Assisted hatching and live births in first-cycle frozen embryo transfers.

The number of in vitro fertilization (IVF) procedures performed in the United States increased annually from 2004 to 2013.( 1 ) Assisted hatching (AH), thinning or creating an opening in the zona pellucida (ZP), prior to implantation is often used during IVF procedures. The majority of the data in randomized clinical trials, reviews, and meta-analyses use clinical pregnancy as the primary outcome.( 2 , 3 ) However, clinical pregnancy is only a surrogate measure for the ultimate end point of live birth. From 1992 to 2009, there have been 15 trials investigating live birth rate (LBR) and AH with only 420 live births reported from these small trials.( 3 , 4 ) Given the limited number of studies, the American Society of Reproductive Medicine guidelines conclude that there is insufficient evidence that AH improves live births.( 4 ) Recent studies have shown that frozen embryo transfers (FETs) have greater LBRs compared to conventional fresh transfers.( 5 , 6 ) Some studies suggest that cryopreserved embryos might be better candidates for AH, but the data is contradictory.( 7 – 10 ) In patients with a previous failed IVF cycle, subgroup analyses have found improved clinical pregnancy rates with AH, but no improvement in live births.( 2 ) Previous analysis of the Society for Assisted Reproductive Technology clinical outcome reporting system (SART CORS) database showed that LBRs were not improved in women with diminished ovarian reserve undergoing fresh cycles with AH and intracytoplasmic sperm injection (ICSI).( 11 ) The goal of this study was to utilize a large, retrospective cohort reported in the SART CORS database between 2004 and 2013 to assess whether AH prior to first-cycle autologous FET improves LBR.

The study population included 151,533 first-cycle, autologous, FETs resulting in 52,773 live births. AH of all transferred embryos occurred in 70,738 cycles and no AH occurred in 80,795 cycles. The overall LBR in the AH cohort and no AH cohort were 34.2% and 35.4%, respectively (p<0.001). The AH and no AH cohorts were significantly different by age at retrieval, race, BMI, gravidity, parity, infertility diagnosis, max FSH level, and numbers of embryos transferred ( Table 1 ). After propensity matching and multiple imputation, logistic regression without regards to AH showed multiple covariates affected the odds of live birth including number of embryos transferred, gravidity, max FSH level, and BMI ( Figure 1 ). In years since 2007, LBR improved compared to 2004. LBR is increased when more than one embryo is transferred compared to single embryo transfer. In all age groups above 35 years old LBR is decreased compared to less than 35 years old. Higher BMIs had a negative effect on LBR. Specific etiologies of infertility including uterine, tubal, hydrosalpinx and endometriosis had lower LBRs compared to male infertility. The not specified other reason for ART (Other RFA) had a higher LBR compared to male infertility ( Figure 1 ). The year of transfer and age at retrieval with AH affected LBR in a matched cohort ( Figure 2 , Supplemental Table 1 , Supplemental Table 2 ). AH resulted in a statistically significant decrease in LBRs in 2012 and 2013 when compared to earlier years in all age groups, with an adjusted odds ratio (AOR) of 0.86 [CI 0.80–0.91] and 0.85 [CI 0.79–0.89] respectively (p 42 years old (p≤0.05, Figure 2 ). In the most recent year (2013), the AH cohort had decreasing LBRs which were more pronounced at each successive age group (p<0.001, Supplemental Table 3 ). Sensitivity analyses using the complete-cases, the matched logistic regression without imputation, and matched and imputed analysis all show the same basic patterns of the association of AH with poor outcome in the older patients in years 2012–2013 ( Supplemental Table 4 ). There was no significant difference in monozygotic twinning between the AH and no AH cohorts (1.1% v 1.1 %, p=0.41). For single embryo FET, the number of monozygotic twins were 164 with AH and 155 without AH. For double embryo FET, the number of monozygotic twins were 182 with AH and 175 without AH. The geographical location of clinics influenced the prevalence of AH in first-cycle FETs ( Table 2 ). Clinics performing AH in greater than 50% of FET cycles were found more frequently in the Midwest (37.8%) and Northeast (33.3%) compared to the South (19.2%) and West (18.8%). The clinics which performed AH in greater than 75% of FET cycles had the lowest LBR (27.9%). Overall the South and West had the highest LBRs in the complete-cases and matched cohort (p<0.001).

