Uterine artery S/D ratio and serum CA125 as predictors of early pregnancy loss in women with ovarian endometriomas: a retrospective case-control study

ROC diagnostic analysis revealed that the AUC of bilateral S/D ratio mean was 0.75 (95% CI: 0.67–0.82), indicating relatively high diagnostic value. With a cutoff value of 6.49, the sensitivity and specificity were 0.61 (95% CI: 0.48–0.73) and 0.77 (95% CI: 0.70–0.84), respectively, suggesting that this indicator had poor diagnostic ability for identifying spontaneous miscarriage patients, but moderate performance for diagnosing non-spontaneous miscarriage cases. The AUC of CA125 was 0.76 (95% CI: 0.69–0.83), also indicating relatively high diagnostic value. With a cutoff value of 23.55, the sensitivity and specificity were 0.84 (95% CI: 0.74–0.93) and 0.60 (95% CI: 0.52–0.68), respectively, demonstrating that this indicator had strong ability in identifying spontaneous miscarriage patients, but weak diagnostic capacity for non-spontaneous miscarriage cases. The multivariate logistic regression model achieved an AUC of 0.85 (95% CI: 0.79–0.90), with a probability cutoff value of 0.223, and sensitivity and specificity of 0.87 (95% CI: 0.78–0.95) and 0.72 (95% CI: 0.65–0.80), respectively. DeLong test results showed that the multivariate logistic regression model demonstrated significantly superior diagnostic performance compared to bilateral S/D ratio mean ( P  = 0.035) and CA125 ( P  = 0.043) (Table  5 ; Fig.  1 ). Table 5 ROC curve analysis Variables AUC (95%CI) Accuracy (95%CI) Sensitivity (95%CI) Specificity (95%CI) Cut off Delong test Bilateral S/D ratio mean 0.75 (0.67–0.82) 0.72 (0.66–0.78) 0.61 (0.48–0.73) 0.77 (0.70–0.84) 6.49 P  = 0.035 CA125 0.76 (0.69–0.83) 0.67 (0.60–0.73) 0.84 (0.74–0.93) 0.60 (0.52–0.68) 23.55 P  = 0.043 Multivariable logistic regression model 0.85 (0.79–0.90) 0.77 (0.70–0.82) 0.87 (0.78–0.95) 0.72 (0.65–0.80) 0.223 - ROC curve analysis Fig. 1 ROC curves for CA125, bilateral mean uterine artery S/D ratio, and logistic model in predicting early spontaneous miscarriage. The ROC curves of diagnostic combination were generated by a logistic regression model that included VAS, CA125, the mean bilateral S/D ratio, and the mean bilateral PI; ROC: receiver operating characteristic; AUC: area under the curve; S/D, systolic/diastolic ratio; CI: confidence interval ROC curves for CA125, bilateral mean uterine artery S/D ratio, and logistic model in predicting early spontaneous miscarriage. The ROC curves of diagnostic combination were generated by a logistic regression model that included VAS, CA125, the mean bilateral S/D ratio, and the mean bilateral PI; ROC: receiver operating characteristic; AUC: area under the curve; S/D, systolic/diastolic ratio; CI: confidence interval Internal validation of the model was performed using Bootstrap resampling with 1000 iterations. After internal validation, the model’s AUC value was 0.830, confirming that the model maintained high discriminative ability in the Bootstrap samples (Fig.  2 ). Fig. 2 Bootstrap internal validation of the logistic model. Blue line shows mean ROC curve from 1000 Bootstrap resampling iterations; gray lines represent ROC curves from 1000 different bootstrap sampling datasets Bootstrap internal validation of the logistic model. Blue line shows mean ROC curve from 1000 Bootstrap resampling iterations; gray lines represent ROC curves from 1000 different bootstrap sampling datasets Calibration curves were used to evaluate the model’s calibration ability. The Hosmer-Lemeshow test results showed: χ²=10.529, P  = 0.230 > 0.05, indicating no difference between predicted probabilities and actual probabilities. After internal validation using Bootstrap resampling with 1000 iterations, the mean absolute error was 0.042. As shown in Fig.  3 , the calibration curve closely approximated both the ideal curve and the original curve, with a slope approaching 1. Therefore, the model was considered to have good accuracy. Fig. 3 Calibration curve of the combined model. Black dashed line shows ideal calibration, orange line shows apparent calibration curve, blue line shows bias-corrected calibration curve Calibration curve of the combined model. Black dashed line shows ideal calibration, orange line shows apparent calibration curve, blue line shows bias-corrected calibration curve

