Association of duration of embryo culture with risk of large for gestational age delivery in cryopreserved embryo transfer cycles.

Thirty-three thousand and thirty (18.2%) FET cycles in the study group (n = 181,592) resulted in LGA infants during the study period of 2014–2019 ( Supplemental Figure 1 , available online). Our primary outcome was the rate of LGA per day of embryo cryopreservation. There was a statistically significant difference in LGA by cryopreservation day ( P <.001). The rate of LGA increased with increasing days of cryopreservation from day 2 (13.7%) to days 3–7 (14.4%, 15.0%, 18.2%, 18.5%, and 18.9%) in the unadjusted analysis. The highest rates of LGA were related to gravidity, full-term births, and preterm births. Further differences in the patient characteristics between the non-LGA and LGA cycles included are listed in Table 1 . Non-Hispanic (NH) White patients had the highest risk of LGA (20.4%), although Asian patients had the lowest risk (10.9%). Age groups of 25–29 and ≥ 40 years had the highest risks of LGA (18.7 and 18.6%, respectively). Not unexpectedly, increasing BMI had a stepwise increase in LGA risk, with BMI >35 kg/m 2 having the highest rates of LGA (29.5% in BMI >35 kg/m 2 ), whereas BMI <18.5 kg/m 2 had the lowest rates of LGA (8.6%) ( Table 1 ). Frozen embryo transfer and linked IVF cycle factors associated with LGA are present in Table 2 . The diagnoses most associated with LGA risk were endometriosis and tubal ligation (19.5% and 19.3%, respectively). Maximum FSH levels 4–10 IU/L were associated with the highest risk of LGA (18.3%), and no PGT-A vs. PGT-A for all or some embryos was also more associated with LGA risk (19.3% vs. 17.0% and 17.4%). Increasing oocyte number (>30) and increasing peak endometrial stripe measurement (≥12 mm) were associated with LGA risk (18.8% and 20.1%, respectively). Poor embryo grade (compared with good) was also associated with LGA risk (19.7% vs. 18.1%). In the multivariable model ( Fig. 1 and Table 3 ), days 5–7 of cryopreservation demonstrated a significant association with LGA (aRR 1.32, 95% CI 1.22–1.44 for day 5, aRR 1.34, 95% CI 1.23–1.46 for day 6, aRR 1.42, 95% CI 1.25–1.62 for day 7, all compared with a group of days 2–3, P <.001). Several other risk factors demonstrated significant associations with LGA. Increasing BMI, as expected, increased the risk of LGA in this model (aRR 1.40, 95% CI 1.36–1.43 for BMI 25.0–29.9 kg/m 2 ; aRR 1.63, 95% CI 1.58–1.68 for BMI 30.0–34.9 kg/m 2 ; aRR 1.94, 95% CI 1.88–2.01 for BMI >35.0 kg/m 2 compared with BMI 18.5–24.9 kg/m 2 , P <.001). Gravidity was associated with a small but significant increased risk of LGA (aRR 1.09, 95% CI 1.06–1.12 for gravidity of 1, aRR 1.12, 95% CI 1.08–1.16 for gravidity of 2–3, and aRR 1.15, 95% CI 1.08–1.22 for gravidity of >3, compared with gravidity of 0, P <.001). Full-term parity was also associated with the risk of LGA (aRR 1.36, 95% CI 1.33–1.40 for full-term parity of 1, aRR 1.39, 95% CI 1.33–1.45 for full-term parity of 2–3, and aRR 1.41, 95% CI 1.29–1.53 for full-term parity of >3, compared with full-term parity of 0, P <.001). Preterm parity demonstrated a significant association with LGA (aRR 1.36, 95% CI 1.31–1.42 for preterm parity of 1, aRR 1.37, 95% CI 1.22–1.52 for preterm parity of 2–3, and aRR 1.82, 95% CI 1.1.24–2.69 for preterm parity of >3, compared with preterm parity of 0, P <.001). Embryo grade of fair compared with good also had a small risk of LGA (1.04, 95% CI 1.01–1.07, P =.008). A decreased risk of LGA was observed in several race and ethnicity groups compared with NH White patients. Asian, Black, and Hispanic patients had a lower risk of LGA (aRR 0.59, 95% CI 0.57–0.62, aRR 0.78, 95% CI 0.74–0.82, aRR 0.86, 95% CI 0.82–0.91, respectively, P <.001). Low BMI compared with normal BMI (<18.5 vs. 18.5–24.9 kg/m 2 ) also demonstrated a lower risk of LGA (aRR 0.61, 95% CI 0.55–0.67, P <.001). Preimplantation genetic testing for aneuploidy of all embryos had a lower risk of LGA compared with no PGT-A (aRR 0.95, 95% CI 0.93–0.97, P <.001).

