Prognostic factors and survival of patients with endometrial cancer: A 10-year retrospective cohort study in a tertiary referral center in Northern Taiwan.

Endometrial cancer (EC) is the most common gynecological cancer worldwide, with over 400,000 newly diagnosed cases per year in 2020 [ 1 ]. In Taiwan, EC ranked fifth in terms of female cancer incidence, and the incidence rate rose rapidly from 1.69 per 100,000 women per year in 1980–11.36 per 100,000 women per year in 2010 [ 2 ]. EC incidence is increasing globally, particularly among young women, with obesity identified as a major risk factor [ 3 4 ]. From 1990 to 2019, the age-standardized incidence rate of EC rose significantly worldwide, while mortality rates decreased globally but increased in many countries [ 3 ]. A comprehensive genetic analysis revealed that 13% of EC patients had germline pathogenic variants (gPVs), with 63% of high-penetrance gPVs exhibiting biallelic inactivation in tumors, primarily affecting mismatch repair and homologous recombination genes [ 5 ]. Patients with gPVs were generally younger, less obese, and more often White compared to those without gPVs [ 5 ]. These findings highlight the importance of considering EC screening at earlier ages, especially for patients with abnormal bleeding and certain ethnic populations, and support germline assessment in EC for treatment and cancer prevention [ 5 6 ]. EC treatment has evolved with advances in molecular classification and targeted therapies. Surgery remains the primary treatment, with minimally invasive approaches preferred [ 7 8 9 10 11 12 13 ]. Adjuvant therapy is based on risk factors, with low-risk cases managed by surgery alone, while high-intermediate risk cases benefit from vaginal brachytherapy [ 14 ]. High-risk EC may require pelvic radiotherapy, chemotherapy, or combined chemoradiation [ 14 15 ]. The Cancer Genome Atlas has identified four molecular subgroups of EC, which have prognostic and therapeutic implications [ 8 14 16 ]. Novel approaches for advanced or recurrent EC include tyrosine kinase inhibitors, statin, HER2-targeted therapies, and checkpoint inhibitors [ 15 17 ]. Ongoing research focuses on integrating molecular profiling into treatment strategies and developing targeted therapies with predictive biomarkers [ 7 14 18 19 20 ]. EC risk factors have been extensively studied, with several key factors identified. Strong evidence supports the association between increased body mass index (BMI) and waist-to-hip ratio with higher EC risk, while parity reduces risk [ 21 ]. Age over 50, obesity (BMI ≥ 25 kg/m 2 ), diabetes, hypertension (HTN), and severe endometrial hyperplasia are significant risk factors for EC in patients with endometrial hyperplasia [ 22 23 24 25 26 ]. Mendelian randomization studies have confirmed causal relationships between EC risk and factors such as type 2 diabetes, uterine fibroids, higher BMI, elevated fasting insulin, and longer telomere length [ 27 ]. The incidence of EC has increased since the Women’s Health Initiative study, possibly due to decreased use of approved estrogen–progestogen therapy and increased use of compounded bioidentical hormone therapy, along with rising obesity and diabetes rates [ 28 ]. This study aimed to comprehensively investigate and characterize various aspects of EC in a tertiary referral center. By addressing these objectives, this study sought to contribute valuable knowledge to the field of EC research, potentially influencing clinical practice and paving the way for further investigations.