A total of 209 patients met the inclusion criteria and were included in the final analysis, with 61 (29.19%) experiencing spontaneous miscarriage and 148 (70.81%) not experiencing miscarriage. No significant differences were observed between groups in terms of age, body mass index, FSH, LH, E2, AMH, gravidity, or parity ( P  > 0.05) (Table  1 ). Table 1 Comparison of baseline characteristics between groups Variables Total ( n  = 209) Miscarriage ( n  = 61) Non-miscarriage ( n  = 148) Statistic P Age, Mean ± SD 30.03 ± 4.99 29.08 ± 4.64 30.42 ± 5.09 t=−1.77 0.078 BMI, Mean ± SD 23.43 ± 2.68 23.15 ± 2.37 23.54 ± 2.80 t = 0.97 0.333 FSH, Mean ± SD 6.29 ± 1.47 6.44 ± 1.66 6.24 ± 1.38 t=−0.92 0.361 LH, Mean ± SD 4.50 ± 1.92 4.82 ± 1.80 4.36 ± 1.95 t=−1.56 0.121 E2, Mean ± SD 35.23 ± 12.08 33.79 ± 10.50 35.83 ± 12.66 t = 1.11 0.269 AMH, Mean ± SD 3.18 ± 1.12 3.19 ± 1.12 3.17 ± 1.12 t=−0.10 0.923 Gravidity, M(Q₁, Q₃) 1.00(0.00, 2.00) 0.00(0.00, 1.00) 1.00(0.00, 2.00) Z=−0.66 0.510 Parity, n(%) Z=−0.75 0.454 0 160 (76.56) 49 (80.33) 111 (75.00) 1 36 (17.22) 8 (13.11) 28 (18.92) 2 13 (6.22) 4 (6.56) 9 (6.08)  t t-test, Z Mann-Whitney test, SD standard deviation M Median, Q ₁ 1 st Quartile, Q₃ 3 st Quartile Comparison of baseline characteristics between groups t t-test, Z Mann-Whitney test, SD standard deviation M Median, Q ₁ 1 st Quartile, Q₃ 3 st Quartile Statistically significant differences were observed in CA125, bilateral mean S/D ratio, bilateral mean PI, bilateral mean RI, and VAS scores ( P  < 0.05), while no significant differences were found in left endometrioma size, right endometrioma size, DE signs, left endometrioma presence, right endometrioma presence, or bilateral endometriomas ( P  > 0.05) (Table  2 ). Table 2 Comparison of clinical characteristics and uterine artery blood flow parameters between groups Variables Total ( n  = 209) Miscarriage ( n  = 61) Non-miscarriage ( n  = 148) Statistic P CA125, Mean ± SD 25.77 ± 12.01 33.53 ± 11.87 22.57 ± 10.54 t=−6.58 < 0.001 Bilateral mean S/D, Mean ± SD 6.24 ± 2.68 7.90 ± 3.26 5.56 ± 2.06 t=−5.19 < 0.001 Bilateral mean RI, Mean ± SD 0.81 ± 0.07 0.84 ± 0.06 0.80 ± 0.06 t=−4.47 < 0.001 Bilateral mean PI, Mean ± SD 2.30 ± 0.46 2.51 ± 0.43 2.22 ± 0.45 t=−4.35 < 0.001 VAS, M(Q₁,Q₃) 1.00(0.00,3.00) 3.00(1.00,4.00) 1.00(0.00,3.00) Z=−4.19 < 0.001 DE signs, n(%) χ²=2.08 0.149  Absent 161 (77.03) 43 (70.49) 118 (79.73)  Present 48 (22.97) 18 (29.51) 30 (20.27) Left endometrioma size, Mean ± SD 1.56 ± 1.30 1.71 ± 1.34 1.49 ± 1.28 t=−1.10 0.272 Right endometrioma size, Mean ± SD 1.43 ± 1.23 1.57 ± 1.44 1.38 ± 1.13 t=−0.92 0.358 Left endometrioma size, n(%) χ²=0.26 0.608  Absent 81(38.76) 22(36.07) 59(39.86)  Present 128(61.24) 39(63.93) 89(60.14) Right endometrioma size, n(%) χ²=0.69 0.407  Absent 80(38.28) 26(42.62) 54(36.49)  Present 129(61.72) 35(57.38) 94(63.51) Bilateral endometrioma size, n(%) χ²=0.01 0.914  Absent 160(76.56) 47(77.05) 113(76.35)  Present 49(23.44) 14(22.95) 35(23.65) t t-test , Z Mann-Whitney test , χ² Chi-square test , SD standard deviation , M Median , Q₁ 1 st Quartile , Q₃ 3rd Quartile Comparison of clinical characteristics and uterine artery blood flow parameters between groups t t-test , Z Mann-Whitney test , χ² Chi-square test , SD standard deviation , M Median , Q₁ 1 st Quartile , Q₃ 3rd Quartile Spontaneous miscarriage was used as the dependent variable, while clinical indicators with statistical significance in the aforementioned univariate analysis (CA125, bilateral S/D ratio mean, bilateral PI mean, and VAS) were included as independent variables in logistic regression analysis for multivariate analysis. Since the variance inflation factor (VIF) of bilateral RI mean was greater than 5, indicating strong multicollinearity with other variables, it was excluded from the multivariate analysis (Table  3 ). Table 3 Multicollinearity assessment results Variable VIF Tolerance VAS 2.305 0.434 CA125 1.339 0.747 Bilateral mean S/D ratio 4.380 0.228  Bilateral mean RI 5.200 0.192  Bilateral mean PI 2.612 0.383 Multicollinearity assessment results Multivariate logistic regression analysis revealed that bilateral S/D ratio mean and CA125 had significant effects on the occurrence of spontaneous miscarriage ( P  < 0.05). For each unit increase in bilateral S/D ratio mean and CA125, the likelihood of spontaneous miscarriage increased by 1.38-fold (95% CI: 1.15–1.66) and 1.08-fold (95% CI: 1.04–1.12), respectively, indicating that bilateral S/D ratio mean and CA125 are risk factors for spontaneous miscarriage. However, this study found no significant association between VAS and bilateral PI mean with spontaneous miscarriage ( P  > 0.05) (Table  4 ). Table 4 Factors associated with spontaneous miscarriage Variables β S.E Z P OR (95%CI) Intercept −5.45 1.11 −4.93 < 0.001 0.00 (0.00ཞ0.04) VAS 0.12 0.10 1.26 0.207 1.13 (0.93ཞ1.37) CA125 0.07 0.02 4.04 < 0.001 1.08 (1.04ཞ1.12) Bilateral S/D ratio mean 0.32 0.09 3.39 < 0.001 1.38 (1.15ཞ1.66) Bilateral PI mean 0.67 0.53 1.3 0.896 1.07 (0.38ཞ3.00) OR Odds Ratio, CI Confidence Interval Factors associated with spontaneous miscarriage OR Odds Ratio, CI Confidence Interval

