Definition of thin endometrium from the perspective of miscarriage following frozen embryo transfer: a retrospective cohort study of 15,979 pregnancies.

A total of 15,979 women with positive serum β-hCG satisfied the inclusion criteria, among whom 548 (3.4%) had an EMT < 7 mm, and 1,933 (12.1%) had an EMT < 8 mm. All patient characteristics are given in Table  1 . Patients with normal endometrial thickness were younger and had a slightly longer infertility duration than those with thin endometrium. When 7 mm was used as the cut-off, TE cases were treated with more natural cycles and less hormone replacement therapy than controls (11.1% versus 7.5%, 70.1% versus 77.2%, respectively). The rate of two embryo transfers was higher in the TE group than in the non-TE group (25.9% versus 20.3%, P  = 0.001). At a cut-off of 8 mm, there was no significant difference in FET preparation protocol, type of embryo transfer, or number of embryo transfers. Table 1 Patients’ general characteristics stratified by endometrial thickness in frozen embryo transfer Variable EMT<7 mm EMT ≥ 7 mm P EMT<8 mm EMT ≥ 8 mm P No. of cycles 548 15,431 1933 14,046 Female age (y) 33(6) 31(5) < 0.001* 32(6) 31(5) < 0.001* Antral follicle count 12(12) 13(12) 0.072 12(11) 13(12) 0.004* Baseline FSH (mIU/ml) 7.32(2.21) 7.18(2.42) 0.136 7.25(2.43) 7.18(2.41) 0.131 AMH (ng/ml) 3.50(4.52) 3.84(4.54) 0.093 3.56(4.24) 3.86(4.61) 0.006* Body mass index (kg/m 2 ) 21.48(3.45) 21.48(3.92) 0.583 21.45(3.52) 21.48(3.99) 0.032* Duration of infertility (y) 2.0(3.0) 3.0(2.5) < 0.001* 2.0(3.0) 3.0(2.0) < 0.001* Infertility causes < 0.001* < 0.001*  Tubal disease 84(15.3%) 3256(21.1%) 394(20.4%) 2950(21.0%)  Ovulation disorders 19(3.5%) 633(4.1%) 54(2.8%) 590(4.2%)  Male factor 38(6.9%) 2037(13.2%) 168(8.7%) 1910(13.6%)  Unexplained 8(1.5%) 772(5.0%) 66(3.4%) 716(5.1%)  Mixed 318(58.0%) 7020(45.5%) 990(51.2%) 6335(45.1%)  Others 81(14.8%) 1713(11.1%) 261(13.5%) 1545(11.0%) FET preparation protocol < 0.001* 0.332  HRT 384(70.1%) 11,920(77.2%) 1467(75.9%) 10,837(77.2%)  NC 61(11.1%) 1156(7.5%) 162(8.4%) 1055(7.5%)  DR + HRT 103(18.8%) 2355(15.3%) 304(15.7%) 2154(15.3%) Type of embryo transfers 0.865 0.176  Cleavage embryo 74(13.5%) 2123(13.8%) 285(14.7%) 1912(13.6%)  Blastocyst 474(86.5%) 13,308(86.2%) 1648(85.3%) 12,134(86.4%) No. of embryo transfers 0.001* 0.089  1 406(74.1%) 12,294(79.7%) 1508(78.0%) 11,192(79.7%)  2 142(25.9%) 3137(20.3%) 425(22.0%) 2854(20.3%) EMT Endometrial thickness, FSH Follicle stimulating hormone, AMH Anti-Müllerian hormone, FET Frozen embryo transfer, HRT Hormone replacement therapy, NC Natural cycle, DR + HRT Downregulation + hormone replacement therapy * P values < 0.05. Continuous data were expressed as medians (interquartile ranges) Categorical data were reported as numbers (percentages) Patients’ general characteristics stratified by endometrial thickness in frozen embryo transfer EMT Endometrial thickness, FSH Follicle stimulating hormone, AMH Anti-Müllerian hormone, FET Frozen embryo transfer, HRT Hormone replacement therapy, NC Natural cycle, DR + HRT Downregulation + hormone replacement therapy * P values < 0.05. Continuous data were expressed as medians (interquartile ranges) Categorical data were reported as numbers (percentages) When 7 mm was used as the cut-off, among the 548 pregnant women with TE, a total of 109 (19.9%) experienced biochemical pregnancy loss, and 118 (21.5%) experienced early miscarriages, which were higher than those in the non-TE group (19.9% versus 12.6%, OR 1.721, 95% CI 1.388–2.133, P  < 0.001; 21.5% versus 15.6%, OR 1.480, 95% CI 1.202–1.823, P  < 0.001, respectively) (Table  2 ). The late pregnancy loss rates of the thin endometrium groups were similar