Abstract
Purpose
This study aimed to characterize recurrence patterns and identify prognostic factors in patients with stage I, grade 2–3 endometrioid endometrial cancer.
Materials and Methods
We retrospectively analyzed 298 patients with 2009 International Federation of Gynecology and Obstetrics stage IA–IB, grade 2–3 endometrioid endometrial cancer treated with curative intent between 2010 and 2020. The primary endpoint was recurrence-free interval (RFI), defined as the time from surgery to documented recurrence or last follow-up. Cox regression analyses were performed to identify prognostic factors. Recursive partitioning analysis (RPA) was used for risk stratification.
Results
With a median follow-up of 80.1 months, 40 patients (13.4%) experienced recurrence. Distant recurrence was the most common recurrence pattern (n = 28, 9.4%), followed by vaginal recurrence (n = 10, 3.4%), para-aortic recurrence (n = 8, 2.7%) and pelvic recurrence (n = 7, 2.3%). In multivariable analysis, older age (hazard ratio [HR] per year, 1.05; p = 0.004), tumor size ≥4 cm (HR, 2.28; p = 0.027), grade 3 histology (HR, 2.20; p = 0.022), and stage IB disease (HR, 2.31; p = 0.043) were significantly associated with inferior RFI. Estrogen receptor (ER) negativity was associated with inferior RFI in exploratory univariable analysis (HR, 5.10; p < 0.001). An age- and stage-based RPA model classified patients into three groups with significantly different RFI (p < 0.001). Post hoc analysis suggested a higher recurrence proportion among ER-negative patients aged ≤70 years with stage IB disease.
Conclusion
Recurrence in stage I, grade 2–3 endometrioid endometrial cancer was infrequent, but distant metastasis was the most common recurrence pattern. Conventional clinicopathologic factors provided clinically useful risk stratification, while ER status may offer additional prognostic information in selected subgroups.
Keywords: Endometrial neoplasms, Recurrence, Prognosis, Estrogen receptors, Radiotherapy
Introduction
Endometrial cancer is one of the most common gynecologic malignancies worldwide, and approximately two-thirds of patients present with disease confined to the uterus at diagnosis [1]. For patients with early-stage endometrioid endometrial cancer, primary treatment consists of hysterectomy, bilateral salpingo-oophorectomy, and nodal staging, most commonly by sentinel lymph node mapping, followed by selective use of adjuvant therapy according to clinicopathologic risk factors [2]. Although prognosis is generally favorable in stage I disease, recurrence still occurs in a subset of patients and remains clinically important because treatment failure may be associated with substantial morbidity and poor survival after distant relapse [1,3].
The role of adjuvant radiation therapy (RT) in early-stage endometrial cancer has been established by landmark randomized trials. GOG-99 and PORTEC-1 showed that pelvic external beam radiation therapy (EBRT) significantly reduces locoregional recurrence (LRR) but does not improve overall survival [4,5]. Subsequently, PORTEC-2 demonstrated that vaginal brachytherapy (VBT) provides vaginal control comparable to pelvic EBRT for high-intermediate-risk disease, with lower gastrointestinal toxicity, thereby establishing VBT as the preferred option for this risk group [6,7]. However, even in these studies, distant metastasis (DM) remained an important pattern of treatment failure, suggesting that clinicopathologic risk stratification alone may not fully account for the biological heterogeneity underlying recurrence risk [4-8].
To guide adjuvant treatment, multiple clinicopathologic risk stratification systems have been developed using factors such as age, myometrial invasion, histologic grade, and lymphovascular space invasion (LVSI). More recently, molecular classification has reshaped the prognostic framework of endometrial cancer. The Cancer Genome Atlas identified four biologically distinct molecular subgroups, and subsequent simplified clinical algorithms enabled their incorporation into routine practice [9,10]. Contemporary European Society of Gynaecological Oncology/European Society for Radiotherapy and Oncology/European Society of Pathology guidelines now integrate molecular classification into risk assessment and adjuvant treatment recommendations [2]. In parallel, additional biomarkers, including hormone receptor expression, may provide further prognostic refinement, particularly within the no specific molecular profile (NSMP) subgroup [11]. Nevertheless, in real-world practice, biomarker availability is often incomplete, and treatment decisions continue to rely on traditional clinicopathologic factors.
Given this background, we conducted a retrospective two-institution study of patients with stage I, grade 2–3 endometrioid endometrial cancer treated with curative intent. The aims of this study were to characterize patterns of recurrence, identify clinicopathologic factors associated with recurrence, and evaluate the prognostic value of hormone receptor status and molecular classification among patients with available biomarker data.
