Association of insulin resistance surrogates with live birth outcomes in women with polycystic ovary syndrome undergoing in vitro fertilization.

This was a single-center retrospective observational study conducted at the Center for Reproductive Medicine, Department of Obstetrics and Gynecology, The Second Hospital of Wenzhou Medical University, Wenzhou City, Zhejiang Province, China, during the period of January 2017 to December 2023 involving patients with PCOS who underwent IVF. The study was approved by the Independent Ethics Committee of the Second Affiliated Hospital and Yuying Children’s Hospital of Wenzhou Medical University (No. 2024-K-277-01). The inclusion criteria for this study were as follows: (1)The diagnosis of PCOS was based on the Rotterdam Criteria, which required that at least two of the following three criteria be met: oligomenorrhea or anovulation, clinical or biochemical hyperandrogenemia, and polycystic ovary ultrasound (defined as an ovarian sinus follicle count of ≥ 12 or an ovarian volume of ≥ 10 cm 3 ); and the exclusion of other causes of hyperandrogenemia and ovulatory dysfunction [ 17 , 18 ]; (2) The female patients were 20–40 years old; (3) Infertile couples undergoing IVF for the first time, ovarian stimulation was performed using gonadotrophin-releasing hormone (GnRH) agonist prolonged protocol. The exclusion criteria were as follows: (1) Males with moderate or severe semen abnormalities or fertilization by ICSI (patients whose sperm source was donor sperm were included in the study) [ 19 ]; (2) Patients with endocrine disorders (e.g., diabetes mellitus and thyroid dysfunction) or taking medications that may affect metabolism or plasma steroid levels; (3) Patients with serious factors affecting embryo implantation, such as uterine malformations, uterine adhesions and adenomyosis; (4) Patient with missing important information (e.g., information on fasting insulin, glucose levels, or pregnancy outcomes). Figure  1 was a flowchart of patients’ selection and exclusions. Fig. 1 Flow chart of patients’ selection and exclusions. IVF, in vitro fertilization; ICSI, intracytoplasmic sperm injection; HOMA-IR, homeostasis model assessment Flow chart of patients’ selection and exclusions. IVF, in vitro fertilization; ICSI, intracytoplasmic sperm injection; HOMA-IR, homeostasis model assessment Ovarian stimulation was performed using GnRH-agonist prolonged protocol. First, 3.75 mg of GnRH agonist was given to the patients on the second to fourth day of menstruation, and then the treatment was followed by luteal-phase ovarian stimulation 30 to 40 days later. Antral follicle size and serum levels of follicle-stimulating hormone (FSH), luteinizing hormone (LH), estradiol (E2), and progesterone (P) were taken into account when adjusting the stimulation. Furthermore, Controlled ovarian hyperstimulation (COH) was also performed using gonadotropins (Gn) such as recombinant FSH and human menopausal gonadotropin (hMG). In general, the starting dose of Gn ranged from 75 to 300 IU, depending on the patient’s age, ovarian function, and previous ovarian stimulation response. The doses of Gn were determined by the growth of ovarian follicles. The final oocyte maturation was triggered by 4000 to 10,000 IU of human chorionic gonadotropin (hCG) when at least two follicles had a mean diameter of 18 mm. Finally, transvaginal ovum retrieval was performed 34–36 h after hCG administration. As detailed in our previously published article [ 20 ], oocytes were fertilized by IVF after ovum retrieval, and normal fertilization was confirmed by the presence of two pronuclear (2PN) 16–20 h after fertilization. In fresh embryo transfer (ET) cycles, ET was performed 3 or 5 days after oocyte retrieval depending on the type of embryo, and all other embryos were frozen according to Gardner’s embryo grading system [ 21 ]. In frozen-thawed embryo transfer (FET) cycles, exogenous estradiol was used to establish an artificial menstrual cycle and FET was carried out 3 or 5 days after the addition of progesterone. Blood