Impact of autoimmune thyroiditis on IVF outcomes in patients with polycystic ovary syndrome and chronic endometritis

A retrospective analysis was conducted on 2,739 patients with infertility who underwent IVF cycles at our hospital from 2018 to 2020, underwent hysteroscopic surgery, and had endometrial pathology. Among these, 1,972 cases had endometritis and 767 cases did not. In the 1,972 endometritis cases, 279 (14.14%) were TPOAb positive and 407 (20.63%) were positive for at least one of TPOAb or TGAb and were classified as AIT-positive. Among the 767 non-endometritis cases, 96 (12.52%) were TPOAb positive, and 147 (19.17%) were AIT-positive. Although the AIT-positive rate and TPOAb-positive rate were numerically higher in the endometritis group, no statistically significant differences were observed (20.63% vs. 19.17%, P  = 0.265), (14.14% vs. 12.52%, P  = 0.389). There was no statistically significant difference in the proportion of PCOS patients between the endometritis and non-endometritis groups (Table  1 ). Table 1 Comparison of diseases between endometritis and non-endometritis groups Item Endometritis group ( n  = 1972) Non-endometritis group ( n  = 767) χ² value P value TPOAb positive 279 (14.14%) 96 (12.52%) 0.743 0.389 AIT positive 407 (20.63%) 147 (19.17%) 1.244 0.265 PCOS patients 315 (15.97%) 102 (13.30%) 3.062 0.080 AIT positive indicates either TPOAb or TGAb positivity Comparison of diseases between endometritis and non-endometritis groups AIT positive indicates either TPOAb or TGAb positivity The study cohort included 417 patients with PCOS, of whom 83 (19.90%) had AIT. The non-PCOS group comprised 2,322 patients, with 471 (20.28%) having AIT. AIT prevalence was not significantly different between the two groups (19.90% vs. 20.28%, P  = 0.859). The Rotterdam phenotype distributions in all PCOS patients, patients with PCOS and CE, and the AIT and non-AIT subgroups of patients with PCOS and CE are summarized in Table  2 . The PCOS phenotype distribution did not differ significantly between the AIT-positive and AIT-negative PCOS with CE groups ( P  = 0.462). Among patients with PCOS and CE, 444 oocyte retrieval cycles were performed, including 96 AIT-positive cycles and 348 AIT-negative cycles. In the AIT group, fresh transfer was performed in 35 cycles, resulting in 8 clinical pregnancies and 2 live births; cumulative live births occurred in 33 cycles. In the non-AIT group, fresh transfer was performed in 131 cycles, resulting in 46 clinical pregnancies and 31 live births; cumulative live births occurred in 144 cycles. Table 2 Distribution of PCOS phenotypes according to the Rotterdam criteria Phenotype Rotterdam features All PCOS patients ( n  = 417) PCOS with CE ( n  = 315) AIT-positive PCOS with CE ( n  = 66) AIT-negative PCOS with CE ( n  = 249) A OA + hyperandrogenism + PCOM 76 (18.2%) 58 (18.4%) 9 (13.6%) 49 (19.7%) B OA + hyperandrogenism 70 (16.8%) 54 (17.1%) 12 (18.2%) 42 (16.9%) C Hyperandrogenism + PCOM 11 (2.6%) 6 (1.9%) 0 (0.0%) 6 (2.4%) D OA + PCOM 260 (62.4%) 197 (62.5%) 45 (68.2%) 152 (61.0%) OA oligo- or anovulation, PCOM polycystic ovarian morphology, CE chronic endometritis, AIT autoimmune thyroiditis The phenotype distribution between the AIT-positive and AIT-negative PCOS with CE groups was compared using Fisher’s exact test  ( P  = 0.462) Distribution of PCOS phenotypes according to the Rotterdam criteria OA oligo- or anovulation, PCOM polycystic ovarian morphology, CE chronic endometritis, AIT autoimmune thyroiditis The phenotype distribution between the AIT-positive and AIT-negative PCOS with CE groups was compared using Fisher’s exact test  ( P  = 0.462) Infertility duration was longer in AIT group: 5.62 ± 3.85 years vs. 4.75 ± 3.15 years, P  = 0.025. Other items showed no significant