Association of pre-pregnancy abnormal glucose metabolism and insulin resistance with assisted reproductive technology outcomes and pregnancy complications in women with PCOS: a retrospective propensity score matching study

Table  1 presents the comparisons of baseline characteristics between the two groups before and after PSM. Marked baseline differences were noted between the two groups pre-matching: the study group had older female and male ages (both P  < 0.001), significantly higher BMI ( P  < 0.001), PLT count ( P  = 0.01), and FPG level ( P  < 0.001), as well as elevated FT3 concentration ( P  = 0.004) and greater endometrial thickness on the day of embryo transfer ( P  = 0.005) compared with the control group. Additionally, the study group had a significantly lower number of transferred embryos ( P  < 0.001), lower AMH level ( P  = 0.005), and reduced TgAb concentration ( P  = 0.007) than the control group. Table 1 Comparisons of the baseline characteristics between PCOS women with and without pre-pregnancy abnormal glucose metabolism/IR undergoing ART Characteristics Before PSM After PSM Control Group ( N  = 1398) Abnormal Glucose Metabolism/IR Group ( N  = 417) P value Control Group ( N  = 378) Abnormal Glucose Metabolism/IR Group ( N  = 378) P value Female age (year) 30.35 ± 3.59 31.17 ± 3.33 < 0.001 30.96 ± 3.59 31.08 ± 3.30 0.291 Male age (year) 32.24 ± 3.94 32.89 ± 3.89 < 0.001 32.99 ± 3.93 32.80 ± 3.91 0.746 Infertility duration (year) 3.93 ± 2.56 3.94 ± 2.57 0.99 4.11 ± 2.62 3.96 ± 2.61 0.354 BMI (kg/m 2 ) 23.54 ± 3.54 24.75 ± 3.72 < 0.001 24.46 ± 3.14 24.22 ± 3.26 0.43 AMH (ng/ml) 7.88 ± 4.12 7.18 ± 3.46 0.005 6.78 ± 3.16 7.30 ± 3.47 0.117 WBC count (×10 9 /L) 6.60 ± 1.79 6.75 ± 1.91 0.216 6.59 ± 1.71 6.65 ± 1.92 0.883 PLT count (×10 9 /L) 270.68 ± 62.16 279.39 ± 63.87 0.01 273.20 ± 60.41 277.22 ± 64.11 0.448 ESR (mm/h) 15.08 ± 11.03 15.47 ± 11.28 0.624 14.82 ± 10.16 14.76 ± 11.03 0.429 FPG (mmol/L) 4.97 ± 0.44 5.56 ± 1.38 < 0.001 4.96 ± 0.41 5.53 ± 1.35 < 0.001 FT3 (pg/mL) 3.51 ± 0.60 3.58 ± 0.64 0.004 3.46 ± 0.57 3.56 ± 0.63 0.001 FT4 (ng/dL) 2.44 ± 4.09 2.75 ± 5.64 0.088 2.45 ± 4.14 2.69 ± 5.68 0.281 TSH (IU/mL) 2.44 ± 1.35 2.60 ± 1.61 0.167 2.41 ± 1.32 2.60 ± 1.61 0.235 TgAb (IU/mL) 46.36 ± 101.52 34.11 ± 72.35 0.007 52.19 ± 103.70 36.02 ± 75.41 < 0.001 Anti-TPO (U/mL) 140.12 ± 320.95 135.45 ± 315.54 0.793 175.46 ± 380.68 137.52 ± 316.67 0.893 Endometrial thickness on the day of embryo transfer (mm) 10.53 ± 2.20 10.82 ± 2.09 0.005 10.80 ± 2.41 10.83 ± 2.12 0.539 Number of transferred embryos 1.58 ± 0.50 1.46 ± 0.50 < 0.001 1.57 ± 0.50 1.47 ± 0.50 0.004 Birth weight (g) 3047.4 ± 619.47 3023.58 ± 645.40 0.782 3044.85 ± 656.67 3032.58 ± 642.31 0.922 Infertility type (n%) Primary infertility 56.5% (790/1398) 54.7% (228/417) 0.508 60.1% (227/378) 54.5% (206/378) 0.123 Secondary infertility 43.5% (608/1398) 45.3% (189/417) 39.9% (151/378) 45.5% (172/378) Embryo transfer cycle type (n%) Fresh embryo transfer 25.8% (360/1398) 23% (96/417) 0.259 28.8% (109/378) 22% (83/378) 0.03 Frozen embryo transfer 74.2% (1038/1398) 77% (321/417) 71.2% (269/378) 78% (295/378) Embryo transfer type (n%) Cleavage embryo 30.3% (423/1397) 28.5% (119/417) 0.495 33.9% (128/378) 27.5% (104/378) 0.058 Blastocyst 69.7% (974/1397) 71.5% (298/417) 66.1% (250/378) 72.5% (274/378) The factors included in PSM: maternal age, infertility duration, infertility type, BMI, AMH, embryo transfer type, number of transferred embryos A 1:1 nearest neighbor matching method with a caliper width of 0.001 was used to match data between the two groups A nominal 2-sided probability value < 0.05 was considered to indicate statistical significance AMH Anti-mullerian hormone, Anti-TPO Anti-thyroid peroxidase, BMI Body mass index, ESR Erythrocyte sedimentation rate, FPG Fasting plasma glucose, FT3 Free triiodothyronine, FT4 Serum free thyroxine, IR Insulin resistance, PLT Platelet, PSM Propensity score matching, SD Standard deviations, TgAb Thyroglobulin antibody, TSH Thyrotropin, WBC White blood cell Comparisons of the baseline characteristics between PCOS women with and without pre-pregnancy abnormal glucose metabolism/IR undergoing ART The factors included in PSM: maternal age, infertility duration, infertility type, BMI, AMH, embryo transfer type, number of transferred embryos A 1:1 nearest neighbor matching method with a caliper width of 0.001 was used to match data between the two groups A nominal 2-sided probability value < 0.05 was considered to indicate statistical significance AMH Anti-mullerian hormone, Anti-TPO Anti-thyroid peroxidase, BMI Body mass index, ESR Erythrocyte sedimentation rate, FPG Fasting plasma glucose, FT3 Free triiodothyronine, FT4 Serum free thyroxine, IR Insulin resistance, PLT Platelet, PSM Propensity score matching, SD Standard deviations, TgAb Thyroglobulin antibody, TSH Thyrotropin, WBC White blood cell Table S1 shows the covariate balance assessment before and after PSM using SMD. After PSM, most baseline covariates were well balanced with SMD < 0.1. However, several variables, including FPG, FT3, TSH, TgAb, anti-TPO, AMH, number of transferred embryos, infertility type, embryo transfer cycle type, and embryo transfer type, remained imbalance (SMD ≥ 0.1). Among them, FPG ( P  < 0.001), FT3 ( P  = 0.001), number of embryos transferred ( P  = 0.004), and TgAb level ( P  < 0.001) remained significantly different post-matching. A significant difference in embryo transfer cycle type was also observed between the two groups after PSM ( P  = 0.03). Before PSM, the study group had significantly lower rates of biochemical pregnancy ( P  = 0.013), clinical pregnancy ( P  = 0.016), and live birth ( P  = 0.01) than the control group. After matching, the live birth rate remained significantly lower ( P  < 0.001) and the miscarriage rate was significantly higher ( P  < 0.001) in the study group compared with the control group. Detailed comparisons of pregnancy outcomes between the two groups are shown in Table  2 . Table 2 Pregnancy outcomes of PCOS women with and without pre-pregnancy abnormal glucose metabolism/IR undergoing ART Outcomes Before PSM After PSM Control Group ( N  = 1398) Abnormal Glucose Metabolism/IR Group ( N  = 417) P value Control Group ( N  = 