Timing of adjuvant radiotherapy for early-stage endometrial carcinoma: a single-center retrospective cohort study

Abstract

Objective

To investigate the appropriate timing of radiotherapy (RT) after hysterectomy in women with early-stage endometrial cancer (EC).

Methods

We analyzed the data of 1,062 patients with early-stage EC who underwent postoperative RT at our hospital between April 1999 and November 2020. Restricted cubic spline were used to explore the relationship between the surgery-radiotherapy interval (SRI) and local recurrence-free survival (LRFS). The maximally selected rank statistics method was used to identify the optimal threshold for SRI. The overall survival (OS), disease-free survival (DFS), LRFS, and distant metastasis-free survival (DMFS) rates were estimated using the Kaplan-Meier method. Multivariate analysis was performed using the Cox proportional hazards regression.

Results

In entire cohort, patients with SRI ≥42 days had worse survival. In multivariate analysis, SRI was an independent prognostic factor for OS (p=0.011), DFS (p=0.019), LRFS (p=0.013) and DMFS (p=0.050). However, in piecewise Cox regression, the significance of SRI for DMFS disappeared. In the subgroup analysis, the optimal cut-off value for SRI in the high-intermediate risk (HIR) and high-risk (HR) groups was 33 days. Multivariate analysis showed that SRI was an independent prognostic factor only for LRFS (p=0.033) and marginally associated with OS (p=0.055).

Conclusion

The timing of postoperative RT is crucial in patients with early-stage EC. Adjuvant RT should be initiated as soon as the vaginal cuff is healed, while for HIR and HR patients, it should be initiated within 33 days.

Synopsis

The delay between surgery and radiotherapy (RT) is associated with poor prognosis. In the high-intermediate risk and high-risk groups, the surgery-RT interval was significantly associated with local recurrence-free survival and marginally associated with overall survival. Therefore, in these groups, adjuvant RT should begin within 33 days after surgery.

INTRODUCTION

Endometrial cancer (EC) is one of the most common malignant tumors in women, with an increasing annual incidence. It was estimated that approximately 78,000 new cancer cases were diagnosed in China in 2022, with an incidence of 11.25/10,000 [1]. It is the sixth most common cancer in women, accounting for an estimated 420,000 new cases globally each year by 2022 [2]. Fortunately, the typical symptom of postmenopausal vaginal bleeding allows for early diagnosis and better survival with standard treatment. Surgery is the cornerstone of treatment for early-stage EC, and radiotherapy (RT) is an effective adjuvant treatment. In recent years, several studies have examined the optimal timing of RT after surgery [3,4,5,6,7,8,9], but the answer remains unclear. In the previous studies, the cut-off value for the interval between hysterectomy and RT was determined empirically or based on earlier studies. As a result, there is still no consensus regarding the impact of RT timing on patient survival outcomes. In addition, no studies have evaluated the optimal timing for RT in different risk groups.

This study aimed to explore the optimal timing of postoperative adjuvant RT in patients with early-stage EC and investigate its impact on patient survival outcomes.

MATERIALS AND METHODS

1. Patient selection

We reviewed the data of 1,269 patients with early-stage EC who underwent postoperative adjuvant RT at our institution between April 1999 and November 2020. One hundred and seventy-six patients who received adjuvant chemotherapy were excluded. Additionally, we excluded 8 patients with other malignancies and 23 patients with missing information about the day of surgery or the start of adjuvant RT. We finally had a cohort of 1,062 patients. Based on the risk stratification defined by the European Society for Medical Oncology EC Clinical Practice Guidelines (2016) [10], patients were classified into low-risk (LR), intermediate-risk, high-intermediate risk (HIR), and high-risk (HR) groups (Table S1). In our institution, patients in the LR group who received RT must meet one of the following criteria: advanced age (≥60 years), big tumors (≥2 cm), or lower uterine segment involvement (LUSI).