differences between the two groups (Table  3 ). Table 3 Comparison of general characteristics between AIT and non-AIT cycles in patients with PCOS and endometritis Item AIT group ( n  = 96) Non-AIT group ( n  = 348) Test statistic P Age 32.08 ± 4.88 31.61 ± 4.14 t = 0.943 0.346 Infertility duration (years) 5.62 ± 3.85 4.75 ± 3.15 t = 2.012 0.025 BMI (kg/m 2 ) 24.37 ± 3.73 24.50 ± 4.43 t = -0.272 0.786 FSH (IU/L) 5.89 ± 1.70 6.36 ± 2.87 t = -1.389 0.166 E2 (pmol/L) 166 (147, 239.5) 190 (138,283) Z = -0.310 0.757 PRL (ug/L) 10.9 (7.89, 15.10) 10.8 (7.81, 17.05) Z = -0.355 0.722 LH (IU/L) 6.23 (3.17, 12.30) 6.82 (3.94, 11.35) Z = -0.463 0.643 T (nmol/L) 0.69 (0.69, 1.34) 0.86 (0.69, 1.36) Z = -1.019 0.308 A (nmol/L) 9.17 ± 4.94 10.19 ± 4.58 t = -1.508 0.133 AMH (ng/mL) 7.89 ± 5.71 8.48 ± 6.28 t = -0.776 0.438 AFC 17.74 ± 9.35 17.42 ± 8.03 t = 0.333 0.739 TSH (mIU/L) 2.23 ± 1.36 2.18 ± 1.19 t = 0.347 0.729 AIT autoimmune thyroiditis, BMI body mass index, FSH follicle-stimulating hormone, LH luteinizing hormone, E2 estradiol, PRL prolactin, AMH anti-Mullerian hormone, TSH thyroid-stimulating hormone, A androstenedione, T testosterone Comparison of general characteristics between AIT and non-AIT cycles in patients with PCOS and endometritis AIT autoimmune thyroiditis, BMI body mass index, FSH follicle-stimulating hormone, LH luteinizing hormone, E2 estradiol, PRL prolactin, AMH anti-Mullerian hormone, TSH thyroid-stimulating hormone, A androstenedione, T testosterone Patients with PCOS and endometritis showed no significant differences in laboratory outcomes between the AIT group and the non-AIT group regarding Gn days, dosage, HCG day E2, LH, endometrial thickness, total oocytes retrieved, number of high-quality embryos, or total blastocysts (Table  4 ). Table 4 Comparison of COS and laboratory outcomes between AIT and non-AIT groups in patients with PCOS and endometritis Item AIT group( n  = 96) Non-AIT group( n  = 348) Test statistic P Gn days 10.14 ± 4.15 10.19 ± 3.69 t = -0.12 0.905 Gn dosage 2189 ± 1191 2144 ± 1123 t = 0.328 0.743 HCG day E2 (pmol/L) 11,288 (6395.5, 16000.75) 9827.5 (5768.75, 15498) Z = -0.265 0.791 HCG day LH (IU/L) 1.88 (0.74, 3.01) 1.99 (1.07, 4.07) Z = -0.508 0.611 Endometrial thickness (mm) 10.11 ± 1.92 9.84 ± 1.81 t = 1.079 0.281 Total number of retrieved oocytes 16.36 ± 9.98 16.21 ± 8.6 t = 0.151 0.880 Number of 2PN 8.78 ± 6.66 9.36 ± 5.91 t = -0.825 0.410 Number of cleavage 11.00 ± 8.37 11.1 ± 6.87 t = -0.148 0.882 Number of 2PN cleavage 8.60 ± 6.68 9.17 ± 5.82 t = -0.827 0.408 Number of Transferable Embryos 4.73 ± 4.24 5.36 ± 4.52 t = -1.218 0.224 Number of high-quality embryos 5.72 ± 4.98 6.62 ± 5.08 t = -1.546 0.123 Total number of blastocysts 1.50 (0.00, 3.75) 1 (0, 3) Z = -0.306 0.760 Negative values in the χ²/t/Z-value column reflect the direction of the t- or Z-statistic and are expected for continuous variables; they do not represent negative P -values LH luteinizing hormone, E2  estradiol Comparison of COS and laboratory outcomes between AIT and non-AIT groups in patients with PCOS and endometritis 11,288 (6395.5, 16000.75) 9827.5 (5768.75, 15498) Negative values in the χ²/t/Z-value column reflect the direction of the t- or Z-statistic and are expected for continuous variables; they do not represent negative P -values LH luteinizing hormone, E2  estradiol Patients with PCOS and CE undergoing fresh embryo transfer had a lower live birth rate in the AIT group than in the AIT-negative group (5.71% [2/35] vs. 23.66% [31/131], P  = 0.021). Among patients undergoing fresh embryo transfer, baseline and transfer-related characteristics did not differ significantly between the AIT and non-AIT groups (Supplementary Table 1). However, CLBR did not differ significantly between groups. The number of cycles required