Materials and Methods
1. Study population
This retrospective cohort study included patients aged 18 years or older with pathologically confirmed endometrial cancer who underwent curative primary treatment at Seoul National University Hospital or Seoul National University Bundang Hospital between 2010 and 2020. Eligible patients had postoperative 2009 International Federation of Gynecology and Obstetrics (FIGO) stage IA–IB disease [12], endometrioid histology, and grade 2 or 3 tumors. Patients were excluded if they had non-endometrioid histology, grade 1 disease, DM at diagnosis, prior cancer treatment, insufficient follow-up duration of less than 6 months, or a history of another malignancy within 5 years. A total of 298 patients were included in the final analysis, including 166 treated at Seoul National University Hospital and 132 treated at Seoul National University Bundang Hospital.
2. Data collection and variables
Clinical, pathologic, treatment, and follow-up data were retrospectively collected from medical records. Baseline clinical variables included age at surgery, body mass index, smoking history, and Eastern Cooperative Oncology Group performance score (ECOG PS). Pathologic variables included tumor size, histologic grade, FIGO stage, LVSI, estrogen receptor (ER)/progesterone receptor (PR) status, and molecular classification. Molecular classification was categorized as mismatch repair-deficient (MMRd), NSMP, or p53-abnormal (p53abn) based on available biomarker data. Because POLE mutation testing was not routinely available in our cohort, POLE-ultramutated tumors were not separately classified. Treatment-related variables included hysterectomy type, salpingo-oophorectomy status, the extent of lymph node dissection, receipt of adjuvant RT, chemotherapy, and RT modality. RT modality was classified as EBRT or VBT.
3. Study endpoints
The primary endpoint of this study was recurrence-free interval (RFI) after primary treatment. Recurrence patterns were classified according to the site of failure as vaginal, pelvic, para-aortic, or distant recurrence. Locoregional recurrence (LRR) was defined as vaginal, pelvic, or para-aortic recurrence. RFI was calculated from the date of surgery to the date of first documented recurrence; patients without recurrence were censored at the date of last follow-up.
4. Statistical analysis
Baseline characteristics were summarized using descriptive statistics. Continuous variables are presented as medians with interquartile ranges, and categorical variables are presented as frequencies and percentages. To identify prognostic factors associated with recurrence, Cox proportional hazards regression analyses were performed. Because immunohistochemical and molecular data were unavailable in a substantial proportion of patients, the primary multivariable analysis for the overall cohort was performed without ER, PR, or molecular classification variables. ER status and molecular classification were evaluated separately in exploratory univariable analyses. Variables meeting the prespecified criterion in univariable analysis (p < 0.05) were included in the multivariable model. Maximally selected rank statistics were used to determine the optimal cutoff value for age and tumor size.
Risk stratification was performed using recursive partitioning analysis (RPA) based on variables identified as significant in multivariable analysis. Recurrence proportions according to ER status were subsequently examined post hoc within each RPA-defined group among patients with available ER results. Survival curves were estimated using the Kaplan-Meier method and compared using the log-rank test. Cumulative LRR and DM rates were compared across RT modalities within each stage–grade subgroup using Fisher-Freeman-Halton exact tests. Patients who received pelvic EBRT, including those who also received VBT, were categorized in the EBRT group, whereas the VBT group included patients treated with VBT alone. A two-sided p-value of less than 0.05 was considered statistically significant. Statistical analyses were performed using Stata version 18.0 (StataCorp LLC, College Station, TX, USA) and R software version 4.2.0 (R Foundation for Statistical Computing, Vienna, Austria).
Results
1. Patient characteristics
A total of 298 patients with stage I, grade 2–3 endometrioid endometrial cancer who underwent curative primary treatment between 2010 and 2020 were included, with a median follow-up of 80.1 months (interquartile range [IQR], 63.7 to 98.9). Table 1 summarizes baseline characteristics. The median age at surgery was 57 years (IQR, 52 to 63). Most patients had good performance status (ECOG PS 0–1, 99.7%). Grade 2 and 3 tumors accounted for 73.5% and 26.5%, respectively. Stage IA disease was present in 66.4% of patients, and stage IB in 33.6%. LVSI was present in 30.5%. ER status was available in 130 of 298 patients (43.6%); among these patients, 117 (90.0%) were ER-positive and 13 (10.0%) were ER-negative. Molecular classification was available in 123 of 298 patients (41.3%); among these patients, 37 (30.1%) had MMRd, 73 (59.3%) had NSMP, and 13 (10.6%) had p53abn.
Table 1.
