence of a significant difference in implantation, pregnancy, or live birth rate between cryopreserved and fresh oocytes cycles when the cycles were restricted to only those proceeding to transfer. Among cycles resulting in pregnancy, if the number of embryos transferred was not considered, there was no evidence of a significant difference in miscarriage or live birth rates between cryopreserved and fresh donor oocyte cycles. However, cryopreserved donor oocyte transfers resulting in pregnancy had lower miscarriage rates (12.1% vs. 16.2%; aRR 0.75, 95% CI 0.58–0.97) and higher live birth rates (86.4% vs. 81.9%; aRR 1.05, 95% CI 1.01–1.09) if only one embryo was transferred but higher miscarriage rates (15.9% vs. 12.4%; aRR 1.28, 95% CI 1.07–1.54) and lower live birth rates (82.6% vs. 85.8%; aRR 0.95, 95% CI 0.92–0.99) if two or more embryos were transferred compared with fresh donor oocyte transfers resulting in pregnancy. In additional analysis, when controlling for oocyte age at retrieval, there was no evidence of a difference in live birth rates among single embryo transfers (results not shown).

Among patients using their own oocytes, we found no evidence of any significant difference in cancellation, implantation, pregnancy, miscarriage, or live birth rates if the oocytes were previously frozen as compared with fresh. Among patients using donor oocytes, although there were some differences in per-cycle outcomes (cancellation, pregnancy, and live birth rates) and per-pregnancy outcomes (miscarriage and live birth) between cryopreserved and fresh oocyte cycles, there was no evidence of any difference in implantation, pregnancy, or live birth rates among transfers. Cryopreserved oocyte cycles among patients using donor oocytes had a decreased risk of cancellation before transfer, as well as a decreased likelihood of pregnancy and live birth compared with fresh oocyte cycles when all cycles were considered; however, there was no evidence of any significant difference in outcomes when the analysis was restricted only to cycles proceeding to transfer. This difference in results between all cycles (including canceled cycles) and only cycles proceeding to ET may reflect the removal of canceled cycles because the cancellation rate differed between cryopreserved and fresh oocyte cycles, or the difference in variables considered for adjustment. Transfer-level variables, such as the number of embryos transferred and the number of embryos cryopreserved, were not considered as possible covariates in the cycle-level models because they had missing values for all cycles that did not proceed to retrieval and/or transfer. Many of these variables are important predictors of success and were retained as significant covariates in the transfer-level models. In addition, for cycles among patients using donor oocytes, there was no evidence of a significant difference in miscarriage or live birth per pregnancy between cryopreserved and fresh oocyte cycles when number of embryos transferred was not considered. However, among all cycles resulting in pregnancy after the transfer of only one embryo, there was a decreased risk of miscarriage and an increased likelihood of live birth, whereas among all cycles resulting in pregnancy after the transfer of two or more embryos, there was an increased risk of miscarriage and a decreased likelihood of live birth for cryopreserved oocyte cycles compared with fresh oocyte cycles. The source of the difference in risk according to the number of embryos transferred is not immediately apparent; however, it could reflect differences in embryo stage or quality, which were not adjusted for during modeling. Goldman et al. ( 8 ) and Almodin et al. ( 9 ) showed that oocyte cryopreservation may negatively affect the development of an embryo to blastocyst stage; however, Garcia et al. ( 10 ) showed no evidence of a difference in cleavage or blastocyst development between fresh and cryopreserved oocyte cycles. When looking for significant differences in ART cycle, transfer, and pregnancy outcomes between fresh and cryopreserved oocyte cycles, it is notable that although many of the associations were not significant after adjustment for other covariates (see footnotes in Table 3 for a complete list of covariates), many of the differences were significant before adjustment. Although the set of covariates included for adjustment for each outcome differed, for autologous cycles and transfers, patient age and the number of prior ART cycles were significant covariates in all models and seemed to explain the variability in outcomes between fresh and cryopreserved oocyte cycles. In addition, for autologous and donor transfers, the number of embryos cryopreserved and the use of assisted hatching were significant covariates in all models and seemed to explain the variability in outcomes between fresh and cryopreserved oocyte cycles. Regardless of whether patients used their own oocytes or donor oocytes, cryopreserved oocyte cycles were generally conducted among older patients, as would be expected because cryopreserved oocyte cycles among patients using their own oocytes are commonly used for planned fertility preservation, whereas cryopreserved oocyte cycles among patients using donor oocytes are commonly used when a patient has diminished ovarian reserve and their