samples and physical examinations are collected at our reproductive center. All basal hormone blood samples including basal FSH, luteinizing hormone (LH), testosterone (T), prolactin (PRL), progesterone (P), estradiol (E2), and anti-Müllerian tubular hormone (AMH) were collected on days 2 to 4 of the menstrual cycle prior to medication, and BMI information was also collected at the same time. Serum glucose, insulin and triglyceride data were measured between 7:30 and 9:30 a.m. after a nighttime fast. Fasting blood glucose was measured by hexokinase assay (Beckman Coulter), triglycerides was determined by automatic biochemical analyzer (AU 5800, Beckman Coulter), and insulin, LH, FSH, PRL, P, E2, T was assessed by electrochemiluminescence immunoassay (Roche, Germany). AMH was measured by enzyme-linked immunosorbent assay (ELISA). The formulas for the different IR surrogates were calculated as follows [ 22 ]: HOMA-IR = fasting glucose (mmol/L) × fasting insulin (µU/mL)/22.5; TyG = Ln[fasting triglycerides (mg/dL) × fasting glucose (mg/dL)/2]; TyG-BMI = TyG index × BMI (kg/m 2 ). IR was defined according to the Homeostatic Model Assessment, defined as HOMA-IR ≥ 2.2 [ 23 ]. The primary outcome variable was the live birth rate, and secondary outcome variables included clinical pregnancy rate, miscarriage rate, neonatal birth weight, number of oocytes retrieved, and number of high-quality embryos. Day 3 (D3) high-quality embryos were generally defined as embryos derived from normally fertilized oocytes with at least 6 cells, cell size consistent with developmental stage, less than 20% fragmentation, and no multinucleation in D3 embryos after fertilization; Pregnancy status was determined by measuring HCG at 14 days post-transplantation. The presence of an intrauterine gestational sac, a yolk sac, fetal pole, and fetal heart pulsations within about four weeks of implantation was considered clinical pregnancy. Live birth was defined as the birth of a live infant at ≥ 24 weeks of gestation. For describing baseline characteristics, we categorized participants into “Non-IR group” (defined as HOMA-IR < 2.2) and “IR group” (defined as HOMA-IR ≥ 2.2). Continuous data were expressed as median (interquartile range), and differences between groups were tested using the Rank sum test; categorical data were expressed as n (%), and differences between groups were tested using the chi-square test or Fisher’s exact probability method. Regression model analysis was used to explore the association between the different IR surrogates and outcomes, using β/OR values and 95% confidence intervals (95% CI) indicating the relationship. In the analysis we applied three models, model I with adjusting none, model II adjusted female age and male age, and model III adjusted female age, male age, duration of infertility, basal LH, basal FSH, basal PRL, basal P, basal E2, basal T, AMH, endometrium thickness, duration of Gn, total dosage of Gn. We then used the restricted cubic spline (RCS) curve based on regression Model III to explore the nonlinear relationship between different IR surrogates and live birth rates in fresh ET cycles and FET cycles. Further, in sensitivity analyses we conducted interaction and stratified analysis according female age (≤ 30 and > 30 years), BMI status (normal or low weight, overweight and obesity) and AMH status (low and high). Lastly, receiver operating characteristic (ROC) curves and their respective areas under the curve (AUC) were used to compare the predictability of different IR surrogates on live birth rates in fresh ET cycles. The data in this study was processed and analyzed using R version 4.4.0. A p-value < 0.05 was considered statistically significant in all analyses.