Patient characteristics
| Characteristics | No. (%) |
|---|---|
| Age (year), median (IQR) | 57 (52–63) |
| BMI (kg/m2), median (IQR) | 24.25 (22.01–26.61) |
| Follow-up (month), median (IQR) | 80.1 (63.7–98.9) |
| Smoking | |
| Never smoker | 295 (99.0) |
| Ex-smoker | 3 (1.0) |
| ECOG PS | |
| 0–1 | 297 (99.7) |
| 2 | 1 (0.3) |
| Hysterectomy type | |
| TAH | 32 (10.7) |
| LAVH | 62 (20.8) |
| TLH | 195 (65.4) |
| ARH | 6 (2.0) |
| LRH | 3 (1.0) |
| Salpingo-oophorectomy | |
| Not done | 3 (1.0) |
| Unilateral | 5 (1.7) |
| Bilateral | 275 (92.3) |
| Other | 15 (5.0) |
| Extent of lymph node dissection | |
| Not done | 35 (11.7) |
| Sentinel only | 32 (10.7) |
| Pelvic | 113 (37.9) |
| Pelvic and para-aortic | 118 (39.6) |
| Tumor size (cm), median (IQR) | 3.4 (2.2–4.8) |
| Histologic type | |
| Endometrioid | 298 (100) |
| Histologic grade | |
| Grade 2 | 219 (73.5) |
| Grade 3 | 79 (26.5) |
| FIGO stage | |
| IA | 198 (66.4) |
| IB | 100 (33.6) |
| LVSI | |
| No | 206 (69.1) |
| Yes | 91 (30.5) |
| Unknown | 1 (0.3) |
| ER | |
| Negative | 13 (4.4) |
| Positive | 117 (39.3) |
| Not tested | 168 (56.4) |
| PR | |
| Negative | 24 (8.1) |
| Positive | 106 (35.6) |
| Not tested | 168 (56.4) |
| Molecular subtype | |
| MMRd | 37 (12.4) |
| NSMP (MMRp, p53-wt) | 73 (24.5) |
| p53abn | 13 (4.4) |
| Not tested | 175 (58.7) |
| Adjuvant therapy | |
| RT alone | 129 (43.3) |
| Chemotherapy alone | 0 (0.0) |
| RT and chemotherapy | 28 (9.4) |
| Not done | 141 (47.3) |
| RT method | |
| EBRTa) | 101 (64.3) |
| VBT | 54 (34.4) |
| Unknown | 2 (1.3) |
| EBRT treatment details | |
| Total dose (Gy), median (range) | 50.4 (45.0–60.0) |
| No. of fractions, median (range) | 28 (23–28) |
| VBT treatment detailsb) | |
| Total dose (Gy), median (range) | 30.0 (25.0–54.4) |
| No. of fractions, median (range) | 5 (5–8) |
IQR, interquartile range; BMI, body mass index; ECOG PS, Eastern Cooperative Oncology Group performance score; TAH, total abdominal hysterectomy; LAVH, laparoscopically assisted vaginal hysterectomy; TLH, total laparoscopic hysterectomy; ARH, abdominal radical hysterectomy; LRH, laparoscopic radical hysterectomy; FIGO, International Federation of Gynecology and Obstetrics; LVSI, lymphovascular space invasion; ER, estrogen receptor; PR, progesterone receptor; MMRd, mismatch repair-deficient; NSMP, no specific molecular profile; MMRp, mismatch repair-proficient; p53-wt, p53 wild-type; p53abn, p53 abnormal; RT, radiation therapy; EBRT, external beam radiation therapy; VBT, vaginal brachytherapy.
One patient who received both EBRT and VBT was categorized in the EBRT group.
One patient received low-dose-rate VBT at a total dose of 65 Gy delivered in two applications.
Surgical procedures included total laparoscopic hysterectomy (65.4%), laparoscopically assisted vaginal hysterectomy (20.8%), and total abdominal hysterectomy (10.7%). Bilateral salpingo-oophorectomy was performed in 92.3% of patients. Pelvic and para-aortic lymph node dissection were performed in 39.6%, and pelvic lymph node dissection alone in 37.9%. Sentinel lymph node evaluation was done in 10.7%. Adjuvant treatment included RT alone in 43.3% and combined RT plus chemotherapy in 9.4%, whereas 47.3% received no adjuvant therapy. Among RT recipients, 101 patients (64.3%) were categorized in the EBRT group, 54 (34.4%) received VBT alone, and two (1.3%) had unclear modality information. The median EBRT dose was 50.4 Gy (range, 45.0 to 60.0), delivered in a median of 28 fractions (range, 23 to 28). Among patients who received high-dose-rate VBT, the median dose was 30 Gy (range, 25.0 to 54.4), delivered in 5 fractions (range, 5 to 8). One patient received low-dose-rate VBT at a total dose of 65 Gy delivered in two applications. One patient received both EBRT and VBT, consisting of EBRT at 46 Gy in 23 fractions and VBT at 15 Gy in 3 fractions.
2. Prognostic factors for RFI
During follow-up, 40 patients (13.4%) experienced recurrence. DM was most common (n = 28, 9.4%), followed by vaginal recurrence (n = 10, 3.4%), para-aortic recurrence (n = 8, 2.7%) and pelvic recurrence (n = 7, 2.3%) (Fig. 1). Univariable and multivariable Cox regression analyses were performed to identify prognostic factors for RFI (Table 2). ER, PR, and molecular classification results were not included in the primary multivariable Cox model because of substantial missing data. Age was modeled as a continuous variable, whereas tumor size was categorized as <4 cm, ≥4 cm, and unknown, using the 4-cm cutoff identified by maximally selected rank statistics.
Fig. 1.