to achieve the first live birth was also similar between groups (Table  5 ). The post-hoc power for the fresh-cycle live birth comparison was approximately 80.0% using a two-proportion Z-test approximation. Table 5 Comparison of pregnancy outcomes between AIT and non-AIT groups in patients with PCOS and endometritis Item AIT group( n  = 96) Non-AIT group( n  = 348) Test statistic P Number of fresh embryo transfer cycles 35 (36.46%) 131 (37.64%) χ² = 0.037 0.847 Fresh-cycle pregnancy rate 22.86% (8/35) 35.11% (46/131) χ² = 1.891 0.169 Live birth rate in fresh-cycles 5.71% (2/35) 23.66% (31/131) χ² = 5.335 0.021 Cumulative live birth rate 34.38% (33/96) 41.38% (144/348) χ² = 1.490 0.222 Number of cycles to achieve first live birth 2 (1,3) 2 (1,3) Z = -1.582 0.114 AIT autoimmune thyroiditis Comparison of pregnancy outcomes between AIT and non-AIT groups in patients with PCOS and endometritis AIT autoimmune thyroiditis Univariate analysis of factors affecting cumulative live birth rate revealed a negative correlation between age and cumulative live birth rate (OR = 0.926, 95% CI = 0.884–0.971, P  = 0.001). HCG day E2 (OR = 1.005, 95% CI = 1.003–1.008, P  < 0.001), number of retrieved oocytes (OR = 1.056, 95% CI = 1.033–1.081, P  < 0.001), number of high-quality embryos (OR = 1.194, 95% CI = 1.140–1.252, P  < 0.001), number of blastocyst formations (OR = 1.321, 95% CI = 1.200-1.455, P  < 0.001) were positively correlated with CLBR. No statistical correlation was observed between AIT, TSH, and CLBR. The type of thyroid antibodies showed no statistical correlation with CLBR (Table  6 ). Table 6 Univariate analysis of factors affecting CLBR in patients with PCOS and endometritis Item OR Value 95% CI P Value Age 0.926 0.884–0.971 0.001 Infertility duration 0.973 0.917–1.033 0.372 Basal FSH 0.992 0.917–1.073 0.841 Basal LH 0.998 0.980–1.017 0.827 Baseline PRL 1.003 0.996–1.011 0.383 Base T 1.023 0.976–1.073 0.344 Base A 1.012 0.962–1.065 0.639 AMH 1.033 0.991–1.077 0.127 AFC 1.006 0.983–1.029 0.637 Gn days 1.025 0.974–1.078 0.346 Gn dosage 0.924 0.788–1.083 0.328 HCG day E2 1.005 1.003–1.008 < 0.001 Endometrial thickness 1.085 0.962–1.223 0.182 Number of retrieved oocytes 1.056 1.033–1.081 < 0.001 Number of high-quality embryos 1.194 1.140–1.252 < 0.001 Number of blastocyst formations 1.321 1.200–1.455 < 0.001 BMI 0.973 0.930–1.019 0.250 Thyroid autoantibodies 1.348 0.840–2.161 0.216 TGAb 0.712 0.425–1.191 0.196 TPOAb 0.648 0.376–1.117 0.118 TSH 1.076 0.921–1.256 0.356 OR odds ratio, CI confidence interval Univariate analysis of factors affecting CLBR in patients with PCOS and endometritis OR odds ratio, CI confidence interval Multivariate analysis showed that the number of high-quality embryos remained independently associated with CLBR (B = 0.130, OR = 1.138, 95% CI 1.053–1.231, P  = 0.001). AIT status and TSH were not significantly associated with CLBR (Table  7 ). Table 7 Multivariate analysis of factors affecting CLBR in patients with PCOS and endometritis Item B SE Wald OR value 95% CI P value Age -0.051 0.035 2.080 0.951 0.887–1.018 0.149 TSH -0.027 0.114 0.055 0.974 0.778–1.218 0.815 TPOAb -0.114 0.523 0.048 0.892 0.320–2.487 0.827 TGAb -0.061 0.486 0.016 0.941 0.363–2.440 0.900 Infertility duration 0.074 0.046 2.626 1.077 0.985–1.179 0.105 BMI -0.019 0.033 0.326 0.981 0.921–1.046 0.568 AMH -0.019 0.026 0.508 0.982 0.933–1.033 0.476 HCG day E2 0.001 0.002 0.192 1.001 0.997–1.005 0.661 Number of retrieved oocytes -0.002 0.020 0.011 0.998 0.961–1.037 0.918 Number of high-quality embryos 0.130 0.040 10.598 1.138 1.053–1.231 0.001 Number of blastocysts 0.116 0.063 3.402 1.123 0.993–1.271 0.065 OR odds ratio, CI confidence interval, B regression coefficient, SE standard error, Wald Wald test Multivariate analysis of factors affecting CLBR in patients with PCOS and endometritis OR odds ratio, CI confidence interval, B regression coefficient, SE standard error, Wald Wald test