378) Abnormal Glucose Metabolism/IR Group ( N  = 378) P value Biochemical pregnancy rate (n%) 73.4% (1026/1398) 67.1% (280/417) 0.013 70.4% (266/378) 66.4% (251/378) 0.241 Clinical pregnancy rate (n%) 62.4% (873/1398) 55.9% (233/417) 0.016 59.5% (225/378) 55.3% (209/378) 0.239 Pregnancy type (n%) Uterine pregnancy 97.5% (851/873) 98.7% (230/233) 0.184 98.2% (221/225) 98.6% (206/209) 0.172 Ectopic pregnancy 1.9% (17/873) 0.4% (1/233) 1.8% (4/225) 0.5% (1/209) Heterotopic pregnancy 0.6% (5/873) 0.9% (2/233) 0 1% (2/209) Miscarriage rate (n%) 30.5% (308/1009) 35.8% (100/279) 0.091 17.2% (45/262) 35.6% (89/250) < 0.001 Miscarriage type (n%) Early miscarriage 90.9% (280/308) 89% (89/100) 0.573 97.8% (44/45) 88.8% (79/89) 0.098 Late miscarriage 9.1% (28/308) 11% (11/100) 1.2% (1/45) 11.2% (10/89) Live birth rate (n%) 50.1% (701/1398) 42.9% (179/417) 0.01 57.4% (217/378) 42.6% (161/378) < 0.001 Preterm birth rate (n%) 26.1% (183/701) 27.9% (50/179) 0.621 26.3% (57/217) 27.3% (44/161) 0.818 Cesarean section rate (n%) 83.7% (586/700) 79.9% (143/179) 0.225 84.8% (184/217) 79.5% (128/161) 0.18 Twin pregnancy rate (n%) 23% (161/701) 16.9% (30/178) 0.077 21.2% (46/217) 15.6% (25/160) 0.171 ART Assisted reproductive technology, IR Insulin resistance, PCOS Polycystic ovary syndrome, PSM Propensity score matching Pregnancy outcomes of PCOS women with and without pre-pregnancy abnormal glucose metabolism/IR undergoing ART ART Assisted reproductive technology, IR Insulin resistance, PCOS Polycystic ovary syndrome, PSM Propensity score matching Table  3 shows that the study group had a significantly higher risk of placental abruption post-matching ( P  = 0.043). Pre-matching, the study group had a significantly higher risk of neonatal jaundice ( P  = 0.03). No significant inter-group differences were noted for other maternal and neonatal complications. During the follow-up, only 1 case (0.7%) of developmental and behavioral disorders was identified among the 153 children in the study group, for which reason no further statistical analyses were performed. Table 3 Comparison of the maternal and neonatal complications in pre-pregnancy abnormal glucose metabolism/IR group and control group Complications Before PSM After PSM Control Group ( N  = 1398) Abnormal Glucose Metabolism/IR Group ( N  = 417) P value Control Group ( N  = 378) Abnormal Glucose Metabolism/IR Group ( N  = 378) P value PIH (n%) 17.2% (91/530) 21.1% (31/147) 0.274 21.7% (35/161) 18.3% (24/131) 0.469 PE (n%) 1.5% (8/530) 3.4% (5/147) 0.169 1.9% (3/161) 2.3% (3/131) 1 Hyperemesis gravidarum (n%) 10.1% (54/532) 5.2% (8/153) 0.062 10.5% (17/162) 5.8% (8/137) 0.147 Premature rupture of fetal membranes (n%) 0.9% (5/532) 1.3% (2/153) 0.656 0 1.5% (2/137) 0.209 Placenta previa (n%) 2.6% (14/532) 0.7% (1/153) 0.211 0.6% (1/162) 0.7% (1/137) 1 Placental abruption (n%) 0.2% (1/532) 2.6% (4/153) 0.01 0 2.9% (4/137) 0.043 Polyhydramnios (n%) 0.6% (3/532) 0.7% (1/153) 1 1.2% (2/162) 0.7% (1/137) 1 Oligohydramnios (n%) 1.7% (9/532) 1.3% (2/153) 1 2.5% (4/162) 1.5% (2/137) 0.691 FGR (n%) 0.6% (3/532) 2% (3/153) 0.129 1.2% (2/162) 2.2% (3/137) 0.664 Postpartum hemorrhage (n%) 1.9% (10/532) 1.3% (2/153) 1 2.5% (4/162) 1.5% (2/137) 0.691 Neonatal hypothyroidism (n%) 0.8% (4/532) 0.7% (1/153) 1 0.6% (1/162) 0.7% (1/137) 1 Neonatal malformation (n%) 2.3% (12/532) 2.6% (4/153) 0.765 3.1% (5/162) 2.9% (4/137) 1 Neonatal jaundice (n%) 39.8% (212/532) 49.7% (76/153) 0.03 43.2% (70/162) 47.4% (65/137) 0.463 Neonatal infection (n%) 3.4% (18/532) 6.5% (10/153) 0.083 3.1% (5/162) 6.6% (9/137) 0.155 Neonatal hypoglycemia (n%) 1.3% (7/532) 2% (3/153) 0.701 1.2% (2/162) 2.2% (3/137) 0.664 NICU (n%) 29.7% (158/532) 35.3% (54/153) 0.187 34% (55/162) 34.3% (47/137) 0.948 Childhood obesity (n%) 1.7% (9/532) 1.3% (2/153) 1 1.2% (2/162) 1.5% (2/137) 1 Delayed physical development (n%) 3.4% (18/532) 2.6% (4/153) 0.797 3.7% (6/162) 2.9% (4/137) 0.759 Speech and language delay (n%) 0.8% (4/532) 2% (3/153) 0.19 1.9% (3/162) 2.2% (3/137) 1 FGR Fetal growth restriction, IR Insulin resistance, NICU Neonatal intensive care unit, PE Preeclampsia, PIH Pregnancy-induced hypertension, PSM Propensity score matching Comparison of the maternal and neonatal complications in pre-pregnancy abnormal glucose metabolism/IR group and control group FGR Fetal growth restriction, IR Insulin resistance, NICU Neonatal intensive care unit, PE Preeclampsia, PIH Pregnancy-induced hypertension, PSM Propensity score matching To further clarify the association between different pre-pregnancy glucose and insulin status and the risk of developing GDM, an exploratory subgroup analysis was performed. Patients were categorized into four groups: normal, pre-pregnancy diabetes, IGT/IFG, and IR. The results demonstrated marked inter-group differences after PSM, with the IGT/IFG group showing the highest risk of progressing to GDM (25.8% vs. 48.5% vs. 17.4%, P  = 0.003). Multivariable logistic regression was performed to identify the independent factors associated with pregnancy outcomes and complications in PCOS patients undergoing IVF/ICSI. Infertility duration (OR = 0.899, 95%CI [0.834,0.968], P  = 0.005) and pre-pregnancy abnormal glucose metabolism/IR (OR = 0.551, 95%CI [0.384,0.792], P  = 0.001) were independently negatively associated with live birth. In contrast, endometrial thickness on the day of embryo transfer (OR = 1.106, 95%CI [1.014,1.208], P  = 0.024) and the number of transferred embryos (OR = 1.704, 95%CI [1.147,2.532], P  = 0.008) were independently positively associated with live birth. Detailed logistic regression results for live birth are shown in Table  4 . Table 4 Logistic regression of factors associated with live birth Characteristics Before matching After matching Adjusted OR (95% CI) P value Adjusted OR (95% CI) P value Female age 0.968 (0.928,1.011) 0.145 0.969 (0.898,1.046) 0.424 Male age 1.008 (0.97,1.047) 0.69 1.005 (0.94,1.074) 0.883 Infertility type 0.845 (0.677,1.055) 0.137 1.059 (0.723,1.552) 0.767 Infertility duration 0.942 (0.902,0.984) 0.008 0.899 (0.834,0.968) 0.005 BMI 0.987 (0.957,1.017) 0.375 1.002 (0.947,1.061) 0.939 Endometrial thickness on the day of embryo transfer 1.073 (1.02,1.129) 0.007 1.106 (1.014,1.208) 0.024 Embryo transfer cycle type 