Patterns of recurrence by site. Venn diagram shows the distribution and overlaps of cumulative recurrence sites during follow-up.
Table 2.
Cox proportional hazards model for recurrence-free interval
| Variable | Univariable | Multivariable | ||
|---|---|---|---|---|
| HR (95% CI) | p-value | HR (95% CI) | p-value | |
| Age at surgery | ||||
| Per 1 year increase | 1.07 (1.04–1.11) | <0.001 | 1.05 (1.02–1.09) | 0.004 |
| Tumor size (cm) | ||||
| <4 | Ref | Ref | ||
| ≥4 | 3.00 (1.48–6.06) | 0.002 | 2.28 (1.10–4.72) | 0.027 |
| Unknown | 1.23 (0.46–3.28) | 0.683 | 1.27 (0.46–3.48) | 0.640 |
| Grade | ||||
| 2 | - | - | ||
| 3 | 2.07 (1.10–3.89) | 0.025 | 2.20 (1.12–4.33) | 0.022 |
| FIGO stage | ||||
| IA | - | - | ||
| IB | 3.23 (1.72–6.08) | <0.001 | 2.31 (1.03–5.19) | 0.043 |
| LVSI | ||||
| Negative | - | |||
| Positive | 1.93 (1.03–3.60) | 0.039 | 1.24 (0.61–2.53) | 0.559 |
| Radiotherapy | ||||
| Not done | - | - | ||
| Done | 1.98 (1.02–3.84) | 0.043 | 0.59 (0.23–1.51) | 0.269 |
| Chemotherapy | ||||
| Not done | - | - | ||
| Done | 0.77 (0.24–2.51) | 0.670 | - | - |
HR, hazard ratio; CI, confidence interval; FIGO, International Federation of Gynecology and Obstetrics; LVSI, lymphovascular space invasion.
In univariable analysis, older age, tumor size ≥4 cm, grade 3 histology, stage IB disease, LVSI positivity, and receipt of adjuvant RT were significantly associated with inferior RFI. In multivariable analysis, older age (hazard ratio [HR] per 1-year increase, 1.05; 95% confidence interval [CI], 1.02 to 1.09; p = 0.004), tumor size ≥4 cm (HR, 2.28; 95% CI, 1.10 to 4.72; p = 0.027), grade 3 histology (HR, 2.20; 95% CI, 1.12 to 4.33; p = 0.022), and stage IB disease (HR, 2.31; 95% CI, 1.03 to 5.19; p = 0.043) remained significantly associated with inferior RFI. Unknown tumor size, LVSI positivity, and receipt of adjuvant RT were not statistically significant in the multivariable analysis.
3. Exploratory univariable Cox analysis of ER status and molecular classification for RFI
Because ER status and molecular classification were unavailable in a substantial proportion of patients, these variables were not included in the multivariable Cox regression model. Instead, they were evaluated in exploratory univariable Cox analyses (Table 3). ER negativity was significantly associated with inferior RFI compared with ER positivity (HR, 5.10; 95% CI, 2.09 to 12.43; p < 0.001), whereas ER-unknown status was not significantly different from ER positivity. Using MMRd as the reference, molecular classification was not significantly associated with RFI: NSMP, HR 1.69 (95% CI, 0.62 to 4.62; p = 0.305); p53abn, HR 1.08 (95% CI, 0.21 to 5.55; p = 0.929); unknown classification, HR 0.71 (95% CI, 0.26 to 1.93; p = 0.501).
Table 3.
Univariable Cox analysis of estrogen receptor status and molecular classification for recurrence-free interval
| Variable | No. | Univariable |
|
|---|---|---|---|
| HR (95% CI) | p-value | ||
| Estrogen receptor status | |||
| Positive | 117 | Ref | |
| Negative | 13 | 5.10 (2.09–12.43) | <0.001 |
| Unknown | 168 | 0.67 (0.34–1.33) | 0.250 |
| Molecular classification | |||
| MMRd | 37 | Ref | |
| NSMP | 73 | 1.69 (0.62–4.62) | 0.305 |
| p53abn | 13 | 1.08 (0.21–5.55) | 0.929 |
| Unknown | 175 | 0.71 (0.26–1.93) | 0.501 |
HR, hazard ratio; CI, confidence interval; MMRd, mismatch repair-deficient; NSMP, no specific molecular profile; p53abn, p53 abnormal.
4. Treatment patterns and cumulative recurrence according to stage, grade, and RT modality
Treatment patterns and cumulative recurrence profiles were summarized according to stage, grade, and RT modality (Table 4). Two patients who received RT but had unclear modality information, one with stage IA grade 2 disease and one with stage IB grade 2 disease, were not displayed as separate modality columns; neither experienced LRR or DM. The distribution of RT modality differed across stage–grade subgroups: most patients with stage IA grade 2 disease did not receive RT (120 of 151, 79.5%), whereas EBRT was used in 19 of 47 patients (40.4%) with stage IA grade 3 disease, 41 of 68 patients (60.3%) with stage IB grade 2 disease, and 31 of 32 patients (96.9%) with stage IB grade 3 disease. One patient with stage IB, grade 3 disease received both EBRT and VBT and was included in the EBRT group.
