Development of a clinical prediction model for pathological upgrade from endometrial atypical hyperplasia to endometrial cancer: a retrospective Eastern Chinese cohort study.

Chong Fan: Writing – original draft, Software, Methodology, Data curation. Xiangjun Zhang: Writing – original draft, Software, Data curation. Mengshan Li: Formal analysis, Data curation. Wenting Wang: Resources, Investigation. Zhenzhen Li: Project administration. Qianqian Lu: Visualization. Fang Teng: Writing – review & editing, Validation, Funding acquisition, Conceptualization. Lili Ge: Writing – review & editing, Supervision, Conceptualization.

Based on the inclusion and exclusion criteria, a total of 519 subjects were ultimately enrolled in this study, comprising 413 (79.58%) in the EAH group and 106 (20.42%) in the EC group. Regarding the methods of endometrial biopsy: in the EC group, 29 cases were obtained via dilatation and curettage, and 77 cases via hysteroscopy. In the EAH group, 171 cases were obtained via dilatation and curettage, 169 cases via hysteroscopy, and for the remaining individuals, the method of acquisition was either not recorded or involved other approaches. The baseline characteristics of enrolled patients are summarized in Table 1 . The mean age was significantly lower in the EAH group (49.63±6.85 years) compared to the EC group (51.44±8.42 years, P =.042). Similarly, preoperative BMI values were significantly more inferior in the EAH group than in the EC group (24.80±3.50 kg/m 2 vs 25.74±4.27 kg/m 2 , P =.038). In the EAH group, the proportions of patients with hypertension, diabetes mellitus, and history of tumors were 22.52% (n=93), 5.81% (n=24), and 4.36% (n=18), respectively, the corresponding proportions were 25.47% (n=29), 10.38% (n=12), and 6.60% (n=7) in the EC group. However, these differences were not statistically significant (all P >.05). For reproductive history, the parity showed a significant difference ( P =.025). Besides, the EC group exhibited a significantly higher proportion of menopausal status (n=47, 44.34%) compared to the EAH group (n=103, 24.94%, P <.0001). Additionally, the endometrial thickness measured by pelvic ultrasound was greater in the EC group (0.93±0.33 mm vs 1.05±0.42 mm, P =.003). There was no significant difference in the history of AUB between the two groups ( P =.069). Also, the incidence of both uterine fibroids (51.33% vs 49.06%, P =.676) and adenomyosis (44.31% vs 38.68%, P =.296) manifested no statistically significant differences. Table 1 Baseline characteristics of the enrolled population Table 1 dummy alt text Variables EAH (n=413) EC (n=106) P value Age (years) 49.63±6.85 51.44±8.42 .042 BMI (kg/m 2 ) 24.80±3.50 25.74±4.27 .038 Hypertension 93 (22.52%) 27 (25.47%) .520 Diabetes mellitus 24 (5.81%) 11 (10.38%) .094 History of tumors 18 (4.36%) 7 (6.60%) .335 Gravidity .252  0–1 92 (22.28%) 29 (27.36%)  2 137 (33.17%) 39 (36.79%)  ≥3 184 (44.55%) 38 (35.85%) Parity .025  0 21 (5.08%) 3 (2.83%)  1 275 (66.59%) 85 (80.19%)  ≥2 117 (28.33%) 18 (16.98%) Menopausal status 103 (24.94%) 47 (44.34%) <.0001 History of AUB 295 (71.43%) 85 (80.19%) .069 Endometrial thickness (cm) 0.93±0.33 1.05±0.42 .003 Uterine fibroids 212 (51.33%) 52 (49.06%) .676 Adenomyosis 183 (44.31%) 41 (38.68%) .296 AUB , abnormal uterine bleeding; BMI , body mass index; EAH , endometrial atypical hyperplasia; EC , endometrial cancer. The bold values are statistically significant ( P < 0.05) Fan et al. Development of a clinical prediction model for pathological upgrade from endometrial atypical hyperplasia to endometrial cancer. Am J Obstet Gynecol 2026. Baseline characteristics of the enrolled population AUB , abnormal uterine bleeding; BMI , body mass index; EAH , endometrial atypical hyperplasia; EC , endometrial cancer. The bold values are statistically significant ( P < 0.05) Among all enrolled patients with EC, the clinicopathological