1.208 (0.919,1.589) 0.176 1.522 (0.962,2.409) 0.073 Embryo transfer type 1.262 (0.959,1.662) 0.097 1.442 (0.913,2.278) 0.116 Number of transferred embryos 1.727 (1.366,2.182) 0 1.704 (1.147,2.532) 0.008 Abnormal glucose metabolism/IR before pregnancy 0.833 (0.647,1.072) 0.155 0.551 (0.384,0.792) 0.001 FT4 1.033 (1.008,1.059) 0.01 - - TgAb - - 1.002 (0.999,1.004) 0.139 BMI Body mass index, CI Confidence interval, FT4 Serum free thyroxine, IR Insulin resistance, OR Odds ratios, PSM Propensity score matching, TgAb Thyroglobulin antibody Adjusting for female age, male age, infertility type, infertility duration, BMI, endometrial thickness on the day of embryo transfer, embryo transfer cycle type, embryo transfer type, number of transferred embryos, abnormal glucose metabolism/IR before pregnancy, and TgAb after PSM Logistic regression of factors associated with live birth BMI Body mass index, CI Confidence interval, FT4 Serum free thyroxine, IR Insulin resistance, OR Odds ratios, PSM Propensity score matching, TgAb Thyroglobulin antibody Adjusting for female age, male age, infertility type, infertility duration, BMI, endometrial thickness on the day of embryo transfer, embryo transfer cycle type, embryo transfer type, number of transferred embryos, abnormal glucose metabolism/IR before pregnancy, and TgAb after PSM Female age was independently inversely associated with preterm birth risk (OR = 0.843, 95%CI [0.739,0.962], P  = 0.011), while GDM (OR = 2.789, 95%CI [1.425,5.458], P  = 0.003), FGR (OR = 11.977, 95%CI [1.034,138.711], P  = 0.047), and the number of embryos transferred (OR = 2.174, 95%CI [1.110,4.257], P  = 0.024) were independently positively associated with preterm birth risk. Male age was marginally associated with preterm birth (OR = 1.107, 95%CI [0.992,1.235], P  = 0.07). However, the wide 95% CI for FGR indicated considerable uncertainty in this estimate, most likely due to the small number of FGR cases in our cohort. Detailed logistic regression results for preterm birth are shown in Table  5 . Table 5 Logistic regression of risk factors associated with preterm birth Characteristics Before matching After matching Adjusted OR (95% CI) P value Adjusted OR (95% CI) P value Female age 0.92 (0.848,0.999) 0.047 0.843 (0.739,0.962) 0.011 Male age 1.029 (0.958,1.105) 0.43 1.107 (0.992,1.235) 0.07 Infertility type 0.809 (0.536,1.223) 0.316 0.865 (0.462,1.62) 0.651 Infertility duration 0.975 (0.888,1.07) 0.593 0.949 (0.819,1.099) 0.481 BMI 1.021 (0.958,1.087) 0.529 1.029 (0.936,1.13) 0.556 WBC 1.129 (1.013,1.259) 0.028 - - ESR 0.994 (0.975,1.014) 0.558 - - PIH 1.867 (1.134,3.076) 0.014 1.652 (0.79,3.453) 0.182 GDM 1.305 (0.812,2.096) 0.271 2.789 (1.425,5.458) 0.003 Premature rupture of fetal membranes* 5.598 (0.96,32.645) 0.056 - - Placental abruption* 4.796 (0.428,53.781) 0.204 - - FGR 13.453 (1.347,134.414) 0.027 11.977 (1.034,138.711) 0.047 Embryo transfer cycle type 1.29 (0.761,2.187) 0.344 2.358 (0.959,5.796) 0.062 Embryo transfer type 1.868 (1.106,3.157) 0.019 1.581 (0.704,3.546) 0.267 Number of transferred embryos 3.362 (2.134,5.297) 0 2.174 (1.11,4.257) 0.024 Abnormal glucose metabolism/IR before pregnancy 1.626 (1.002,2.637) 0.049 1.669 (0.894,3.115) 0.108 BMI Body mass index, CI Confidence interval, ESR Erythrocyte sedimentation rate, FGR Fetal growth restriction, GDM Gestational diabetes mellitus, IR Insulin resistance, OR Odds ratios, PIH Pregnancy-induced hypertension, PSM Propensity score matching, WBC White blood cell Adjusting for female age, male age, infertility type, infertility duration, PIH, GDM, premature rupture of fetal membranes, placental abruption, FGR, embryo transfer cycle type, embryo transfer type, number of transferred embryos, abnormal glucose metabolism/IR before pregnancy, and BMI after PSM *Due to the low event rate after matching, premature rupture of fetal membranes and placental abruption were not included in the multivariate regression analysis Logistic regression of risk factors associated with preterm birth BMI Body mass index, CI Confidence interval, ESR Erythrocyte sedimentation rate, FGR Fetal growth restriction, GDM Gestational diabetes mellitus, IR Insulin resistance, OR Odds ratios, PIH Pregnancy-induced hypertension, PSM Propensity score matching, WBC White blood cell Adjusting for female age, male age, infertility type, infertility duration, PIH, GDM, premature rupture of fetal membranes, placental abruption, FGR, embryo transfer cycle type, embryo transfer type, number of transferred embryos, abnormal glucose metabolism/IR before pregnancy, and BMI after PSM *Due to the low event rate after matching, premature rupture of fetal membranes and placental abruption were not included in the multivariate regression analysis Preterm birth was significantly associated with adverse neonatal and childhood outcomes (Table  6 ), including higher rates of NICU admission ( P  < 0.001), delayed physical development ( P  = 0.001), and childhood developmental and behavioral disorders ( P  = 0.024). Table 6 Association between preterm birth and complications in newborns and childhood Complications Term delivery Preterm birth ( N  = 512) ( N  = 173) P value Neonatal hypothyroidism (n%) 0.8% (4/512) 0.6% (1/173) 1 Neonatal malformation (n%) 2% (10/512) 3.5% (6/173) 0.253 Neonatal jaundice (n%) 41.2% (211/512) 44.5% (77/173) 0.447 Neonatal infection (n%) 3.9% (20/512) 4.6% (8/173) 0.68 Neonatal hypoglycemia (n%) 1.4% (7/512) 1.7% (3/173) 0.719 NICU (n%) 20.7% (106/512) 61.3% (106/173) < 0.001 Childhood obesity (n%) 2% (10/512) 0.6% (1/173) 0.307 Delayed physical development (n%) 2% (10/512) 6.9% (12/173) 0.001 Developmental and behavioral disorders (n%) 1.8% (9/512) 5.2% (9/173) 0.024 Speech and language delay (n%) 0.6% (3/512) 2.3% (4/173) 0.072 NICU Neonatal intensive care unit Association between preterm birth and complications in newborns and childhood NICU Neonatal intensive care unit For GDM, female age (OR = 1.305, 95%CI [1.074,1.585], P  = 0.007) and FPG level (OR = 1.865, 95%CI [1.014,3.430], P  = 0.045) were identified as independent positive associated indicators. For PIH, BMI (OR = 1.506, 95%CI [1.198,1.894], P  < 0.001), embryo transfer type (OR = 6.406, 95%CI [1.158,35.440], P  = 0.033), and FPG level (OR = 2.224, 95%CI [1.088,4.544], P  = 0.028) were independent positive associated indicators. Detailed logistic regression results for GDM and PIH are shown in Tables  7 and 8 , respectively. Table 7 Logistic regression of risk factors associated with GDM Characteristics Before matching After matching Adjusted OR (95% CI) P value Adjusted OR (95% CI) P value Female age 1.194 (1.061,1.344) 0.003 1.305 (1.074,1.585) 0.007 Male age 0.911 (0.822,1.01) 0.076 0.891 (0.763,1.04) 0.143 Infertility type 0.958 (0.548,1.676) 0.881 1.309 (0.575,2.979) 0.521 Infertility duration 0.815 (0.704,0.944) 0.006 0.933 (0.762,1.144) 0.507 BMI 1.055 (0.971,1.147) 0.205 1.1 (0.966,1.252) 0.151 Embryo transfer cycle type 1.095 (0.539,2.226) 0.802 0.769 (0.261,2.266) 0.634 Embryo transfer type 0.932 (0.447,1.943) 0.851 1.174 (0.357,3.861) 0.792 Number of transferred embryos 0.791 (0.439,1.424) 0.434 0.722 (0.298,1.75) 0.471 FPG 2.307 (1.395,3.814) 0.001 1.865 (1.014,3.43) 0.045 Anti-TPO 1.001 (1,1.001) 0.016 1.001 (1,1.002) 0.052 WBC 1.107 (0.952,1.288) 0.187 0.89 (0.707,1.12) 0.321 Anti-TPO Anti-thyroid peroxidase, BMI Body mass index, CI Confidence interval, FPG Fasting plasma glucose, GDM Gestational diabetes mellitus, OR Odds ratios, PSM Propensity score matching, WBC White blood cell Adjusting for female age, male age, infertility type, infertility duration, BMI, embryo transfer cycle type, Embryo transfer type, number of transferred embryos, FPG, anti-TPO, and WBC after PSM Logistic regression of risk factors associated with GDM Anti-TPO Anti-thyroid peroxidase, BMI Body mass index, CI Confidence interval, FPG Fasting plasma glucose, GDM Gestational diabetes mellitus, OR Odds ratios, PSM Propensity score matching, WBC White blood cell Adjusting for female age, male age, infertility type, infertility duration, BMI, embryo transfer cycle type, Embryo transfer type, number of transferred embryos, FPG, anti-TPO, and WBC after PSM Table 8 Logistic regression of risk factors associated with PIH Characteristics Before matching After matching Adjusted OR (95% CI) P value Adjusted OR (95% CI) P value Female age 1.062 (0.908,1.243) 0.449 0.899 (0.674,1.2) 0.471 Male age 0.979 (0.855,1.12) 0.756 0.971 (0.774,1.218) 0.8 Infertility type 0.755 (0.349,1.631) 0.474 0.663 (0.209,2.101) 0.485 Infertility duration 0.945 (0.794,1.124) 0.521 0.911 (0.689,1.205) 0.516 BMI 1.348 (1.18,1.539) 0 1.506 (1.198,1.894) 0 Embryo transfer cycle type 0.902 (0.353,2.3) 0.829 1.558 (0.347,7) 0.563 Embryo transfer type 2.239 (0.828,6.057) 0.112 6.406 (1.158,35.44) 0.033 Number of transferred embryos 2.187 (0.981,4.877) 0.056 1.965 (0.586,6.585) 0.274 FPG 1.179 (0.773,1.797) 0.445 2.224 (1.088,4.544) 0.028 Anti-TPO 1 (0.999,1.001) 0.464 1 (0.998,1.002) 0.88 WBC 1.184 (0.953,1.472) 0.127 1.086 (0.774,1.524) 0.633 PLT 0.997 (0.99,1.004) 0.385 0.996 (0.986,1.006) 0.433 Basal LH 0.988 (0.923,1.058) 0.732 1.001 (0.924,1.085) 0.974 Anti-TPO Anti-thyroid peroxidase, BMI Body mass index, CI Confidence interval, FPG Fasting plasma glucose, LH Luteinizing hormone OR Odds ratios, PIH Pregnancy-induced hypertension, PLT Platelet, WBC White blood cell, PSM Propensity score matching Adjusting for female age, male age, infertility type, infertility duration, BMI, embryo transfer cycle type, embryo transfer type, number of transferred embryos, FPG, anti-TPO, WBC, PLT, and basal LH after PSM Logistic regression of risk factors associated with PIH Anti-TPO Anti-thyroid peroxidase, BMI Body mass index, CI Confidence interval, FPG Fasting plasma glucose, LH Luteinizing hormone OR Odds ratios, PIH Pregnancy-induced hypertension, PLT Platelet, WBC White blood cell, PSM Propensity score matching Adjusting for female age, male age, infertility type, infertility duration, BMI, embryo transfer cycle type, embryo transfer type, number of transferred embryos, FPG, anti-TPO, WBC, PLT, and basal LH after PSM ROC curve analyses for GDM and PIH were performed in both the total cohort before PSM and the matched cohort after PSM. A binary logistic regression exploratory prediction model was constructed in both pre‑PSM and post‑PSM populations using female age, BMI, FPG, infertility duration, and anti-TPO to calculate individual predicted probability of GDM. Then, ROC curves were plotted based on the predicted probability. For prediction before PSM (Fig.  1 A), the prediction model yielded an AUC of 0.746 (95%CI [0.689, 0.803], P <0.001). Using the maximum Youden index (0.414), the optimal threshold for the combined predictive probability was determined as 0.18 specific to this cohort, with 78.9% sensitivity and 62.5% specificity. After PSM, the model achieved an AUC of 0.712 (95%CI [0.614, 0.809], P <0.001) for GDM prediction (Fig.  1 B). The optimal cutoff value of the combined predicted probability was 0.24, with a sensitivity of 60.0%, a specificity of 75.6%, and a Youden index of 0.356. All statistically derived thresholds are cohort-specific exploratory indicators rather than definitive diagnostic criteria, given the lack of external validation. Fig. 1 A Receiver operating characteristic (ROC) curve analysis for gestational diabetes mellitus (GDM) prediction by predicted probability before propensity score matching (PSM). Predictive variables included female age, body mass index (BMI), fasting plasma glucose (FPG), infertility duration, and anti-thyroid peroxidase (anti-TPO). The red diagonal line represents the reference line (AUC = 0.500), indicating no discriminatory ability. The blue line represents the prediction line, yielding an AUC of 0.746 (95% confidence interval: 0.689–0.803, P <0.001). Using the maximum Youden index (0.414), the optimal threshold for the combined predictive probability was determined as 0.18 specifically in this cohort, with 78.9% sensitivity and 62.5% specificity. B Receiver operating characteristic (ROC) curve analysis for gestational diabetes mellitus (GDM) prediction by predicted probability after propensity score matching (PSM). Predictive variables included female age, body mass index (BMI), fasting plasma glucose (FPG), infertility duration, and anti-thyroid peroxidase (anti-TPO). The red diagonal line represents the reference line (AUC = 0.500), indicating no discriminatory ability. The blue line