to those of the control group, including the late pregnancy loss rates due to malformation (0.7% versus 0.7%, OR 1.073, 95% CI 0.394–2.924, P  = 0.890), spontaneous late miscarriage (1.8% versus 1.2%, OR 1.507, 95% CI 0.793–2.863, P  = 0.210), and stillbirth (0.0% versus 0.2%, P  = 0.310). The binary multiple logistic regression analysis showed a 59.2% increase in the biochemical pregnancy loss rate (aOR 1.592, 95% CI 1.260–2.012, P  < 0.001) and a 28.7% increase in the early miscarriage rate (aOR 1.287, 95% CI 1.018–1.626, P  = 0.035) in women with an EMT < 7 mm compared with an EMT ≥ 7 mm after adjusting for age, BMI, AFC, duration of infertility, infertility cause, FET preparation protocol, type of embryo transfers and number of embryo transfers (Table  3 ). Table 2 Pregnancy loss outcomes in the non-TE group and the TE group Variable EMT<7 mm EMT ≥ 7 mm P EMT<8 mm EMT ≥ 8 mm P No. of cycles 548 15,431 1933 14,046 Biochemical pregnancy loss 109(19.9%) 1946(12.6%) < 0.001* 321(16.6%) 1734(12.3%) < 0.001* EPL 118(21.5%) 2413(15.6%) < 0.001* 332(17.2%) 2199(15.7%) 0.086 LPL for malformation 4(0.7%) 105(0.7%) 0.890 18(0.9%) 91(0.6%) 0.156 Spontaneous LPL 10(1.8%) 188(1.2%) 0.207 31(1.6%) 167(1.2%) 0.122 Stillbirth 0(0.0%) 29(0.2%) 0.310 2(0.1%) 27(0.2%) 0.390 TE Thin endometrium, EMT Endometrial thickness, EPL Early pregnancy loss, LPL Late pregnancy loss * P values < 0.05 Categorical data are reported as numbers (percentages) Pregnancy loss outcomes in the non-TE group and the TE group TE Thin endometrium, EMT Endometrial thickness, EPL Early pregnancy loss, LPL Late pregnancy loss * P values < 0.05 Categorical data are reported as numbers (percentages) Table 3 Logistic regression of the endometrial thickness and pregnancy loss outcomes OR (95% CI) P aOR (95% CI) a P EMT ≥ 7 mm as REF  Biochemical pregnancy loss 1.721(1.388,2.133) < 0.001* 1.592(1.260,2.012) < 0.001*  EPL 1.480(1.202,1.823) < 0.001* 1.287(1.018,1.626) 0.035*  LPL for malformation 1.073(0.394,2.924) 0.890 1.043(0.327,3.326) 0.944  Spontaneous LPL 1.507(0.793,2.863) 0.210 1.309(0.606,2.827) 0.494 EMT ≥ 8 mm as REF  Biochemical pregnancy loss 1.414(1.242,1.610) < 0.001* 1.333(1.159,1.534) < 0.001*  EPL 1.117(0.984,1.268) 0.086 1.015(0.881,1.169) 0.841  LPL for malformation 1.441(0.867,2.395) 0.158 1.307(0.737,2.318) 0.360  Spontaneous LPL 1.355(0.921,1.993) 0.124 1.253(0.800,1.962) 0.325 OR Odds ratio, CI Confidence interval, aOR Adjusted odds ratio, REF Reference, EMT Endometrial thickness, EPL Early pregnancy loss, LPL Late pregnancy loss a: The multiple logistic regression model was adjusted for age, BMI, AFC, duration of infertility, infertility cause, FET preparation protocol, type of embryo transfers, and number of embryo transfers P values < 0.05 Logistic regression of the endometrial thickness and pregnancy loss outcomes OR Odds ratio, CI Confidence interval, aOR Adjusted odds ratio, REF Reference, EMT Endometrial thickness, EPL Early pregnancy loss, LPL Late pregnancy loss a: The multiple logistic regression model was adjusted for age, BMI, AFC, duration of infertility, infertility cause, FET preparation protocol, type of embryo transfers, and number of embryo transfers P values < 0.05 When 8 mm was used as the cut-off, women with thin endometrium still had higher odds of biochemical pregnancy loss (16.6% versus 12.3%, aOR 1.333, 95% CI 1.159–1.534, P  < 0.001), but a similar trend of early miscarriage (17.2% versus 15.7%, aOR 1.015, 95% CI 0.881–1.169, P  = 0.841) when compared with women in the non-TE group (Tables  2 and 3 ). The late miscarriage rates in the groups with thin endometrium were also not significantly different from those in the control group. Figure 2 reveals the probability