This retrospective case-control study was conducted at the Fujian Maternity and Child Health Hospital between January 2022 and June 2024. The study included women with ovarian endometriomas who achieved natural conception and confirmed intrauterine pregnancy. Participants were classified into spontaneous miscarriage group and non-miscarriage group based on early pregnancy outcomes. Eligible patients were identified through systematic electronic medical record review and comprehensive database screening using diagnostic codes and ultrasound reports. Cases and controls were classified based on documented pregnancy outcomes at 12 weeks of gestation. Due to the retrospective design, detailed screening logs typical of prospective studies were not available, but all included cases ( n  = 209) had complete data for study variables as specified in exclusion criteria. The study was approved by the Ethics Committee of Fujian Maternity and Child Health Hospital (2025KY023) and conducted in accordance with the principles of the Declaration of Helsinki. Women aged 20–38 years. Menstrual cycle 21–35 days. Pre-pregnancy ultrasound showing unilateral or bilateral ovarian endometrioma (measuring < 4 cm in diameter). Natural conception. Confirmed intrauterine pregnancy. Women aged 20–38 years. Menstrual cycle 21–35 days. Pre-pregnancy ultrasound showing unilateral or bilateral ovarian endometrioma (measuring < 4 cm in diameter). Natural conception. Confirmed intrauterine pregnancy. Concurrent uterine fibroids (diameter ≥ 3 cm or submucosal fibroids), adenomyosis, intrauterine adhesions, uterine malformations, polycystic ovary syndrome, bilateral fallopian tube obstruction. Recurrent pregnancy loss, thrombophilia, hypertension, diabetes, thyroid dysfunction, antiphospholipid syndrome, or other autoimmune diseases. AMH < 1.1 ng/mL. Recent use of hormonal medications (oral contraceptives, GnRH agonists, etc.) or anticoagulants within 3 months. Abnormal male semen analysis (according to WHO 5th edition criteria). Missing data for study variables. Concurrent uterine fibroids (diameter ≥ 3 cm or submucosal fibroids), adenomyosis, intrauterine adhesions, uterine malformations, polycystic ovary syndrome, bilateral fallopian tube obstruction. Recurrent pregnancy loss, thrombophilia, hypertension, diabetes, thyroid dysfunction, antiphospholipid syndrome, or other autoimmune diseases. AMH < 1.1 ng/mL. Recent use of hormonal medications (oral contraceptives, GnRH agonists, etc.) or anticoagulants within 3 months. Abnormal male semen analysis (according to WHO 5th edition criteria). Missing data for study variables. Patients underwent blood testing for baseline hormone levels (FSH, LH, E2, AMH) on cycle days 2–5. Serum CA125 testing was performed during the non-menstrual phase within 3 months before pregnancy. Starting from cycle days 9–10, transvaginal ultrasound monitoring of follicular development, endometrial thickness, and ovarian endometrioma size was performed, along with evaluation of uterine artery blood flow parameters. Uterine artery blood flow parameter measurement was performed according to the methods described by Choi et al. and ISUOG Practice Guidelines [ 11 , 19 ]. Patients were positioned supine with empty bladders for transvaginal color Doppler ultrasound examination. The uterine artery was identified at the level of the internal cervical os using color Doppler, with the ascending branch of the uterine artery identified 1–2 cm lateral to the cervix. The pulsed Doppler sample volume was set to 2 mm with the sampling angle adjusted to < 30°. The timing of uterine artery Doppler measurements was standardized to the follicular phase (cycle days 9–10). This timing represents established methodology in reproductive research and provides optimal baseline conditions by minimizing hormonal fluctuations that occur during other cycle phases, which can significantly affect uterine artery impedance indices. At least three consecutive waveforms were obtained to measure the pulsatility index (PI), resistance index (RI), and peak systolic velocity/end-diastolic velocity ratio (S/D) of both left and right uterine arteries. The mean values of both sides were used for analysis. All ultrasound examinations were performed by two experienced ultrasonographers using the same equipment (GE Voluson E10, General Electric Company, USA) and probe (RIC5-9-D transvaginal