Study design, setting, and participants: This retrospective cohort study was conducted at the Reproductive Medicine Center, Department of Obstetrics and Gynecology, Peking University Third Hospital, Beijing, China. We reviewed infertile women who underwent IVF/ICSI treatment and hysteroscopic surgery with available endometrial pathology between January 2018 and December 2020. Patients were consecutively included using a non-probability sampling method if they had complete medical records, thyroid antibody testing, ovarian stimulation and embryological data, and follow-up information. Exclusion criteria were congenital uterine malformations, moderate-to-severe endometriosis, adenomyosis, recurrent implantation failure, and diminished ovarian reserve. The study size was determined by the number of eligible cycles available from January 2018 to December 2020. Patients were first divided into CE and non-CE groups according to endometrial pathology. Patients with both PCOS and CE were then identified for subgroup analysis and further classified into AIT and non-AIT groups according to thyroid antibody status (Fig. 1 ). CE was defined as CD138 positivity (visible CD138-positive cells under high-power microscopy), with or without CD38 positivity. AIT was defined as positivity for either thyroid peroxidase antibody (TPOAb) or thyroglobulin antibody (TGAb). PCOS was diagnosed according to the Rotterdam criteria, namely the presence of at least two of the following three features after exclusion of other etiologies that mimic PCOS: oligo- or anovulation (OA), clinical or biochemical hyperandrogenism, and polycystic ovarian morphology (PCOM) on ultrasound. PCOS phenotypes were classified as phenotype A (OA + hyperandrogenism + PCOM), phenotype B (OA + hyperandrogenism), phenotype C (hyperandrogenism + PCOM), or phenotype D (OA + PCOM) [ 7 , 8 ]. The main exposure was AIT status. The main clinical outcome was cumulative live birth rate (CLBR), while other outcomes included fresh-cycle clinical pregnancy rate, fresh-cycle live birth rate, ovarian stimulation parameters, and embryological outcomes. Laboratory and histological assessments: Serum TPOAb, TGAb, and TSH levels were measured using commercial kits from Siemens Healthcare Diagnostics on a fully automatic chemiluminescence immunoassay analyzer, ADVIA Centaur platform (Siemens Healthcare Diagnostics, Germany). According to the manufacturer’s instructions and the institutional laboratory reference range, TPOAb and TGAb positivity were defined as values ≥ 60 IU/mL. Endometrial CD138 immunohistochemistry was performed using a rabbit anti-human CD138 monoclonal antibody, clone GR106 (Gene Tech Company Limited, Shanghai, China; Cat# GT245101 /02/04/07/14/28 and GT245121 /29/11), according to the manufacturer’s instructions. CE was defined as the presence of visible CD138-positive plasma cells in the endometrial stroma under high-power microscopy. Fig. 1 Grouping flowchart of the study cohort population Grouping flowchart of the study cohort population IVF, in vitro fertilization, AIT, autoimmune thyroiditis, PCOS, polycystic ovary syndrome. IVF/ICSI treatment and embryo transfer strategy: Ovarian stimulation protocols, including the GnRH-a ultra-long protocol, GnRH-a long protocol, antagonist protocol, mild stimulation protocol, or IVM protocol, were selected according to patient age, body mass index (BMI), ovarian reserve, and clinical judgment. Recombinant follicle-stimulating hormone (FSH; Merck Serono, Switzerland) was initiated at 150–300 IU/day and adjusted according to serum hormone levels and follicular development. When two follicles reached 18 mm in