Table 4.
Treatment patterns and cumulative recurrence according to stage, grade, and RT modality
| Stage/Grade | No. | No RT | EBRT | VBT | p-valuea) |
|---|---|---|---|---|---|
| Stage IA/Grade 2b) | 151 | 120 (79.5) | 10 (6.6) | 20 (13.2) | |
| LRR: 3/120 (2.5) | LRR: 0/10 (0.0) | LRR: 1/20 (5.0) | 0.589 | ||
| DM: 6/120 (5.0) | DM: 0/10 (0.0) | DM: 0/20 (0.0) | 0.737 | ||
| Stage IA/Grade 3 | 47 | 14 (29.8) | 19 (40.4) | 14 (29.8) | |
| LRR: 2/14 (14.3) | LRR: 1/19 (5.3) | LRR: 1/14 (7.1) | 0.808 | ||
| DM: 1/14 (7.1) | DM: 0/19 (0.0) | DM: 1/14 (7.1) | 0.506 | ||
| Stage IB/Grade 2b) | 68 | 6 (8.8) | 41 (60.3) | 20 (29.4) | |
| LRR: 0/6 (0.0) | LRR: 5/41 (12.2) | LRR: 2/20 (10.0) | >0.99 | ||
| DM: 1/6 (16.7) | DM: 8/41 (19.5) | DM: 3/20 (15.0) | >0.99 | ||
| Stage IB/Grade 3 | 32 | 1 (3.1) | 31c) (96.9) | 0 (0.0) | |
| LRR: 0/1 (0.0) | LRR: 7/31 (22.6) | LRR: - | N/A | ||
| DM: 0/1 (0.0) | DM: 8/31 (25.8) | DM: - | N/A |
Values are presented as number/total (%).
RT, radiation therapy; EBRT, external beam radiation therapy; VBT, vaginal brachytherapy; LRR, locoregional recurrence; DM, distant metastasis; N/A, not available.
p-values were calculated using Fisher-Freeman-Halton exact tests comparing cumulative LRR or DM rates across No RT, EBRT, and VBT within each stage–grade subgroup. Patients with unclear RT modality were excluded from these tests.
Two patients who received RT but had unclear modality information, one with stage IA grade 2 disease and one with stage IB grade 2 disease, were omitted from the modality-specific columns; neither experienced LRR or DM. Percentages for RT modality were calculated within each stage–grade subgroup, and LRR/DM rates were calculated within each RT modality subgroup.
One patient with stage IB grade 3 disease received both EBRT and VBT and was categorized in the EBRT group.
In stage IA grade 2 disease, cumulative LRR and DM rates were low across modality subgroups, with no significant differences according to RT modality (LRR, p = 0.589; DM, p = 0.737). In stage IA grade 3 disease, the no-RT subgroup showed a numerically higher LRR rate than the EBRT or VBT subgroups (14.3% vs. 5.3% vs. 7.1%), but this difference was not statistically significant (p = 0.808); DM rates also did not differ significantly across RT modalities (7.1% vs. 0.0% vs. 7.1%; p = 0.506). In stage IB grade 2 disease, cumulative LRR rates were 0.0%, 12.2%, and 10.0%, whereas cumulative DM rates were 16.7%, 19.5%, and 15.0% in the no-RT, EBRT, and VBT groups, respectively; neither outcome differed significantly across RT modalities (LRR, p > 0.99; DM, p > 0.99). In stage IB grade 3 disease, nearly all patients received EBRT, and cumulative LRR and DM remained frequent despite EBRT use, occurring in seven of 31 patients (22.6%) and eight of 31 patients (25.8%), respectively.
5. RPA using clinicopathologic factors and post hoc analysis of ER status
RPA was performed using age, tumor size, histologic grade, and FIGO stage, which were significantly associated with RFI in the multivariable analysis. The final tree classified patients into three groups: group 1 comprised patients aged ≤70 years with stage IA disease (n = 189), group 2 comprised patients aged ≤70 years with stage IB disease (n = 85), and group 3 comprised patients aged >70 years regardless of stage (n = 24) (Fig. 2A). Recurrence occurred in 12 patients (6.3%), 18 patients (21.2%), and 10 patients (41.7%) in groups 1, 2, and 3, respectively. The estimated 3-year RFI rates were 96.0%, 79.7%, and 62.5%, and the corresponding 5-year RFI rates were 93.2%, 79.7%, and 55.6% in groups 1, 2, and 3, respectively. The RFI curves differed significantly among the three groups (overall log-rank p < 0.001) (Fig. 2B).
Fig. 2.
Clinicopathologic recursive partitioning analysis-based recurrence-risk stratification. (A) Recursive partitioning analysis using clinicopathologic variables. (B) Kaplan–Meier curves for recurrence-free interval according to the three recursive partitioning analysis-defined groups.