characteristics were described ( Table 2 ). The final carcinoma histocytes included 104 (98.12%) endometrioid EC, 1 (0.94%) undifferentiated carcinoma EC, and 1 (0.94%) carcinosarcoma. According to the degree of differentiation, EC was classified into grade 1 (G1), G2, and G3, which comprised 58 (54.72%), 32 (30.19%), and 1 (0.94%) patients, respectively, in this study. The presence of deep myometrial invasion was observed in 13 cases (12.26%). The number of EC patients with stage Ia, Ib, II, and III were 91 (85.85%), 10 (9.43%), 1 (0.94%), and 4 (3.78%). Besides, the present of EC patients with lymphovascular space invasion (LVSI) were 5 in 59 assessed individuals. Meanwhile, the present of positive lymph node status were 3 in 55 assessed individuals. Moreover, there were 28, 26, 18, and 4 patients in 76 assessed individuals with a maximum tumor diameter of 0 to 1 cm, 1.1 to 2 cm, 2.1 to 3 cm, and over 3 cm, respectively. Of the 519 patients, 255 (49.13%) underwent preoperative magnetic resonance imaging (MRI), comprising 73 patients in the EC group and 182 patients in the EAH group. Among those preoperatively diagnosed as EAH by MRI, 39 (79.59%) were confirmed as EAH and 10 (20.41%) as EC postoperatively; whereas for cases with preoperative MRI indicating EC, postoperative pathology confirmed EAH in 12 (22.22%) and EC in 42 (77.77%) cases. Table 2 Clinicopathological characteristics in patients with endometrial cancer at time of final pathology Table 2 dummy alt text Variables Frequency Percent (%) Final carcinoma histotype  Endometrioid 104 98.12%  Undifferentiated 1 0.94%  Carcinosarcoma 1 0.94% Carcinoma grade  1 58 54.72%  2 32 30.19%  3 1 0.94%  Not assessed 15 14.15% Carcinoma stage  Ia 91 85.85%  Ib 10 9.43%  II 1 0.94%  IIIC 4 3.78% Lymphovascular space invasion  Absent 54 50.94%  Present 5 4.72%  Not assessed 47 44.34% Lymph node status  Negative 52 49.06%  Positive 3 2.83%  Not assessed 51 48.11% Tumor size (cm)  0–1.0 28 26.41%  1.1–2.0 26 24.53%  2.1–3.0 18 16.98%  >3.0 4 3.78%  Not assessed 30 28.30% Fan et al. Development of a clinical prediction model for pathological upgrade from endometrial atypical hyperplasia to endometrial cancer. Am J Obstet Gynecol 2026. Clinicopathological characteristics in patients with endometrial cancer at time of final pathology Subsequently, we compared preoperative hematologic test between the two groups ( Table 3 ). First, comparison of complete blood count parameters revealed significantly higher hemoglobin (130.16±15.69 g/L vs 121.74±16.77 g/L, P <.0001) and platelet count (255.71±93.19 × 10 9 /L vs 232.47±71.49 × 10 9 /L, P =.006) in the EC group than the EAH group. However, no statistically significant differences were observed in white blood cell (WBC) counts, neutrophils, lymphocyte, and monocyte between the groups (all P >.05). Then, the results of fasting blood glucose (FBG) displayed higher in the EC group (5.65±1.00 mmol/L vs 5.30±1.04 mmol/L, P =.002). Meanwhile, total cholesterol (TC) in the EC group also higher than that in the EAH group (5.34±1.03 mmol/L vs 4.82±0.91 mmol/L, P <.0001), but the EC group and the EAH group exhibited comparable triglyceride (1.58±0.97 mmol/L vs 1.51±1.30 mmol/L, P =.620). Regarding coagulation function tests, fibrinogen (2.81±0.64 g/L vs 2.44±0.49 g/L, P <.0001) were significantly elevated in the EC group compared with the EAH group. Nevertheless, the results of activated partial thromboplastin time (APTT), prothrombin time (PT), and D-dimer indicated no significant differences between the two groups. As for tumor biomarker tests, only human epididymis protein 4 (HE4) demonstrated lower in the EAH group (62.32±32.69 pmol/L vs 51.38±18.24 pmol/L, P =.003). Table 3 Preoperative hematologic testing of the enrolled population Table 3 dummy alt text Variables EAH (n=413) EC (n=106) P value WBC (× 10 9 /L) 5.73±1.73 5.79±1.39 .742 Hemoglobin (g/L) 121.74±16.77 130.16±15.69 <.0001 Platelet (× 10 9 /L) 232.47±71.49 255.71±93.19 .006 Neutrophils (× 10 9 /L) 3.64±1.56 3.67±1.25 .819 Lymphocyte (× 10 9 /L) 1.63±0.53 1.68±0.51 .353 Monocyte (× 10 9 /L) 0.34±0.24 0.31±0.09 .107 FBG (mmol/L) 5.30±1.04 5.65±1.00 .002 TC (mmol/L) 4.82±0.91 5.34±1.03 <.0001 Triglyceride (mmol/L) 1.51±1.30 1.58±0.97 .620 PT (s) 11.47±0.70 11.55±0.57 .247 APTT (s) 27.08±2.91 26.79±2.33 .292 Fibrinogen (g/L) 2.44±0.49 2.81±0.64 <.0001 D-dimer (mg/L) 0.29±0.34 0.34±0.30 .167 HE4 (pmol/L) 51.38±18.24 62.32±32.69 .003 AFP (ng/mL) 2.86±1.61 3.09±1.64 .233 CEA (ng/mL) 1.49±0.97 1.57±2.12 .632 CA125 (U/mL) 30.23±72.69 26.47±48.34 .649 CA153 (U/mL) 10.27±6.11 11.50±6.11 .107 CA199 (U/mL) 11.98±13.55 43.27±217.45 .178 NLR 2.46±1.52 2.45±1.61 .948 LMR 5.28±2.01 5.78±1.88 .021 PLR 154.96±63.88 163.42±70.79 .237 SII 48.42±29.34 47.93±24.87 .875 SIRI 0.07±0.05 0.06±0.03 .021 TyG 1.17±0.64 1.33±0.61 .034 TyG-BMI 29.67±17.83 34.92±17.86 .009 AFP , alpha-fetoprotein; APTT , activated partial thromboplastin time; CA , cancer antigen; CEA , carcinoembryonic antigen; FBG , fasting blood glucose; HE4 , human epididymis protein 4; NLR , neutrophil-to-lymphocyte ratio; LMR , lymphocyte-to-monocyte ratio; PLR , platelet-to-lymphocyte ratio; PT , prothrombin time; SII , systemic immune-inflammation Index; SIRI , systemic inflammation response index; TC , total cholesterol; WBC , white blood cells; TyG , triglyceride glucose. The bold values are statistically significant ( P < 0.05) Fan et al. Development of a clinical prediction model for pathological upgrade from endometrial atypical hyperplasia to endometrial cancer. Am J Obstet Gynecol 2026. Preoperative hematologic testing of the enrolled population AFP , alpha-fetoprotein; APTT , activated partial thromboplastin time; CA , cancer antigen; CEA , carcinoembryonic antigen; FBG , fasting blood glucose; HE4 , human epididymis protein 4; NLR , neutrophil-to-lymphocyte ratio; LMR , lymphocyte-to-monocyte ratio; PLR , platelet-to-lymphocyte ratio; PT , prothrombin time; SII , systemic immune-inflammation Index; SIRI , systemic inflammation response index; TC , total cholesterol; WBC , white blood cells; TyG , triglyceride glucose. The bold values are statistically significant ( P < 0.05) Immediately following this, the calculated indexes were compared. No significant differences were detected in NLR, PLR, and SII between the two groups. However, LMR (5.78±1.88 vs 5.28±2.01, P =.021), TyG (1.33±0.61 vs 1.17±0.64, P =.034), and TyG-BMI (34.92±17.86 vs 29.67±17.83, P =.009) were all elevated in the EC group than those in the EAH group, while SIRI was lower in the EC group. To identify risk factors for pathological upgrade from preoperative EAH to EC, multivariate logistic regression analysis on the parameters above mentioned that demonstrated statistically significant differences were performed. Parameters showing significant differences in the logistic regression analysis are displayed in Table 4 . Six risk factors were identified, including preoperative endometrial thickness (OR = 2.894, 95% CI=1.293–6.475, P =.010), hemoglobin (OR = 1.025, 95% CI=1.004–1.047, P =.020), HE4 (OR = 1.023, 95% CI=1.010–1.036, P <.0001), TC (OR = 1.607, 95% CI=1.179–2.189, P =.003), fibrinogen (OR = 2.573, 95% CI=1.502–4.406, P =.001), and LMR (OR = 1.210, 95% CI=1.040–1.408, P =.013). Table 4 The results of logistic regression analysis Table 4 dummy alt text Variables B SE OR 95% CI P value Endometrial thickness (mm) 1.063 0.411 2.894 1.293–6.475 .010 Hemoglobin (g/L) 0.025 0.011 1.025 1.004–1.047 .020 HE4 (pmol/L) 0.023 0.006 1.023 1.010–1.036 <.0001 TC (mmol/L) 0.474 0.158 1.607 1.179–2.189 .003 Fibrinogen (g/L) 0.945 0.274 2.573 1.502–4.406 .001 LMR 0.191 0.077 1.210 1.040–1.408 .013 CI , confidence interval; LMR , lymphocyte-to-monocyte ratio; OR , odds ratio; SE , standard error; TC , total cholesterol. Fan et al. Development of a clinical prediction