Age does not affect maximal endometrial thickness achieved in frozen embryo transfer cycles: a SARTCORS study.

A total of 95,331 NOD recipient cycles and 25,247 GC cycles were included in our analysis. Among the NOD recipient cycles, 9,005 (9.45%) were age < 35 with a mean (SD) age of 31.3 (2.6) years and 86,326 (90.55%) were age ≥ 35 with a mean (SD) age of 43.2 (4.2) years. Among the GC cycles, 17,198 (68.12%) were age < 35 with a mean (SD) age of 29.3 (3.4) years and 8,049 (31.88%) were age ≥ 35 with a mean (SD) age of 38.0 (3.0) years (Table  1 ). Reasons for ART are reported to SART and multiple reasons can be selected for the same patient. Among the NOD recipient cycles, the most common reasons for ART were diminished ovarian reserve (< 35: 57.9%; ≥35: 79.9%), premature ovarian failure (< 35: 17.8%; ≥35: 3.4%) and male factor (< 35: 14.3%; ≥35: 16.4%). Among the GC cycles, the most common reasons were ovarian insufficiency (< 35: 36.1%; ≥35: 31.4%), uterine factor infertility (< 35: 17.6%; ≥35: 18.6%) and medical contraindication to pregnancy (< 35: 10.9%; ≥35: 11.1%). The median [IQR] number of cycles completed by each patient was 2 [1,3]. Table 1 Characteristics of non-identified oocyte donor recipient and gestational carrier cycles Measure or outcome Young NOD recipients ( n  = 9,005) Young GCs ( n  = 17,198) Older NOD recipients ( n  = 86,326) Older GCs ( n  = 8,049) Patient* age in years at cycle start Mean (SD) 31.3 (2.6) 29.3 (3.4) 43.2 (4.2) 38.0 (3.0) Patient* race/ethnicity N (%)  American Indian/Alaska Native 11 (0.12) 22 (0.13) 137 (0.16) 6 (0.07)  Asian 684 (7.60) 178 (1.04) 9361 (10.84) 159 (1.98)  Black/African American 270 (3.00) 597 (3.47) 5757 (6.67) 210 (2.61)  Hispanic/Latino 394 (4.38) 1778 (10.34) 4608 (5.34) 493 (6.12)  Native Hawaiian/Other Pacific Islander 3 (0.03) 20 (0.12) 118 (0.14) 6 (0.07)  White 4702 (52.22) 6204 (36.07) 35,204 (40.78) 3366 (41.82)  Mixed Race/Ethnicity 51 (0.57) 118 (0.69) 974 (1.13) 64 (0.80)  Unknown 2406 (26.72) 6158 (35.81) 24,579 (28.47) 2840 (35.28) BMI* (kg/m 2 ) Mean (SD) 26.3 (6.3) - 26.4 (5.9) - Most common reasons for ART N (%)  Male infertility 1289 (14.31) 1747 (10.16) 14,182 (16.43) 847 (10.52)  Endometriosis 591 (6.56) 649 (3.77) 4165 (4.82) 334 (4.15)  Polycystic ovary syndrome 427 (4.74) 721 (4.19) 1868 (2.16) 412 (5.12)  Tubal factor 447 (4.96) 723 (4.20) 6310 (7.31) 356 (4.42)  Uterine 362 (4.02) 3910 (22.74) 6184 (7.16) 1836 (22.81)  Medical contraindication to pregnancy 31 (0.34) 1876 (10.91) 239 (0.28) 893 (11.09)  Ovarian insufficiency 5214 (57.90) 6211 (36.11) 68,932 (79.85) 2524 (31.36)  Unexplained 463 (5.14) 974 (5.66) 3230 (3.74) 466 (5.79) NOD = non-identified oocyte donor; GC = gestational carrier; BMI = body mass index; *age, race/ethnicity, and BMI are reported by SART for the gestational carrier in GC cycles, and for the NOD recipient in NOD recipient cycles Characteristics of non-identified oocyte donor recipient and gestational carrier cycles Patient* age in years at cycle start Mean (SD) Patient* race/ethnicity N (%) BMI* (kg/m 2 ) Mean (SD) Most common reasons for ART N (%) NOD = non-identified oocyte donor; GC = gestational carrier; BMI = body mass index; *age, race/ethnicity, and BMI are reported by SART for the gestational carrier in GC cycles, and for the NOD recipient in NOD recipient cycles Among cycles where an embryo transfer was attempted, endometrial thickness was between 8 and 18 mm in 88% of GC cycles and 84% of NOD recipient cycles. There was a statistically significant difference in the mean EMT between younger and older gestational carriers (10.2 vs. 10.4, p  < 0.001) though this difference was only 0.2 mm. There was no difference in EMT between younger and older NOD recipients (9.7 vs. 9.7, p  = 0.081). The NOD recipients consistently achieved slightly less endometrial thickness in both age groups compared with gestational carrier controls. In a simple linear regression model including all NOD cycles, mean endometrial thickness decreased slightly with age (β = -0.008, 95% CI [-0.012, -0.005], p  < 0.001) while in a model including all GC cycles, mean endometrial thickness increased slightly with age (β = 0.032, 95% CI [0.024, 0.040], p  < 0.001). Mean endometrial thickness across age groups in NOD and GC cycles are shown in Fig.  1 . Fig. 1 Mean endometrial thickness by age group in NOD and GC cycles. EM = endometrial; NOD = non-identified oocyte donor; GC = gestational carrier Mean endometrial thickness by age group in NOD