A total of 543 PCOS patients were finally included in the study and divided into two groups: “Non-IR group” and “IR group”. The baseline characteristics of the two groups are presented in Table  1 . In the IR group, participants were more likely to have higher BMI, TyG, TyG-BMI, triglyceride, fasting glucose, insulin, and basal testosterone levels, as well as more duration and total amount of Gn, and additionally, were more likely to have lower basal LH, FSH, and AMH levels (all P  < 0.05). Table 1 Baseline characteristics according to different insulin resistance groups based on HOMA-IR index Characteristics Total ( n  = 543) Non-IR ( n  = 236) IR ( n  = 307) P -value Female age (years) 29.00 (27.00, 32.00) 29.00 (27.00, 32.00) 29.00 (27.00, 32.00) 0.507 Male age (years) 31.00 (29.00, 34.00) 31.00 (28.75, 34.00) 31.00 (29.00, 34.00) 0.195 Duration of infertility (years) 3.00 (2.00, 5.00) 3.00 (2.00, 4.00) 3.00 (2.00, 5.00) 0.048 BMI (kg/m 2 ) 23.42 (20.96, 26.30) 21.34 (19.53, 23.17) 25.24 (22.89, 27.34) < 0.001 TyG 8.50 (8.13, 8.94) 8.17 (7.93, 8.53) 8.76 (8.39, 9.11) < 0.001 TyG-BMI 202.02 (170.92, 229.33) 174.47 (156.37, 198.65) 220.83 (198.85, 245.67) < 0.001 Triglycerides (mmol/L) 1.20 (0.85, 1.87) 0.91 (0.74, 1.27) 1.55 (1.06, 2.25) < 0.001 Fasting blood glucose (mmol/L) 4.99 (4.74, 5.30) 4.83 (4.58, 5.10) 5.10 (4.86, 5.40) < 0.001 Insulin (uU/mL) 10.97 (7.22, 16.05) 6.60 (4.90, 8.29) 15.20 (12.20, 20.15) < 0.001 Basal LH (IU/L) 6.52 (3.86, 9.96) 6.86 (4.33, 10.97) 6.22 (3.68, 9.13) 0.015 Basal FSH (IU/L) 6.31 (5.33, 7.35) 6.45 (5.74, 7.58) 6.17 (5.18, 7.23) 0.009 Basal PRL (ng/ml) 10.52 (8.04, 14.47) 10.65 (8.08, 14.68) 10.42 (8.02, 14.22) 0.928 Basal P (ng/ml) 0.49 (0.35, 0.67) 0.49 (0.35, 0.67) 0.48 (0.34, 0.67) 0.499 Basal E2 (pg/ml) 42.30 (34.20, 53.40) 42.22 (35.67, 53.20) 42.60 (33.60, 53.84) 0.668 Basal T (ng/ml) 0.38 (0.28, 0.55) 0.36 (0.26, 0.51) 0.40 (0.29, 0.56) 0.022 AMH (ng/ml) 6.19 (4.21, 9.12) 6.53 (4.47, 10.19) 5.86 (3.99, 8.48) 0.032 Endometrium thickness (mm) 11.00 (9.80, 12.70) 11.20 (10.00, 13.00) 11.00 (9.80, 12.50) 0.189 Duration of Gn (days) 12.00 (10.00, 14.00) 11.00 (10.00, 13.00) 12.00 (10.00, 14.00) < 0.001 Total dosage of Gn (IU) 1813.00 (1375.00, 2419.00) 1656.50 (1200.00, 2025.00) 2025.00 (1575.00, 2700.00) < 0.001 Fresh embryo transfer 0.912 No 185 (34.97%) 81 (34.32%) 104 (33.88%) Yes 358 (65.93%) 155 (65.68%) 203 (66.12%) OHSS 0.882 No 527 (96.96%) 229 (97.03%) 299 (97.39%) Yes 16 (3.04%) 7 (2.97%) 8 (2.61%) Data are expressed as median (interquartile range) or n (%) Abbreviations: BMI, Body Mass Index; FSH, follicle-stimulating hormone; LH, luteinizing hormone; E2,estradiol; PRL, Prolactin; P, progesterone; T, testosterone; AMH, anti-Mullerian hormone; Gn, gonadotropin; HOMA-IR, homeostasis model assessment of insulin resistance; TyG, triglyceride glucose index; TyG-BMI, triglyceride glucose-body mass; OHSS, ovarian hyperstimulation syndrome Baseline characteristics according to different insulin resistance groups based on HOMA-IR index Data are expressed as median (interquartile range) or n (%) Abbreviations: BMI, Body Mass Index; FSH, follicle-stimulating hormone; LH, luteinizing hormone; E2,estradiol; PRL, Prolactin; P, progesterone; T, testosterone; AMH, anti-Mullerian hormone; Gn, gonadotropin; HOMA-IR, homeostasis model assessment of insulin resistance; TyG, triglyceride glucose index; TyG-BMI, triglyceride glucose-body mass; OHSS, ovarian hyperstimulation syndrome Among the participants, 358 fresh ET cycles were performed, resulting in the birth of 218 newborns, and another 409 FET cycles were performed, resulting in the birth of 242 newborns. Table  2 presents the correlation between different IR surrogates and outcomes in fresh ET cycles, all three IR surrogates showed stable negative correlations with live birth rate in all three regression models (in Model III: TyG-BMI OR = 0.99, 95% CI: 0.98 ~ 0.99; TyG OR = 0.47, 95% CI: 0.27 ~ 0.82; HOMA-IR OR = 0.84, 95% CI: 0.72 ~ 0.97; all P  < 0.05). Moreover, in the fully adjusted regression model III, TyG-BMI and TyG were also negatively associated with clinical pregnancy rate and the number of oocytes retrieved, but HOMA-IR was not associated with these (Table  2 ). In addition, the three IR surrogates did not correlate with the number of high-quality embryos, miscarriage rate, or neonatal birth weight when fully adjusted for