model for pathological upgrade from endometrial atypical hyperplasia to endometrial cancer. Am J Obstet Gynecol 2026. The results of logistic regression analysis CI , confidence interval; LMR , lymphocyte-to-monocyte ratio; OR , odds ratio; SE , standard error; TC , total cholesterol. The ROC curve was constructed based on the above findings. The results ( Table 5 ) showed that the AUC for endometrial thickness was 0.577, corresponding 95% CI was 0.509–0.644. Besides, the AUC for hemoglobin was 0.662 (95% CI: 0.596–0.727). Regarding the other results of blood tests, the AUC for HE4, TC, fibrinogen, and LMR were 0.619 (95% CI: 0.543–0.694, sensitivity: 0.386, specificity: 0.844), 0.655 (95% CI: 0.588–0.722), 0.654 (95% CI: 0.585–0.722), and 0.630 (95% CI: 0.563–0.696,), respectively. In addition, the combined diagnostic model of these biomarkers achieved an AUC of 0.794 (95% CI: 0.739–0.849, sensitivity: 0.759, specificity: 0.730). Table 5 The results of ROC curve analysis Table 5 dummy alt text Variables AUC Cut-off Sensitivity Specificity 95% CI P value Endometrial thickness (mm) 0.577 10.300 0.819 0.338 0.509–0.644 .035 Hemoglobin (g/L) 0.662 123.500 0.795 0.468 0.596–0.727 <.0001 HE4 (pmol/L) 0.619 59.800 0.386 0.844 0.543–0.694 .001 TC (mmol/L) 0.655 4.860 0.669 0.559 0.588–0.722 <.0001 Fibrinogen (g/L) 0.654 2.751 0.530 0.726 0.585–0.722 <.0001 LMR 0.630 5.690 0.542 0.681 0.563–0.696 <.0001 Combined 0.794 0.759 0.730 0.739–0.849 <.0001 AUC , aera under the curve; CI , confidence interval; HE4 , human epididymis protein 4; LMR , lymphocyte-to-monocyte ratio; ROC , receiver operating characteristic; TC , total cholesterol. Fan et al. Development of a clinical prediction model for pathological upgrade from endometrial atypical hyperplasia to endometrial cancer. Am J Obstet Gynecol 2026. The results of ROC curve analysis AUC , aera under the curve; CI , confidence interval; HE4 , human epididymis protein 4; LMR , lymphocyte-to-monocyte ratio; ROC , receiver operating characteristic; TC , total cholesterol. Additionally, we performed further analysis focusing on women who underwent preoperative pelvic MRI scans. Those without a preoperative MRI were excluded. The final study population comprised 73 patients in the EC group and 182 patients in the EAH group. Logistic regression analysis was performed, followed by ROC curve analysis using the methodology described above. The final results ( Table 6 ) manifested that the AUC for MRI was 0.779, with a sensitivity of 0.580 and a specificity of 0.929, corresponding 95% CI was 0.706–0.851. The combined model integrating MRI with hemoglobin, TC, and fibrinogen levels achieved an AUC of 0.871 (95% CI: 0.819–0.922), with a sensitivity of 0.696 and a specificity of 0.894. The optimal cut-off values were >135.500 g/L for hemoglobin, >4.855 mmol/L for TC, and >2.910 g/L for fibrinogen. Table 6 The results of ROC curve analysis (included preoperative MRI) Table 6 dummy alt text Variables AUC Cut-off Sensitivity Specificity 95% CI P value Preoperative MRI 0.779 0.580 0.929 0.706–0.851 <.0001 Hemoglobin (g/L) 0.680 135.500 0.507 0.788 0.606–0.753 <.0001 TC (mmol/L) 0.631 4.855 0.667 0.565 0.553–0.708 .002 Fibrinogen (g/L) 0.665 2.910 0.435 0.829 0.587–0.743 <.0001 Combined 0.871 0.696 0.894 0.819–0.922 <.0001 AUC , aera under the curve; CI , confidence interval; HE4 , human epididymis protein 4; MRI , magnetic resonance imaging; ROC , receiver operating characteristic. Fan et al. Development of a clinical prediction model for pathological upgrade from endometrial atypical hyperplasia to endometrial cancer. Am J Obstet Gynecol 2026. The results of ROC curve analysis (included preoperative MRI) AUC , aera under the curve; CI , confidence interval; HE4 , human epididymis protein 4; MRI , magnetic resonance imaging; ROC , receiver operating characteristic.