2. Survival and follow-up

The surgery-radiotherapy interval (SRI) was defined as the duration between surgery and adjuvant RT. Follow-up was initiated at the end of RT, every 3–6 months for the first 2 years and every 6–12 months thereafter. Overall survival (OS) was defined as the time from surgery to the date of death or last follow-up. Cancer-specific survival (CSS) was defined as the time from surgery to death due to EC or the last follow-up. Disease-free survival (DFS) was defined as the time from surgery to the date of treatment failure, death, or the last follow-up. Local recurrence-free survival (LRFS) was defined as the time from surgery to local recurrence, death, or the last follow-up. Regional recurrence-free survival (RFS) was defined as the time from surgery to regional lymph node (LN) metastasis, death, or the last follow-up. Distant metastasis-free survival (DMFS) was defined as the time from surgery to the date of distant metastasis (DM) (site outside the locoregional area), death, or last follow-up.

3. Statistical analysis

Restricted cubic spline (RCS) can be used in Cox proportional hazard models to explore the non-linear associations (p<0.050) between continuous predictors and hazard ratios of event [11]. We used it to evaluate the correlations between the SRI and the hazard ratio for LRFS. If the RCS indicated no significant nonlinearity, the maximally selected rank statistics (MSRS) method was applied to find the optimal threshold. MSRS is used to identify optimal cut-points in continuous variables that best separate groups based on a rank-based test statistic. Bootstrap resampling (1,000 repetitions) was used to evaluate the reproducibility and stability of the identified threshold. The Kaplan-Meier method was used to analyze survival data, and survival curves were compared using log-rank tests. Cox proportional hazard regression models were used for univariate and multivariate analyses of prognostic factors. In order to analyze the influencing factors of survival more comprehensively, factors that emerged significant (p<0.200) in the univariate analysis were entered into the multivariate analysis. The landmark method is a survival analysis technique used to address time-dependent biases when evaluating the effect of a time-varying factors. Piecewise univariate and multivariate Cox regression analyses were used to explore prognostic factors. All statistical analyses were performed using the SPSS software (version 25.0; IBM Corp., Armonk, NY, USA) and the ‘survival,’ survminer,’ ‘rms,’ “maxstat,” “boot,” “jskm,” and “ggplot2” packages of the R software (version 4.3.1; R Foundation for Statistical Computing, Vienna, Austria).

RESULTS

1. Patient characteristics

The baseline characteristics of the entire cohort (n=1,062) and subgroups are presented in Table 1. The median age at diagnosis was 57 (range: 23–85) years. A total of 1,053 patients (99.2%) had a Karnofsky Performance Status (KPS) score ≥80. Nine hundred and ninety-seven (93.9%) patients had stage I disease, 65 (6.1%) had stage II disease. All patients underwent hysterectomy, of which 728 (68.5%) underwent LN evaluation (sentinel LN biopsy or LN dissection). Postoperative pathology indicated endometrioid adenocarcinoma in 1,028 (96.8%) patients, 896 (84.4%) were G1 or G2. Deep myometrial invasion was seen in 455 (42.8%) patients, 193 (18.2%) were LVSI-positive, and 315 (29.7%) had LUSI. All patients received adjuvant RT. Vaginal brachytherapy (VBT) alone was administered to 789 (74.3%) patients using ¹⁹²Ir at a high dose rate (HDR) with 20–30 Gy in 4–6 fractions. A combination of external beam radiotherapy (EBRT) and VBT was administered to 228 (21.5%) patients. The dose for the HDR brachytherapy boost was 10–20 Gy in 2–4 fractions. EBRT alone was administered to 45 (4.2%) patients at a dose of 45–50.4 Gy in 25–28 fractions.

Table 1. Baseline characteristics of the entire cohort and subgroups.