diameter (17 mm in antagonist protocols), and the mean estradiol concentration per dominant follicle was approximately 200–300 ng/L, 250 µg recombinant hCG and/or 0.2 mg short-acting GnRH agonist was administered. Oocytes were retrieved 36–38 h later. Fertilization was performed using IVF or intracytoplasmic sperm injection (ICSI), followed by embryo culture. Fresh embryo transfer was generally considered in patients with relatively low OHSS risk, no premature progesterone elevation, suitable embryos and suitable endometrial thickness and morphology. However the final decision was individualized. In clinical practice, fresh transfer was usually not performed when the trigger day estradiol level was > 20,000 pmol/L, or when the trigger day estradiol level was > 15,000 pmol/L together with more than 20 oocytes retrieved. For fresh transfer, one to two day 3 embryos, or one day 5 embryo, were transferred into the uterine cavity. All patients diagnosed with CE received empirical antibiotic therapy with doxycycline 0.1 g twice daily orally or levofloxacin 0.5 g once daily orally plus metronidazole 0.2 g twice daily orally for 14 days. Repeat hysteroscopy or endometrial biopsy was not routinely performed after antibiotic treatment. Therefore, IVF outcomes were analyzed based on treated CE, as routine post-treatment histological confirmation was not available. Luteal support was administered for 8 weeks after embryo transfer. Intrauterine pregnancy was confirmed by transvaginal ultrasound 4 weeks after embryo transfer. Clinical outcomes were followed up by telephone. Outcome definitions: Clinical pregnancy was defined as the presence of a gestational sac in the uterine cavity on transvaginal ultrasound 4 weeks after embryo transfer. Live birth was defined as delivery of at least one live infant after 24 weeks of gestation. Fresh-cycle pregnancy outcomes, including fresh-cycle clinical pregnancy rate and fresh-cycle live birth rate, were analyzed only among cycles that underwent fresh embryo transfer. CLBR was defined as the number of live births from all embryo transfers following a single oocyte retrieval cycle divided by the number of oocyte retrieval cycles multiplied by 100%. Only the first live birth after each oocyte retrieval cycle was counted. Cycles without live birth but with remaining frozen embryos were excluded from the CLBR calculation. Covariates included age, infertility duration, BMI, basal hormones, anti-Mullerian hormone (AMH), antral follicle count (AFC), thyroid-stimulating hormone (TSH), ovarian stimulation parameters, number of retrieved oocytes, number of high-quality embryos, and number of blastocysts. This study was approved by the Medical Science Research Ethics Committee of Peking University Third Hospital (Approval No. 2008013) and was conducted in accordance with the Declaration of Helsinki. Statistical analysis: Statistical analysis was performed using SPSS 27.0 software (IBM, USA). Continuous variables with normal distribution were expressed as mean ± standard deviation and compared using Student’s t-test. Variables not normally distributed were presented as median (25th percentile, 75th percentile) and compared using the Mann-Whitney U test. Categorical variables were expressed as frequencies and percentages and compared using the chi-square test or Fisher’s exact test, as appropriate. Logistic regression was used to identify factors associated with CLBR. Variables with clinical relevance or significance in univariate analysis were entered into the multivariate model. A post-hoc power analysis was performed for the comparison of fresh-cycle live birth rates between the AIT and non-AIT groups using the observed proportions and a two-sided alpha level of 0.05. P  < 0.05 was considered statistically significant.

