Endometrial cancer represents a growing threat in gynecologic oncology, with rising incidence worldwide and approximately 97,000 annual deaths [1]. This malignancy primarily spreads through lymphatic pathways, with lymph node metastasis present in 10%–15% of apparently early-stage cases [2]. Lymphadenectomy provides critical staging information for treatment planning, though its survival benefit remains debated. For selected cases where lymphadenectomy is performed, postoperative lymphocele formation represents a significant clinical challenge, affecting between 1%–58% of patients [3].

Lymphoceles significantly impair recovery and reduce quality of life [4, 5]. These fluid collections result from ineffective closure of lymphatic vessel stumps during surgery [6] and can lead to serious complications including infections and prolonged hospitalization [7, 8]. Key risk factors include extent of lymphatic dissection, obesity, and surgical drain management [9].

Current predictive models for surgical complications in endometrial cancer show important limitations. The american society of anesthesiologists (ASA) classification demonstrates value for general complications but lacks specificity for lymphatic complications [10]. While the frailty Index effectively predicts morbidity in elderly patients [11], its complexity limits routine clinical application. Similarly, the surgical complexity score has been primarily validated for ovarian cancer [12], with insufficient evidence for endometrial cancer complications.

The estimation of physiologic ability and surgical stress (E-PASS) system offers a promising alternative by uniquely integrating both patient factors and surgical parameters. Through its comprehensive risk score calculation [13], E-PASS potentially provides a more complete approach to predicting specific complications like lymphocele formation [14]. Though valuable across surgical specialties, its application in gynecologic oncology remains underexplored [15].

Within the evolving landscape of endometrial cancer management, where lymphadenectomy is increasingly applied selectively rather than routinely, this retrospective cohort study evaluated whether E-PASS scores can effectively predict lymphocele formation following endometrial cancer surgery. Our objective was to establish a clinically useful risk stratification tool to guide preventive strategies and support individualized surgical approaches.

Overall, 180 patients with endometrial cancer who underwent radical surgery were evaluated. Postoperative lymphocele developed in 62 patients (34.4%), while 118 patients (65.6%) remained lymphocele-free during follow-up.

Demographic and perioperative characteristics were comparable between groups. Neither patient-specific variables (age, body mass index [BMI], comorbidity burden, postoperative hemoglobin levels) nor surgical parameters (anesthetic technique, drainage management, prior abdominal operations) exhibited significant differences (all p $>$ 0.050).

Notably, tumor staging analysis revealed that advanced-stage disease (FIGO III–IV) occurred more frequently in patients who developed lymphoceles (40.3% vs. 20.3%, $\chi$² = 8.193, p = 0.004). However, histological classification showed similar distribution patterns between groups ($\chi$² = 1.178, p = 0.278).

Analysis of E-PASS parameters revealed slight differences between groups that did not reach statistical significance. The lymphocele cohort showed marginally higher values in preoperative risk scores (0.35 [0.08] vs. 0.33 [0.05], t = 1.536, p = 0.128), surgical stress scores (0.36 [0.10] vs. 0.34 [0.06], t = 1.444, p = 0.152), and comprehensive risk scores (0.43 [0.15] vs. 0.40 [0.13], t = 1.419, p = 0.159). Of the examined variables, FIGO staging showed a significant association with lymphocele formation risk (p = 0.004) (Table 1).

We performed Spearman correlation analysis between FIGO staging and E-PASS components due to the binary nature of FIGO stage. No significant correlations were found between FIGO staging and PRS (Spearman correlation coefficients [rs] = –0.027, p = 0.715), SSS (rs = 0.076, p = 0.310), or CRS (rs = –0.009, p = 0.909), indicating cancer stage independence from physiological risk assessment. Strong correlations existed among E-PASS components: CRS correlated significantly with PRS (rs = 0.706, p 0.001) and SSS (rs = 0.542, p 0.001). PRS and SSS also showed moderate correlation (rs = 0.256, p = 0.001), suggesting patients with poor preoperative status experience higher surgical stress (Table 2). CRS’s superior performance reflects its integrated approach, combining preoperative physiological status and surgical stress factors. This comprehensive assessment offers better risk evaluation than individual components alone. The independence between FIGO staging and E-PASS suggests these systems provide complementary information for surgical risk assessment.