Variables		Entire cohort				HR + HIR
		Total (n=1,062)	SRI <42 days (n=424)	SRI ≥42 days (n=638)	p-value	Total (n=353)	SRI <33 days (n=51)	SRI ≥33 days (n=302)	p-value
Age (yr)					0.004				0.007
	<60	661 (62.2)	286 (67.5)	375 (58.8)		224 (63.5)	41 (80.4)	183 (60.6)
	≥60	401 (37.8)	138 (32.5)	263 (41.2)		129 (36.5)	10 (19.6)	119 (39.4)
KPS					0.009				0.064
	90	691 (65.1)	256 (60.4)	435 (68.2)		247 (70.0)	30 (58.8)	217 (71.9)
	80	362 (34.1)	164 (38.7)	198 (31.0)		105 (29.7)	21 (41.2)	84 (27.8)
	70	9 (0.8)	4 (0.9)	5 (0.8)		1 (0.3)	0 (0.0)	1 (0.3)
Tumor size (cm)					0.134				0.007
	<2	163 (15.3)	59 (13.9)	104 (16.3)		55 (15.6)	5 (9.8)	50 (16.6)
	≥2	653 (61.5)	254 (59.9)	399 (62.5)		213 (60.3)	25 (49.0)	188 (62.3)
	Unknown	246 (23.2)	111 (26.2)	135 (21.2)		85 (24.1)	21 (41.2)	64 (21.2)
Histological grade					0.310				0.050
	G1+G2	896 (84.4)	356 (84.0)	540 (84.6)		194 (55.0)	20 (39.2)	174 (57.6)
	G3	132 (12.4)	58 (13.7)	74 (11.6)		131 (37.1)	26 (51.0)	105 (34.8)
	Gx	34 (3.2)	10 (2.3)	24 (3.8)		28 (7.9)	5 (9.8)	23 (7.6)
DMI					0.129				0.221
	<1/2	607 (57.2)	230 (54.2)	377 (59.1)		203 (57.5)	25 (49.0)	178 (58.9)
	≥1/2	455 (42.8)	194 (45.8)	261 (40.9)		150 (42.5)	26 (51.0)	124 (41.1)
LVSI					0.009				0.023
	Yes	193 (18.2)	61 (14.4)	132 (20.7)		185 (52.4)	19 (37.3)	166 (55.0)
	No	869 (81.8)	363 (85.6)	506 (79.3)		168 (47.6)	32 (62.7)	136 (45.0)
LUSI					0.158				0.573
	Yes	315 (29.7)	138 (32.5)	177 (27.7)		96 (27.2)	12 (23.5)	84 (27.8)
	No	732 (68.9)	282 (66.5)	450 (70.5)		252 (71.4)	38 (74.5)	214 (70.9)
	Unknown	15 (1.4)	4 (1.0)	11 (1.7)		5 (1.4)	1 (2.0)	4 (1.3)
PWC					<0.001				0.037
	(−)	821 (77.3)	364 (85.8)	457 (71.6)		261 (73.9)	45 (88.2)	216 (71.5)
	(+)	36 (3.4)	6 (1.5)	30 (4.7)		12 (3.4)	0 (0.0)	12 (4.0)
	Unknown	205 (19.3)	54 (12.7)	151 (23.7)		80 (22.7)	6 (11.8)	74 (24.5)
FIGO stage (2009)					0.236				1.000
	I	997 (93.9)	403 (95.0)	594 (93.1)		288 (81.6)	42 (82.4)	246 (81.5)
	II	65 (6.1)	21 (5.0)	44 (6.9)		65 (18.4)	9 (17.6)	56 (18.5)
Risk group					0.075				0.354
	LR	398 (37.5)	154 (36.3)	244 (38.2)		-	-	-
	IR	311 (29.3)	141 (33.3)	170 (26.6)		-	-	-
	HIR	211 (19.9)	82 (19.3)	129 (20.2)		211 (59.8)	27 (52.9)	184 (60.9)
	HR	142 (13.3)	47 (11.1)	95 (15.0)		142 (40.2)	24 (47.1)	118 (39.1)
LN evaluation					0.893				0.611
	Yes	728 (68.5)	385 (74.3)	433 (79.6)		256 (72.5)	35 (68.6)	221 (73.2)
	No	334 (31.5)	132 (31.1)	111 (20.4)		97 (27.5)	16 (31.4)	81 (26.8)
Patterns of radiotherapy					0.001				<0.001
	VBT	789 (74.3)	290 (68.4)	499 (78.2)		220 (62.3)	22 (43.1)	198 (65.6)
	EBRT+VBT	228 (21.5)	103 (24.3)	125 (19.6)		112 (31.7)	20 (39.2)	92 (30.5)
	EBRT	45 (4.2)	31 (7.3)	14 (2.2)		21 (6.0)	9 (17.6)	12 (4.0)

Values are presented as number (%).