probe, frequency 5–9 MHz). Deep endometriosis (DE) signs were determined through detailed gynecological examination, including bimanual and tri-manual examinations when necessary. Examination particularly focused on the uterosacral ligaments, rectovaginal septum, and posterior vaginal fornix for nodules, tenderness, ligament thickening, and structural irregularities [ 20 ]. Regarding folic acid supplementation, some patients began taking folic acid (0.4–0.8.4.8 mg daily) 3 months before pregnancy, while others started during the cycle of follicular monitoring. All patients received folic acid supplementation throughout pregnancy. Sexual intercourse guidance was provided based on follicular development: sexual intercourse every 2–3 days from the end of menstruation until ovulation, with additional intercourse within 2 days post-ovulation. Post-ovulation, patients received oral dydrogesterone (Duphaston ® ; Abbott B.V., The Netherlands) 10 mg twice daily (total 20 mg/day) continuing until 10 weeks of gestation. Blood β-hCG and progesterone levels were tested approximately 14 days post-ovulation to confirm pregnancy, with ultrasound confirmation of intrauterine pregnancy performed 28–35 days post-ovulation. Patients were followed until early pregnancy outcome determination. The primary outcome measure was early pregnancy loss, defined as a nonviable intrauterine pregnancy with either an empty gestational sac or a gestational sac containing an embryo or fetus without fetal heart activity within the first 12 6/7 weeks of gestation [ 21 ]. Gestational age was primarily determined by early pregnancy ultrasound dating using crown-rump length (CRL) measurement performed at 6–8 weeks of gestation, with last menstrual period (LMP) dating used for initial assessment but corrected when ultrasound measurements differed by >7 days. Assessment parameters included: Basic characteristics: age, body mass index (BMI), gravidity, parity, FSH, LH, E2, AMH. Clinical characteristics: dysmenorrhea VAS score, CA125, DE signs, left ovarian endometrioma size, right ovarian endometrioma size, presence of left endometrioma, presence of right endometrioma, presence of bilateral endometriomas. Uterine artery blood flow parameters (bilateral mean PI, bilateral mean RI, bilateral mean S/D). Basic characteristics: age, body mass index (BMI), gravidity, parity, FSH, LH, E2, AMH. Clinical characteristics: dysmenorrhea VAS score, CA125, DE signs, left ovarian endometrioma size, right ovarian endometrioma size, presence of left endometrioma, presence of right endometrioma, presence of bilateral endometriomas. Uterine artery blood flow parameters (bilateral mean PI, bilateral mean RI, bilateral mean S/D). Post-hoc power analysis was performed using G*Power 3.1.9.7 software with Type I error (α) set at 0.05 and Type II error (β) set at 0.20. With our achieved sample size of 209 patients (61 miscarriage cases and 148 controls), the study achieved > 85% statistical power for the logistic regression analysis, ensuring adequate power to detect clinically meaningful differences. All 209 included cases had complete datasets for study variables. No imputation methods were required, as cases with missing data were excluded during the selection process according to predefined exclusion criteria. Continuous variables following normal distribution were expressed as mean ± standard deviation (Mean ± SD), and independent samples t-test was used to analyze intergroup differences. Non-normally distributed variables were presented as median and interquartile range [M (Q₁, Q₃)], and nonparametric Mann-Whitney U test was employed for intergroup comparison. Categorical variables were described as frequency and percentage [n(%)]. Chi-square test was applied for unordered categorical variables, while nonparametric Mann-Whitney U test was used for ordinal categorical variables (i.e., ranked variables) to assess intergroup differences. Variables with statistical significance in univariate analysis were included in logistic regression for multivariate analysis to identify the final statistically significant factors. Receiver operating characteristic (ROC) curves were plotted using the proc package to calculate the area under the curve (AUC), optimal cutoff value, specificity, and sensitivity. Model accuracy was evaluated using the Hosmer-Lemeshow test and calibration curves. Internal validation of the model was conducted using Bootstrap resampling method. Logistic regression analysis was chosen because the primary outcome was a binary variable (miscarriage vs. non-miscarriage). Variables with p  < 0.10 in univariate analysis or of known clinical relevance were included in multivariate analysis to control for potential confounders. All model assumptions were assessed, including linearity, multicollinearity (VIF < 5), and sufficient events per variable. Model fitness and calibration were evaluated by Hosmer-Lemeshow test and calibration curve. A P-value < 0.05 was considered statistically significant. All analyses were performed using SPSS 25.0 and R version 4.4.0.