Logistic regression analysis was performed with CRS as a continuous variable, FIGO stage (I–II = 0, III–IV = 1) as a categorical variable, and postoperative lymphocele formation (yes = 1, no = 0) as the dependent variable (Table 3). Binary logistic regression analysis was performed to evaluate the predictive value of CRS and FIGO stage for postoperative lymphocele formation in patients with endometrial cancer (Table 4). The results demonstrated that both CRS and FIGO stage were independent risk factors for lymphocele formation. Specifically, for each unit increase in CRS, the risk of developing lymphocele increased by 16.1% (OR = 1.161, 95% CI: 1.110–1.214, p 0.001), indicating a significant positive correlation between surgical complexity and postoperative lymphocele formation. Additionally, patients with advanced FIGO stage (III–IV) showed a substantially higher risk of lymphocele formation compared to those with early-stage disease (I–II) (OR = 3.211, 95% CI: 1.262–8.172, p = 0.014). This suggests that patients with advanced-stage endometrial cancer are more than three times more likely to develop postoperative lymphocele compared to early-stage patients, highlighting the importance of careful postoperative monitoring in this high-risk group.

The predictive performance of different E-PASS scoring systems for postoperative lymphocele in patients with endometrial cancer was evaluated using ROC curve analysis (Fig. 2). The CRS demonstrated superior discriminative ability with the highest AUC of 0.930 (95% CI: 0.893–0.966), with optimal sensitivity of 80.60% and specificity of 91.50%. The SSS showed good predictive performance with an AUC of 0.873 (95% CI: 0.818–0.927), achieving 69.41% and 92.41% sensitivity and specificity, respectively. In contrast, the PRS showed relatively poor predictive capability, with AUC of 0.686 (95% CI: 0.600–0.772). The significant difference in AUCs between CRS and other E-PASS components (all p 0.001) suggests that CRS is the most reliable predictor for postoperative lymphocele development in patients with endometrial cancer undergoing radical surgery. For FIGO stage, as a binary categorical variable (I–II vs. III–IV), ROC analysis is not applicable. However, its diagnostic performance showed sensitivity of 40.3% and specificity of 79.7% for predicting lymphocele formation. Although only CRS and FIGO stage emerged as independent risk factors in multivariable analysis, we present the diagnostic performance of all E-PASS components in Table 5 to provide comprehensive information about each parameter’s individual predictive capability. The combination of CRS with FIGO stage improves the overall predictive performance compared to either parameter alone (Table 5).

Fig. 2.

Receiver operating characteristic (ROC) curves comparing different predictive models for postoperative lymphocele in patients with endometrial cancer. ROC curves showing the discriminative ability of four different predictive models: comprehensive risk score (CRS, black line), surgical stress score (SSS, red line), physiological risk score (PRS, blue line). The area under the curve (AUC) values with 95% confidence intervals were: CRS 0.930 (0.893–0.966), SSS 0.873 (0.818–0.927), PRS 0.686 (0.600–0.772). The diagonal reference line (grey) represents random prediction (AUC = 0.500). CRS demonstrated superior predictive performance with optimal sensitivity of 80.60% and specificity of 91.50%. The curves were derived from a retrospective analysis of 180 patients who underwent radical surgery for endometrial cancer between 2012 and 2023.

This retrospective investigation spanning 11 years examined how E-PASS scoring components relate to lymphocele formation after endometrial cancer surgery. Our analysis revealed that the CRS offers the most robust predictive capability among the assessed parameters. This finding suggests that an integrated assessment combining both physiological capacity and surgical stress provides more accurate risk stratification than either component alone or traditional staging systems.