confounders. Table 2 Associations between different insulin resistant surrogates and laboratory data and outcomes in fresh ET cycles Variable Model I β/OR(95%CI) P -value Model II β/OR(95%CI) P -value Model III β/OR(95%CI) P -value No. of oocytes retrieved TyG-BMI index -0.03 (-0.05 ~ -0.02) < 0.001 -0.03 (-0.05 ~ -0.01) < 0.001 -0.03 (-0.05 ~ -0.01) 0.016 TyG index -1.54 (-2.66 ~ -0.43) 0.007 -1.47 (-2.60 ~ -0.35) 0.010 -1.96 (-3.48 ~ -0.45) 0.011 HOMA-IR index -0.32 (-0.61 ~ -0.03) 0.032 -0.31 (-0.60 ~ -0.02) 0.039 -0.02 (-0.42 ~ 0.39) 0.941 No. of High-quality embryos TyG-BMI index -0.01 (-0.01 ~ -0.01) < 0.001 -0.01 (-0.02 ~ -0.01) < 0.001 -0.01 (-0.02 ~ 0.00) 0.052 TyG index -0.48 (-0.86 ~ -0.11) 0.012 -0.51 (-0.89 ~ -0.13) 0.009 -0.35 (-0.86 ~ 0.17) 0.186 HOMA-IR index -0.09 (-0.19 ~ 0.00) 0.062 -0.10 (-0.19 ~ 0.00) 0.057 -0.02 (-0.16 ~ 0.12) 0.769 Clinical pregnancy rate TyG-BMI index 0.99 (0.98 ~ 0.99) 0.001 0.99 (0.98 ~ 0.99) 0.002 0.99 (0.98 ~ 0.99) 0.014 TyG index 0.59 (0.39 ~ 0.89) 0.011 0.59 (0.39 ~ 0.90) 0.014 0.48 (0.27 ~ 0.86) 0.014 HOMA-IR index 0.95 (0.86 ~ 1.04) 0.278 0.95 (0.86 ~ 1.05) 0.305 0.91 (0.80 ~ 1.05) 0.206 Miscarriage rate TyG-BMI index 1.01 (1.01 ~ 1.02) 0.008 1.01 (1.01 ~ 1.02) 0.019 1.01 (0.99 ~ 1.02) 0.286 TyG index 2.39 (1.29 ~ 4.40) 0.005 2.27 (1.22 ~ 4.23) 0.010 1.94 (0.78 ~ 4.83) 0.152 HOMA-IR index 1.20 (1.06 ~ 1.35) 0.005 1.19 (1.06 ~ 1.35) 0.005 1.15 (0.97 ~ 1.36) 0.098 Live birth rate TyG-BMI index 0.99 (0.98 ~ 0.99) < 0.001 0.99 (0.98 ~ 0.99) < 0.001 0.99 (0.98 ~ 0.99) 0.010 TyG index 0.48 (0.33 ~ 0.72) < 0.001 0.50 (0.34 ~ 0.74) < 0.001 0.47 (0.27 ~ 0.82) 0.008 HOMA-IR index 0.85 (0.76 ~ 0.95) 0.003 0.85 (0.76 ~ 0.95) 0.005 0.84 (0.72 ~ 0.97) 0.019 Neonatal birth weight TyG-BMI index 1.55 (-0.48 ~ 3.57) 0.136 1.62 (-0.43 ~ 3.67) 0.122 -0.20 (-3.00 ~ 2.61) 0.890 TyG index -33.28 (-184.37 ~ 117.81) 0.666 -35.54 (-186.90 ~ 115.82) 0.646 -90.94 (-266.88 ~ 85.00) 0.313 HOMA-IR index 33.87 (-8.79 ~ 76.53) 0.121 32.17 (-10.68 ~ 75.01) 0.143 -7.94 (-56.93 ~ 41.05) 0.751 Model I adjust for: None Model II adjust for: Female age, Male age Model III adjust for: Female age, Male age, Duration of infertility, Basal LH, Basal FSH, Basal PRL, Basal P, Basal E2, Basal T, AMH, Endometrium thickness, Duration of Gn, Total dosage of Gn Associations between different insulin resistant surrogates and laboratory data and outcomes in fresh ET cycles Model I adjust for: None Model II adjust for: Female age, Male age Model III adjust for: Female age, Male age, Duration of infertility, Basal LH, Basal FSH, Basal PRL, Basal P, Basal E2, Basal T, AMH, Endometrium thickness, Duration of Gn, Total dosage of Gn Table  3 shows the correlation between different IR surrogates and outcomes in FET cycles, we found no correlation between TyG and HOMA-IR with clinical pregnancy rate ( P  > 0.05), whereas TyG-BMI showed a robust negative correlation with clinical pregnancy rate (OR = 0.99, 95% CI: 0.98 ~ 0.99; P  = 0.007). Furthermore, there was no correlation between these three IR surrogates and the rate of miscarriage, live births and neonatal birth weight in FET cycles (all P  > 0.05). Table 3 Associations between different insulin resistant surrogates and outcomes in FET cycles Variable Model I β/OR(95%CI) P -value Model II β/OR(95%CI) P -value Model III β/OR(95%CI) P -value FET clinical pregnancy rate TyG-BMI index 0.99 (0.99 ~ 0.99) 0.002 0.99 (0.99 ~ 0.99) 0.006 0.99 (0.98 ~ 0.99) 0.007 TyG index 0.64 (0.45 ~ 0.92) 0.015 0.67 (0.46 ~ 0.96) 0.028 0.67 (0.45 ~ 1.01) 0.057 HOMA-IR index 0.91 (0.83 ~ 0.99) 0.045 0.91 (0.83 ~ 1.00) 0.052 0.91 (0.82 ~ 1.01) 0.081 FET miscarriage rate TyG-BMI index 1.00 (0.99 ~ 1.01) 0.494 1.00 (0.99 ~ 1.01) 0.398 0.99 (0.98 ~ 1.01) 0.284 TyG index 0.60 (0.29 ~ 1.23) 0.164 0.56 (0.27 ~ 1.19) 0.130 0.49 (0.21 ~ 1.15) 0.099 HOMA-IR index 0.96 (0.79 ~ 1.16) 0.656 0.95 (0.79 ~ 1.16) 0.630 0.96 (0.76 ~ 1.20) 0.702 FET live birth rate TyG-BMI index 0.99 (0.99 ~ 0.99) 0.019 0.99 (0.99 ~ 1.00) 0.056 0.99 (0.99 ~ 1.00) 0.090 TyG index 0.78 (0.55 ~ 1.11) 0.173 0.82 (0.58 ~ 