This study was conducted in accordance with the Declaration of Helsinki and approved by the Medical Ethics Committee of Nanjing Women and Children’s Healthcare Hospital (Approval Number: 2024KY-090). This study included patients admitted to our hospital for EAH who underwent hysterectomy from January 1, 2016 to June 30, 2024. Based on postoperative pathological results, all patients were classified into either the EAH group or the EC group. Following this stratification, we systematically compared preoperative general characteristics, clinical manifestations, laboratory test results, imaging examination, and postoperative pathological features between the EAH and EC groups. Differential indicators identified through these comparisons were subsequently subjected to regression analysis and receiver operating characteristic (ROC) curve analysis, ultimately identified preoperative high-risk factors associated with pathological upgrade from EAH to EC. Eligibility criteria were defined as: (1) patients who underwent hysterectomy from January 1, 2016 to June 30, 2024; (2) histologically confirmed diagnosis of EAH through preoperative endometrial sampling; (3) hysterectomy performed as primary EAH management; and (4) availability of complete clinical and pathological data sets verified through medical records. Exclusion criteria consisted of: (1) preoperative histologic diagnosis other than EAH or confirmed EC; (2) preoperative EAH diagnosis with subsequent diagnostic procedures (diagnostic curettage or hysteroscopy) without hysterectomy; (3) missing data in electronic medical record system; and (4) active systemic infections within a month prior to surgery. Clinical data were systematically collected from electronic medical record system, capturing: demographic parameters including age (years), height (m), and weight (kg); clinical characteristics encompassing gravidity, parity, menopausal status, history of abnormal uterine bleeding, history of progesterone therapy, and comorbidities (malignancy, hypertension, and diabetes); imaging parameters comprising pelvic ultrasound measurements results; preoperative laboratory test indicators (blood routine examination, serum biochemical index, coagulation parameters, and tumor markers); and postoperative histopathological findings. Besides, the mode of biopsy was recorded by reviewing the electronic medical records . Derived metabolic indices and inflammatory markers were algorithmically computed, including BMI, Neutrophil-to-Lymphocyte Ratio (NLR), Platelet-to-Lymphocyte ratio (PLR), Lymphocyte-to-Monocyte Ratio (LMR), Systemic Immune-inflammation Index (SII), Systemic Inflammation Response Index (SIRI), Triglyceride Glucose (TyG), and TyG-BMI. We calculated the above indices according to the following equations: BMI = weight / (height) 2 ; NLR = neutrophil count / lymphocyte count; PLR = platelet count / lymphocyte count; LMR = lymphocyte count / monocyte count; SIRI = (neutrophil count × monocyte count) / lymphocyte count; SII = (neutrophil count × platelet count) / lymphocyte count; TyG = Ln [fasting TG (mg/dl) × FBG (mg/dl)] / 2; TyG-BMI = TyG × BMI. Statistical analyses were performed using IBM SPSS Statistics® 25.0 (Armonk, NY, USA: IBM Corp.). Parameters with more than 30% missing data were excluded from the analysis except for those deemed clinically significant, while the remaining missing values in other parameters were imputed using SPSS. Continuous variables underwent normality assessment via Shapiro-Wilk tests. Data meeting normality assumptions were compared using independent Student's t -tests, while non-normally distributed data were analyzed with Mann-Whitney U tests. Categorical variables were evaluated through chi-square tests. Variables demonstrating significant intergroup differences were subsequently incorporated into logistic regression analysis and corresponding standard error (SE), odds ratio (OR), and 95% confidence interval (CI) were calculated. The diagnostic efficacy of individual and combined parameters was quantified using ROC curve analysis, with area under the curve (AUC) calculations. A two-tailed P -value <.05 was defined statistical significance throughout all analyses.