and GC cycles. EM = endometrial; NOD = non-identified oocyte donor; GC = gestational carrier When adequate endometrial thickness is not achieved, FET cycles are often canceled. Given this, we hypothesized that cycles with lower endometrial thickness had been cancelled. To explore this, cycle cancellation rates were examined to determine if differential cancellation rates could explain similarities in the final endometrial thickness reported to SART. The highest cancellation rates were seen in GC cycles (9.1% and 7.7% for those < 35 and ≥ 35 years, respectively) compared with NOD recipient cycles (5.4% and 7.0% for those < 35 and ≥ 35 years, respectively). The most common reason for cycle cancellation was “inadequate endometrial response” (defined at the discretion of the reporting clinic), which accounted for 64.2% and 60.0% of cancellations in younger and older GCs, and 60.3% and 50.4% of cancellations in young and older NOD recipients, respectively. In a multivariate logistic regression model including all NOD recipient cycles, there was an increased chance of cycle cancellation for inadequate endometrial response with increasing age (OR 1.026, 95% CI 1.018, 1.033), whereas in a multivariate logistic regression model including all GC cycles, there was a decreased chance of cancellation for endometrial response with increasing age (OR 0.976, 95% CI 0.966, 0.987). This may be due to the slight changes in endometrial thickness with age noted in the two groups (decreasing thickness with age in NOD recipients, increasing thickness with age in GC cycles). Clinical pregnancy rates were examined for cycles in which the transfer of a single good quality blastocyst (defined by SART as “an embryo free of or with only minor imperfection”) was performed. For the years covered by this SART dataset, the ploidy of the transferred embryo was not reported for FET cycles. We therefore restricted this analysis to transfers of good-quality single embryos. Pairwise assessment of clinical pregnancy rates revealed significant differences between GCs and NOD recipients, with NOD recipients having a significantly lower clinical pregnancy rate compared to GC controls in both age groups (younger NOD recipients vs younger GCs: 57.8% vs 68.0%, p < 0.001; older NOD recipients vs older GCs: 56.1% vs 64.6%, p < 0.001). Clinical pregnancy rates across age groups in NOD and GC cycles are shown in Fig.  2 . Fig. 2 Clinical pregnancy rate by age group in NOD and GC cycles. NOD = non-identified oocyte donor; GC = gestational carrier Clinical pregnancy rate by age group in NOD and GC cycles. NOD = non-identified oocyte donor; GC = gestational carrier To assess the impact of age on clinical pregnancy and live birth rates, logistic regression models stratified by GC status were generated. There was a decrease in clinical pregnancy rate with increasing age in GC (OR 0.98, 95% CI 0.97, 0.99) and NOD cycles (OR 0.99, 95% CI 0.98, 0.99). There was also a decreased live birth rate with increasing age in GC (OR 0.98, 95% CI 0.96, 0.995) and NOD cycles (OR 0.97, 95% CI 0.96, 0.98). To visually assess the impact of endometrial thickness on the clinical pregnancy rate, the clinical pregnancy rate was plotted against the endometrial thickness in both GC and NOD recipient cycles. In both groups, clinical pregnancy rate was found to increase with increasing endometrial thickness up to 8 mm. There was no change in clinical pregnancy rate with endometrial thicknesses of 8–18 mm (Fig.  3 ). Fig. 3 Clinical pregnancy rate by endometrial thickness in NOD and GC cycles. EM = endometrial; NOD = non-identified oocyte donor; GC = gestational carrier Clinical pregnancy rate by endometrial thickness in NOD and GC cycles. EM = endometrial; NOD = non-identified oocyte donor; GC = gestational carrier

This was a US population-based retrospective cohort study assessing FET cycles reported to SART between 2016 and 2020 to determine if there is a difference in maximal endometrial thickness achieved based on age. Approval was obtained from the Rutgers University Institutional Review Board as well as the SART Research Committee. Endometrial thickness was not recorded in SART prior to 2016; thus, cycles prior to 2016 were excluded. The data used for this study were obtained from the SART CORS. Data were collected through voluntary submission, verified by SART, and reported to the Centers for Disease Control and Prevention (CDC) in compliance with the Fertility Clinic Success Rate and Certification Act