1.17) 0.281 0.86 (0.58 ~ 1.29) 0.464 HOMA-IR index 0.93 (0.85 ~ 1.02) 0.134 0.93 (0.85 ~ 1.03) 0.161 0.93 (0.84 ~ 1.04) 0.198 FET neonatal birth weight TyG-BMI index 1.29 (-1.38 ~ 3.96) 0.345 1.39 (-1.32 ~ 4.11) 0.316 -1.08 (-4.52 ~ 2.36) 0.537 TyG index -34.64 (-220.95 ~ 151.67) 0.716 -29.74 (-219.46 ~ 159.98) 0.759 -197.05 (-402.58 ~ 8.48) 0.062 HOMA-IR index -24.96 (-74.02 ~ 24.10) 0.320 -23.98 (-73.69 ~ 25.73) 0.345 -59.36 (-113.85 ~ -4.87) 0.054 Model I adjust for: None Model II adjust for: Female age, Male age Model III adjust for: Female age, Male age, Duration of infertility, Basal LH, Basal FSH, Basal PRL, Basal P, Basal E2, Basal T, AMH, Endometrium thickness, Duration of Gn, Total dosage of Gn Associations between different insulin resistant surrogates and outcomes in FET cycles Model I adjust for: None Model II adjust for: Female age, Male age Model III adjust for: Female age, Male age, Duration of infertility, Basal LH, Basal FSH, Basal PRL, Basal P, Basal E2, Basal T, AMH, Endometrium thickness, Duration of Gn, Total dosage of Gn In addition, Fig.  2 also suggests stable negative correlations between the three IR surrogates and live birth rates in the fresh ET cycles, whereas such correlations do not exist in the FET cycles. Then, based on all the above results, we conducted stratified analyses to assess the effects of the three IR surrogates on live birth rate (in the fresh ET cycles). As shown in Fig.  3 , the negative correlations between the three IR surrogates and live birth rate were stable and similar across all stratified populations (P-interaction > 0.05). Fig. 2 ( A ) Restricted cubic spline fitting for the association between different IR surrogates with live birth rate in the fresh ET cycles. ( B ) Restricted cubic spline fitting for the association between different IR surrogates with live birth rate in the FET cycles ( A ) Restricted cubic spline fitting for the association between different IR surrogates with live birth rate in the fresh ET cycles. ( B ) Restricted cubic spline fitting for the association between different IR surrogates with live birth rate in the FET cycles Fig. 3 Stratified associations between different IR surrogates ( A . TyG-BMI index; B . TyG index; C . HOMA-IR index) and live birth rate in the fresh ET cycles according to baseline characteristics Stratified associations between different IR surrogates ( A . TyG-BMI index; B . TyG index; C . HOMA-IR index) and live birth rate in the fresh ET cycles according to baseline characteristics Table  4 ; Fig.  4 show the results of the ROC curves. The TyG-BMI demonstrated superior predictive ability for insulin resistance (defined as HOMA-IR ≥ 2.2) with an AUC of 0.81, compared to the TyG’s AUC of 0.77 ( P  < 0.001). As for the live birth in fresh ET cycles, the AUCs of TyG-BMI, TyG and HOMA-IR were 0.64 (95% CI: 0.58, 0.69) vs. 0.61 (95% CI: 0.55, 0.67) vs. 0.60 (95% CI: 0.55, 0.67), respectively. This suggested that the TyG-BMI was significantly better than the TyG and HOMA-IR in the prediction of live birth in fresh ET cycles. Table 4 Comparison of ROC curves for different surrogates to predict insulin resistance and live birth rate in fresh ET cycles Objects/Surrogates Cutoff (Sensitivity, Specificity ) AUC (95% CI) P -value Insulin resistance TyG-BMI index 206.4 (0.69, 0.83) 0.81 (0.78, 0.85) < 0.001 TyG index 8.37 (0.77, 0.64) 0.77 (0.74, 0.81) ET live birth rate TyG-BMI index 216.7 (0.73, 0.54) 0.64 (0.58, 0.69) < 0.05 TyG index 8.41 (0.55, 0.69) 0.61 (0.55, 0.67) HOMA-IR index 2.43 (0.58, 0.62) 0.60 (0.55, 0.67) AUC, area under the curve; CI, confidence interval Z-test was used to compare statistically significant differences between AUCs Comparison of ROC curves for different surrogates to predict insulin resistance and live birth rate in fresh ET cycles AUC, area under the curve; CI, confidence interval Z-test was used to compare statistically significant differences between AUCs Fig. 4 ( A ) ROC curves for different surrogates to predict insulin resistance. ( B ) ROC curves for different surrogates to predict live birth rate in the fresh ET cycles ( A ) ROC curves for different surrogates to predict insulin resistance. ( B ) ROC curves for different surrogates to predict live birth rate in the fresh ET cycles