Patients with EC had a mean age of 57.55 and a BMI of 25.77 [ Table 1 ]. The average tumor size was 40.87 mm. The disease factors included DM (16.31%), HTN (36.88%), and smoking (0.71%). Predominant Stage 1 patients (75.18%) underwent treatments such as chemotherapy (21.99%) and radiotherapy (65.25%). Histological analysis confirmed EC in 97.16% of cases. Clinical parameters included anemia (34.04%), thrombocytosis (9.22%), and leukocytosis (6.38%). Surgical methods included laparoscopy (34.75%), open procedures (65.25%), myometrial invasion (82.98%), and partial omentectomy in 49 patients. Additional observations were lymphovascular space (34.75%), cervical stromal (11.35%), and fallopian tube (7.8%) invasion. The LN data showed an average of 26.01 nodes, with 9.22% metastases. The tumor grades were 25.53%, 34.04%, and 40.42% for grades 1, 2, and 3, respectively. Other risk factors included adenomyosis (19.15%), myoma (44.68%), ovarian invasion (9.93%), and recurrence (19.86%). Basic characteristics of the patients ( n =141) SD: Standard deviation, BMI: Body mass index, DM: Diabetes mellitus, HTN: Hypertension, CT: Chemotherapy, RT: Radiotherapy, LVSI: Lymphovascular space invasion, OP: Operation, LN: Lymph node This study evaluated factors influencing PFS in univariate and multivariate analyses (Models 1–3) [ Table 2 ]. Univariate analysis (Model 1) identified chemotherapy (HR = 2.07, P = 0.024), myometrium invasion >1/2 (HR = 1.92, P = 0.035), and fallopian tube invasion (HR = 2.16, P = 0.049) as significant risk factors, but these lost significance in multivariate models after adjusting for all variables and grade (Model 2) and grade level (Model 3). The only consistent predictor of worse PFS was total LN numbers (adjusted HR [aHR] = 1.05, P = 0.006 in Model 2; P = 0.004 in Model 3). Other factors – including age, BMI, comorbidities, histology type, and surgical approach – showed no significant association. Chemotherapy retained borderline significance (aHR = 3.90, P = 0.045 in Model 3), possibly reflecting treatment selection bias in high-risk patients. Overall, LN numbers emerged as the strongest independent prognostic marker for PFS. The univariate and multivariate analysis of the progression-free survival Model 1: Univariate analysis, Model 2: Multivariate analysis (all variables+grade), Model 3: Multivariate analysis (all variables+low/high grade). HR: Hazard ratio, aHR: Adjusted HR, CI: Confidence interval, BMI: Body mass index, DM: Diabetes mellitus, HTN: Hypertension, CT: Chemotherapy, RT: Radiotherapy, LVSI: Lymphovascular space invasion, OP: Operation, LN: Lymph node This study examined factors influencing OS through univariate (Model 1) and multivariate analyses (Models 2–3) [ Table 3 ]. In univariate analysis, significant predictors of worse OS included older age (HR = 1.03, P = 0.167; significant in multivariate), larger tumor size (HR = 1.03, P < 0.0001), advanced stage (HR = 3.59, P = 0.005), chemotherapy (HR = 4.39, P = 0.001), lymphovascular space invasion (HR = 2.69, P = 0.032), open surgery (HR = 2.75, P = 0.028), deep myometrial invasion (HR = 4.86, P = 0.001), cervical stromal invasion (CSI) (HR = 6.48, P = 0.0002), fallopian tube invasion (HR = 3.81, P = 0.018), LN invasion (HR = 3.01, P = 0.050), ovarian invasion (HR = 6.44, P < 0.0001), and recurrence (HR = 3.04, P = 0.048). However, in multivariate analysis, only age (aHR = 1.19, P = 0.004), CSI (aHR = 21.16, P = 0.008), open surgery (aHR = 6.90, P = 0.022), and recurrence (aHR = 9.25, P = 0.012) remained significant, while HTN was protective (aHR = 0.09, P = 0.032). Other factors, including tumor size, LN involvement, and histologic grade, lost significance after adjustment. These findings highlight age, CSI, open surgery, and recurrence as key independent predictors of mortality, whereas HTN may confer a survival advantage, possibly due to treatment effects or confounding factors. The univariate and multivariate analysis of the overall survival Model 1: Univariate analysis, Model 2: Multivariate analysis (all variables + grade), Model 3: Multivariate analysis (all variables + low/high grade). HR: Hazard ratio; CI: Confidence interval, BMI: Body mass index, DM: Diabetes mellitus, HTN: hypertension, CT: Chemotherapy, RT: Radiotherapy, LVSI: Lymphovascular space invasion, OP: Operation, LN: Lymph node, aHR: Adjusted HR Figure 1 presents Kaplan–Meier survival curves for PFS (A) and overall OS (B) over a follow-up period of up to 12 years in EC. The 5-year PFS and OS were 70.8% and 86%, respectively [ Figure 1 ]. The PFS curve [ Figure 1a ] showed an initial sharp decline within the first 2 years, followed by a gradual decrease, indicating early disease progression in some patients while others remained progression-free. The OS curve [ Figure 1b ] demonstrated a more sustained decline, with a high survival rate maintained beyond 10 years, suggesting that while some patients experienced disease recurrence, they continued to survive with treatment. The Kaplan–Meier plot of the progression-free survival and overall survival. The 5-year progression-free survival was 70.8% (a) and overall survival was 86% (b) Figure 2 presents Kaplan–Meier