This study was reviewed and approved by the Madigan Army Medical Center Institutional Review Board and was conducted in compliance with applicable regulations regarding human subject research. The study was also approved by the SARTCORS research committee before its release of data. We examined factors associated with the risk of LGA after FET using a retrospective cohort of FET cycles performed during 2014–2019 that resulted in singleton live births (n = 188,294). We included the patient’s first FET cycle only, and we excluded cycles that with unknown birth weight (n = 4,564), those with gestational age at birth of 42 weeks (n = 867), those with no birth weight or gestational age (n = 46), those with fetal weights >5,500 g (n = 57), those with no live birth (n = 30), those with use of a gestational carrier (n = 40), those cycles with >1 infant born but only one live birth (n = 116), and those with missing data for day of cryopreservation (n = 982). Mean birth weights per gestational age and the calculation of the 90 th percentile for the gestational age were calculated using the INTERGROWTH-21 st methodology ( 18 ). The INTEGROWTH-21 st includes fetal gender in the calculation, so a gender variable was not used in the model below. There were 182,574 singleton FET cycles included in our final analysis. The primary outcome was the rate of LGA per day of embryo cryopreservation. To this end, we compared all patients and linked IVF cycle and FET cycle data as possible risk factors for LGA (age, race, ethnicity, body mass index [BMI], year of IVF cycle, gravidity, parity, smoking status, infertility diagnosis, maximum follicle-stimulating hormone [FSH] levels measurement, use of PGT-A, number of oocytes retrieved, endometrial thickness, number of embryos transferred, embryo grade, and day of cryopreservation of the embryo) between LGA and non-LGA outcomes. Missing data or data not consistent with normative values (BMI >60 kg/m 2 , oocytes retrieved >60, and others) were listed as missing and included in our analysis. To further characterize patient and IVF-FET cycles as independent risk factors for LGA after FET, we performed a multivariable generalized linear regression (assuming log link and binomial error variance), including patient factors (race and ethnicity, BMI, parity [full-term and preterm], gravidity, embryo morphology grade, peak endometrial measurement, use of PGT-A, and day of embryo cryopreservation). Pearson chi-square tests were used to compare categorical variables between LGA and non-LGA outcomes for FET cycles. Factors that were selected a priori or that were significantly associated with LGA in univariate analyses were included in the multivariable analysis. We estimated adjusted relative risks (aRRs) and 95% confidence intervals (CIs) to determine patient and IVF-FET cycle factors associated with LGA. We estimated a sample size estimation, with 90% power and an alpha of 0.01 (using a Bonferroni correction of 0.05/5 comparisons), and assuming a 3% decrease in LGA risk per day from days 6–5 of embryo cryopreservation, we would need 6,058 patients per cohort to demonstrate this difference. All statistical analyses were performed using SPSS statistical software, version 28 (IBM Corporation), and results were considered significant at a P value <.05.

Our study demonstrates that the extension of embryos in culture before cryopreservation is an independent risk factor for LGA. The aim of this study was to examine the association between the day of transfer of frozen embryos after cryopreservation and LGA-age infants resulting from those pregnancies. Both the unadjusted and multivariable models demonstrated an associated risk. The observed rates demonstrated a stepwise increase in LGA risk from day 2 (13.7%) to day 7 (18.9%). However, in models adjusted for important confounders, the risk of LGA increased from days 5–7. As more clinics are extending culture to days 5–7 for trophectoderm biopsy, it is important to understand the impact of extended culture on the risk of LGA infants after FET cycles ( 19 ). Several hypotheses have been proposed for the known association between FET and the increased risk of LGA infants compared with fresh embryo transfers, including epigenetic changes to the embryo during cryopreservation, culture media type, and exposure, as well as the absence of a corpus luteum ( 8 ). The increased risk of LGA with each additional day of transfer demonstrated in our study posits that increased and prolonged exposure to culture media may impact this risk. Although specific cryopreservation protocols (type of culture media used, vitrification techniques used by each individual laboratory and others) were not available within the SARTCORS database, the prolonged exposure of the embryo to culture media, regardless of type or method, appears to contribute to the increased risk of LGA. Previous literature has suggested an association between prolonged exposure to culture media and epigenetic changes within the embryo ( 20 ). Zhao et al. ( 21 ) used mouse oocytes and demonstrated that the vitrification process can alter the expression of DNA methyltransferase. This was particularly important because subsequent studies linked the expression of DNA methyltransferases with abnormal fetal and placental weights ( 22 , 23 ). Although many studies have been performed in fresh embryo transfer cycles, few studies have evaluated birth weights for infants born from FET cycles in the context of culture media use, and those that were available had relatively low FET numbers for analysis. Two of these studies ( 24 , 25 ) found no difference between different mediums and infant birth weight, although a third study showed a trend of increased LGA pregnancies in Sage media compared with human tubal fluid (15.1% vs. 6.3%; P =.09) ( 26 ). Culture media constituents, including amounts of certain proteins within the culture media (namely serum albumin), have been also shown to influence the size of offspring in