Polycystic Ovary Syndrome (PCOS) is a common and multi-system disease in the female population, and its global incidence is about 8–13% depending on the population studied and definitions used, which has been on the rise in recent years [ 1 , 2 ]. PCOS is characterized by abnormal ovarian function due to absent or sparse ovulation, irregular menstruation, and polycystic changes in the ovaries, as well as hyperandrogenism, and can be variable depending on body weight, race, age, and environmental factors [ 3 , 4 ]. Infertility is one of the common reasons for patients with PCOS to have medical consultations, and the prevalence of PCOS can even be as high as 70% in patients with infertility due to ovulation disorders [ 5 ]. For some patients who fail to achieve pregnancy after standardized fertility guidance or interventions such as ovulation induction therapy, assisted reproductive techniques like in vitro fertilization/intracytoplasmic sperm injection (IVF/ICSI) are required to achieve fertility. Insulin resistance (IR) is a pathological state in which insulin’s ability to regulate glucose metabolism is impaired, and it is also an important part of the pathophysiologic mechanism of PCOS, and IR is present in about 50–70% of PCOS patients [ 6 ]. Also, IR has a negative impact on the outcomes of PCOS patients undergoing assisted reproductive fertilization [ 7 ]. Li et al. conducted a high-quality meta-analysis to assess the IR on IVF/ICSI outcomes in patients with PCOS, which included a total of 12 observational studies, and found that IR had a significantly negative impact on metaphase II (MII) oocytes counts, total embryo counts, clinical pregnancy rates, and miscarriage rates in patients with PCOS [ 8 ]. Song et al. retrospectively analyzed data from 329 women undergoing IVF and found a negative correlation between homeostasis model assessment (HOMA-IR) index, body mass index (BMI), and clinical pregnancy rate in patients with PCOS combined with infertility [ 9 ]. Similarly, another large retrospective cohort study from China enrolled a total of 2055 women with PCOS undergoing their first fresh IVF cycle, suggesting that IR decreases ovarian response in these patients, especially in the lean subgroup. Also IR may lead to a higher risk of early miscarriage [ 10 ]. Therefore, active assessment of IR levels in patients with PCOS can help improve reproductive outcomes. Traditional methods of assessing IR levels, such as the hyperinsulinemic-euglycemic clamp (HIEC) and HOMA-IR, have high accuracy but are complex and time-consuming to perform, and thus have limited application in research and clinical settings [ 11 ]. For this reason, researchers have proposed methods to assess IR based on simple indices such as fasting triglycerides and glucose. The triglyceride-glucose index (TyG) was first proposed by South American researchers as an index for IR assessment and has good correlation with IR [ 12 , 13 ]. Compared with traditional methods, the TyG index has the advantage of being easily accessible and less expensive [ 14 ]. Based on the TyG index, the researchers further proposed the triglyceride-glucose-body mass index (TyG-BMI), which incorporates the obesity-indicator body mass index (BMI), thus improving its correlation with IR levels [ 15 , 16 ]. However, as far as we know, there have been no studies investigating the relationship between different insulin resistance surrogates and IVF outcomes in patients with PCOS. Hence, in this large-scale study, we determined to examine the effect of IR (assessed by different IR surrogates) on IVF outcomes among PCOS patients. Our findings were intended to provide evidence for individualized ovarian stimulation with an emphasis on early assessment and intervention prior to in vitro fertilization to improve pregnancy outcomes in patients with PCOS.