autologous oocytes will not work. The vast majority of donor cryopreserved oocyte cycles, and donor cycles in general, were conducted with oocytes retrieved from women younger than 30 years, because younger oocytes are associated with higher success rates. In addition, both autologous and donor cryopreserved oocyte cycles used ICSI and assisted hatching more frequently than autologous and donor fresh oocyte cycles, because ICSI has been identified as possibly beneficial for ART procedures involving oocytes that have been cryopreserved ( 11 ). Finally, fewer embryos were cryopreserved in cycles using autologous and donor cryopreserved oocytes compared with fresh oocyte cycles. It is likely that fewer cryopreserved oocytes are thawed for fertilization as compared with the average oocyte yield in a fresh cycle. The only characteristic significantly associated with live birth for cryopreserved oocyte cycles among patients using their own oocytes was having at least one embryo cryopreserved, an indicator that multiple good-quality embryos were available for transfer; this was also a significant predictor of success for all other cycle types. Cryopreserved oocyte cycles among patients using donor oocytes had one other significant predictor of success: having no prior ART cycles. In comparison, fresh oocyte cycles had a variety of significant predictors of success both among patients using their own oocytes or donor oocytes. Differences in characteristics significantly associated with live birth between cryopreserved and fresh oocyte cycles may be due to the smaller number of cryopreserved oocyte cycles; they may also be due to the preservation of fertility when oocytes are frozen. This study is the first national study exploring cryopreserved oocyte cycles with the ability to control for other factors that may influence success. Previous studies, many of which were clinic-specific and therefore lacking generalizability, vary in their results. Kushnir et al. ( 12 ) found that donor cryopreserved oocyte cycles had worse live birth rates than fresh oocyte cycles; however, because the study used aggregate data, it did not adjust for any factors that might affect live birth rates. Several other studies found that cryopreserved oocyte cycles had success rates that were approaching or not different from those of fresh oocyte cycles ( 8 – 10 , 13 , 14 ). Some additional studies compared fresh and cryopreserved oocyte cycles for sibling donations (i.e., those from the same donor but given to different recipients), such that they inherently controlled for characteristics of the donor. These studies also found that frozen oocytes cycles had similar success rates as fresh oocyte cycles ( 15 – 18 ). This study is subject to several limitations. Because cryopreserved oocyte cycles only became nonexperimental in October 2012, we were only able to analyze the most recent year of ART data collected in NASS, 2013, when these cycles were first reported to the Centers for Disease Control and Prevention. As a result, and because of the low prevalence of cryopreserved oocyte cycles, the sample size was small, resulting in a lack of power to detect differences, particularly for autologous cycles. Additionally, because of some limitations in adapting an existing surveillance system to collect a new type of ART cycle, some important information, such as stage of ET, was not available for the cryopreserved oocyte cycles. Other important variables, such as embryo quality, method of cryopreservation, and mechanism for endometrial preparation also were not included in the analysis, because they are not currently collected for any cycle type in NASS. A recent review of studies comparing the method of cryopreservation found that vitrification may improve pregnancy rates over slow freezing, and therefore may be an important variable for adjustment ( 19 ). Comparisons between autologous fresh and autologous frozen oocyte cycles should be interpreted with caution owing to the difference in group sizes and our inability to account for the mechanism of endometrial preparation. As previously noted, age of the woman providing the oocyte was not controlled because of missing data; however, for donor cycles, donor eggs are usually selected from healthy, young donors, as evidenced by the similar age distributions seen in Table 1 . Finally, we were unable to look at differences in results based on the reason for oocyte freezing because this field is not captured in NASS. Oocyte cryopreservation has expanded ART treatment options for many patients, including those needing fertility preservation for medical reasons and those choosing to delay child-bearing who do not want or are unable to fertilize oocytes before freezing. With only some differences in per-cycle and per-pregnancy outcomes between donor cryopreserved and fresh oocyte cycles, our data suggest that after adjusting for other significant factors, there is limited evidence of any difference in outcomes between fresh and cryopreserved oocyte cycles, both for fresh embryo cycles using a patient’s own oocytes or donor oocytes. Because oocyte cryopreservation is expected to increase, more work is needed to further explore the impact and safety, both in the short term and longer term, of fertility preservation for the purpose of deferred childbearing ( 1 ).