ER status was subsequently evaluated post hoc within each RPA-defined group, excluding patients with unknown ER status (Table 5). In group 1, recurrence occurred in five of 73 ER-positive patients (6.8%) and one of five ER-negative patients (20.0%). In group 2, recurrence occurred in six of 33 ER-positive patients (18.2%) and five of six ER-negative patients (83.3%). In group 3, recurrence occurred in six of 11 ER-positive patients (54.5%) and one of two ER-negative patients (50.0%).
Table 5.
Post hoc analysis of recurrence according to ER status within RPA-defined groups
| RPA group | Definition | ER assessed | Recurrence among ER-positive | Recurrence among ER-negative |
|---|---|---|---|---|
| Group 1 | Age ≤70 years, stage IA | 78/189 (41.3) | 5/73 (6.8) | 1/5 (20.0) |
| Group 2 | Age ≤70 years, stage IB | 39/85 (45.9) | 6/33 (18.2) | 5/6 (83.3) |
| Group 3 | Age >70 years, regardless of stage | 13/24 (54.2) | 6/11 (54.5) | 1/2 (50.0) |
Values are presented as number/total (%).
ER, estrogen receptor; RPA, recursive partitioning analysis.
Discussion and Conclusion
In this retrospective two-institution cohort of patients with stage I, grade 2–3 endometrioid endometrial cancer, recurrence occurred in 13.4% of patients, with distant recurrence of 9.4%, accounting for 70% of all recurrence events as the predominant pattern of failure. In the Cox regression analyses excluding ER, PR, and molecular classification variables because of substantial missing data, older age, tumor size ≥4 cm, grade 3 histology, and stage IB disease remained significantly associated with inferior RFI after adjustment. RPA based on these clinicopathologic variables further classified patients into three groups defined by age and stage. In a post hoc analysis within the RPA-defined groups, ER-negative tumors showed a higher proportion of recurrence among patients aged ≤70 years with stage IB disease. However, this observation was based on a limited number of ER-negative patients and should be interpreted as exploratory.
The predominance of distant recurrence in our cohort is clinically important. Historical randomized trials established that adjuvant pelvic RT improves locoregional control in early-stage endometrial cancer but does not clearly improve overall survival, and later studies similarly suggested that DM remains a major mode of treatment failure in higher-risk early-stage disease [4-7]. More recent retrospective series have also reported that distant relapse accounts for a substantial proportion of recurrences despite adjuvant treatment [3,13], although these studies included broader stage and grade populations than ours, which limits direct comparisons. Our findings therefore support the view that postoperative RT mainly contributes to locoregional control, whereas systemic recurrence remains a major driver of treatment failure even in early-stage, grade 2–3 endometrioid endometrial cancer.
The identification of age, tumor size, grade, and stage as prognostic factors is consistent with the conventional clinicopathologic framework used to guide adjuvant treatment in early-stage endometrial cancer [4,5,14]. Importantly, these clinicopathologic factors remained significantly associated with RFI in the multivariable model, supporting their practical value in real-world settings where biomarker testing, including molecular classification, is incomplete or selectively performed. The RPA findings further support the utility of conventional clinicopathologic variables. Among patients aged ≤70 years, stage differentiated patients with lower and higher recurrence risk, whereas patients older than 70 years formed the highest-risk group regardless of stage. The age threshold of 70 years was derived from maximally selected rank statistics and should therefore be regarded as a pragmatic, data-driven cutoff rather than a biologically fixed threshold.
ER negativity was significantly associated with inferior RFI in the exploratory univariable analysis. When ER status was examined after clinicopathologic risk stratification, ER-negative tumors showed higher recurrence rates than ER-positive tumors in both groups aged ≤70 years, with the largest difference observed in patients with stage IB disease. In contrast, recurrence rates were similar according to ER status among patients older than 70 years. These findings suggest that ER status may provide additional prognostic information. This observation is biologically plausible and is consistent with emerging evidence suggesting that ER expression may refine prognostic stratification, particularly within NSMP endometrial cancer [11]. However, ER status was available in only 43.6% of the cohort, and only 13 patients were ER-negative. The post hoc findings are therefore vulnerable to selection bias and random variation and should be regarded as hypothesis-generating.
Contemporary international guidelines have incorporated molecular classification into endometrial cancer risk assessment and adjuvant treatment planning [2,15]. In addition, recent studies and trials have shown that molecular classification can alter adjuvant treatment recommendations and support individualized postoperative management [10,16]. Despite this established clinical relevance, molecular classification was not significantly associated with RFI in our exploratory univariable analysis. This lack of a significant association may partly reflect the incomplete implementation of contemporary molecular classification in our cohort, including the absence of routine POLE mutation testing due to limited availability, the large proportion of patients with unknown molecular status, and the relatively small number of molecularly characterized patients, particularly those with p53abn tumors. Therefore, our findings should not be interpreted as contradicting contemporary guidelines or recent studies supporting molecularly informed risk stratification and individualized adjuvant treatment planning.