In this large-scale study, we revealed that the three IR surrogates (TyG-BMI, TyG and HOMA-IR) were negatively associated with live birth rates in fresh ET cycles, and this association was stable across all subgroups of the population. However, this relationship did not exist in FET cycles. Furthermore, our study found that TyG-BMI was superior to TyG and HOMA-IR in predicting the rate of live birth in fresh ET cycles. In addition, TyG-BMI and TyG were negatively correlated with clinical pregnancy rate and number of oocytes retrieved in fresh ET cycles, but HOMA-IR was not associated with these variables. As far as we know, this was the first study to investigate the relationship between different IR surrogates and IVF outcomes in patients with PCOS. The current study considers that insulin resistance is the main pathophysiologic mechanism leading to infertility in PCOS and has a definite adverse effect on pregnancy outcomes in patients with PCOS [ 24 ]. This negative impact is thought to be caused by different mechanisms. First, IR is usually characterized by hyperinsulinemia, which has the potential to disrupt the regulation of the hypothalamic-pituitary-ovarian (HPO) axis in patients with PCOS, resulting in elevated LH secretion and decreased FSH secretion, which may inhibit follicular maturation and ovulation [ 25 ]. In addition, hyperinsulinemia can lead to a decrease in hepatic production of sex hormone-binding globulin (SHBG), which leads to increased levels of circulating steroid hormones (e.g., estrogens and androgens), so that in patients with PCOS hyperinsulinemia tends to be associated with high LH levels and high androgens, which in turn can trigger adipose tissue dysfunction [ 26 , 27 ]. Another possible mechanism for elevated androgen levels is that insulin binds to receptors on follicular membrane cells, inducing an increase in luteinizing hormone activity at the ovarian level, which increases the production of androgens, ultimately impairing ovulatory function [ 28 ]. Secondly, previous studies have found that IR increases oxidative stress in mouse follicles and disrupts mitochondrial function in mouse oocytes, while IR also reduces glucose uptake and utilization in ovarian granulosa cells by decreasing the expression of glucose transporters type 4 (GLUT4), all of these effects ultimately have a negative impact on oocyte quality [ 29 – 31 ]. Furthermore, in PCOS, IR can induce an inflammatory response with increased nuclear factor-κB (NF-κB) activation, oxidative stress, and TNF release from circulating mononuclear cells (MNCs). It has been shown that oxidative stress plays an important role in low-grade chronic inflammation (LGCI) in PCOS, and oxidative stress can be significantly increased by the expression of proinflammatory cytokines [ 32 ]. Consequently, IR, adipose tissue dysfunction, hyperandrogenemia, and LGCI may cooperate with each other in the vicious circle leading to the pathogenesis of PCOS [ 33 ]. There have been some previous studies exploring the relationship between IR surrogates and ovarian response and pregnancy outcomes in IVF, but the results have been inconsistent. The Ovarian Sensitivity Index (OSI) is an accurate indicator of ovarian sensitivity to exogenous gonadotropins during IVF and is calculated as follows: OSI = [(Number of retrieved oocytes/Total gonadotropin dose) ×1000]. Li et al. [ 34 ] conducted a retrospective cohort study aimed at exploring the association between HOMA-IR and OSI, which included a total of 1508 women with PCOS aged 20–39 years old who underwent their first oocyte retrieved cycle, and found that women with PCOS had a negative correlation between Ln HOMA-IR and Ln OSI. And, patients in the highest tertile of HOMA-IR had lower OSI values compared to patients in the lowest tertile of HOMA-IR. The study by Luo et al. similarly found a significant negative effect of HOMA-IR on OSI, especially in lean PCOS patients [ 10 ]. Also in this study, IR was found to be independently and positively associated with an increased risk of early miscarriage in fresh embryo transfer cycles, but had no significant effect on pregnancy and live birth rates [ 10 ]. In contrast, a single-center, retrospective, observational cohort study from Chen et al., which included a total of 948 female patients with PCOS who underwent their first embryo transfer (both fresh and frozen transfer cycles), found that the rates of early miscarriage, macrosomia, and gestational diabetes mellitus increased significantly with increasing HOMA-IR, whereas the rate of live births decreased significantly. After adjusting for confounders, HOMA-IR was an