OS curves for various prognostic factors in EC. Figure 2a demonstrates that anemia is significantly associated with worse OS than nonanemic patients ( P = 0.0347). Figure 2b shows a significant reduction in OS for patients with disease recurrence ( P = 0.0384), indicating its strong impact on survival. Figure 2c examines the effect of DM on OS but did not find a statistically significant difference ( P = 0.096). Figure 2d compares OS between patients with and without HTN, showing no significant difference ( P = 0.2848). Figure 2e evaluated the impact of age (<57 vs. ≥57 years) on OS, but no significant effect was observed ( P = 0.6939). Overall, anemia and disease recurrence were significant prognostic factors for worse OS in EC, whereas DM, HTN, and age did not show statistically significant associations with survival outcomes. The Kaplan–Meier plot of the overall survival stratification by various factors. (a) Anemia. (b) Recurrence. (c) Diabetes mellitus. (d) Hypertension. (e) Age Supplement Figure 1 presents Kaplan–Meier PFS curves for various prognostic factors in EC. Supplement Figure 1a showed a significant difference in PFS based on cancer stage ( P = 0.0194), with advanced-stage patients (stage 3–4) having worse survival. Supplement Figure 1b highlighted a significantly worse PFS in patients with CSI compared to those without ( P = 0.0438). Supplement Figure 1c evaluated cytology results but found no significant difference in survival ( P = 0.3934). Supplement Figure 1d compares PFS between patients with and without fallopian tube invasion, showing no statistical significance ( P = 0.2320). Supplement Figure 1e assessed the impact of HTN on survival, also finding no significant difference ( P = 0.1904). Supplement Figure 1f examined LN involvement, comparing patients with ≥12 versus <12 LNs examined but did not show a significant impact on PFS ( P = 0.1563). Overall, cancer stage and CSI appear to be significant prognostic factors for disease progression, while other variables do not show a statistically significant effect on survival outcomes.

This study of EC patients (mean age: 57.55 years, BMI 25.77) identified key prognostic factors through multivariate analysis: Total LN count emerged as the strongest independent predictor of worse PFS (aHR = 1.05, P = 0.004), whereas age (aHR = 1.19, P = 0.004), CSI (aHR = 21.16, P = 0.008), open surgery (aHR = 6.90, P = 0.022), and recurrence (aHR = 9.25, P = 0.012) significantly predicted poorer OS. Surprisingly, HTN showed a protective effect (aHR = 0.09, P = 0.032). Survival analysis revealed 5-year PFS and OS rates of 70.8% and 86%, respectively, with early progression but sustained long-term survival. The findings highlight the importance of LN evaluation and high-risk pathological features in EC prognosis, though the study’s limitation of not assessing the LNMR suggests future research directions. These results emphasize the need for comprehensive surgical staging and personalized treatment approaches in EC management. The datasets generated during and/or analyzed during the current study are available from the corresponding author on reasonable request. Dr. Dah-Ching Ding, an editorial board member at Tzu Chi Medical Journal , had no role in the peer review process of or decision to publish this article. The other authors declared no conflicts of interest in writing this paper.

EC patients (most were stage 1 [75.18%]) underwent various treatments, and histological analysis confirmed type 1 EC in 88.65% of cases. Noteworthy findings in the multivariate analyses included the total LN number significantly affecting PFS. The significant influences of age, CSI, open surgery, and recurrence were noted on the OS. Age can be a significant factor in the prognosis and survival of individuals with EC [ 29 ]. The median age of diagnosis for EC is 62 years [ 29 ]. Generally, older age is associated with a higher risk of developing EC, and older age at diagnosis is associated with a lower survival rate, including serous histology, advanced stage, and less favorable clinical characteristics [ 29 ]. Age >80 years was associated with poor survival [ 30 ], and age >70 years had a poor outcome, associated with more high-risk histology, high grade, and more than 50% myometrial invasion [ 31 ]. Agreeing with previous studies, our study also showed that older age was associated with poor survival. Our study found that a higher total number of LNs examined was correlated with worse PFS in EC. The relationship between LN examination and prognosis in EC is complex. While a higher LN ratio (metastatic to total nodes) is associated with worse PFS and OS [ 32 33 ], the total number of LNs examined alone does not consistently predict outcomes. Machine learning analysis suggests that the number of positive nodes is a stronger prognostic factor than para-aortic node involvement, with 6+ positive nodes associated with significantly worse survival [ 34 ]. Interestingly, the number of nodal stations sampled, rather than total node count, may be a more accurate predictor of LN metastasis [ 35 ]. These findings highlight the importance of comprehensive lymphadenectomy and suggest that