DMI, depth of myometrial invasion; EBRT, external beam radiotherapy; FIGO, International Federation of Gynecology and Obstetrics; HIR, high intermediate-risk; HR, high-risk; IR, intermediate-risk; KPS, Karnofsky performance status; LN, lymph node; LR, low-risk; LUSI, lower uterine segment involvement; LVSI, lymphovascular space invasion; PWC, peritoneal washing cytology; SRI, surgery-radiotherapy interval; VBT, vaginal brachytherapy.

2. Follow-up and survival outcomes

In the entire cohort, the median follow-up period was 52.0 (range, 3–244) months. The follow-up rate was 87.4%, 134 patients were lost to follow-up. Seventy-two (6.8%) patients had a relapse, while 41 (56.9%) had only DM, 19 (26.4%) had only locoregional recurrence, and 12 (16.7%) had both. Sixty patients (5.6%) died. Of those, 23 (38.3%) died due to EC, 13 (21.7%) died of cardiovascular and cerebrovascular diseases, 6 (10.0%) died of second primary cancer, 6 (10.0%) died of non-malignancies and 12 (20.0%) died of unknown causes. The 3-year OS, DFS, LRFS, and DMFS rates were 98.4%, 94.5%, 97.1%, and 95.3%, respectively. The 5-year OS, DFS, LRFS, and DMFS rates were 95.7%, 88.9%, 93.8%, and 91.0%. In the HIR+HR group, 38 (10.8%) patients had a relapse, while 22 (57.8%) had only DM, 8 (21.1%) had only locoregional recurrence, and 8 (21.1%) had both. Twenty patients died (5.7%) and 11 (55.0%) died due to EC. The 3-year OS, DFS, LRFS, and DMFS rates were 96.5%, 89.5%, 98.0%, and 91.5%. The 5-year OS, DFS, LRFS, and DMFS rates were 95.9%, 83.7%, 94.7%, and 87.1%, respectively.

3. SRI cut-off value and survival analyses for the entire cohort

The median SRI for the entire cohort was 46 (range: 14–150) days. The SRI was less than 30, 30–60, and >60 days in 12.1%, 67.1%, and 20.8% of the patients, respectively. RCS was used to evaluate the nonlinear relationship between SRI and LRFS, adjusted for age, histologic grade, myometrial invasion depth, stromal invasion, LVSI, and LUSI. Although the nonlinearity test for SRI and LRFS was not statistically significant (p=0.134), the RCS plot suggested an increased hazard ratio for SRI ≥42 days (Fig. 1). The optimal SRI cutoff, determined by the MSRS, was 42 days (Bootstrap reproducibility rate: 0.892). Patients with SRI <42 days (median follow-up: 77 months) versus ≥42 days (median follow-up: 42 months) exhibited significantly higher 5-year OS (96.0% vs. 95.1%, p=0.007; Fig. 2A), DFS (91.4% vs. 86.3%, p=0.007; Fig. 2B), LRFS (94.6% vs. 93.1%, p=0.003; Fig. 2C), and DMFS (92.6% vs. 89.1%, p=0.025; Fig. 2D). For DMFS curves showing crossover, landmark analysis (cutoff: 100 months) revealed that shorter SRI was associated with lower risk within 100 months post-surgery (p=0.040), but this significance diminished beyond 100 months (p=0.344) (Fig. S1).

Fig. 1. Nonlinear relationship of SRI with LRFS.

LRFS, local recurrence-free survival; SRI, surgery-radiotherapy interval.

Fig. 2. A comparison of survival curves among the SRI of <42 days and ≥42 days in the entire cohort. (A) OS, (B) DFS, (C) LRFS, and (D) DMFS.

DFS, disease-free survival; DMFS, distant metastasis-free survival; LRFS, local recurrence-free survival; OS, overall survival; SRI, surgery-radiotherapy interval.