We analyzed 105,517 fresh embryo cycles initiated in 2013, including 422 autologous cryopreserved oocyte cycles, 93,181 autologous fresh oocyte cycles, 2,223 donor cryopreserved oocyte cycles, and 9,691 donor fresh oocyte cycles, as reported to the Centers for Disease Control and Prevention’s National ART Surveillance System (NASS). The NASS contains data from 467 US clinics that performed assisted reproductive technology (ART) cycles in 2013 and accounts for approximately 98% of all ART cycles conducted in the United States ( 5 ). Fresh embryo cycles are cycles with the intent to transfer an embryo created during the current cycle; fresh oocyte cycles are cycles with the intent to transfer embryos derived from oocytes retrieved during the current cycle; and cryopreserved oocyte cycles are cycles in which oocytes that were previously frozen are thawed with the intent of fertilizing the oocytes and transferring the resulting embryo(s) to the patient during the current cycle. Autologous cycles use oocytes belonging to the patient, whereas donor cycles use oocytes provided for the patient (the recipient) by a donor ( 6 , 7 ). Frozen embryo cycles, or cycles with the intent to transfer a thawed embryo that was cryopreserved during a previous cycle, were excluded from analysis, because the state of the oocyte (fresh or cryopreserved) is not collected in NASS for these cycles (n = 55,506). Donor embryos, or those embryos left over from a patient’s ART treatment that are then donated to another patient, were excluded from analysis (n = 1,460) ( 5 ). Additionally, a small number of cycles of mixed type were removed from analysis (n = 726). We compared characteristics and outcomes of cycles utilizing fresh and cryopreserved oocytes, both autologous and donor, among all cycles started, among all cycles proceeding to retrieval (i.e., excluding cycles canceled before retrieval), among all cycles proceeding to transfer (i.e., excluding cycles canceled before retrieval or transfer), and among all cycles resulting in pregnancy (i.e., excluding canceled cycles or cycles for which a pregnancy was not achieved). Assisted reproductive technology cycle-level characteristics explored in this study included clinic size, patient age, obstetric history, and reason for ART. Retrieval characteristics included age of the woman providing the oocyte at time of retrieval; number of oocytes retrieved; use of intracytoplasmic sperm injection (ICSI, the injection of a single sperm directly into an oocyte); and number of embryos cryopreserved at the end of the ART cycle for potential future use. Transfer characteristics included use of assisted hatching (a micromanipulation technique designed to enhance implantation); number of embryos transferred; and embryo stage at transfer, either cleavage (2 to 3 days after fertilization) or blastocyst (5 to 6 days after fertilization) ( 6 , 7 ). Outcomes of ART cycles explored in this study included cancellation, in which an ART cycle is stopped after the start of ovarian stimulation or monitoring but before the retrieval of an oocyte and/or the transfer of an embryo; the number of embryos achieving implantation (number of fetuses) per the number of embryos transferred; pregnancy; miscarriage; and birth of at least one live-born infant ( 6 , 7 ). We compared cycle-, retrieval-, and transfer-level characteristics of cryopreserved and fresh oocyte cycles, both among autologous and donor cycles, using a Rao-Scott χ 2 test, which adjusts for clustering of the data by clinic. We calculated live birth rates for cryopreserved and fresh oocyte cycles, both among autologous and donor cycles, according to the cycle-and transfer-level characteristics listed above. We explored whether these characteristics were significantly associated with live birth using unadjusted log binomial regression models with generalized estimating equations to account for clustering by clinic. We assessed ART treatment outcomes, including cancellations, pregnancies, and live births among all ART cycles; implantation, pregnancy, and live births among all ART transfers; and miscarriage and live births among all ART pregnancies. Comparisons were made between fresh and cryopreserved oocyte cycles stratified by oocyte source (autologous or donor). Outcomes were modeled using unadjusted and adjusted log binomial or Poisson regression with generalized estimating equations. For outcomes among all cycles, we adjusted for significant risk factors from the set of all possible cycle-level risk factors; for outcomes among transfers and pregnancies, we adjusted for significant risk factors from the set of all possible cycle-, retrieval-, and transfer-level risk factors. Significant risk factors were determined using backward model building. Finally, we considered all interactions between the independent variable of interest (fresh or cryopreserved oocyte) and each of the other independent variables. We retained those interactions that were significant in the adjusted models and that affected the interpretation of the comparison between fresh and cryopreserved oocyte cycles. We did not include age of the woman providing the oocyte at time of retrieval in the adjusted models owing to 24% and 21% missing values among cryopreserved autologous and donor oocyte cycles, respectively; however, we did check to determine whether the inclusion of this variable affected results and reported differences within the text. We did not adjust for variables that were not available for all cycle types, such as embryo stage at transfer, which was only available for fresh oocyte cycles. All statistical analysis was conducted in SAS v. 9.3 (SAS Institute). Statistical significance was determined using α = 0.05. This study was approved by the Institutional Review Board of the Centers for Disease Control and Prevention.