To our knowledge, this is the first study to systematically evaluate the combined predictive value of CA125 levels and uterine artery blood flow parameters for early pregnancy loss in women with endometriosis with natural conception. Our findings demonstrated that elevated CA125 levels and increased uterine artery S/D ratio are independent risk factors for early spontaneous miscarriage, with their combined application providing good risk stratification for miscarriage assessment. CA125 is an important biomarker reflecting endometriosis activity, primarily secreted by endometrial epithelial cells and mesoderm-derived epithelial cells [ 9 , 22 ]. A comprehensive review by Cao Y et al. suggested that CA125 has high diagnostic value in predicting pregnancy outcomes in threatened abortion, particularly excelling in identifying high-risk patients [ 23 ]. Our study also demonstrated that CA125 levels in the miscarriage group were significantly higher than in the non-miscarriage group, consistent with current research findings [ 24 ]. CA125 not only reflected disease activity but may also have been associated with pregnancy outcomes through several possible mechanisms: ①elevated CA125 levels reflect inflammation and immune responses in endometriosis patients, which may interfere with early embryonic development and implantation [ 9 , 25 ]; ②CA125 is associated with changes in the endometrial environment, affecting embryo implantation and early development [ 26 ]; ③elevated CA125 levels may be related to increased oxidative stress, leading to decreased embryo quality [ 27 ]. The uterine artery S/D ratio is an important indicator for evaluating uterine perfusion, reflecting uterine artery resistance and blood flow status [ 28 ]. Uterine artery Doppler examination serves as an important indicator for assessing placental blood flow perfusion, with S/D ratio closely related to pregnancy outcomes. Early studies established normal reference ranges for uterine artery S/D ratio, with classic research by Fleischer et al. [ 29 ] and Schulman et al. [ 30 ] determining S/D ratio >2.6 as the diagnostic criterion for abnormality. Multiple recent studies have confirmed the correlation between elevated uterine artery S/D ratio and adverse pregnancy outcomes: Adekanmi et al. [ 31 ] found that the second-trimester uterine artery S/D ratio was significantly higher in the adverse pregnancy outcome group compared to the normal group ( p = 0.001); Duragkar et al. [ 32 ] showed that elevated S/D ratio was positively correlated with adverse pregnancy outcomes. More direct evidence comes from Tian et al. [ 28 ], whose study clearly demonstrated that pregnant women who continued pregnancy had significantly lower uterine artery S/D ratios compared to the miscarriage group, consistent with our findings. Possible explanations include: ①increased S/D ratio indicates increased uterine blood flow resistance, leading to insufficient uterine perfusion and affecting early embryonic development [ 28 ]; ②endometriosis-related inflammatory factors may cause uterine vascular constriction and microcirculation disorders, manifesting as increased S/D ratio [ 33 ]; ③increased S/D ratio may reflect abnormal uterine smooth muscle function, affecting uterine peristalsis and blood flow perfusion [ 34 ]. Elevated levels of CA125 and increased uterine artery S/D ratio may represent different aspects of systemic pathological changes in patients with ovarian endometriomas < 4 cm in diameter. The former reflects biochemical markers of inflammatory damage, while the latter represents physical changes in vascular function. As a glycoprotein, CA125 is primarily produced by epithelial cells in body cavities and is elevated in patients with ovarian endometriomas-type endometriosis, reflecting the activity level of ectopic endometrial tissue on the cyst wall and local inflammatory status [ 9 , 22 ]. Concurrently, the increase in the uterine artery S/D ratio directly reflects the rise in vascular resistance and abnormalities in hemodynamics. Studies have shown that even small ovarian cysts can cause significant increases in contralateral uterine artery blood flow resistance. Waratani et al. found