Live birth rates from FETs have been increasing since 2004. In all first-cycle FETs analyzed, there was a decrease in LBR with AH compared to no AH (34.2% v 35.4%, p=0.001). AH was correlated with a reduction of LBR in later reporting years and in older patients. One of the largest differences was seen in women > 42 years old with AH (LBR 14% versus 30% with no AH; AOR 0.55 [CI 0.43 to 0.69], p<0.001). However, this group of older patients is a small proportion of the total patients in the data set, and this large effect is hidden within the 1% overall difference in LBR. There was no significant increase in monozygotic twinning due to AH in this FET cohort, which is in contrast to prior reports showing up to a 3 times risk with fresh embryo transfers.( 17 ) Previously published data showed that the SART database is not adequate for recording fetal anomalies; therefore, anomalies were not analyzed in this study.( 15 ) Age is a major determinant of infertility and is predictive of live birth with fertility treatments.( 18 ) Older women were more likely to receive AH than younger women in this study ( Table 1 ). In all years analyzed, women 38 years old and above undergoing AH had a decreased chance of live birth compared to those without AH. In the 2013 cohort, all age groups had decreased live birth rates with AH, with the most pronounced harm in the older age groups. It is unclear whether AH is affecting the embryo or poorer quality embryos are selected for AH. More studies are needed in this older age group with similar quality embryos to further assess this observation. The number of patients receiving AH increased every year, along with the total number of first-cycle, FETs performed. Prior to 2012, the percentages of cycles with AH were significantly less than those without AH. In the last two years of the analysis (2012–2013), the number of cycles with and without AH is similar. Although there was a steady increase in LBR in all first-cycle FET in the matched cohort over time, the LBR in relationship to AH was significantly decreased in 2012 and 2013 compared to earlier years. The recent negative effect of AH could be due to different AH techniques performed or other changes in the IVF process including the use of embryo vitrification. The different AH techniques including mechanical, chemical, and laser methods are not recorded in the SART database, so we were unable to account for this in our analysis. Vitrification of embryos has been in use since the early 2000s and found to have similar implantation and pregnancy rates as compared to fresh embryo transfer.( 19 ) However, individual clinic use of cryopreservation techniques may vary over time and this is not reported to SART. Another possible confounder during the process of preparing an embryo for freezing is blastocele puncture (assisted shrinkage), which is also not standardized.( 20 ) Poor prognosis patients were more likely to have AH. FETs with 3 or more embryos were more likely to receive AH. Also, when diminished ovarian reserve was the etiology for infertility, patients were more likely to receive AH compared to the other etiologies. Asian, African Americans, and Hispanic ethnicities were more likely to receive AH compared to Caucasians. Previous studies focused on “poor prognosis” patients are limited by small numbers and heterogeneity in the definition of “poor prognosis.”( 3 , 8 ) Further studies are needed to evaluate these influences on AH. Pre-implantation genetic screening and diagnosis have become a new addition to the armamentarium considered during IVF.( 21 ) We elected to exclude data from these cycles at this time because pre-implantation genetic testing necessitates a breech in the ZP similar to AH. A clinic’s geographical location has variable results on LBR. Patient population and insurance coverage may be partly responsible.( 22 ) Compared to the Midwest reference group, AH is performed more commonly in the Midwest and Northeast, however LBRs in all cycles were higher in the South and West locations. Clinics that reported the highest percentage of IVF cases with AH (>75%) had the lowest LBRs with those cycles, but overall these cases were few and it is important to note that the statistical significance of these numbers may not have clinical relevance. This study differs from others given the large number of cycles available for study through the SART CORS database. While prospective, randomized studies are the gold standard in reducing bias, this large-scale retrospective analysis remains important to expose specific patient characteristics that may benefit and/or be harmed by certain procedures. A recent meta-analysis identified 36 randomized studies assessing AH, however only 8 studies pertained to AH with frozen-thawed embryos and they found no significant difference in clinical pregnancy or live birth.( 3 ) In the current study, propensity matching allowed for a matched set of comparable recipients with and without AH. Several limitations exist in the present study. The SART CORS database is an excellent repository of cycle information and outcomes; however, its retrospective nature limits our ability to acquire information that is not available with FET cycles, such as stage and quality of the embryo. There are also several variables that are missing or known to be unreliable in this dataset.( 15 ) Multiple imputations were performed for missing variables, such as BMI and FSH values. Additionally, de-identification of the database prevents the linkage of thawed cycles to the original retrieval and insemination technique. Thus, we were unable to determine the extent to which insemination technique, embryo quality at the time of cryopreservation and cryopreservation techniques play in LBR with AH in first-cycle, FETs. We excluded cycles marked as ICSI to limit the impact of ZP penetration, as thaw cycle data did not reflect this information prior to 2014. In conclusion, AH resulted in a decreased LBR in more recent years and a decrease in LBR in older age groups. Prospective, clinical studies are needed to improve our knowledge on the impact of AH on FETs.