Current evidence indicates that EAH may represent a precursor lesion to EC. This study examined patients undergoing hysterectomy for EAH, stratified by postoperative pathology into EAH and EC cohorts, to systematically investigate risk factors for EC. Our findings suggest significant differences in clinical characteristics and laboratory parameters between groups, with endometrial thickness, menopausal status, platelet, TC, fibrinogen, and LMR emerging as potential predictors of pathological upgrade from EAH to EC. Multivariate analysis identified six independent predictors: endometrial thickness, hemoglobin, HE4, total cholesterol, fibrinogen, and Lymphocyte-to-Monocyte Ratio. The six-factor combined model achieved an AUC of 0.794 (sensitivity: 75.9%, specificity: 73.0%). Among the 255 patients (49.13%) who underwent preoperative MRI, 49 were diagnosed with endometrial atypical hyperplasia and 54 with endometrial cancer. The postoperative pathological concordance rates were 77.55% (38/49) for endometrial atypical hyperplasia and 77.78% (42/54) for endometrial cancer. The combined model integrating MRI with hemoglobin, total cholesterol, and fibrinogen levels achieved an AUC of 0.871 (sensitivity: 69.6%, specificity: 89.4%). Endometrial thickness has been established as a critical surveillance parameter during the occurrence and development of EC. A systematic review by Li et al proposed 12 mm as the most accurate diagnostic threshold for EC in asymptomatic postmenopausal women, while 11 mm best distinguished EC from EAH. 18 Canadian clinical guidelines similarly indicate that an endometrial thickness <11 mm is rarely a serious problem in nonbleeding individuals. 19 A Chinese population-based study established 10.5 mm as the optimal predictive threshold for EC and precancerous lesions in asymptomatic patients. 20 These findings collectively underscore the diagnostic significance of endometrial thickness in EC. In our cohort, the mean endometrial thickness was 10.5 mm in the EC group vs 9.3 mm in the EAH group. Logistic regression analysis identified endometrial thickness as an independent predictor of pathological upgrade from preoperative EAH diagnosis to postoperative EC confirmation, with an optimal cut-off value of 10.3mm. While this threshold approximates previously reported values, the observed discrepancies may reflect population-specific variations. High-quality prospective studies or meta-analyses that include data from multiple countries and regions worldwide are warranted to provide an in-depth investigation of this issue. Prior projects have consistently identified age and menopausal status as potential risk factors for EC. Colombo et al. reported the average EC diagnosis age of 63 years, with over 90% of cases occurring in women aged more than 50 years. 21 Goonewardene et al. indicated approximately 75% of both EAH and EC cases arise in women aged over 55 years. 22 Matsuo et al further demonstrated that age is a significant predictor of concurrent EC among American women with EAH. 23 Their work also showed that the risk of EC was 1.02% in women aged 50 years. In our study, intergroup comparisons revealed significantly older age in the EC cohort vs the EAH cohort. However, multivariate analysis did not identify age as an independent risk factor in this research. Meanwhile, postmenopausal women with EAH demonstrate a higher incidence of EC, with menopause typically occurring around age 50 years. 