of pregnancy (pregnancy probability = 1 - pregnancy loss rate) in the Kaplan-Meier survival curves at gestational weeks by different ranges of EMT, divided into three groups of  8 mm. We observed that the curve for EMT of 7–8 mm was much closer to that for EMT of > 8 mm, which was consistent with the previous data. The early pregnancy loss rate decreased from 21.5% at an EMT < 7 mm to 17.2% at an EMT < 8 mm, which resulted in reduced odds from 1.287 (95% CI 1.018–1.626, P  = 0.035) at an EMT < 7 mm to 1.015 (95% CI 0.881–1.169, P  = 0.841) at an EMT < 8 mm in multiple logistic regression analysis (Table  3 ). Therefore, an EMT of 7 mm seemed to be more appropriate to define TE from the perspective of miscarriage. Fig. 2 Kaplan-Meier survival curve for probabilities for pregnancy (pregnancy probability = 1 - pregnancy loss rate) at gestational weeks by different ranges of EMT, divided into three groups of  8 mm. EMT: endometrial thickness Kaplan-Meier survival curve for probabilities for pregnancy (pregnancy probability = 1 - pregnancy loss rate) at gestational weeks by different ranges of EMT, divided into three groups of  8 mm. EMT: endometrial thickness Among the 15,979 pregnancies included in this study, there were 9 cases with EMT less than 4 mm, 29 with EMT less than 5 mm, and 159 with EMT less than 6 mm, respectively. The number of cases with EMT less than 4 mm is too small to be analyzed. The miscarriage outcomes of the cutoffs of 5 mm and 6 mm are shown in the supplementary tables (Table S1 and Table S2), from which the thin endometrium group exhibited higher biochemical pregnancy loss and early pregnancy loss rates, similar to the group with EMT less than 7 mm. To evaluate adverse obstetric and neonatal complications, patients with singleton live births were selected, and their demographic data are displayed in Table  4 . When 7 mm was used as the cut-off, among 10,044 patients with singleton live births, those with thin endometrium had a nearly doubled risk of preterm birth (16.5% versus 8.6%, P  < 0.001) and low birth weight (10.0% versus 4.4%, P  < 0.001) compared with those without TE. There were no significant differences in the rates of hypertensive disorders of pregnancy, gestational diabetes mellitus, placenta previa, placental abruption, or macrosomia between the groups. For the cut-off of 8 mm, women with thin endometrium had a higher risk of preterm birth (11.1% versus 8.6%, P  = 0.005) and low birth weight (6.5% versus 4.3%, P  = 0.001) than women with EMT > 8 mm, while the other indicators were similar to those of the control group (Table  4 ). Table 4 Obstetric-neonatal outcomes between the non-TE group and TE group of singletons Variable EMT<7 mm EMT ≥ 7 mm P EMT<8 mm EMT ≥ 8 mm P No. of cycles 280 9764 1119 8925 Female age (y) 32(7) 31(5) < 0.001* 32(6) 31(5) < 0.001* Baseline FSH (mIU/ml) 7.34(2.19) 7.16(2.40) 0.179 7.24(2.37) 7.16(2.39) 0.212 Body mass index (kg/m 2 ) 21.23(3.41) 21.48(3.90) 0.314 21.23(3.61) 21.48(3.91) 0.015* Duration of infertility (y) 2.0(3.0) 3.0(2.5) 0.186 2.0(3.0) 3.0(2.5) 0.059 Type of embryo transfer 0.337 0.964  Cleavage embryo 38(13.6%) 1142(11.7%) 131(11.7%) 1049(11.8%)  Blastocyst 242(86.4%) 8622(88.3%) 988(88.3%) 7876(88.2%) No. of embryo transfers 0.012* 0.582  1 223(79.6%) 8306(85.1%) 944(84.4%) 7585(85.0%)  2 57(20.4%) 1458(14.9%) 175(15.6%) 1340(15.0%)  Gestational hypertension 14(5.0%) 293(3.0%) 0.054 38(3.4%) 269(3.0%) 0.475  Gestational diabetes mellitus 7(2.5%) 312(3.2%) 0.518 39(3.5%) 280(3.1%) 0.522  Placenta previa 6(2.2%) 220(2.3%) 0.908 27(2.4%) 199(2.2%) 0.688  Placental abruption 0(0.0%) 10(0.1%) 0.593 1(0.1%) 9(0.1%) 0.911  Preterm birth < 37 wk 46(16.5%) 841(8.6%) < 0.001* 124(11.1%) 