independent risk factor for early miscarriage and macrosomia rates [ 35 ]. In addition, a study conducted by Gao et al. found that pre-treatment with the insulin sensitizer metformin prior to IVF/ICSI and embryo transfer cycles improved the clinical pregnancy rate in FET cycles when combined with HOMA-IR ≥ 2.71 in patients with PCOS, but not in fresh embryo transfer cycles [ 36 ]. It is worth noting that all of these studies above that assessed IR levels used the HOMA-IR index. Li et al. [ 37 ] attempted to explore the relationship between TyG-BMI and IVF reproductive outcomes in patients with PCOS, a total of 966 women with PCOS who underwent their first FET cycle were enrolled, and the final results found that the number of embryos available and the number of high-quality embryos were lower in the highest quartile of TyG-BMI compared to those in the lowest quartile of TyG-BMI. However, TyG-BMI did not correlate with the pregnancy rate or the live birth rate. The conclusions of these previous related studies were inconsistent, and we consider this to be due to the fact that the fertilization methods in these studies included both IVF and ICSI, and the adoption of ICSI is often due to severe sperm abnormalities in men, which means that the conclusions of these studies are inevitably influenced by sperm-related confounding factors. In our study, in order to avoid this situation, all of our fertilization methods were IVF. In addition, our study found that although the three IR surrogates (TyG-BMI, TyG and HOMA-IR) were negatively associated with live birth rates in fresh ET cycles, this relationship was not present in FET cycles. It was similar to the findings of Li et al. who tried to explore the relationship between TyG-BMI and IVF pregnancy outcomes in PCOS patients, while in their study the overall transplantation strategy was IVF-FET, and the final results similarly suggested that there was no correlation between TyG-BMI and live birth rate [ 37 ]. We speculate that this may be due to the fact that in our study the IR surrogates were measured on day 2 to 4 of the menstrual cycle prior to IVF, whereas FET is usually performed several months or even half a year after oocyte retrieval, during which time the patient’s IR surrogates are constantly changing due to the influence of other factors such as lifestyle or weight loss, thus resulting in the absence of a correlation between IR surrogates and the live birth rates in the FET cycles. Further studies in the future may consider following up the changes in IR surrogates during IVF, which may help to analyze the correlation between IR surrogates and IVF outcomes. However, this study has some limitations. First, as a single-center retrospective study, the influence of potential confounders on the results is unavoidable, and thus future large prospective cohort studies are still needed to further validate our conclusions. Second, as our study participants were limited to Chinese and PCOS females, this may limit the generalization of our findings to other races and non-PCOS females. Finally, we did not conduct follow-up investigations of changes in BMI, blood glucose, triglycerides, and insulin during IVF, which may be useful to analyze the correlation between changes in different IR surrogates and IVF outcomes. Regardless of these limitations, our study has several strengths. To our knowledge, this is the first study to simultaneously investigate the relationship between different IR surrogates and IVF outcomes in patients with PCOS in one study, whereas most of the previous studies have used a single HOMA-IR index to assess the level of IR, which provides favorable evidence to guide patients with PCOS to take the necessary interventions prior to IVF. Second, to avoid the influence of potential male confounders, we excluded subjects whose spouses had moderate or severe semen abnormalities or who were fertilized through ICSI. Lastly, on the basis of multivariate regression analyses we conducted sensitivity analyses with stratification and interaction tests, which improved the reliability of our conclusions.

In this large-scale study, our study revealed that the three IR surrogates (TyG-BMI, TyG and HOMA-IR) were negatively associated with the live birth rates in fresh ET cycles. It is recommended that clinicians should emphasize using IR surrogates for early assessment of IR levels in PCOS patients prior to IVF, and that targeted metabolic interventions may be appropriate for patients with high IR levels to improve IVF pregnancy outcomes. Our study provides a new perspective on the management of reproductive health in women with PCOS.