Physiological capacity reflects a patient’s overall health status and recovery potential, while surgical stress scoring quantifies both psychological and physiological impacts of surgery [14, 19]. The E-PASS scoring system, comprising PRS, SSS, and CRS, provides clinicians with precise patient assessment by comprehensively evaluating surgical risk and stress response [20]. Our findings align with several previous studies while also revealing some unique insights. Kayra et al. [19] demonstrated in their prospective study of 156 patients with endometrial cancer that E-PASS scoring components effectively predicted postoperative complications, with CRS showing the highest predictive value. Similarly, Chen et al. [14] reported that elevated CRS correlated with increased lymphatic complications in gynecologic surgery. However, our study demonstrated enhanced predictive accuracy for CRS compared to these previous reports, possibly due to our larger sample size and more comprehensive assessment protocol. Our findings regarding the correlation between FIGO staging and lymphocele formation corroborate Pan et al.’s [21] retrospective analysis, though with some variations that might be attributed to differences in surgical techniques and postoperative management. This difference might be attributed to variations in surgical techniques and postoperative management protocols. The mechanistic basis for our observations can be explained through several pathways. As demonstrated by Chen et al. [22] and Kitano et al. [23], elevated PRS often reflects compromised physiological reserves and reduced healing capacity, while Norimatsu et al. [24] and Dai et al. [25] established that higher SSS indicates more extensive surgical trauma and lymphatic disruption. The superior predictive performance of CRS in our study, consistent with Haga et al.’s [26] foundational work, likely stems from its comprehensive integration of both physiological and surgical stress parameters, particularly relevant in complex procedures like endometrial cancer surgery.

The clinical implications of our findings extend beyond mere risk prediction, offering a practical framework for personalized surgical planning in endometrial cancer treatment. The strong performance of CRS enables clinicians to identify high-risk patients and implement targeted interventions. Based on our ROC analysis, we propose CRS-based risk stratification for clinical decision-making, though specific thresholds require further validation in larger cohorts. For surgical approach, high-risk patients would benefit from sentinel lymph node mapping when oncologically appropriate and advanced vessel-sealing devices to minimize lymphatic disruption, while intermediate-risk patients should receive standard lymphadenectomy with careful attention to lymphatic preservation. Drainage strategy should be tailored to risk level: high-risk patients receiving extended closed-suction drainage (7–10 days) with daily output monitoring, intermediate-risk patients having standard drainage with removal upon output reduction below 50 mL/24 h, and low-risk patients potentially qualifying for early drain removal or selective placement. Postoperative monitoring should likewise be risk-stratified: high-risk patients undergoing enhanced ultrasound assessment on days 7, 14, and 21, intermediate-risk patients receiving evaluation on day 14, and low-risk patients following standard clinical protocols. Additional preventive measures for high-risk patients should include early mobilization with compression therapy, nutritional optimization (especially with low preoperative albumin), and thorough patient education regarding lymphocele symptoms. Notably, endometrial cancer lymphadenectomy guidelines have undergone significant changes in recent years. Both the European Society of Gynecological Oncology/European Society for Radiotherapy and Oncology/European Society of Pathology 2020 (ESGO/ESTRO/ESP 2020) consensus and the latest National Comprehensive Cancer Network (NCCN) guidelines recommend more individualized lymph node assessment strategies [27, 28]. Research from AC Camargo Cancer Center revealed that sentinel lymph node (SLN) mapping reduced lymphocele incidence from 14.1% with systematic lymphadenectomy to 3.4% with SLN alone (p = 0.009). Multivariate analysis identified systematic lymphadenectomy as an independent risk factor for lymphocele (OR 3.68) [29].

This study’s unique contribution lies in its comprehensive integration of both physiological and surgical parameters through the E-PASS system, providing a more nuanced risk assessment than traditional staging-based approaches. We recommend incorporating CRS calculation into routine preoperative assessment workflows, with particular attention to patients showing elevated values. Future research should pursue several key directions: multicenter prospective validation to confirm CRS threshold generalizability across diverse populations; integration of CRS with molecular biomarkers (such as vascular endothelial growth factor [VEGF]-C/D or inflammatory markers) to enhance prediction and reveal pathophysiological mechanisms; combination with radiological parameters like lymphatic mapping or ultrasound elastography to create stronger predictive models; development of artificial intelligence (AI) algorithms that integrate these clinical, molecular, and radiological factors; and interventional studies evaluating whether CRS-based preventive strategies actually reduce lymphocele incidence and improve outcomes.