The descriptive analysis of RT modality by stage–grade subgroups illustrates substantial treatment selection according to baseline pathologic risk. Patients with stage IA grade 2 disease were predominantly managed without adjuvant RT and had low cumulative LRR and DM rates. In contrast, EBRT use was more common in higher-risk subgroups and was nearly universal in stage IB grade 3 disease. Within stage–grade subgroups, no statistically significant differences in cumulative LRR or DM were observed across RT modalities. However, these comparisons were limited by small modality-specific subgroup sizes and substantial confounding by indication. Therefore, the observed recurrence rates should be interpreted as descriptive patterns rather than direct estimates of the comparative efficacy of EBRT, VBT, or no RT.
The recurrence profile of stage IB grade 3 disease is notable. Despite near-universal EBRT use, cumulative LRR and DM remained frequent in this subgroup. This finding suggests that conventional locoregional treatment may not fully offset the adverse prognosis associated with deep myometrial invasion and high-grade histology. The substantial DM rate in this subgroup also indicates that systemic relapse risk, rather than locoregional failure alone, contributes meaningfully to treatment failure. These observations support the need for further risk refinement using additional biologic markers and, potentially, systemic treatment strategies in selected high-risk early-stage patients.
This study has several limitations. First, its retrospective design introduces the possibility of selection bias and treatment heterogeneity. Second, biomarker availability was incomplete, particularly for ER and molecular classification, which limited the interpretability of biomarker-based analyses. Furthermore, the post hoc ER comparisons within the RPA-defined groups included very small ER-negative subgroups and were intended to be descriptive rather than definitive subgroup analyses. Third, LVSI was not assessed using a standardized three-tiered approach, which may have reduced its prognostic significance, given previous data suggesting that substantial LVSI is particularly relevant for pelvic and distant recurrence [17]. Fourth, this cohort was staged according to the 2009 FIGO system. Because the revised 2023 FIGO staging system incorporates additional histologic and molecular features that may result in stage migration, some patients in our cohort might be classified differently under current criteria, which may limit the direct applicability of our findings to contemporary practice [18,19]. Finally, the number of recurrence events was modest, especially in subgroup and RPA analyses, increasing the risk of model instability.
In conclusion, recurrence in stage I, grade 2–3 endometrioid endometrial cancer was infrequent, but DM was the most common recurrence pattern. In the multivariable clinicopathologic Cox model, older age, tumor size ≥4 cm, grade 3 histology, and stage IB disease were significantly associated with inferior RFI. An age- and stage-based RPA model provided clinically useful risk stratification. Post hoc findings suggested that ER status may further refine recurrence risk within selected clinicopathologic groups, particularly among younger patients with stage IB disease, although this observation requires validation because of incomplete ER testing and the small number of ER-negative patients. These results support the continued value of conventional clinicopathologic risk assessment while highlighting the need for improved biologic risk refinement and systemic relapse prevention strategies in selected high-risk early-stage patients.
Footnotes
Statement of Ethics
This study was approved by the Institutional Review Board at Seoul National University Hospital (No. J-2306-052-1436) and Seoul National University Bundang Hospital (No. B-2305-829-103). The committee granted an informed consent waiver based on the retrospective nature and anonymity of the data used in this work.
Conflict of Interest
No potential conflict of interest relevant to this article was reported.
Funding
None.
Author Contributions
Conceptualization, JSS, KYE; Investigation and methodology, SML, JSS, KSK, HCK, KYE; Formal analysis, SML, JSS, KYE; Data curation, JSS, KYE; Visualization, SML, JSS, KYE; Writing of the original draft, SML, KYE; Writing of the review and editing, SML, JSS, KSK, HCK, KYE; Approval of the final manuscript, all authors.
Data Availability Statement
Results of the analyzed data can be provided according to reasonable request.