763(8.6%) 0.005*  Low birthweight < 2,500 g 28(10.0%) 427(4.4%)  4,000 g 10(3.6%) 563(5.8%) 0.118 52(4.6%) 521(5.8%) 0.106 EMT Endometrial thickness, FSH Follicle stimulating hormone, FET Frozen embryo transfer, HRT Hormone replacement therapy, NC Natural cycle, DR + HRT Downregulation + hormone replacement therapy * P values < 0.05. Continuous data are expressed as medians (interquartile ranges) Categorical data are reported as numbers (percentages) Obstetric-neonatal outcomes between the non-TE group and TE group of singletons EMT Endometrial thickness, FSH Follicle stimulating hormone, FET Frozen embryo transfer, HRT Hormone replacement therapy, NC Natural cycle, DR + HRT Downregulation + hormone replacement therapy * P values < 0.05. Continuous data are expressed as medians (interquartile ranges) Categorical data are reported as numbers (percentages) Table  5 shows the binary multiple logistic regression analysis of obstetric and neonatal complications after adjusting for eight confounders. Compared with the control group, in the groups with thin endometrium, the adjusted odds ratio (aOR) for the rate of preterm birth increased from 1.281 (95% CI: 1.022–1.606) at an EMT < 8 mm to 2.050 (95% CI: 1.427–2.945) at an EMT < 7 mm, and that for the rate of low birth weight increased from 1.394 (95% CI: 1.041–1.867) to 2.488 (95% CI: 1.611–3.843). It is worth noting that the rate of HDP was slightly higher in the EMT < 7 mm group (aOR 1.843, 95% CI 1.002–3.387, P  = 0.049) after adjusting for confounders. Other indicators, including GDM, placenta previa, and macrosomia, in the groups with thin endometrium were similar to those in the control group, consistent with the results in Table  4 . Table 5 Odds ratios and adjusted odds ratios of singleton abnormal obstetric-neonatal outcomes OR (95% CI)  P aOR (95% CI) a   P EMT ≥ 7 mm as REF  Gestational hypertension 1.707(0.985,2.958) 0.057 1.843(1.002,3.387) 0.049*  Gestational diabetes mellitus 0.779(0.365,1.664) 0.519 0.276(0.068,1.120) 0.072  Placenta previa 0.953(0.420,2.164) 0.908 0.841(0.307,2.302) 0.736  Preterm birth < 37 wk 2.093(1.514,2.894) < 0.001* 2.050(1.427,2.945) < 0.001*  Low birthweight < 2,500 g 2.430(1.625,3.633) < 0.001* 2.488(1.611,3.843)  4,000 g 0.605(0.320,1.144) 0.122 0.576(0.268,1.235) 0.156 EMT ≥ 8 mm as REF  Gestational hypertension 1.134(0.803,1.602) 0.476 1.253(0.864,1.816) 0.234  Gestational diabetes mellitus 1.118(0.795,1.572) 0.522 0.902(0.592,1.377) 0.634  Placenta previa 1.087(0.724,1.633) 0.688 1.020(0.627,1.657) 0.938  Preterm birth < 37 wk 1.336(1.093,1.633) 0.005* 1.281(1.022,1.606) 0.032*  Low birthweight  4,000 g 0.786(0.587,1.053) 0.106 0.911(0.662,1.254) 0.568 OR Odds ratio, CI Confidence interval, aOR Adjusted odds ratio, REF Reference. a: The multiple logistic regression model was adjusted for age, BMI, AFC, duration of infertility, infertility cause, FET preparation protocol, type of embryo transfers, and number of embryo transfers. Odds ratios and adjusted odds ratios of singleton abnormal obstetric-neonatal outcomes OR Odds ratio, CI Confidence interval, aOR Adjusted odds ratio, REF Reference. a: The multiple logistic regression model was adjusted for age, BMI, AFC, duration of infertility, infertility cause, FET preparation protocol, type of embryo transfers, and number of embryo transfers. We further examined the pregnancy loss and obstetric-neonatal outcomes in patients with an endometrial thickness of 6–7 mm or 7–8 mm. As presented in Table S3, no statistically significant difference was observed in outcomes between the 6 mm ≤ EMT < 7 mm group and the 7 mm ≤ EMT < 8 mm group, despite higher biochemical pregnancy and early miscarriage rates in the 6 mm ≤ EMT < 7 mm group.