Methodological rigor was ensured through: (1) 11-year cohort (n = 180) providing temporal validity; (2) standardized lymphocele imaging protocols; (3) multifactorial analysis integrating clinical/E-PASS parameters. Our statistical framework combined correlation-multivariate modeling with ROC analysis and confounder adjustment. Stage-inclusive enrollment (IA–IV) and $\ge$6-month survival criteria enhanced generalizability while controlling attrition bias

Study limitations include: (1) single-center Chinese cohort requiring multiethnic validation; (2) exclusion of high-risk subgroups (comorbidities/immunodeficiency/neoadjuvant); (3) observational design precludes causal inference. Residual confounding persists despite multivariate adjustment. Nevertheless, CRS maintained strong predictive validity (AUC = 0.930) supporting clinical translation for endometrial cancer lymphocele risk stratification.

Our study confirms E-PASS scoring, particularly CRS, as an effective predictor of lymphocele formation after endometrial cancer surgery. We propose risk stratification (low 0.500, intermediate 0.5–0.8, high $>$0.800) to guide management, where high-risk patients benefit from modified surgical techniques and enhanced surveillance. Despite single-center limitations, this approach offers a practical framework for personalized surgical planning that balances oncologic outcomes with complication prevention. Future multicenter studies should validate these findings and evaluate whether CRS-guided strategies effectively reduce lymphocele incidence in diverse patient populations.

E-PASS, estimation of physiologic ability and surgical stress; PRS, preoperative risk score; SSS, surgical stress score; CRS, comprehensive risk score; NYHA, New York heart association; VC, vital capacity; FEV1.0%, forced expiratory volume in 1 second; ASA, american society of anesthesiologists; IQR, interquartile range; SD, standard deviation; OR, odds ratios; CI, confidence intervals; ROC, receiver operating characteristic; AUC, area under the curve; BMI, body mass index; FIGO, International Federation of Gynecology and Obstetrics; SE, standard error; ESGO/ESTRO/ESP 2020, European Society of Gynecological Oncology/European Society for Radiotherapy and Oncology/European Society of Pathology 2020; NCCN, National Comprehensive Cancer Network; VEGF, vascular endothelial growth factor; AI, artificial intelligence.

The datasets generated and analyzed during the current study are available from the corresponding author upon reasonable request.

HL and ZH conceived the research idea. HL, ZH, and JD jointly designed the research methodology and study framework. HL and ZH performed data collection, statistical analysis, and interpretation of results. HL and ZH drafted the initial manuscript. JD supervised the entire study and critically revised the manuscript for important intellectual content. All authors participated in manuscript revision and editing. All authors read and approved the final manuscript. All authors have participated sufficiently in the work and agreed to be accountable for all aspects of the work. All authors participated in manuscript revision and editing. All authors read and approved the final manuscript. All authors have participated sufficiently in the work and agreed to be accountable for all aspects of the work.

The study was carried out in accordance with the guidelines of the Declaration of Helsinki. This study was approved by the Ethics Committee of Dongyang People’s Hospital of Wenzhou Medical University (approval number: DRY-2024-YX-390). The requirement for informed consent was waived by the institutional review board due to the retrospective nature of this study.

We would like to thank Editage (https://www.editage.cn/) for English language editing.

This research received no external funding.

The authors declare no conflict of interest.

The authors acknowledge the use of language editing tools, including AI-assisted software, for grammar checking and linguistic refinement only; all scientific content, data analysis, results interpretation, and conclusions were exclusively generated and validated by the human authors. During the preparation of this work the authors used ChatGPT and DeepSeek in order to check spelling and grammar. After using these tools, the authors reviewed and edited the content as needed and take full responsibility for the content of the publication.