that 90% of patients with ovarian endometriomas exhibited an abnormal uterine artery resistance index ≥ 0.8 [ 17 ], indicating the prevalence of vascular dysfunction in ovarian endometriomas-type endometriosis. This alteration in vascular function may stem from the release of inflammatory mediators around the cyst, affecting the endothelial function and contractility of adjacent vessels [ 33 ]. In patients with < 4 cm cysts, the simultaneous elevation of these two indicators holds particular clinical significance. Although small cysts are less than 4 cm, their biological activity may still be considerable, and persistent inflammatory responses can not only lead to CA125 release but also impact local vascular function [ 6 , 9 ]. The combined detection of these parameters provides a more comprehensive assessment of the physiological status in patients with < 4 cm ovarian endometriomas, offering more reliable evidence for individualized pregnancy monitoring and management in clinical practice. Our study systematically evaluated the diagnostic performance of bilateral mean S/D ratio and CA125 through ROC curve analysis, providing insights into their clinical utility for predicting early pregnancy loss in women with endometriosis. The bilateral mean S/D ratio achieved an AUC of 0.75 (95% CI: 0.67–0.82) with an optimal cutoff value of 6.49, demonstrating sensitivity of 0.61 (95% CI: 0.48–0.73) and specificity of 0.77 (95% CI: 0.70–0.84). This performance profile suggests moderate diagnostic capability with acceptable specificity. CA125 demonstrated comparable diagnostic value with an AUC of 0.76 (95% CI: 0.69–0.83) and an optimal cutoff value of 23.55 U/mL, exhibiting relatively high sensitivity of 0.84 (95% CI: 0.74–0.93) but moderate specificity of 0.60 (95% CI: 0.52–0.68). When combining both parameters through multivariate logistic regression modeling, we observed improved risk stratification capability. The combined model demonstrated an AUC of 0.85 (95% CI: 0.79–0.90), achieving sensitivity of 0.87 (95% CI: 0.78–0.95) and specificity of 0.72 (95% CI: 0.65–0.80) at a probability cutoff of 0.223. DeLong test analysis showed statistically significant improvement compared to bilateral S/D ratio alone ( P  = 0.035) and CA125 alone ( P  = 0.043). This combined approach may help address some limitations of individual parameters: CA125’s higher sensitivity with lower specificity can be complemented by bilateral mean S/D ratio’s better specificity, potentially resulting in more balanced diagnostic performance while reducing some false-positive results. Bootstrap resampling validation with 1000 iterations demonstrated reasonably stable performance with an AUC of 0.830. Calibration analysis using the Hosmer-Lemeshow test (χ²=10.529, P  = 0.230) indicated acceptable agreement between predicted probabilities and actual outcomes, with a mean absolute error of 0.042 after internal validation. These validation results suggest the predictive model has reasonable reliability, though further external validation remains necessary. These findings may have potential clinical implications for women with small ovarian endometriomas (< 4 cm) seeking natural conception. Based on our single-center retrospective study, we propose preliminary clinical management considerations, though these suggestions should be interpreted with caution and require prospective validation before any routine implementation. The identified cutoff values (CA125 ≥ 23.55 U/mL and bilateral mean S/D ratio ≥ 6.49) might potentially assist in developing risk-stratified monitoring protocols. For standardized assessment, CA125 evaluation during the non-menstrual phase and uterine artery Doppler during follicular phase (cycle days 9–10) may provide appropriate timing for these measurements. Women with both parameters below the identified thresholds might be candidates for standard antenatal care monitoring. Those with one elevated parameter could potentially benefit from enhanced surveillance, including bi-weekly β-hCG and progesterone monitoring during pregnancy weeks 4–12. Women presenting with both parameters above the cutoff values might be considered for more intensive monitoring strategies, potentially including weekly hormone assessments, individualized consideration for extended luteal phase support duration, and closer ultrasound surveillance beginning at 6 weeks gestation. Additional considerations for higher-risk patients might include reasonably increasing the monitoring frequency of β-hCG and progesterone levels, appropriately extending luteal phase support therapy based on individual risk assessment, considering