of 1992 (Public Law 102–493). SART maintains HIPAA-compliant business associate agreements with reporting clinics. In 2004, following a contract change with the CDC, SART gained access to the SART CORS data system for the purposes of conducting research. Over 90% of all assisted reproductive technology (ART) cycles in the United States are performed at SART-member clinics. SART annually selects up to 10 clinics, approximately 2.5% of SART clinics, for an on-site validation visit utilizing metrics and a blinded selection process to identify outlier clinics. Medical records are reviewed during the validation visit to verify the designation, outcome, and reporting of cycles. Clinics with significant systematic reporting errors undergo data correction. Six primary metrics and twenty-six secondary metrics are used for clinic selection. The metrics include low prospective reporting for both egg retrieval cycles and total cycles, high live birth rates in the various age groups, low cancellation rate, high percentage of total fertility preservation cycles, high percentage of embryo banking and oocyte banking cycles, high percentage of fertility preservation cycles where oocytes were thawed or embryos were transferred within a year, high percentage of deleted cycles, high percentage of cycles converted from IUI, and low percentage of cycles in which no embryos were suitable for transfer with and without PGT. SART does not validate the accuracy of data entry fields such as gonadotropin dosage, number of oocytes retrieved, number of fertilized oocytes, number of embryos cryopreserved, PGT results, or demographic fields such age and diagnosis. In order to assess the impact of age on maximal endometrial thickness and to isolate endometrial effects, NOD recipients aged < 35 and ≥ 35 years were compared. In order to control for the effects of infertility, gestational carriers < 35 and ≥ 35 years were included as the respective controls. The age cutoff of 35 years was chosen as it is a commonly used cutoff in studies of ovarian aging, as well as in studies evaluating obstetric risks and outcomes. For analyses of clinical pregnancy rate (CPR) and live birth rate (LBR), only cycles were included in which a high-quality blastocyst-stage single embryo transfer was performed. Age was also analyzed as a continuous variable to examine associations between age and endometrial thickness, cancellation rates, clinical pregnancy and live birth rates. In addition to patient age, maximal endometrial thickness and gestational carrier status, additional variables were collected: race and ethnicity, reason for ART, cycle cancellation rates and reasons for cancellation, and cycle outcomes including pregnancy outcomes. BMI is reported to SART only in non-GC cycles and was collected for the NOD recipients. The data were collected by cycle; while a single patient may be represented more than once in the dataset, the majority of patients are represented by 1 to 3 cycles. Suspected data entry errors were excluded, including endometrial thickness  30 mm and BMI  70. Cycles with a missing patient ID, missing patient or gestational carrier age, or where patient sex was listed as male were also excluded. Cycles in which the ART source was unknown and NOD recipient cycles in which autologous oocytes were used were excluded. Additionally, embryo banking cycles, donor embryo cycles, and cycles in which both donor and autologous oocytes were fertilized, were excluded. Chi-Square, Fisher Exact, or Kruskal-Wallis tests were used to compare measures across the four study groups, where appropriate. Post-hoc comparisons utilizing Chi-Square, Fisher Exact, and Wilcoxon Rank Sum tests were then run to further identify significantly different pairwise outcomes. Gestational carrier status stratified simple linear regression models were then used to look at the change in mean endometrial thickness by numeric age. Additionally, gestational carrier status stratified multivariable logistic regression models were used to examine the odds of cycle cancellation, clinical pregnancy status, and resulting live birth status. Any two-sided p-value < 0.05 was considered statistically significant unless a post-hoc Bonferroni correction for multiple comparisons was needed. All statistical analyses were completed in SAS version 9.4.