Records were reviewed for all women with a positive serum β-human chorionic gonadotropin (β-hCG) test after undergoing the first frozen-thaw embryo transfer in the Department of Reproductive Medicine Center at Tongji Hospital from January 2018 to December 2022. The exclusion criteria were as follows: the presence of known uterine abnormalities (e.g., uterine malformations, adenomyosis, untreated submucosal fibroids and endometrial polyps, and uterine effusion), PGT, recurrent spontaneous abortion, oocyte donation and incomplete data (e.g., null for endometrial thickness, number of embryo transfers and type of embryo transfers). Follow-up continued until July 2023 for all patients. We extracted data on demographic characteristics and clinical and laboratory data from electronic medical records. The study was performed in accordance with the Declaration of Helsinki and was approved by the Institutional Research Ethics Review Board of Tongji Hospital, Tongji Medical College, Huazhong University of Science and Technology (approval number TJ-IRB20230921). The requirement for informed consent was waived by the Institutional Research Ethics Review Board of Tongji Hospital because of the retrospective nature of the study. Figure  1 exhibits the flowchart of patient inclusion and clinical outcome of thin endometrium while defined by the EMT cutoff of 8 mm. Fig. 1 Flow chart of the patient inclusion and clinical outcome of thin endometrium defined by the EMT cutoff of 8 mm. FET, frozen embryo transfer. PGT: preimplantation genetic test. RSA: recurrent spontaneous abortion. β-hCG: β-human chorionic gonadotropin. EMT: endometrial thickness. TE: thin endometrium Flow chart of the patient inclusion and clinical outcome of thin endometrium defined by the EMT cutoff of 8 mm. FET, frozen embryo transfer. PGT: preimplantation genetic test. RSA: recurrent spontaneous abortion. β-hCG: β-human chorionic gonadotropin. EMT: endometrial thickness. TE: thin endometrium Conventional FET preparation protocols include natural cycle (NC), hormone replacement therapy (HRT), and downregulation before hormone replacement therapy (DR + HRT). Complete descriptions of the specific endometrial preparation protocols, luteal support, and laboratory procedures were reported previously [ 14 ]. In brief, in NC, the timing of progesterone supplementation was determined by mature follicle size and hormone levels. In HRT cycles, estradiol valerate tablets (PROGYNOVA, Bayer, Germany) were started on days 2–4 of menstruation, and endometrial development was monitored by transvaginal ultrasound. Progesterone supplementation was initiated when EMT reached ≥ 8 mm. For patients with persistently thin endometrium after extended estrogen administration, progesterone was initiated when a triple-line endometrial pattern was present and EMT was close to the patient’s previous trigger-day EMT or had reached an individual plateau. EMT was recorded on the day of progesterone initiation and used for grouping in the present analysis; therefore, women with EMT < 8 mm or < 7 mm were not excluded solely because EMT remained below these thresholds. DR + HRT was defined as the administration of 3.75 mg GnRH-a (leuprorelin acetate [BEIYI, Lizhu, China]) on the second day of menstruation, followed by the initiation of oral estrogen 28 days later. The standard luteal phase support protocol was continued until ten weeks’ gestation. Endometrial thickness was measured by transvaginal ultrasound in the midsagittal plane of the uterus on the day of progesterone supplementation as the maximum distance between the two endometrial-myometrial junction interfaces. Patients were permitted to transfer only if the echogenic pattern of the endometrium was determined as a triple-line pattern, with a central hyperechoic line surrounded by two hypoechoic layers. The analysis mainly compared the definitions of thin endometrium (an EMT < 7 mm or < 8 mm) to determine which definition was more appropriate based on the risk of miscarriage. In addition, obstetric and neonatal complications were also included in this study, considering the importance of offspring safety. The sample size was calculated using R according to the difference in clinical pregnancy rate (37.7% versus 41.4%) and live birth rate (27.0% versus 30.0%) when the cut-off value was 8 mm, as published previously [ 15 ]. To provide a two-tailed significance level of 0.05 and a power of 80%, 2,048 participants were required in the EMT < 8 mm group. However, the number of cycles presented in Table 2 for each group corresponds to the number of β-hCG-positive cycles after FET in each group, which constitutes a subset of the total cycles. In this study, the EMT < 8 mm group had a total of 4,046 cycles, fulfilling the sample size requirement