animal models ( 27 , 28 ). Two studies using human embryos have also suggested that added protein sources may also have a role in fetal weight ( 29 , 30 ). Finally, 2 recent meta-analyses demonstrate a mildly increased risk of LGA in programmed cycles vs. natural cycles (an advanced odds ratio of 1.08–1.10), suggesting the corpus luteum and/or use of exogenous estrogen and progesterone may influence LGA ( 31 , 32 ). Our study is not the first to evaluate the risk of LGA regarding extended embryo culture. Mäkinen et al. ( 33 ) found that the length of embryo culture was a significant independent factor for determining birth weight after fresh IVF cycles, with an increased risk of LGA on day 5 or 6 compared with day 3 transfers. This study was limited, however, by a small number of day 5 or 6 transfers. Studies performed in frozen cycles have revealed conflicting results, likely because of low numbers of day 6 or day 7 embryos and/or including fresh transfers ( 13 – 15 , 34 ). Our study likely demonstrates the highest number of patients in cohorts of day 2, day 4, and days 6–7 in this type of study. Several other variables demonstrated a significant association with LGA within the multivariable model. Increasing BMI, as expected, increased the risk of LGA in this model in a stepwise manner, with increasing LGA for overweight and obese patients when compared with patients of normal BMI. Increasing gravity, term and preterm parity, the number of oocytes (>30), and fair embryo quality were also independent risk factors for LGA. Both BMI and increasing parity have been demonstrated previously to increase the risk of LGA after FET ( 35 ). Prepregnancy overweight and obese patients are well known to have an increased risk of gestational diabetes (a known risk factor for excessive fetal growth), and infants born from overweight and obese mothers have a higher percentage of fat when compared with normal-weight mothers (even in nondiabetic pregnancies) ( 36 , 37 ). Increasing parity and gravity are associated with an increased risk of gestational diabetes and a higher prepregnancy BMI, both known to increase fetal weight in subsequent pregnancies ( 38 , 39 ). It is not clear the pathological rationale for why increased numbers of oocytes and fair embryos compared with good embryo grades would increase the chances of LGA. We can only hypothesize that patients with increased numbers of oocytes and embryos with both good and fair embryo grades may have more embryos frozen on days 5–7, whereas those with low oocyte numbers are likely to cryopreserve embryos on days 2 and 3. Race and ethnicity of Asian, Black, and Hispanic (compared with NH White), PGT-A all embryos (vs. none), and BMI <18.5 kg/m 2 (compared with normal BMI) were found to be protective against LGA pregnancies. The relationship between race and ethnicity as well as fetal weight disorders are not well described. One student demonstrated that despite excess weight gain in multiple race and ethnicity groups, NH White and Asian patients were more likely to have LGA infants than Black patients ( 40 ). It is not clear from this study and others the cause of these relationships. Preimplantation genetic testing for aneuploidy is a known factor in reducing the risk of LGA infants, as a recent study from Li et al. ( 41 ) demonstrated trophectoderm biopsy significantly decreased the risk of macrosomic and LGA infants born after PGT-A and FET cycles compared with nonbiopsied FET cycles ( 41 ). Low prepregnancy BMI (<18.5 kg/m 2 ) is a well-known risk factor for smaller fetal growth ( 42 ). Despite studies suggesting FET cycle-associated birth weight has increased from 1991–2015, one study demonstrated the risk of LGA births resulting from FET cycles has decreased over the period of 2004–2018 ( 35 , 43 ). Our study found no association with the year of embryo transfer and the risk of LGA from 2014–2019 in our study cohort. This study has some key strengths and limitations. Because the SARTCORS database comprises >90% of all IVF cycles in the United States, this is one of the largest and most comprehensive studies to evaluate the risk of LGA associated with the day of transfer after cryopreservation for FET cycles. However, limitations inherent to observational studies such as the risk of selection bias, their retrospective nature, and lack of complete information on the specific cryopreservation protocols used in each cycle were present. In addition, our power analysis was conducted using a 3% difference per day of cryopreservation; therefore, we would need approximately 6,000 patients per day of cryopreservation. Although days 5 and 6 had >6,000 patients per cohort, other days did not meet this goal. The differences found between these possible underpowered cohorts, however, were still statistically significant in the final logistic binomial model, strongly suggesting the findings are correct in this study. Lastly, this study was not able to address the effect of vitrification on the rates of LGA in FET cycles. By limiting the analysis to start in 2014, our analysis hopefully restricted the analysis to embryos that were vitrified, but likely there are a small number of embryos that were slow-frozen. We can draw no conclusions from the method of freezing embryos for this study. In summary, we have demonstrated that there is an association between the day of cryopreservation and the risk of LGA-age infants for frozen embryos. Although our study supports previously known variables (such as BMI and parity) that increase the risk of LGA infants after FET cycles, our study also demonstrates extended time in culture as an independent risk factor for LGA. Several aforementioned hypotheses have been proposed to explain the physiology responsible for this phenomenon, although more research is required to investigate them further. By trying to understand the causes, we can reduce the risk of LGA infants after FET, which will help prevent both fetal and maternal morbidity associated with LGA.