The overarching goal of this study was to evaluate the impact of age on maximum endometrial thickness achieved in an FET cycle and to correlate that with clinical outcomes in two distinct groups, oocyte donor recipients and gestational carriers. To our knowledge, this is the first study to compare outcomes in these groups to isolate the effects of age on the endometrium (NOD recipients) and to control for the effects of infertility (GC). Our principal finding is that pregnancy rates are significantly lower in NOD recipients compared to GC controls and are lower in older compared to younger women in both groups, despite very minimal differences in endometrial thickness across groups. Additionally, we have found that most FETs are performed within a narrow range of endometrial thicknesses (8–18 mm) with high cancellation rates owing mostly to insufficient thickness. While our data suggest a weak association between age and endometrial thickness, as well as endometrial thickness and pregnancy outcomes, we found a significantly higher clinical pregnancy rate in GCs compared to NOD recipients in both the young and older age groups, as well as a decreasing clinical pregnancy rate with age. These data support the conclusions of prior studies regarding the challenge of using endometrial thickness alone to predict endometrial receptivity and embryo implantation, and highlight the importance of factors other than endometrial thickness in predicting embryo transfer success. In particular, our study highlights the difference in pregnancy rates between fertile women and infertile women in which a uterine or endometrial factor was not clinically suspected. We found that despite very slight changes in endometrial thickness with age (less than half of a millimeter), pregnancy rates decline substantially with advancing age and are significantly lower in women with infertility compared to fertile controls, thus implicating an endometrial factor that mediates the chance of pregnancy and cannot be assessed by endometrial thickness alone. We also observed that, despite the limitations in the use of endometrial thickness to predict endometrial receptivity, embryo transfer cycles are nevertheless tightly regulated with a low threshold for cancellation when endometrial thickness goals are not met. Given the higher rates of cancellation in GC cycles, there appears to be even more stringency applied to these cycles in terms of achieving “optimal” endometrial thickness, which may stem from the higher costs associated with the use of a GC. Until recently, studies of reproductive aging have focused primarily on ovarian and oocyte aging. There is a well-established fund of literature surrounding the decline in oocyte quantity and quality with age [ 20 – 23 ] which has led to fertility preservation solutions such as oocyte and ovarian tissue cryopreservation [ 24 ] as well as novel investigations into the possibilities of generating germ cells from stem cells [ 25 ]. While it is has long been understood that the pace of reproductive aging is dictated primarily by ovarian function [ 20 ], there is a growing appreciation for the contribution of the endometrium to the age-related decline in fertility [ 26 ]. Recent retrospective studies have observed a lower pregnancy rate following euploid embryo transfer in women of advanced reproductive age [ 27 , 28 ]. A systematic review and meta-analysis including 7 large studies found a decline in ART success with increasing maternal age following euploid embryo transfer [ 27 ], and an examination of the SART database for live birth rates in cycles using oocyte donors from different sources found a decline in live birth in recipients aged 40 and older [ 28 ]. Additionally, despite the widespread use of endometrial thickness to predict endometrial receptivity in FET cycles, data supporting this practice is conflicting [ 29 ]. A recent critical appraisal of studies assessing endometrial thickness and embryo transfer outcomes highlights important limitations of this assessment of endometrial receptivity including the accuracy and reproducibility of ultrasound measurements, the quality of ultrasound equipment, the arbitrary nature of the 7 mm cutoff, the differences between analyses performed in continuous vs. categorical terms, and the retrospective nature of many of these studies [ 30 ]. Attempts to study the impact of endometrial thickness on pregnancy outcomes are often complicated by “cancellation bias”—pregnancy rates at endometrial thicknesses < 7 mm are not well studied due to the high likelihood of cancellation when the thickness is perceived to be inadequate. Transfers performed in this