The SART CORS database, consisting of data from the majority of clinics performing Assisted Reproductive Technology (ART) in the United States, was used for this study.( 12 , 13 ) The research proposal was approved by the SART research committee. Data were collected and verified by SART and reported to the Centers for Disease Control and Prevention in compliance with the Fertility Clinic Success Rate and Certification Act of 1992 (Public Law 102-493). The data in the SART CORS are validated annually with some clinics having on-site visits for chart review based on an algorithm for clinic selection. During each visit, data reported by the clinic were compared with information recorded in patients’ charts. Ten out of 11 data fields selected for validation were found to have discrepancy rates of ≤5%.( 14 ) All data were de-identified and the Office of Regulatory Affairs at the University of Texas Health Science Center at San Antonio allowed exemption from Institutional Review Board approval. Data analyzed was from 2004 to 2013 and included first, not banked, thawed, autologous IVF cycles in women undergoing FET. Cycles with preimplantation genetic screening or diagnosis (PGS/PGD), ICSI, and donor oocytes were excluded. The main outcome was LBR in cycles where all transferred embryos received assisted hatching (AH) and those in which none received assisted hatching (no AH). The cycles where only some transferred embryos received AH were excluded from analysis. Study populations were analyzed in regards to etiology of infertility, previous gravidity, age, race, max follicle stimulating hormone (FSH) level, body mass index (BMI), reporting year, and number of embryos transferred. Secondary outcome was monozygotic twinning.( 15 ) Demographics, prior pregnancy history, and etiologic factors of infertility were compared for AH relative to no AH groups using chi-squared tests for categorical variables and Mann-Whitney U tests for continuous variables. Factors associated with a lower rate of live birth, i.e. older age and higher number of embryos transferred, were often associated with having AH. Hence, propensity matching was used to account for these confounding covariates and a logistic regression model was constructed to identify the predictors of AH. The propensity model included the following covariates: etiology of infertility, previous gravidity, age at retrieval, race, max FSH level, region, reporting year, and total embryos transferred. The embryos that underwent AH were matched to those without AH using similar characteristics based upon the estimated probability of receiving AH. This matched set represents comparable recipients with and without AH that formed the main analytical dataset used to estimate the effect of AH. This dataset was summarized in terms of demographics, prior pregnancy history, and etiological factors as previously described. This analytic set was used to estimate the effects of AH on live-birth, again using logistic regression adjusting for covariates. The interactions between AH and other predictors of live birth were considered (e.g., patient age, year of procedure). In order to account for potential biases of missing data, multiple imputations were performed to fill in the missing values to account for the variation due to uncertainty. Imputation was conducted with the Amelia R package.( 16 ) The effects of AH were averaged over 10 imputed datasets. A sensitivity analysis was performed to account for propensity matching and missing data. The propensity score matched data set with imputation was compared to 1) a logistic regression analysis of the complete-cases (those without missing data) and 2) logistic regression analysis of the matched cohort. This data analysis was conducted within an accountable data analysis process. All data analyses were conducted in R (R Foundation for Statistical Computing, Vienna, Austria), and the threshold for statistical significance was a two-sided p-value of 0.05. The SART dataset included the percent reported cycles of AH per de-identified clinic. Location of the clinics was grouped by Midwest, Northeast, South, and West as determined by SART CORS. Clinics use of AH was analyzed by clinic location in relationship to live birth. Differences in LBRs were analyzed by Chi-square by R (Vienna, Austria).

AH, assisted hatching; LB R, live birth rate; AO R, adjusted odds ratio; Asterik denotes p-value less than 0.05 AOR for live birth for each age group by year in a matched cohort. AOR less than 1 indicates reduced LBR and greater than 1 indicates increased LBR. AH, assisted hatching; LBR, live birth rate; AOR, adjusted odds ratio; CI, 95% Confidence Interval. AOR for live birth for each age group in the year 2013. AOR less than 1 indicates reduced LBR. P value <0.05 denotes a difference in groups. FET, frozen embryo transfer; AH, assisted hatching; FSH, follicle-stimulating hormone; BMI, body mass index Data are mean +/− standard deviation (SD) or number (%) unless otherwise specified. Values are the adjusted odds ratio (AOR) for live birth using all clinical covariates. Asterik denotes p-value less than 0.5 na - not available due to large amounts of missing variables