represents the prediction line, yielding an AUC of 0.712 (95% confidence interval: 0.614–0.809, P <0.001). The optimal cutoff value of the combined predicted probability was 0.24, with a sensitivity of 60.0%, a specificity of 75.6%, and a Youden index of 0.356 A Receiver operating characteristic (ROC) curve analysis for gestational diabetes mellitus (GDM) prediction by predicted probability before propensity score matching (PSM). Predictive variables included female age, body mass index (BMI), fasting plasma glucose (FPG), infertility duration, and anti-thyroid peroxidase (anti-TPO). The red diagonal line represents the reference line (AUC = 0.500), indicating no discriminatory ability. The blue line represents the prediction line, yielding an AUC of 0.746 (95% confidence interval: 0.689–0.803, P <0.001). Using the maximum Youden index (0.414), the optimal threshold for the combined predictive probability was determined as 0.18 specifically in this cohort, with 78.9% sensitivity and 62.5% specificity. B Receiver operating characteristic (ROC) curve analysis for gestational diabetes mellitus (GDM) prediction by predicted probability after propensity score matching (PSM). Predictive variables included female age, body mass index (BMI), fasting plasma glucose (FPG), infertility duration, and anti-thyroid peroxidase (anti-TPO). The red diagonal line represents the reference line (AUC = 0.500), indicating no discriminatory ability. The blue line represents the prediction line, yielding an AUC of 0.712 (95% confidence interval: 0.614–0.809, P <0.001). The optimal cutoff value of the combined predicted probability was 0.24, with a sensitivity of 60.0%, a specificity of 75.6%, and a Youden index of 0.356 The predictive models for PIH incorporated female age, BMI, FPG, infertility duration, anti‑TPO, number of transferred embryos and embryo transfer type. For PIH prediction before PSM (Fig.  2 A), the model yielded an AUC of 0.741 (95%CI [0.676,0.807], P <0.001), with the statistically derived optimal cutoff value of the combined predicted probability being 0.17 (sensitivity 73.2%, specificity 64.9%, Youden index 0.381). After matching (Fig.  2 B), the combined model achieved an AUC of 0.711 (95%CI [0.606,0.815], P <0.001), and the optimal cutoff value was 0.22 (sensitivity 71.4%, specificity 69.9%, Youden index 0.413). Similar to the GDM model, these cutoff values for PIH are specific to this retrospective single-center cohort and require cautious interpretation. Fig. 2 A Receiver operating characteristic (ROC) curve analysis for pregnancy-induced hypertension (PIH) prediction by predicted probability before propensity score matching (PSM). Predictive variables included female age, body mass index (BMI), fasting plasma glucose (FPG), infertility duration, anti-thyroid peroxidase (anti-TPO), number of transferred embryos, and embryo transfer type. The red diagonal line represents the reference line (AUC = 0.500), indicating no discriminatory ability. The blue line represents the prediction line, yielding an AUC of 0.741 (95% confidence interval: 0.676–0.807, P <0.001), with the optimal cutoff value of the combined predicted probability being 0.17 (sensitivity 73.2%, specificity 64.9%, Youden index 0.381). B Receiver operating characteristic (ROC) curve analysis for pregnancy-induced hypertension (PIH) prediction by predicted probability after propensity score matching (PSM) . Predictive variables included female age, body mass index (BMI), fasting plasma glucose (FPG), infertility duration, anti-thyroid peroxidase (anti-TPO), number of transferred embryos, and embryo transfer type. The red diagonal line represents the reference line (AUC = 0.500), indicating no discriminatory ability. The blue line represents the prediction line, yielding an AUC of 0.711 (95% confidence interval: 0.606–0.815, P <0.001), and the optimal cutoff value was 0.22 (sensitivity 71.4%, specificity 69.9%, Youden index 0.413) A Receiver operating characteristic (ROC) curve analysis for pregnancy-induced hypertension (PIH) prediction by predicted probability before propensity score matching (PSM). Predictive variables included female age, body mass index (BMI), fasting plasma glucose (FPG), infertility duration, anti-thyroid peroxidase (anti-TPO), number of transferred embryos, and embryo transfer type. The red diagonal line represents the reference line (AUC = 0.500), indicating no discriminatory ability. The blue line represents the prediction line, yielding an AUC of 0.741 (95% confidence interval: 0.676–0.807, P <0.001), with the optimal cutoff value of the combined predicted probability being 0.17 (sensitivity 73.2%, specificity 64.9%, Youden index 0.381). B Receiver operating characteristic (ROC) curve analysis for pregnancy-induced hypertension (PIH) prediction by predicted probability after propensity score matching (PSM) . Predictive variables included female age, body mass index (BMI), fasting plasma glucose (FPG), infertility duration, anti-thyroid peroxidase (anti-TPO), number of transferred embryos, and embryo transfer type. The red diagonal line represents the reference line (AUC = 0.500), indicating no discriminatory ability. The blue line represents the prediction line, yielding an AUC of 0.711 (95% confidence interval: 0.606–0.815, P <0.001), and the optimal cutoff value was 0.22 (sensitivity 71.4%, specificity 69.9%, Youden index 0.413) Overall, these findings suggest that the combined assessment of the identified factors may have moderate exploratory predictive value for GDM and PIH. However, given the retrospective nature of this study, these models and their corresponding thresholds should be considered preliminary exploratory tools and require external validation in prospective cohorts before clinical application.

A total of 417 PCOS patients with pre-pregnancy abnormal glucose metabolism or IR and 1398 patients without either condition were enrolled in the study, all of whom underwent IVF/ICSI and embryo transfer. After PSM, 378 patients were included in each group. We evaluated baseline characteristics, pregnancy outcomes, maternal and neonatal complications, influencing factors for key pregnancy outcomes, and high-risk subgroups for GDM and PIH before and after PSM.