range often reflect a decision to proceed with an inadequate endometrium when attempts to reach greater thickness have not been successful. A recent publication [ 31 ] utilizing the SART CORS dataset notes that among the > 244,000 cycles included in the study, only 1.1% were performed with an endometrial thickness < 6 mm, and only 2.9% with a thickness of 6–7 mm. Although this study found an increase in the odds of live birth with increasing endometrial thickness up to 9 mm, this study too is unable to account for cancelled cycles. Given the relative dearth of data in the lower ranges of endometrial thickness and the fact that transfers performed in this range often reflect a decision to proceed under conditions perceived to be suboptimal, the association with pregnancy outcomes in these ranges are difficult to interpret. Though some evidence supports the use of endometrial thickness to guide embryo transfers, with recommendations to avoid embryo transfers when the endometrium has not achieved sufficient thickness [ 15 , 32 ], other studies question the utility of endometrial thickness in guiding cycle management and determining timing of an embryo transfer [ 33 ]. Given both the decline in pregnancy rate with age following euploid embryo transfers, and the possible association of endometrial thickness and pregnancy outcomes, we sought to elucidate whether age affects maximal endometrial thickness achieved in an FET cycle. Our goal was to understand whether endometrial thickness mediates the observed decline in pregnancy rates with age. We found minimal effect of age on endometrial thickness, as well as little association between endometrial thickness and pregnancy outcomes; however, we found significant differences in pregnancy rates when comparing presumably fertile gestational carriers to presumably infertile oocyte donor recipients, as well as when comparing younger to older women in both groups. These findings strongly suggest an optimal endometrial microenvironment in younger fertile women that is not captured by measurements of thickness alone, and implicate an adverse effect of age on endometrial function. Our study has many strengths, including the large number of cycles analyzed in a database that includes clinics across the country with a variety of providers, protocols and practice patterns. Although prior studies have evaluated the impact of age on FET cycle outcomes, the principal value of our study is in the use of a very large national database and the application of a novel comparison model utilizing GCs to control for the effects of infertility. Our study not only compares young and older oocyte donor recipients to study the effects of age on endometrial thickness but is also novel in comparing these two groups to gestational carriers to control for the effects of infertility. Some limitations in our study include the inherent differences, such as age distribution, between the NOD recipient group and the GC group. The NOD recipient group tended to be older and accounted for most of the cycles in our dataset, while the GC group tended to be younger and accounted for fewer cycles. For this reason, our analyses were stratified by GC status. Additionally, although “uterine factor” can be selected as an infertility diagnosis when entering data into the SART database, this is a broad category that may include both corrected and uncorrected factors. However, given the costs associated with GC and NOD cycles, it is highly likely that if there was any pre-existing uterine pathology it was corrected prior to embryo transfer. Although the strength of the SART database is the large number of available cycles from many different clinics, the limitation is the inability to assess data entry errors and account for outliers, as well as the lack of information on FET protocol. However, given the relatively widespread practice of programmed FET cycles during this study’s timeframe, it is likely that the majority of the cycles included were programmed FET cycles. There is also a lack of information about the age of the oocyte in all cycles included in our analyses. Additionally, for the years covered by this SART dataset, the ploidy of the transferred embryo was not reported for FET cycles; however we restricted our analyses to good quality, single embryo transfers. Finally, while the data was analyzed on a per-cycle basis, and a single patient may be represented by multiple cycles, most patients are represented by 1 to 3 cycles.