for statistical analysis. The primary outcomes were miscarriage rates, including the biochemical pregnancy loss rate, early miscarriage rate, and late miscarriage rate. Late miscarriages included miscarriages induced due to abnormal antenatal testing and spontaneous miscarriages. The secondary outcomes were the rates of obstetric and neonatal complications, including preterm birth, macrosomia, low birth weight, hypertensive disorders of pregnancy (HDP), gestational diabetes mellitus (GDM), placenta previa, and placental abruption. Pregnancy was defined as a β-hCG level > 10 IU/litre 12 days after blastocyst transfer or 14 days after embryo transfer. Clinical pregnancy was defined as the presence of a gestational sac in the uterus at the 6th-7th week of gestation. Biochemical pregnancy loss was defined as the failure to achieve clinical pregnancy. Early miscarriage was defined as miscarriage before 13 weeks of gestation after achieving clinical pregnancy. Late miscarriage was defined as a clinical pregnancy with miscarriage between 13 and 24 weeks of gestation. Late miscarriages were divided into two categories: miscarriages induced due to fetal structural malformations or chromosomal abnormalities detected on prenatal examination between 13 and 24 weeks of gestation and those for which the prenatal examination was normal but spontaneous miscarriage occurred in this period. Stillbirth was defined as intrauterine or postpartum death after 24 weeks of gestational age. Twin pregnant women were also included in the analysis due to the small number, i.e., if a patient had two sacs in the uterus at 6–7 weeks of gestation but ended up delivering only one baby, which was not considered as pregnancy loss. Ectopic pregnancies were also not considered to be miscarriages. Obstetric and neonatal complications were further analysed in patients with singleton live births. Preterm birth was defined as birth at 4000 g. Low birthweight was defined as a birthweight < 2500 g. The diagnostic criteria for GDM were based on the 75-g 2-hour oral glucose tolerance test between 24 and 28 weeks’ gestation. Under normal conditions, the blood glucose levels should be lower than 5.1 mmol/L before glucose intake, 10.0 mmol/L one hour after intake, and 8.5 mmol/L two hours after intake, respectively. Gestational diabetes mellitus was diagnosed if any of the measured blood glucose values met or exceeded these thresholds, provided that the pre-pregnancy blood glucose level was within the normal range [ 16 , 17 ]. HDP included gestational hypertension and preeclampsia with a systolic blood pressure ≥ 140 mm Hg and/or a diastolic blood pressure ≥ 90 mm Hg on two occasions at least 4 h apart after 20 weeks of gestation in a woman with a previously normal blood pressure [ 18 , 19 ]. Patients undergoing IVF-ET sought consultation in our reproductive department before 12 weeks of pregnancy and in the obstetrics department after 12 weeks of pregnancy. Therefore, we obtained examination reports for early miscarriage outcomes from electronic health records, and maternal-neonatal outcomes were followed up by telephone. In addition, we also collected paper versions of the Anomaly Scan report and archived them. All statistical analyses were performed using SPSS version 26.0 (IBM, New York, USA). The normal distribution of continuous parameters was tested by the Kolmogorov-Smirnov test. Continuous data that were not normally distributed were expressed as medians (interquartile ranges), and categorical data were shown as numbers (percentages). We divided all enrolled populations into TE and non-TE groups using 7 mm and 8 mm as EMT cutoffs. We compared baseline data, the likelihood of miscarriage, and obstetric-neonatal complications between the TE and non-TE groups, and adjusted for these outcomes using multivariate logistic analysis. First, baseline characteristics were compared; baseline data encompassed female age, body mass index (BMI), ovarian function indicators, duration of infertility, FET preparation protocol, number of embryo transfers, and type of embryo transfer. The Mann-Whitney U test was conducted to assess continuous variables, and categorical variables were analysed using Pearson’s chi-squared test or Fisher’s exact test. Binary logistic regression models were used to assess the risk of pregnancy loss and obstetric-neonatal complications, adjusting for baseline factors. The antral follicle count (AFC) was selected as an ovarian function indicator. FET preparation protocol, number, and type of embryo transfer were categorical variables with dummy variable conversion. The Kaplan-Meier method was used to evaluate pregnancy loss by gestational age according to different ranges of EMT. All tests were two-sided, with a significance level of 0.05.