low-molecular-weight heparin therapy following individualized thrombophilia evaluation, and early referral to specialized high-risk pregnancy management. The relatively accessible nature of both CA125 testing and uterine artery Doppler assessment makes this screening approach feasible in most healthcare settings, potentially enabling more targeted resource allocation by identifying patients who might benefit from intensive monitoring while avoiding unnecessary interventions in lower-risk patients. Our findings may be applicable to women with confirmed small ovarian endometriomas (< 4 cm) in routine clinical practice settings where comprehensive surgical staging through laparoscopy is not routinely performed for patients seeking natural conception, representing a common clinical scenario encountered in reproductive medicine practice. However, several important limitations must be acknowledged. Our assessment of deep endometriosis was limited to clinical examination findings rather than advanced imaging evaluation or laparoscopic confirmation, reflecting real-world clinical management but potentially underestimating disease extent. Some patients in our cohort may have had concurrent minimal peritoneal endometriosis that remained undiagnosed due to the absence of systematic laparoscopic evaluation. Additionally, our predictive model addresses the practical clinical question of risk assessment in patients with known small endometriomas, which represents one specific clinical scenario, and may not be generalizable to other patient populations. It is crucial to emphasize that these are preliminary observations from a single-center cohort with modest diagnostic performance and should complement, not replace, comprehensive clinical assessment. The retrospective nature of our study limits our ability to establish causation, and the generalizability of these findings requires validation through larger prospective multicenter studies across diverse populations. Clinicians should exercise careful judgment when considering these preliminary findings, taking into account individual patient factors and maintaining established clinical protocols until robust prospective validation is achieved. The identified risk stratification parameters should be interpreted within the specific context of this study population when considering implementation of risk-based monitoring protocols. The strengths of this study include: ①relatively strict inclusion and exclusion criteria were used to minimize major confounding factors that might affect pregnancy outcomes; ②multiple potential risk factors were evaluated, with particular attention to the combined application of uterine artery blood flow parameters and CA125, providing reference data for this field; ③cutoff values for CA125 and uterine artery S/D ratio were determined through ROC curve analysis, providing clinical reference; ④all study subjects achieved natural conception, avoiding potential confounding factors from assisted reproductive technology. The limitations of this study include: ①inherent limitations of retrospective study design may lead to selection bias and inability to completely control all potential confounding factors; ②as a single-center study, the patient population is relatively concentrated, potentially limiting external validity and generalizability, requiring multicenter studies for further validation; ③relatively short follow-up period, focusing primarily on early pregnancy outcomes; ④study subjects were limited to endometriosis patients with ovarian endometriomas < 4 cm in diameter, with deep endometriosis assessment limited to clinical examination and no systematic evaluation for peritoneal disease, potentially limiting generalizability to all endometriosis subtypes; ⑤relatively small sample size ( n  = 209), which may affect statistical power and requires larger sample studies to validate these conclusions; ⑥the combined predictive model requires external validation to confirm its clinical applicability. It should be noted that the retrospective case-control design of this study limits our ability to establish causal relationships between the measured parameters and pregnancy outcomes. The observed associations between elevated CA125 levels, increased uterine artery S/D ratio, and spontaneous miscarriage may reflect correlation rather than causation. While our findings suggest these parameters are independently associated with increased miscarriage risk, future prospective studies with larger sample sizes are needed to confirm potential causal mechanisms and validate these associations in different populations.