Together, our data highlight an age-related decline in pregnancy rates in groups in which neither significant uterine factor nor oocyte factor are anticipated, suggesting an important impact of endometrial function in the success of IVF. While this impact is smaller than the age-related decline in oocyte quality and quantity, our data support the need for further research into the mechanisms that may underlie an adverse impact of age on the endometrium.

Age is known to affect the success of assisted reproductive technology (ART) treatment as women with advanced maternal age (AMA) have been found to have lower pregnancy and live birth rates and higher miscarriage rates [ 1 ]. Significant research efforts have been directed at investigating the effects of aging on oocytes and reproductive outcomes as reproductive aging has largely been attributed to age-related decline in oocyte quality [ 2 , 3 ]. Despite improvements in understanding and assessing embryo quality, cycles using donor oocytes still only have a live birth rate of approximately 50% per transfer [ 4 ], suggesting that other factors are contributing to cycle outcomes. To date, the few studies that have examined the effect of aging on the endometrium have produced inconsistent results [ 5 – 7 ]. A recent analysis of endometrial gene expression demonstrated altered transcriptomic profiles in women 35 years and older [ 8 ]. Though this study was limited by sample size and inability to control for other patient characteristics, it suggests an important role for age-related changes within the endometrium in adversely affecting reproductive success. Therefore, the impact of endometrial aging on infertility warrants further investigation. In recent years, frozen embryo transfers (FET) have largely replaced fresh embryo transfers due to concern for adverse effects of ovarian stimulation on the endometrium and increase in utilization of preimplantation genetic testing (PGT) technology [ 9 , 10 ]. While our current method of assessing endometrial adequacy by measuring thickness in one dimension is crude and limited, it remains our only validated tool. In clinical practice, endometrial thickness is assessed periodically during an FET cycle with the goal of achieving an endometrial thickness of > 7–10 mm in order to optimize pregnancy outcomes [ 11 , 12 ]. Decreased endometrial thickness (< 8 mm) has been associated with reduced implantation and clinical pregnancy rates [ 13 – 15 ]. Furthermore, a thin endometrium is correlated with adverse neonatal and pregnancy outcomes such as lower birthweight, preterm delivery, placenta previa, and cesarean delivery [ 16 – 18 ]. It is largely unknown whether standard endometrial thickness cutoffs can be applied to older women, and whether older women are as likely as younger women to achieve an optimal endometrial thickness when undergoing ART. One study has evaluated the influence of female age on endometrial thickness, reporting a linear increase in ongoing pregnancy rate for every additional 1 mm of endometrial thickness achieved in women ≥35. However, this was a single-center retrospective study that included all women undergoing FET; as they did not limit their analysis to non-identified oocyte donor (NOD) recipients they were unable to isolate endometrial factor in their analysis [ 19 ]. In this study, we used the SART Clinic Outcome Reporting System (SART CORS) cycle data for 2016–2020 to evaluate the impact of age on maximum endometrial thickness achieved in an FET cycle and to correlate that with clinical outcomes. To isolate the effects of age on the endometrium, we compared young (< 35yo) and older (≥35yo) NOD recipients, and to control for the effects of infertility, young (< 35yo) and older (≥35yo) gestational carriers (GCs) served as the respective controls. We hypothesized that older women (NOD recipients and GCs) would achieve less endometrial thickness and would have a decreased rate of live birth.