In this large retrospective cohort study involving 15,979 pregnant women, from the perspectives of miscarriage and obstetric-neonatal complications, an EMT < 7 mm seemed to be a more appropriate definition of thin endometrium in patients undergoing frozen embryo transfer. The group with an EMT < 8 mm exhibited a risk of early miscarriage similar to that of the control group (17.2% versus 15.7%, P  = 0.086). However, the thin endometrium group (EMT < 7 mm) had significantly higher rates of early miscarriage, preterm birth, and low birth weight than the control group (EMT ≥ 7 mm). To the best of our knowledge, for the first time, we explored the definition of a thin endometrium in patients undergoing FET cycles from the unique perspectives of miscarriage and obstetric-neonatal complications. Moreover, we performed multiple logistic regression in serum β-hCG-positive patients. We found that patients with thin endometrium should continue to be monitored closely after achieving biochemical pregnancy; the risk of miscarriage in early pregnancy, as well as the significantly higher risk of preterm birth and low birth weight in the third trimester of pregnancy, may be clinically important, especially in patients with an EMT of < 7 mm. Understanding thin endometrium has been challenging and controversial recently [ 3 ]. Most studies found clinical pregnancy and live birth rates associated with an EMT < 7 mm [ 5 – 7 , 20 – 25 ]. However, there were some dissenting voices, probably owing to the small sample sizes, with 36 and 57 patients with an EMT < 7 mm, respectively [ 26 – 27 ]. How thin is too thin from the perspectives of spontaneous miscarriage and maternal-fetal complications? A thin endometrium increases the risk of miscarriage, and uterine evacuation for miscarriage may further damage the endometrium, worsening endometrial receptivity. In addition, consistent with our finding regarding the miscarriage rate, a recent study with the largest sample size of 96,000 patients found that the pregnancy loss rate was approximately 41% at an EMT of 7–11.9 mm, 46% at an EMT of 6–6.9 mm and 62.2% at an EMT < 6 mm, which suggested that the miscarriage rate starts to increase at an EMT below 7 mm [ 15 ]. Our study further analyzed pregnancy loss by adjusting for confounding factors such as age and BMI. In summary, compared with a cut-off of 8 mm, a cut-off of 7 mm better predicted an EMT that would allow embryos to develop throughout the first trimester. Evidence on endometrial thickness and IVF safety, including obstetric and neonatal outcomes, is still limited but increasing. Women with an EMT < 7 mm delivered two weeks earlier than those with an EMT ≥ 7 mm [ 28 ]. A recent meta-analysis that included 76,404 cycles from 19 studies found that thin endometrium was associated with higher risks of low birth weight, preterm birth, placental abruption, and gestational hypertension [ 29 ]. More recently, a retrospective study also reported associations between a thinner endometrium and adverse perinatal and obstetric-neonatal outcomes after FET [ 30 ]. The risks of placenta previa, gestational diabetes mellitus, and macrosomia in the thin endometrium group were comparable to those of the normal thickness group. The findings of this study were partially consistent with ours. Plausible explanations for the partial inconsistency are listed as follows. First, the definition of thin endometrium varied widely. We mainly compared cut-off values of 7 mm and 8 mm, but the meta-analysis involved studies with different diagnostic criteria with cut-off values including 7.5, 8.0, 9.0, 9.7, 10.0 mm, etc. Additionally, our patient cohort study only included patients undergoing frozen embryo transfer versus the meta-analysis that included patients undergoing fresh and frozen embryo transfer, while a high oestrogenic environment in fresh embryo transfer may result in a possible difference in endometrial receptivity after FET. Our investigation was designed only to identify patients with singleton live births, as obstetric and neonatal complications are higher in twin pregnant women than in singleton pregnant women, which also differed from the meta-analysis. The mechanisms underlying poor outcomes in patients with a thin endometrium remain unclear. One plausible pathway is that EMT < 7 mm may indicate a more severe defect in endometrial receptivity, vascularization, and stromal decidualization than EMT of 7–8 mm. Because decidualization and trophoblast invasion are closely linked to spiral artery remodeling, an initially thin and potentially functionally impaired endometrium may compromise early placental development and uteroplacental perfusion [ 31 , 32 ]. Consistent with this hypothesis, a recent study examining placental pathology revealed a significant reduction in placental thickness and an increased incidence of bilobed placentas in the thin endometrium