The results of univariate analysis are summarized in Table S2. Multivariate analysis (Table 2) identified SRI (p=0.011), advanced age (p=0.005), G3 tumor (p=0.028), LUSI (p=0.030) and LN evaluation (p=0.002) as independent prognostic factors for OS. For CSS, non-endometrioid histology (p=0.024), G3 tumor (p=0.005), LUSI (p=0.001) and LN evaluation (p=0.044) were independent prognostic factors. Independent predictors of DFS included SRI (p=0.019), advanced age (p=0.007), G3 tumor (p=0.012) and LN evaluation (p=0.001). For LRFS, SRI (p=0.013), advanced age (p=0.002), G3 tumor (p=0.001), LVSI (p=0.041) and LN evaluation (p<0.001). For DMFS, SRI (p=0.050) and LN evaluation (p=0.004) were independent prognostic factors, but piecewise Cox regression (cutoff: 100 months) showed only LN evaluation retained significance (p=0.004), while SRI lost significance in both time segments (<100 months: p=0.087; ≥100 months: p=0.271).

Table 2. Multivariate analysis of risk factors for survival outcomes in the entire cohort.

Variables		OS		CSS		DFS		LRFS		DMFS
		Hazard ratio (95% CI)	p-value	Hazard ratio (95% CI)	p-value	Hazard ratio (95% CI)	p-value	Hazard ratio (95% CI)	p-value	Hazard ratio (95% CI)	p-value
SRI (<42 vs. ≥42 days)		2.09 (1.18–3.68)	0.011	2.25 (0.9–5.63)	0.082	1.65 (1.08–2.5)	0.019	1.88 (1.14–3.11)	0.013	1.56 (1–2.44)	0.050
Age (<60 vs. ≥60 years)		2.12 (1.25–3.6)	0.005	2.00 (0.81–4.93)	0.132	1.7 (1.15–2.51)	0.007	2.1 (1.33–3.33)	0.002	1.46 (0.97–2.21)	0.072
Histological grade
	G1+G2 vs. G3	2.07 (1.08–3.97)	0.028	4.08 (1.53–10.88)	0.005	1.85 (1.14–3.01)	0.012	2.58 (1.5–4.44)	0.001	1.6 (0.94–2.73)	0.086
	G1+G2 vs. Gx	2.25 (0.68–7.42)	0.183	5.86 (1.26–27.25)	0.024	1.76 (0.7–4.39)	0.228	2.24 (0.81–6.22)	0.121	2.00 (0.8–5.03)	0.140
DMI (<1/2 vs. ≥1/2)		-	-	1.98 (0.79–4.95)	0.145	1.35 (0.91–1.99)	0.138	-	-	1.29 (0.85–1.97)	0.228
LUSI (No vs. Yes)		1.85 (1.06–3.23)	0.030	4.1 (1.72–9.78)	0.001	-	-	-	-	-	-
PWC (Negative vs. Positive)		-	-	-	-	2.27 (0.97–5.33)	0.060	2.98 (1.05–8.45)	0.041	2.2 (0.87–5.59)	0.097
FIGO stage (I vs. II)		1.75 (0.61–5.02)	0.297	-	-	1.69 (0.81–3.52)	0.165	-	-	1.98 (0.94–4.16)	0.071
LN evaluation (No vs. Yes)		0.43 (0.25–0.73)	0.002	0.41 (0.17–0.98)	0.044	0.53 (0.36–0.77)	0.001	0.42 (0.26–0.66)	<0.001	0.55 (0.37–0.83)	0.004

CI, confidence interval; CSS, cancer-specific survival; DFS, disease-free survival; DMI, depth of myometrial invasion; DMFS, distant metastasis-free survival; FIGO, International Federation of Gynecology and Obstetrics; LN, lymph node; LRFS, local recurrence-free survival; LUSI, lower uterine segment involvement; LVSI, lymphovascular space invasion; OS, overall survival; PWC, peritoneal washing cytology; SRI, surgery-radiotherapy interval.