This retrospective case-control study demonstrates that serum CA125 levels and uterine artery S/D ratio are independently associated with early spontaneous miscarriage risk in women with small ovarian endometriomas (< 4 cm) who conceive naturally. In our cohort of 209 women, CA125 ≥ 23.55 U/mL demonstrated good discriminatory ability (AUC = 0.76, sensitivity = 0.84, specificity = 0.59) for identifying women at increased miscarriage risk. Similarly, uterine artery S/D ratio ≥ 6.49 showed comparable predictive performance (AUC = 0.75, sensitivity = 0.70, specificity = 0.77). Combined assessment of both parameters significantly enhanced predictive accuracy (AUC = 0.85, sensitivity = 0.87, specificity = 0.72), indicating superior risk stratification capability compared to individual markers. Multivariate analysis confirmed that both CA125 and uterine artery S/D ratio remained independently associated with early spontaneous miscarriage risk after adjusting for potential confounders, demonstrating their independent value in risk assessment. These findings have important clinical implications for the management of women with small endometriomas seeking natural pregnancy. The identified biomarkers are readily accessible through standard clinical assessments and could facilitate implementation of risk-stratified monitoring protocols. Women with elevated parameters may benefit from enhanced surveillance, optimized luteal phase support, and specialized management strategies, while those with normal values can receive standard antenatal care. Given the retrospective single-center design of this study, findings require further validation in larger prospective multicenter studies. In conclusion, combined evaluation of serum CA125 and uterine artery S/D ratio provides a practical approach for early identification of women at high risk for early spontaneous miscarriage among those with small ovarian endometriomas and natural pregnancy. These findings support the development of evidence-based risk stratification protocols that could improve pregnancy outcomes through targeted interventions.

Endometriosis is a common gynecological condition characterized by the presence of endometrial-like tissue outside the uterine cavity, affecting approximately 10% of women of reproductive age worldwide [ 1 ]. This condition not only causes dysmenorrhea, chronic pelvic pain, and dyspareunia but is also closely associated with increased risks of infertility and miscarriage [ 2 , 3 ]. Studies have demonstrated that the spontaneous miscarriage rate in women with endometriosis is higher than in the general population, with those achieving natural conception being at particularly elevated risk, although the specific mechanisms and risk factors remain incompletely understood [ 4 , 5 ]. Endometriosis-associated inflammation and immune dysregulation may interfere with early embryonic development and implantation processes [ 6 ], with extracellular vesicle-mediated immune dysfunction potentially exerting adverse effects on embryo implantation [ 7 ]. Additionally, factors such as altered endometrial environment, abnormal uterine smooth muscle contractions, and vascular dysfunction associated with endometriosis may contribute to increased miscarriage risk [ 8 , 9 ]. The uterine artery, as the primary vascular supply to the uterus, plays a crucial role in endometrial growth and pregnancy maintenance. Doppler ultrasound evaluation of uterine artery blood flow provides important hemodynamic information, including pulsatility index (PI), resistance index (RI), and systolic/diastolic ratio (S/D) [ 10 ]. Research by Choi et al. indicated that uterine artery blood flow parameters could serve as potential predictors of endometrial receptivity and pregnancy outcomes [ 11 ]. Fan et al. further demonstrated that uterine artery blood flow characteristics are closely associated with embryo implantation and early pregnancy maintenance [ 12 ]. Cancer Antigen 125 (CA125) is an important biomarker for the diagnosis and monitoring of endometriosis, with levels reflecting disease activity and inflammatory status [ 13 , 14 ]. Although the specificity of CA125 for endometriosis diagnosis remains controversial, it continues to hold significant value in endometriosis management [ 15 ]. Recent studies on biomarkers predicting assisted reproductive technology outcomes have shown that CA125 levels on the day of hCG administration or oocyte retrieval correlate with pregnancy outcomes [ 16 ]. However, the role of CA125 in predicting early pregnancy miscarriage risk in endometriosis patients has not been fully evaluated. Research on uterine artery blood flow parameters in endometriosis patients remains relatively limited. Waratani et al. [ 17 ] found that the pulsatility index and resistance index of the uterine artery ipsilateral to the endometrioma were significantly higher than those on the contralateral side in non-pregnant endometriosis patients, suggesting increased uterine artery vascular resistance in these patients. Regarding pregnancy, Salmeri et al. [ 18 ] reported abnormal uterine artery pulsatility index in patients with moderate-to-severe endometriosis, associated with impaired placental perfusion in the second and third trimesters. However, studies investigating changes in uterine artery blood flow parameters during early pregnancy in endometriosis patients and their predictive value for spontaneous miscarriage risk remain scarce, with even fewer reports on the combined use of blood flow parameters and CA125 for predicting early miscarriage risk. Therefore, this study aimed to investigate the risk factors for early miscarriage in endometriosis patients with natural conception, with particular focus on the predictive value of uterine artery blood flow parameters and CA125 levels for early miscarriage risk, providing clinicians with more precise assessment tools for early intervention in high-risk patients.