References
- 1.Makker V, MacKay H, Ray-Coquard I, et al. Endometrial cancer. Nat Rev Dis Primers. 2021;7:88. doi: 10.1038/s41572-021-00324-8. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2.Concin N, Matias-Guiu X, Cibula D, et al. ESGO-ESTRO-ESP guidelines for the management of patients with endometrial carcinoma: update 2025. Lancet Oncol. 2025;26:e423–35. doi: 10.1016/s1470-2045(25)00167-6. [DOI] [PubMed] [Google Scholar]
- 3.Francis SR, Ager BJ, Do OA, et al. Recurrent early stage endometrial cancer: patterns of recurrence and results of salvage therapy. Gynecol Oncol. 2019;154:38–44. doi: 10.1016/j.ygyno.2019.04.676. [DOI] [PubMed] [Google Scholar]
- 4.Keys HM, Roberts JA, Brunetto VL, et al. A phase III trial of surgery with or without adjunctive external pelvic radiation therapy in intermediate risk endometrial adenocarcinoma: a Gynecologic Oncology Group study. Gynecol Oncol. 2004;92:744–51. doi: 10.1016/j.ygyno.2003.11.048. [DOI] [PubMed] [Google Scholar]
- 5.Creutzberg CL, van Putten WL, Koper PC, et al. Surgery and postoperative radiotherapy versus surgery alone for patients with stage-1 endometrial carcinoma: multicentre randomised trial. PORTEC Study Group. Post Operative Radiation Therapy in Endometrial Carcinoma. Lancet. 2000;355:1404–11. doi: 10.1016/s0140-6736(00)02139-5. [DOI] [PubMed] [Google Scholar]
- 6.Nout RA, Smit VT, Putter H, et al. Vaginal brachytherapy versus pelvic external beam radiotherapy for patients with endometrial cancer of high-intermediate risk (PORTEC-2): an open-label, non-inferiority, randomised trial. Lancet. 2010;375:816–23. doi: 10.1016/s0140-6736(09)62163-2. [DOI] [PubMed] [Google Scholar]
- 7.Wortman BG, Creutzberg CL, Putter H, et al. Ten-year results of the PORTEC-2 trial for high-intermediate risk endometrial carcinoma: improving patient selection for adjuvant therapy. Br J Cancer. 2018;119:1067–74. doi: 10.1038/s41416-018-0310-8. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Bendifallah S, Canlorbe G, Collinet P, et al. Just how accurate are the major risk stratification systems for early-stage endometrial cancer? Br J Cancer. 2015;112:793–801. doi: 10.1038/bjc.2015.35. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Cancer Genome Atlas Research Network. Kandoth C, Schultz N, et al. Integrated genomic characterization of endometrial carcinoma. Nature. 2013;497:67–73. doi: 10.1038/nature12113. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.Leon-Castillo A. Update in the molecular classification of endometrial carcinoma. Int J Gynecol Cancer. 2023;33:333–42. doi: 10.1136/ijgc-2022-003772. [DOI] [PubMed] [Google Scholar]
- 11.Vermij L, Jobsen JJ, Leon-Castillo A, et al. Prognostic refinement of NSMP high-risk endometrial cancers using oestrogen receptor immunohistochemistry. Br J Cancer. 2023;128:1360–8. doi: 10.1038/s41416-023-02141-0. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Pecorelli S. Revised FIGO staging for carcinoma of the vulva, cervix, and endometrium. Int J Gynaecol Obstet. 2009;105:103–4. doi: 10.1016/j.ijgo.2009.02.012. [DOI] [PubMed] [Google Scholar]
- 13.Wang W, Wang T, Liu Z, et al. Overview of adjuvant radiotherapy on survival, failure pattern and toxicity in stage I to II endometrial carcinoma: a long-term multi-institutional analysis in China. BMC Cancer. 2022;22:266. doi: 10.1186/s12885-022-09343-4. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.Harkenrider MM, Abu-Rustum N, Albuquerque K, et al. Radiation therapy for endometrial cancer: an American Society for Radiation Oncology Clinical Practice Guideline. Pract Radiat Oncol. 2023;13:41–65. doi: 10.1016/j.prro.2022.09.002. [DOI] [PubMed] [Google Scholar]
- 15.National Comprehensive Cancer Network . Plymouth Meeting, PA: National Comprehensive Cancer Network; 2026. Uterine neoplasms (version 2.2026) [Internet] [cited 2026 Apr 15]. Available from: https://www.nccn.org/professionals/physician_gls/pdf/uterine.pdf. [Google Scholar]
- 16.van den Heerik AS, Horeweg N, Haverkort MA, et al. Molecular profile-based adjuvant treatment for women with high-intermediate risk endometrial cancer (PORTEC-4a): results of a randomised, open-label, phase 3, multicentre, non-inferiority trial. Lancet Oncol. 2026;27:23–35. doi: 10.1016/s1470-2045(25)00612-6. [DOI] [PubMed] [Google Scholar]
- 17.Bosse T, Peters EE, Creutzberg CL, et al. Substantial lymph-vascular space invasion (LVSI) is a significant risk factor for recurrence in endometrial cancer: a pooled analysis of PORTEC 1 and 2 trials. Eur J Cancer. 2015;51:1742–50. doi: 10.1016/j.ejca.2015.05.015. [DOI] [PubMed] [Google Scholar]
- 18.Berek JS, Matias-Guiu X, Creutzberg C, et al. FIGO staging of endometrial cancer: 2023. Int J Gynaecol Obstet. 2023;162:383–94. doi: 10.3802/jgo.2023.34.e85. [DOI] [PubMed] [Google Scholar]
- 19.Schwameis R, Fanfani F, Ebner C, et al. Verification of the prognostic precision of the new 2023 FIGO staging system in endometrial cancer patients: an international pooled analysis of three ESGO accredited centres. Eur J Cancer. 2023;193:113317. doi: 10.1016/j.ejca.2023.113317. [DOI] [PubMed] [Google Scholar]