24 All these findings suggest that age or menopausal status may play a role in the pathogenesis and progression of EC. Collectively, we speculate that postmenopausal women have thickened endometrium due to the lack of progesterone antagonistic effects, and with the increase of age, EAH patients have endometrial malignancy under the combined effect of various factors such as decreased physical fitness, changes in diet and lifestyle, and accumulation of other systemic diseases. Emerging evidences have established a correlation between EC and metabolic disorders, such as obesity, diabetes, dyslipidemia, and hypertension. A statistically significant correlation exists between BMI and EC risk, with obesity (BMI≥30 kg/m 2 ) being recognized as a chief pathogenic driver. 25 , 26 Meta-analytic data confirm hypertension as an independent risk factor (RR=1.37, 95% CI: 1.27–1.47). 27 While our study found no significant differences in hypertension or diabetes prevalence between the EAH and EC groups, the EC cohort demonstrated significantly elevated fasting glucose, TC, and BMI values. Existing mechanistic studies reveal that hyperglycemia promotes endometrial proliferation and invasion through ERα/GLUT4 and AMPK/mTOR/S6 signaling pathways. 28 The discrepant findings between our study and previous reports may stem from differing comparator groups: while prior research primarily contrasted EC patients with healthy controls, our study compared EC with EAH cases. Although EAH represents a precancerous condition, its presence indicates established endometrial abnormalities and underlying molecular alterations preceding malignant transformation. These observations underscore the critical importance of metabolic optimization (glycemic control, blood pressure management, and lipid regulation) for EC prevention and progression delay. Through multivariate analysis, we developed a combined predictive model incorporating endometrial thickness, hemoglobin, HE4, TC, fibrinogen, and LMR. The AUC of this model was 0.794 (95% CI: 0.739–0.849), with sensitivity of 0.759 and specificity of 0.730, showing superior predictive accuracy compared to individual parameters alone. Meanwhile, this model highlights the prognostic significance of metabolic parameter (TC), coagulation markers (platelet and fibrinogen), and inflammatory index (LMR), suggesting their potential synergistic roles in EC pathogenesis. This study developed a prediction model pathological upgrade from EAH to EC. However, this merely represents the prediction model for the population in eastern China. Thus, multicenter studies encompassing diverse geographic regions are needed in order to promote it to a wider audience. Despite the contributions of this investigation, several limitations should be considered. Firstly, half of the patients did not undergo preoperative MRI in this study, because MRI was not a mandatory routine examination prior to surgery and complete imaging data were unavailable for patients who underwent MRI at external institutions. But as is well known, MRI possesses unique advantages in EC diagnosis, therapeutic decision-making, and prognostic evaluation. 29 In the present research, Logistic analysis of the patients who underwent MRI scans revealed that preoperative MRI demonstrated a high specificity of 0.929 for diagnosing EC. Furthermore, the combined diagnostic predictive model incorporating MRI indicated an AUC of 0.871, with sensitivity and specificity of 0.696 and 0.894, respectively. These results collectively suggest that MRI possesses an inherent strength in the diagnosis of EC. Prospective studies incorporating MRI findings are warranted to enhance the predictive model for pathological upgrade from EAH to EC. Secondly, while our cohort predominantly represented eastern China populations, the findings may not be generalizable to all Chinese demographics. Variations in ethnic background, socioeconomic status, and environmental exposures are known to influence EC progression. 30 Large-scale multicenter studies encompassing diverse geographic regions are needed to validate and refine our predictive model for nationwide application.