both the extent and location of nodal involvement are crucial factors in determining prognosis for EC patients. Our study showed that CSI was associated with poor prognosis in EC. Deep CSI (outer third) is an independent predictor of death in Stage II ECs, with patients showing significantly reduced survival compared to those with no invasion or inner two-thirds CSI [ 36 ]. CSI is linked to a 44% increase in progression risk and 33% increase in death risk [ 37 ]. Diagnosis of CSI can be challenging, as MRI may miss it when no cancer is found in the uterine body [ 38 ]. Abnormal Pap smears are strongly associated with occult CSI in EC patients, with an adjusted odds ratio of 2.65 [ 39 ]. The presence of CSI and the increasing number of positive pelvic nodes are associated with poor prognosis, though chemotherapy (adriamycin plus cisplatin) shows improved progression-free and OS compared to whole abdominal irradiation in patients with CSI [ 37 ]. Preoperative anemia may be associated with poor survival in patients with EC. A previous study showed that preoperative anemia was associated with a lower OS and PFS than patients with EC but no anemia [ 40 ]. Another meta-analysis also showed that preoperative anemia was associated with lower survival than patients with EC but no anemia [ 41 ]. Wilairat and Benjapibal reported that preoperative anemia is associated with poor prognostic factors (including advanced FIGO stage, higher tumor grade, deeper myometrial invasion, and lymphovascular space invasion) but not survival in patients with EC [ 40 ]. Consistent with previous studies, our study also showed that preoperative anemia was associated with poor survival. Recurrence is associated with poor survival in patients with EC. The OS of EC reaches up to 80% [ 42 ]. Nevertheless, recurrent EC has a significantly lower OS rate, falling below 20% [ 43 ]; hence, recurrent EC is associated with poor survival [ 44 ]. A recent population-based cohort study unveiled that the recurrence of endometrioid EC is linked to a survival rate below 50% [ 45 ]. Our study was consistent with previous studies that state that recurrence of EC is associated with a higher HR of OS. Research suggests a strong association between DM and EC. A highly suggestive link between DM and EC incidence has been reported [ 24 25 46 ]. Patients with DM have a 97% increased risk of developing EC compared to nondiabetics [ 47 ]. Hyperglycemia, a common feature of DM, may contribute to EC pathogenesis independently of obesity [ 48 ]. However, the impact of antidiabetic medications on EC risk remains unclear. While some studies found no significant association between metformin, sulfonylureas, or insulin use and EC risk, sensitivity analyses suggested metformin might increase EC incidence [ 49 ]. The mechanisms underlying the DM-EC association are not fully understood, but increased glucose metabolism may play a role in EC growth and progression [ 48 ]. However, our study did not identify a correlation between DM and EC, possibly due to the limited number of DM cases included. Multiple studies have found a significant association between HTN and increased risk of EC. A recent meta-analysis of 26 observational studies reported a 37% higher risk of EC in hypertensive women [ 50 ]. This finding is consistent with an earlier meta-analysis that found a 61% increased risk overall, with stronger associations in case–control studies compared to cohort studies [ 51 ]. A large pooled analysis of 29 studies also found a 14% increased risk of EC in hypertensive women, with stronger associations observed in pre-/perimenopausal women and never users of postmenopausal hormone therapy [ 52 ]. The association between HTN and EC risk appears to be independent of other factors such as obesity and diabetes, though these may amplify the risk [ 50 53 ]. These findings suggest that HTN may be an independent risk factor for EC, warranting further research into underlying biological mechanisms. However, our study did not identify a correlation between HTN and EC, possibly due to the limited number of HTN cases included. These results need cautious interpretation and further validation in larger, prospective cohorts with more comprehensive control of comorbidities and confounders, particularly age. Recent studies have highlighted the prognostic significance of positive peritoneal cytology (PPC) in EC. PPC is associated with decreased recurrence-free survival and OS, particularly in p53abn and NSMP molecular subgroups [ 54 ]. It correlates with higher recurrence rates, especially distant recurrences [ 55 ]. A prospective multicenter trial found that PPC conversion during minimally invasive surgery with intrauterine manipulators was linked to worse oncological outcomes [ 56 ]. Furthermore, a large multicenter study based on the ESGO/ESTRO/ESP risk classification demonstrated that PPC significantly impacted prognosis, especially in intermediate and high-intermediate risk groups [ 57 ]. These findings suggest that PPC remains an important prognostic factor in EC, despite its removal from the FIGO staging system in 2009. The studies emphasize the need for further research on PPC and its