This retrospective study included women with PCOS who underwent IVF/ICSI and embryo transfer at Renmin Hospital of Wuhan University from January 2019 to January 2025. The study was approved by the Ethics Committee of Renmin Hospital of Wuhan University and ethical exemption was granted due to the anonymous retrospective analysis of clinical data without individual intervention (2026ACCR-M089). PCOS was diagnosed according to the Rotterdam criteria: patients met at least two of the following three criteria, after excluding other endocrine causes: (1) Clinical and/or biochemical evidence of hyperandrogenism; (2) Polycystic ovarian morphology on transvaginal ultrasound; (3) Oligo-ovulation or anovulation [ 17 ]. Patients with the following conditions were excluded: (1) Uterine malformations; (2) Severe cerebrovascular, cardiac, hepatic, renal, or hematopoietic system diseases; (3) Other medical conditions contraindicating ART and/or pregnancy. Most patients underwent at least part of the following glucose metabolic screening prior to ART initiation: (1) Fasting plasma glucose (FPG); (2) 75 g oral glucose tolerance test (OGTT) with measurement of 2-hour plasma glucose; (3) Fasting insulin for the calculation of homeostasis model assessment of insulin resistance (HOMA-IR). The following clinical and laboratory variables were extracted from the electronic medical record system: female age, male age, infertility type, infertility duration, body mass index (BMI), anti-mullerian hormone (AMH), white blood cell (WBC) count, platelet (PLT) count, erythrocyte sedimentation rate (ESR), FPG, free triiodothyronine (FT3), free thyroxine (FT4), thyrotropin (TSH), thyroglobulin antibody (TgAb), anti-thyroid peroxidase (anti-TPO), endometrial thickness on the day of embryo transfer, embryo transfer cycle type, embryo transfer type, number of transferred embryos, pregnancy type, birth weight, and primary pregnancy outcomes. Primary pregnancy complications (PIH, PE, placental and membrane abnormalities, fetal growth restriction [FGR]), neonatal complications (congenital malformation, jaundice, neonatal intensive care unit [NICU] admission), and offspring childhood development were collected via questionnaire surveys and telephone follow-ups. Childhood outcomes were assessed by parental report rather than clinical examination, which may introduce potential reporting bias. The primary study outcome was live birth. Secondary outcomes included biochemical pregnancy, clinical pregnancy, miscarriage, preterm birth and pregnancy complications. The primary outcome was live birth rate. Based on local clinical data, the estimated live birth rate was 50% in PCOS patients without abnormal glucose metabolism or IR and 35% in those with abnormal glucose metabolism/IR. Using a two-sided test with α = 0.05 and 1 − β = 80%, and a 10% missing data adjustment, the minimum required sample size was 380 (190 per group). A total of 1815 eligible patients were enrolled from January 2019 to January 2025, which was far larger than calculated, ensuring adequate statistical power. Based on pre-pregnancy glucose and insulin status, patients were stratified into two groups: (1) Control group: no pre-existing abnormal glucose metabolism or IR; (2) Study group: presence of either abnormal glucose metabolism or IR. Abnormal glucose metabolism was defined as the presence of pre-pregnancy diabetes, IGT, or impaired fasting glucose (IFG). Pre-pregnancy diabetes was diagnosed by any of the following: FPG ≥ 7.0 mmol/L, or 2-hour OGTT ≥ 11.1 mmol/L, or glycated hemoglobin ≥ 6.5%, or random serum glucose ≥ 11.1 mmol/L with classic hyperglycemic symptoms or a hyperglycemic crisis. In this study, pre-pregnancy diabetes was diagnosed primarily based on FPG/OGTT results. In the absence of a prior diabetes diagnosis, IGT was defined as 2-hour OGTT 7.8–11.0 mmol/L, and IFG was defined as fasting plasma glucose 6.1–6.9 mmol/L with 2 h OGTT < 7.8 mmol/L. Normal glucose tolerance was defined as FPG < 6.1 mmol/L and 2-hour OGTT  1.78 [ 18 , 19 ]. Patients in the IR group were strictly screened to exclude those with glucose metabolic abnormalities (pre-pregnancy diabetes, IGT, IFG). All subgroups were mutually exclusive with no overlapping cases. According to the developmental stage of the embryo, embryo transfer type was categorized as cleavage-stage embryo (day 2–3 after fertilization) and blastocyst (day 5–6 after fertilization). Reproductive outcomes were defined as follows: Live birth was the delivery of a viable infant after 28 weeks of gestation, with the live birth rate calculated as the proportion of live birth cycles among all transfer cycles. Biochemical pregnancy was a positive serum beta human chorionic gonadotropin (β-hCG) test before ultrasound evidence of a gestational sac. Clinical pregnancy was confirmed by the presence of a gestational sac on transvaginal ultrasound. Miscarriage was defined as pregnancy loss before 28 weeks of gestation, further stratified into early miscarriage (< 12 weeks) and late miscarriage (12–28 weeks). Preterm birth was the live birth of an infant between 28 and 37 weeks of gestation. All statistical analyses were performed using SPSS 26.0 (IBM Corp., Armonk, NY, USA). Categorical variables were presented as frequencies and percentages, and continuous variables were described as means ± standard deviation (SD) for normally distributed data or medians (interquartile range) for non-normally distributed data. The chi-square test or Fisher’s exact test was used to compare categorical data between the two groups, while independent t-test/Mann‑Whitney U test was applied for continuous variables. PSM was used to balance baseline characteristics between the two groups, with the following covariates included: female age, infertility duration, infertility type, BMI, AMH, embryo transfer type, and number of transferred embryos. A 1:1 nearest neighbor matching method with a caliper width of 0.001 was adopted for PSM. Standardized mean difference (SMD) was used to evaluate the balance of covariates before and after PSM. An absolute SMD value < 0.1 indicated a well-balanced covariate. Logistic regression analyses were performed to identify the associated factors for live birth, preterm birth, GDM and PIH before and after PSM, with results presented as odds ratios (OR) and 95% confidence intervals (CI). A preliminary exploratory multivariate combined prediction model was conducted and receiver operating characteristic (ROC) curve was plotted to evaluate the overall predictive efficacy and discrimination ability of the model for GDM and PIH in PCOS patients undergoing ART. A two-sided P value < 0.05 was considered statistically significant. For post-hoc multiple comparisons following chi-square or Kruskal-Wallis tests, the statistical significance threshold was adjusted to P  < 0.017.