In summary, this study developed a composite predictive model incorporating endometrial thickness, menopausal status, platelet count, TC, fibrinogen, and LMR to preoperatively assess the risk of pathological upgrade from EAH to EC in patients undergoing hysterectomy. When clinically feasible, performing an MRI and integrating its findings with hemoglobin, TC, and fibrinogen levels yields a prediction model with superior performance. The models provide clinicians with an evidence-based reference, offering them an additional option when making clinical decisions.

Endometrial cancer (EC), one of the three major malignancies in the female reproductive system, ranks as the sixth most common cancer among women 1 and represents the most prevalent gynecological malignancy in industrialized nations. 2 According to global statistics from 2022, EC accounted for over 420,000 new cases and nearly 100,000 deaths worldwide, with an annual mortality rate increase of 1.9%. 1 , 3 EC is typically classified into estrogen-dependent and non–estrogen-dependent subtypes. The estrogen-dependent category primarily includes endometrioid EC (EEC), while non–estrogen-dependent subtypes encompass serous EC (SEC), clear cell EC (CCEC), and uterine carcinosarcoma (UCS). 4 EEC constitutes more than 80% of newly diagnosed EC cases, with endometrial atypical hyperplasia (EAH), defined as abnormal proliferation of endometrial glands, is considered its precancerous lesion. 5 Previous studies have demonstrated that 17% to 52% of EAH patients present concurrent EC, 6 while the annual risk of EAH progression to EC is approximately 8.2%. 7 The standard treatment for EAH patients without fertility requirements is total hysterectomy. 8 For premenopausal patients, the standard surgical procedure is total hysterectomy with bilateral salpingectomy while preserving the ovaries. If intraoperative frozen section is sent and it confirm malignancy, oophorectomy and comprehensive staging surgery should be performed. 9 However, reports indicate that the sensitivity of frozen section analysis for detecting endometrial cancer ranges from 27% to 75%, which raises significant clinical concerns. 10 Patients whose intraoperative pathology indicates no malignancy but final postoperative pathology reveals cancer will require a second surgery. The Mayo criteria suggest that lymph node dissection may be omitted in early-stage EC. 11 Therefore, if EAH with concurrent EC can be predicted preoperatively, simultaneous oophorectomy may be recommended for perimenopausal patients to obviate the need for reoperation. Based on the above, identifying the high-risk factors for concomitant EC in patients with EAH is critical to guiding clinical management. However, existing studies regarding high-risk factors associated with pathological upgrading from preoperative EAH diagnosis to postoperative EC remain inconsistent. Giannella L et al. identified age and Body Mass Index (BMI) as predictive high-risk factors for EC coexistence in EAH patients. 12 Consistent with these findings, Lee N et al . demonstrated that age serves as a predictive factor for EC comorbidity in EAH patients, providing clinical guidance for treatment strategies in complex EAH cases aged greater than or equal to 51 years. 13 Conversely, Rajadurai VA et al proposed that severe EAH and menopausal status constitute independent risk factors for EC, though age showed limited predictive value. 14 Zhou L et al . analyzed variables including age, cancer antigen 125 (CA125), CA199, and abnormal uterine bleeding, identifying only CA125 and CA199 as reliable predictors of EC comorbidity in EAH patients. 15 Additional studies showed predictive value of endometrial stripe thickness for concurrent EC detection in EAH patients, though standardized thickness criteria remain undefined. 16 , 17 These findings collectively indicate that while existing research explores high-risk factors for postoperative EC upgrading in EAH patients, substantial heterogeneity in study criteria, nonuniform inclusion populations, and inconsistent research parameters have precluded definitive conclusions regarding EC. Consequently, systematic investigation of high-risk factors for postoperative EC progression in EAH patients warrants more scholarly attention. This retrospective cohort study systematically analyzes demographic characteristics, laboratory parameters, and clinicopathological profiles of patients undergoing hysterectomy with preoperative diagnoses of EAH. All included populations were divided into EAH group and EC group according to postoperative pathology, we aim to comprehensively identify high-risk factors associated with histological upgrading from EAH to EC. This study is expected to provide evidence-based clinical guidance for decision-making in EAH management.