Autoimmune thyroid antibodies are among the most common organ specific autoantibodies in women of reproductive age, but their relationship with CE remains unclear. Previous studies have detected the expression of thyroid peroxidase (TPO) and thyroglobulin (TG) at both mRNA and protein levels in human endometrial glands, luminal epithelium, and full-term placental tissue, and TPO antibodies have been detected in endometrial tissue, suggesting that TPO on the endometrium may serve as a binding target for anti-TPO antibodies [ 9 ]. Xie et al. [ 10 ] used single-cell RNA sequencing to analyze endometrial samples from unexplained infertility patients with AIT and euthyroid status, and found altered immune cell composition and cytokine related pathways during the window of implantation. These findings suggest that AIT may be associated with local endometrial immune dysregulation. CE mostly results from infectious inflammation caused by pathogens, while a minority arise from other chronic stimuli. Research on the correlation between CE and autoimmune markers is limited. Kushnir et al. [ 11 ] compared endometrial pathology and plasma inflammatory markers and autoimmune markers (total immunoglobulin, ANA, thyroid antibodies, antiphospholipid antibodies) in 55 patients with recurrent implantation failure (RIF)/recurrent spontaneous abortion (RSA). Results indicated no differences in systemic inflammation or autoimmunity between CE and non-CE patients. In this study, the prevalence of autoimmune thyroid antibodies was 20.63% in endometritis patients versus 19.17% in non-endometritis patients ( P = 0.265), the prevalence of anti-TPOAb was 14.14% in endometritis patients versus 12.52% in non-endometritis patients ( P = 0.389), AIT showed an increase in CE patients by numerical values, but no statistically significant difference. This suggests that AIT may influence endometrial immune function at a cellular or molecular level, but it may not lead to the development of histologically diagnosed CE. The relationship between PCOS and CE also remains controversial. One cohort study matched 189 PCOS patients with 697 non-PCOS patients using propensity score matching [ 6 ] and compared CE incidence. After PSM, no significantly higher CE incidence was observed among infertile women with PCOS. However, other studies define CE as the presence of more than five CD138-positive cells under high power microscopy. Wang et al. [ 12 ] found a significantly higher prevalence of endometritis (hyperemia, micropolyps, edema) diagnosed by hysteroscopy in PCOS patients (44.4% vs. 22.2%, P < 0.05), with most presenting as hyperemic strawberry endometrium (71.55%). However, the CD138 positivity rate in these hyperemic cases was only 10.14%, showing no difference from the non-PCOS group. Different PCOS phenotypes may also influence endometritis occurrence. Seungbaek Lee et al. [ 13 ] used an AI-driven CD138 identification software to compare CD138-positive rates between PCOS and control groups. Within the same cycle phase, there was no significant difference in CD138 cell proportions between PCOS cases and controls. The phenotype combining anovulation, hyperandrogenism, and ultrasound-detected polycystic ovarian morphology (PCOM) exhibited a significantly higher CD138-positive proportion than the phenotype showing only anovulation and PCOM ( P < 0.001). In our cohort, the proportion of PCOS patients among endometritis versus non-endometritis patients was 15.97% versus 13.30%, P  = 0.08, showing a numerical increase but no statistically significant difference. Further studies need to be carried out in different PCOS subtypes and different CE diagnostic criteria. Because PCOS, CE, and AIT are all associated with endocrine, inflammatory, or autoimmune abnormalities, their coexistence may have combined effects on reproductive outcomes. The impact of AIT on female fertility has been widely investigated, but the findings remain inconsistent. Most studies suggest that thyroid autoimmunity in euthyroid women does not significantly reduce IVF pregnancy or live birth rates, whereas other studies have reported an association between thyroid autoantibodies and miscarriage or preterm birth [ 2 , 14 , 15 ]. AIT has also been reported to be slightly more frequent in patients with PCOS or endometriosis, and shared inflammatory or autoimmune pathways may contribute to these associations [ 16 , 17 ]. In the PCOS with CE subgroup, AIT was not associated with CLBR, and ovarian stimulation parameters and most embryological outcomes were comparable between the AIT and non-AIT groups. This finding is broadly consistent with previous studies showing no significant reduction in live birth rates after embryo transfer or in CLBR among infertile women with AIT (50.6% vs. 52.1%, OR 0.94, 95% CI 0.86–1.02, adjusted OR 0.97, 95% CI 0.89–1.06) [ 3 , 4 ]. The number of retrieval cycles required to achieve the first live birth also showed no significant variation. In multivariate analysis, the number of high-quality embryos remained independently associated with CLBR (OR = 1.138, 95% CI: 1.053–1.231, P = 0.001), whereas AIT status and TSH were not significantly associated with CLBR. These findings suggest that, in patients with PCOS and CE, CLBR may