group versus the control group, which was associated with an increased prevalence of perfusion-related disorders [ 33 ]. This pathway may help explain why EMT < 7 mm showed stronger associations with preterm birth and low birth weight than EMT 7–8 mm in our cohort. However, this interpretation remains speculative and requires confirmation in future prospective studies with placental and mechanistic assessments. For reproductive clinicians and their patients, it remains a troublesome issue and a difficult conversation about whether to perform embryo transfer with thin endometrium. Based on our findings, 7 mm may be a clinically useful cutoff. Compared with EMT < 8 mm, EMT < 7 mm identified patients with higher risks of early miscarriage and preterm birth. Compared with an EMT < 6 mm threshold, an EMT < 7 mm cutoff may avoid missing patients who still warrant close counselling and follow-up, because EMT < 6 mm accounted for only 1.0% of pregnancies in this cohort, whereas EMT < 7 mm accounted for 3.4%. This issue is clinically relevant because evidence suggests that FET, particularly HRT cycles, may increase the risk of HDP [ 34 , 35 ]. Nevertheless, not all studies have found a consistent association between endometrial thickness and clinical outcomes, including studies in which embryo transfer was continued without an EMT-based cancellation threshold [ 27 , 36 ]. Our research has some limitations. First, to further elucidate whether the cutoff value for thin endometrium should be defined as 7–8 mm, we examined the pregnancy loss rates in the 6–7 mm and 7–8 mm groups, following the grouping method outlined in a previously published study [ 37 ]. While the pregnancy loss rate was higher in the 6–7 mm group, no statistically significant difference was observed. Future prospective studies with larger sample sizes and more sophisticated methodologies are needed to yield more robust and conclusive findings. Second, this was a retrospective study conducted in a single medical centre; however, we included a large sample size and adjusted for a wide array of confounding factors in multivariate logistic regression to render the results relatively reliable. In addition, our findings may not be generalizable to patients undergoing intrauterine insemination, those with natural pregnancies, and those undergoing fresh IVF cycles. Finally, subjective EMT measurements inevitably led to measurement errors, but experienced sonographers completed all measurements in accordance with the standard operation protocol in ISO-9001 documentation.

From the point of view of miscarriage and obstetric-neonatal complications, given the widespread use of 7–8 mm EMT cut-offs, we found that an EMT < 7 mm seems to be a more appropriate cut-off for thin endometrium. The findings of this study could provide information to counsel thin endometrium patients undergoing FET. Given the retrospective nature of this study, prospective studies are warranted to confirm the conclusions.

During in vitro fertilization-embryo transfer (IVF-ET), embryo and endometrium are principal factors for achieving pregnancy and live birth. With the developed embryo culture and preimplantation genetic testing (PGT) techniques that make obtaining good-quality embryos possible, the endometrium becomes more important and optimal preparation of the uterine lining increases the live birth rate per transfer [ 1 ]. Endometrial thickness reflects endometrial functional status and predicts endometrial receptivity [ 2 ]. However, thin endometrium is still highly controversial, especially regarding the ambiguous definition [ 3 – 4 ]. Some studies have suggested that the risk of a poor IVF outcome is increased at an endometrial thickness less than 7 mm [ 5 – 7 ]; increased pregnancy complications have also been reported at an endometrial thickness below 8–10 mm [ 8 – 9 ]. Gallos et al. reported that an optimal threshold of 10 mm maximized live births and minimized pregnancy losses [ 10 ], but pregnancies have occurred in patients with an endometrial thickness as low as 4 mm [ 11 – 12 ]. It should be noted, however, that many clinicians define a thin endometrium as less than 7–8 mm based on clinical pregnancy and live birth rates. How thin is too thin from the perspectives of spontaneous miscarriage and maternal-fetal complications? The impact of miscarriage on thin endometrium is nonnegligible, especially managing by dilation and curettage, further making endometrium thinner and diminishing the success of the next cycle of embryo transfer [ 13 ]. To explore the definition of thin endometrium, we carried out a large retrospective cohort study on the risk of pregnancy loss and obstetric-neonatal complications based on endometrial thickness using 7 mm and 8 mm as the cut-offs in pregnant women following first frozen embryo transfer (FET) cycles.

Supplementary Material 1. Supplementary Material 1.