integration with molecular classifications and biomarkers like L1CAM for improved risk stratification in EC. Conversely, our study found that PPC was associated with a lower HR for PFS, warranting further investigation. This study has several strengths. First, it provides a detailed dataset encompassing various aspects, including patient demographics, disease characteristics, treatment modalities, and outcomes, which offers a holistic view of EC. Second, multivariate analyses of both PFS and OS enhanced the robustness of the findings. This approach considers multiple variables simultaneously and provides a nuanced understanding of their interplay. Third, the study explored clinically relevant factors such as HTN, staging, surgical methods, and histology, contributing valuable insights that can inform medical decision-making and patient care. Fourth, including diverse parameters, such as tumor grading, LN involvement, and cytology, provides depth to the analysis and enables a more comprehensive evaluation of EC. This study has several limitations. First, its retrospective design may have introduced biases and limited the ability to establish causal relationships. Prospective studies are more robust for causal inferences. Second, since the study relied on data from a single center, the findings may lack generalizability to broader populations. Third, unmeasured or unknown confounding variables could have influenced the observed associations. The high HRs for diabetes and recurrence, along with their wide CIs, suggest potential overfitting or imprecision due to limited event numbers, warranting cautious interpretation and validation in larger cohorts. Our study showed that CSI was associated with poor prognosis in EC. However, we acknowledge that this association may be influenced by the relatively small number of patients with stage II disease. Our analysis did not evaluate the LN metastatic ratio (LNMR, metastatic-to-total nodes), which has been shown to better predict outcomes in EC. The observed association between higher total LN counts and worse PFS and OS may be influenced by confounding factors, including surgical extent or underlying high-risk disease. Future studies should incorporate LNMR and correlate it with metastatic patterns to clarify its prognostic utility.

The study was conducted by the Declaration of Helsinki and was approved by the Research Ethics Committee of Taipei Tzu Chi Hospital (approval number: IRB 11-C-069, date of approval: 2022.08.08). The requirement for informed consent was waived owing to the low risk to the patients. This retrospective study included 141 patients with EC who underwent a hysterectomy-based surgical intervention at Taipei Tzu Chi Hospital between January 2011 and January 2020. Patients who did not undergo hysterectomy or lacked complete pathology and laboratory data regarding complete blood counts were excluded from the study. A pathologist confirmed the diagnosis of EC. Patients diagnosed with malignant endometrial neoplasm (C54.1 International Classification of Diseases, Tenth Revision, Clinical Modification, ICD-10-CM) who underwent hysterectomy-based surgical intervention at Taipei Tzu Chi Hospital were selected for the study. In addition, (1) patients who did not undergo surgery, (2) those lost to follow-up, and (3) those with incomplete data were excluded from this study. Patient information was obtained from our hospital’s electronic medical records. We collected the following clinical information: (1) basic information such as age, height, weight, BMI, and history of HTN and diabetes mellitus (DM); (2) laboratory data, including preoperative complete blood counts; (3) pathological information including histological subtype, tumor grade, lymph node (LN) invasion, lymphovascular space invasion, clinical staging, International Federation of Gynecology and Obstetrics staging, ascites cytology results, immunohistological staining for estrogen receptor and progesterone receptor, and additional pathological findings such as adenomyosis and myoma; and (4) surgical history, including date, type, and route. The anemia threshold was established at a hemoglobin level of 10,000/μL. For thrombocytosis, the cutoff value was defined as a platelet count of >400,000/μL. The primary outcomes were progression-free survival (PFS) and overall survival (OS), whereas the secondary outcomes assessed the association between various factors and EC survival. We used the Chi-square test to evaluate categorical variables and the independent t -test to analyze continuous variables. Survival outcomes were estimated using the Kaplan–Meier method. Cox proportional hazard regression analysis was performed to investigate the factors associated with cancer mortality. Hazard ratio (HR) and 95% confidence intervals (CIs) were calculated. Statistical analyses were performed using the SPSS software (version 24.0; IBM Corp., Armonk, NY, USA). Statistical significance was set at P < 0.05.

The Kaplan–Meier plot of the progression-free survival stratification by various factors. (a) Stage. (b) Cervical stromal invasion. (c) Cytology. (d) Fallopian tube invasion. (e) Hypertension. (f) Lymph node numbers