Polycystic ovary syndrome (PCOS) is one of the most common endocrine and metabolic disorders among women of childbearing age, with a prevalence rate of approximately 5% to 20% [ 1 ]. The core characteristics of PCOS include elevated androgen levels, ovulatory disorders, and polycystic ovarian morphology, with anovulation being the primary cause of PCOS-related infertility [ 2 ]. In addition to infertility, women with PCOS commonly present with insulin resistance (IR), hyperinsulinemia (HI), and obesity, which elevate their risk of metabolic and cardiovascular disorders including impaired glucose tolerance (IGT) and type 2 diabetes mellitus (T2DM) [ 3 – 5 ]. The widespread application of assisted reproductive technology (ART) has enabled an increasing number of PCOS patients to achieve pregnancy via in vitro fertilization (IVF) or intracytoplasmic sperm injection (ICSI). However, the inherent pathophysiological features of PCOS may increase perinatal risks for both mothers and offspring. This contributes to adverse pregnancy complications such as miscarriage, gestational diabetes mellitus (GDM), pregnancy-induced hypertension (PIH), preeclampsia (PE), preterm birth, and cesarean delivery, findings supported by numerous studies and meta-analyses [ 6 , 7 ]. Consistently, studies focusing on PCOS patients undergoing ART have confirmed that this population faces significantly higher risks of maternal complications such as GDM, PIH, and PE. Additionally, they are also at increased risk of adverse perinatal outcomes, including preterm birth, miscarriage, low birth weight, and neonatal complications [ 8 , 9 ]. Current evidence indicates that these adverse ART outcomes are linked not only to the underlying metabolic syndrome of PCOS but also to comorbidities including obesity, IR, and HI [ 10 – 12 ]. Notably, IR and subsequent metabolic abnormalities not only participate in the pathogenesis of PCOS but also impair reproductive function by affecting follicular development, oocyte quality, and endometrial receptivity, thereby increasing the risk of adverse pregnancy outcomes such as miscarriage [ 11 , 13 , 14 ]. Furthermore, the prevalence of IGT and T2DM in PCOS patients is significantly higher than in the general population. Even after weight matching, this risk remains approximately 2- to 4- fold higher [ 15 , 16 ], which is closely associated with adverse maternal and infant outcomes. However, the adverse effects of pre-pregnancy abnormal glucose metabolism and IR on long-term pregnancy complications in PCOS patients undergoing ART remain poorly elucidated, and whether such metabolic derangements potentially influence maternal and neonatal outcomes needs further cohort exploration. Therefore, this study aimed to explore the association of pre-pregnancy abnormal glucose metabolism and IR with ART outcomes and pregnancy complications in women with PCOS. Propensity score matching (PSM) was performed to improve baseline comparability, and a logistic regression model was established to evaluate the main factors correlated with live birth, preterm birth, GDM and PIH. Preliminary exploratory multivariate prediction models were conducted to evaluate the predictive efficacy and discrimination for GDM and PIH among patients with PCOS. These findings may provide exploratory references for risk stratification and assist in selecting PCOS patients who require closer antenatal monitoring to optimize maternal and neonatal outcomes.

This study highlights a significant association between pre-pregnancy metabolic abnormalities and adverse ART outcomes in PCOS patients. We identified independent risk factors and established prediction models for GDM and PIH with moderate discriminative ability. These findings underscore the importance of routine pre-ART metabolic screening and targeted interventions. However, given the retrospective nature of the study, our prediction models remain exploratory and warrant rigorous validation in large-scale prospective cohorts before clinical implementation.

In this retrospective PSM study of PCOS patients undergoing IVF/ICSI and embryo transfer, we explored statistical correlations between pre-pregnancy abnormal glucose metabolism/IR and ART outcomes, pregnancy complications, as well as neonatal and childhood developmental outcomes. After PSM to minimize confounding bias, our findings suggest that pre-pregnancy abnormal glucose metabolism/IR is correlated with relatively lower live birth rate, higher miscarriage rate, and elevated risk of placental abruption in PCOS women undergoing ART. We further identified associated factors for key pregnancy outcomes and established preliminary exploratory moderate predictive efficacy models for GDM and PIH, providing exploratory references for the management of PCOS patients undergoing ART, but not for definitive clinical decision-making. Our results demonstrated that PCOS patients with pre-pregnancy abnormal glucose metabolism/IR had a significantly lower live birth rate and a higher miscarriage rate than those without these metabolic abnormalities, which is consistent with the majority of previous studies. Chen et al. observed in a retrospective cohort of PCOS patients undergoing their first embryo transfer cycle that the early miscarriage rate increased from 7.14% to 16.22% and the live birth rate decreased from 63.03% to 47.88% with increasing HOMA-IR values [ 20 ]. A meta-analysis also indicated that high BMI (OR = 1.48, 95%CI [1.32,1.67]) and IR (MD = 0.32, 95%CI [0.15,0.49]) were associated with an increased risk of spontaneous abortion in PCOS patients undergoing ART [ 12 ]. By contrast, Luo et al. [ 11 ] reported no significant association between IR and live birth rate in PCOS patients. However, their study was retrospective and failed to adjust for confounding factors related to clinical interventions. These unadjusted factors may underlie the discrepant findings between their study and ours. Our study also found that endometrial thickness on the day of embryo transfer and the number of transferred embryos were independent positive associated indicators for live birth, while longer infertility duration was an independent negative associated indicator. Zhu et al. showed in a study of 1242 PCOS patients that female age, high BMI, infertility duration, number of transferred embryos, and embryo transfer type exert significant independent impacts on live birth in fresh embryo transfer cycles using antagonist protocols [ 21 ]. A retrospective analysis of 1920 PCOS patients undergoing fresh embryo transfer found that endometrial thickness on the trigger day, number of transferred embryos, and infertility duration were associated with live birth in univariate analysis, but only infertility duration remained significant in the multivariable model [ 22 ]. Another study of 1018 PCOS patients using the gonadotropin-releasing hormone antagonist protocol indicated endometrial thickness as a positive factor for live birth, but found no significant independent association between infertility duration and live birth [ 23 ]. These findings indicate that the relationship between infertility duration and live birth remains controversial, underlining the need for further large-scale prospective studies to validate its independent contribution and clarify potential causal mechanisms. GDM and PIH are the two most common pregnancy complications in PCOS patients undergoing ART. Our logistic regression analysis indicated that female age and FPG are independent associated indicators of GDM, while BMI, embryo transfer type, and FPG are independent associated indicators of PIH. These findings are consistent with previous research and further highlight the importance of pre-pregnancy metabolic optimization in this population as a potential preventive strategy. Kouhkan et al. identified higher BMI (25–29.9 kg/m², OR = 3.27; ≥30 kg/m², OR = 5.14) and elevated FPG (OR = 1.40) as significant risk associated indicators for GDM in a nested case-control study of 270 pregnant women undergoing ART, and the GDM group had a significantly higher mean female age than the non-GDM group [ 24 ]. Mustaniemi et al. also found in a multicenter case-control study that obesity (BMI ≥ 30 kg/m², OR = 5.14) and advanced maternal age (≥ 35 years, OR = 2.58) are associated with an increased risk of GDM in PCOS patients [ 25 ]. Li et al. demonstrated in a prospective cohort of 248 PCOS women that elevated early pregnancy FPG (≥ 4.86 mmol/L) is independently associated with a higher risk of subsequent GDM (OR = 2.80), further supporting the potential indicative value of FPG for GDM [ 26 ]. However, FPG screening alone may be insufficient to detect abnormal glucose metabolism in PCOS patients, and the OGTT remains the gold standard for diagnosis [ 27 , 28 ]. Therefore, a combined screening approach of FPG and OGTT is recommended for the detection of prediabetes in PCOS patients before ART. A positive correlation between BMI and PIH has been consistently demonstrated in numerous studies. Jia and colleagues found that PCOS women with GDM complicated by PIH had significantly higher BMI and FPG levels than healthy controls, and these variables were identified as independent associated indicator of both GDM and hypertensive disorders of pregnancy [ 29 ]. A meta-analysis by Palomba et al. reported that PCOS women have a 3–4 fold increased risk of PIH without BMI adjustment [ 30 ]. Our study identifies embryo transfer type as an independent associated indicator of PIH, a finding not previously reported specifically in PCOS populations. While a recent meta-analysis [ 31 ] noted a higher PIH risk in programmed frozen embryo transfer (FET) cycles compared to natural cycles (OR = 1.40)—specifically limited to blastocyst transfers (OR = 1.87)—no prior study has directly examined the association between transfer stage (cleavage vs. blastocyst) and PIH risk within the PCOS subgroup. Notably, although opinions remain inconsistent, most current high‑quality evidence shows no significant difference in PIH risk between cleavage‑stage and blastocyst transfer [ 32 , 33 ]. By contrast, FET, particularly artificial cycle FET without a functional corpus luteum, was strongly linked to an increased risk of PIH and PE [ 34 , 35 ]. The absence of corpus luteum–derived progesterone and relaxin may impair vascular adaptation and elevate PIH risk, particularly in metabolically vulnerable PCOS women [ 36 , 37 ]. In our study population after PSM, 86.1% of blastocyst transfers were FET cycles, among which 58.3% used artificial cycles and 26.5% used down‑regulated artificial cycles for endometrial preparation. Those may account for the substantially elevated risk of PIH observed with blastocyst transfer in our study. Our data suggest this association requires further validation. The identification of BMI, FPG, and embryo transfer type as associated indicators highlights the complex interplay between maternal metabolism and ART procedures in driving hypertensive disorders. For GDM prediction, the model included female age, BMI, FPG, infertility duration, and anti-TPO; for PIH, the model further incorporated the number and type of transferred embryos. All models showed moderate discriminative ability before and after PSM, with AUCs ranging from 0.71 to 0.75. Consistent evidence has linked advanced age, elevated BMI, and higher FPG levels to an increased risk of GDM, both in the general obstetric population and among women with PCOS. A previous study of 434 PCOS women established a multivariable logistic model for GDM with age, BMI, HbA1c, total cholesterol and family history, achieving an AUC of 0.937 and optimal cutoff of 0.186 (sensitivity 83.3%, specificity 92.3%) [ 38 ]. Another prospective multicenter pre-pregnancy cohort in 189 PCOS women established a multivariable logistic model incorporating first-degree relatives with T2DM, FPG, fasting insulin, and other metabolic factors, with an AUC of 0.87, supporting feasible and effective GDM prediction [ 39 ] in their respective cohorts. Zhou et al. identified that age ≥ 27 years, BMI ≥ 24 kg/m², family history of hypertension or hypertensive disorders complicating pregnancy (HDCP), adverse pregnancy history, and oral contraceptive use were independent risk associated indicators for HDCP in women with PCOS, with these factors showing good predictive value on ROC analysis ( n = 188) [ 40 ]. Furthermore, incorporating ART-related variables—specifically, the number and type of embryos transferred—into our PIH prediction model aligns with recent evidence. This evidence suggests that ART protocols and embryo transfer strategies may independently be associated with an increased risk of hypertensive and metabolic complications in PCOS pregnancies [ 8 , 41 ]. Collectively, these findings emphasize that precise pre-ART management—specifically targeting weight control, rigorous glucose screening (combined FPG/OGTT), and careful consideration of embryo transfer strategies—is essential to potentially reduce the risks of GDM and PIH in women with PCOS. Our regression analysis revealed an inverse correlation between female age and preterm birth risk. This contrasts with most previous studies, which show that advanced maternal age (especially ≥ 35 or ≥ 40 years) increases preterm birth risk [ 42 , 43 ]. A large cohort study of over 160,000 deliveries identified a U-shaped relationship between maternal age and preterm birth, with increased risk in women aged ≥ 40 years and younger women (20–24 years), and the lowest risk in the 30–34 years age group [ 44 ]. The median age of PCOS patients in our study was only 30 years, and the inverse association between age and preterm birth is likely attributable to the limited age range of our cohort, where the well-established risks of advanced maternal age have not yet emerged. Furthermore, the mean female age was 30.73 years in the double-embryo transfer group and 29.80 years in the twin delivery group, both lower than in the single-embryo transfer (31.34 years) and singleton delivery (30.91 years) groups. Preterm patients also showed a younger mean age compared to term birth patients (30.02 vs. 30.94 years). Therefore, the inverse association may be partly attributed to the higher prevalence of double‑embryo transfer and twin pregnancies in younger patients in our cohort, both of which are strong risk associated indicators for preterm birth [ 45 , 46 ]. However, there were also studies suggesting that a higher risk of preterm birth among women aged 30–34 years [ 47 , 48 ]. Inconsistencies in previous studies may also stem from differences in study design, population characteristics, and adjustment for confounding factors. Importantly, our study found that preterm birth is significantly associated with adverse neonatal and long-term childhood outcomes, including higher NICU admission rates, delayed physical development, and childhood developmental and behavioral disorders. Preterm birth is a well-recognized risk factor for offspring neurological development. Compared to full-term infants, preterm infants face higher rates of behavioral problems (e.g., attention deficit hyperactivity disorder, depression, autism) and greater impairments in language, cognitive, sensory, emotional, and motor domains [ 49 , 50 ]. These findings highlight the long-term developmental consequences of preterm delivery and the importance of implementing preventive strategies for preterm birth in high-risk PCOS pregnancies. Our findings have several clinical implications for the management of PCOS patients undergoing ART. First, despite PSM, residual confounding cannot be fully excluded due to incomplete baseline balance and retrospective design; thus, our results indicate an association rather than causation between pre-pregnancy metabolic abnormalities and adverse outcomes. Second, the prediction models and ROC-derived cutoffs are strictly exploratory and lack external validation. They should not be used as diagnostic criteria but may serve as theoretical references for risk stratification in research settings. Third, targeted pre-ART interventions (e.g., glucose control, weight management) are recommended to optimize metabolic status and reduce perinatal risks. Finally, given the link between preterm birth and long-term developmental risks, future prospective studies should include regular offspring follow-up for early intervention. This study has several limitations. First, despite applying PSM to balance baseline characteristics, residual differences remained, indicating incomplete balance. Combined with the retrospective design, this precludes causal inference; our findings demonstrate a statistical association between pre-pregnancy metabolic abnormalities and adverse outcomes, but cannot prove causality. Second, the single-center setting may limit the generalizability of our findings to other populations. Third, the small sample size for rare outcomes (e.g., FGR, placental abruption) resulted in wide 95% CIs and unstable estimates. Fourth, reliance on parental reports for childhood developmental data may introduce recall bias. Fifth, data on concurrent autoimmune diseases and endometriosis—known risk factors for adverse outcomes—were not systematically screened. Sixth, combining abnormal glucose metabolism and IR into a single exposure group may reduce the interpretability of our results. Large-scale prospective studies are warranted to validate these findings and explore the underlying mechanisms.

Supplementary Material 1. Supplementary Material 1. Supplementary Material 2. Supplementary Material 2.