be more strongly influenced by embryo availability and embryo quality than by thyroid antibody status alone. This is also consistent with previous studies showing that cumulative outcomes after IVF are closely related to the number and quality of embryos available for transfer [ 18 , 19 ]. Although CLBR did not differ significantly between groups, the fresh-cycle live birth rate was lower in the AIT group than in the non-AIT group: 5.71% (2/35) vs. 23.66% (31/131), P = 0.021. One possible explanation is that AIT may exert a more pronounced effect on endometrial receptivity under the supraphysiological endocrine environment of fresh embryo transfer. Notably, the distribution of PCOS phenotypes according to the Rotterdam criteria was comparable between the AIT-positive and AIT-negative PCOS with CE groups, suggesting that the observed difference in fresh-cycle live birth rate was unlikely to be primarily driven by an imbalance in PCOS phenotype distribution. However, given the potential heterogeneity in endocrine, metabolic, and inflammatory characteristics across PCOS phenotypes, further studies are needed to determine whether PCOS phenotype modifies the reproductive impact of thyroid autoimmunity. Previous studies [ 10 ] have suggested that thyroid autoimmunity may be associated with immune dysregulation in the endometrium during the window of implantation, including alterations in endometrial immune cell composition and cytokine pathways. Experimental evidence in mice also indicates that AIT may impair embryo implantation even in the euthyroid state by compromising endometrial morphology, pinopode development, and receptivity marker expression [ 20 ]. In patients with PCOS and CE, the endometrium may already be exposed to a chronic inflammatory and hormonally disturbed environment. Therefore, the coexistence of AIT may further amplify local immune or inflammatory dysregulation, particularly in fresh cycles after controlled ovarian stimulation. By contrast, frozen embryo transfer may allow recovery from the supraphysiological hormonal environment after ovarian stimulation and provide more controlled endometrial preparation. The possible advantage of frozen embryo transfer in selected PCOS patients is also supported by previous studies. A meta-analysis suggested that frozen embryo transfer was associated with a higher live birth rate than fresh transfer in PCOS patients with high ovarian response (RR: 1.18, 95% CI: 1.06–1.31; n = 3398), while no significant difference was observed in normal response PCOS patients (RR: 1.13, 95% CI: 0.90–1.41; n = 1608) [ 21 ]. A 2016 RCT involving 1508 infertile women with PCOS found higher live birth rates in frozen cycles (49.3% vs. 42%), and lower risks of pregnancy loss and OHSS than fresh transfer, although the risk of preeclampsia was higher [ 22 ]. This may partly explain why the observed difference was limited to fresh-cycle live birth rate and was not reflected in CLBR. However, despite broadly comparable fresh transfer characteristics between groups, residual selection bias could not be completely excluded because fresh embryo transfer was individualized rather than randomly assigned. Given the limited fresh transfer subgroup size, this finding should be considered exploratory. Whether frozen embryo transfer may be a preferable strategy for PCOS patients with CE and AIT requires confirmation in larger prospective studies. Our findings may help guide counseling for patients with PCOS and CE who are also positive for thyroid autoantibodies, especially when considering fresh versus frozen embryo transfer strategies. This study has several limitations. First, CE resolution after antibiotic therapy was not routinely confirmed by repeat hysteroscopy or endometrial biopsy. Therefore, the actual post-treatment CE resolution rate in this cohort was unknown, and some patients classified as having treated CE may have had persistent or recurrent histological inflammation at the time of embryo transfer. This blindspot could introduce residual confounding into embryo transfer outcomes, particularly fresh-cycle outcomes, because unresolved CE may independently impair endometrial receptivity and implantation. Second, although key baseline characteristics were broadly comparable within the fresh-cycle subgroup, residual selection bias could not be excluded because transfer strategy was individualized. The limited sample size also precluded formal interaction analyses for potential effect modifiers. Third, the fresh-transfer subgroup was small, and the observed difference in fresh-cycle live birth rate should therefore be interpreted cautiously. Finally, although the main clinical protocols were generally consistent from 2018 to 2020, subtle changes in clinical practice or individualized physician therapy could not be completely excluded. In summary, AIT was not associated with the prevalence of CE or with CLBR in patients with PCOS and CE. The lower fresh-cycle live birth rate observed in the AIT group should be interpreted cautiously. Embryo quality remains a key determinant of CLBR in this population, whereas the effect of AIT on fresh-cycle outcomes requires confirmation in larger prospective studies.

Supplementary Material 1. Supplementary Material 1.