4. SRI cut-off value and survival analyses in subgroups

In the HIR+HR group, the median SRI was 47 (range: 16–137) days. The SRI was less than 30, 30–60, and >60 days in 9.4%, 67.7%, and 22.9% of the patients, respectively. RCS analysis for SRI and LRFS showed no significant nonlinearity (p=0.366). The optimal SRI cutoff was 33 days (Bootstrap reproducibility rate: 0.713). The 5-year OS and LRFS for the 2 groups (SRI <33 days vs. SRI ≥33 days) were 97.7% versus 95.5% (p=0.014; Fig. 3A) and 95.0% versus 91.8% (p=0.016; Fig. 3C), but no significant differences in DFS (p=0.081; Fig. 3B) or DMFS (p=0.179; Fig. 3D).

Fig. 3. A comparison of survival curves among the SRI of <33 days and ≥33 days in the high intermediate-risk and high-risk group. (A) OS, (B) DFS, (C) LRFS, and (D) DMFS.

DFS, disease-free survival; DMFS, distant metastasis-free survival; LRFS, local recurrence-free survival; OS, overall survival; SRI, surgery-radiotherapy interval.

Univariate analysis showed in Table S3. Multivariate analysis (Table 3) identified G3 as an independent prognostic factor for OS (p=0.035), while SRI (p=0.055), age (p=0.067), and PWC (p=0.073) showed marginal significance. For LRFS, SRI (p=0.033), age (p=0.029), and G3 (p=0.006) were independent prognostic factors. In the HIR+HR subgroup, EBRT did not improve pelvic LN control (p=0.506; Fig. S2), and recurrence patterns did not differ significantly between EBRT±VBT and VBT groups (p=0.463; Fig. S3), with DM remaining predominant (83.3% vs. 65.2%). Pelvic LN recurrence rates were 11.1% and 17.4%, respectively.

Table 3. Multivariate analysis of risk factors for survival outcomes in the high intermediate-risk and high-risk group.

Variables		OS		DFS		LRFS		DMFS
		Hazard ratio (95% CI)	p-value	Hazard ratio (95% CI)	p-value	Hazard ratio (95% CI)	p-value	Hazard ratio (95% CI)	p-value
SRI (<32 vs. ≥32 days)		7.56 (0.96–59.52)	0.055	2.12 (0.81–5.57)	0.126	5.00 (1.14–21.87)	0.033	1.91 (0.71–5.16)	0.199
Age (<60 vs. ≥60 years)		2.39 (0.94–6.06)	0.067	1.86 (1.02–3.4)	0.043	2.37 (1.09–5.13)	0.029	1.83 (0.95–3.53)	0.069
Histological grade
	G1+G2 vs. G3	3.19 (1.08–9.41)	0.035	-	-	3.46 (1.43–8.4)	0.006	1.04 (0.48–2.23)	0.928
	G1+G2 vs. Gx	3.3 (0.77–14.09)	0.107	-	-	3.19 (0.92–11.04)	0.067	1.48 (0.5–4.39)	0.480
DMI (<1/2 vs. ≥1/2)		-	-	1.52 (0.84–2.75)	0.165	-	-	1.86 (0.96–3.58)	0.066
LVSI (No vs. Yes)		-	-	-	-	-	-	0.55 (0.26–1.18)	0.124
PWC (Negative vs. Positive)		4.07 (0.88–18.8)	0.073	-	-	-	-	-	-
LN evaluation (No vs. Yes)		0.57 (0.22–1.44)	0.232	0.67 (0.36–1.23)	0.196	0.57 (0.26–1.23)	0.152	-	-

CI, confidence interval; DFS, disease-free survival; DMFS, distant metastasis-free survival; DMI, depth of myometrial invasion; LN, lymph node; LRFS, local recurrence-free survival; LVSI, lymphovascular space invasion; OS, overall survival; PWC, peritoneal washing cytology; SRI, surgery-radiotherapy interval.

DISCUSSION

We used a large institutional dataset to explore the relationship between the SRI and clinical outcomes in early-stage EC. Using RCS modelling and the MSRS method, we identified the optimal threshold for SRI and demonstrated its prognostic significance. In the overall cohort, an SRI ≥42 days was associated with worse OS, DFS, LRFS, and DMFS. However, piecewise COX multivariate analysis revealed that SRI lost its significant effect on DMFS, while LN evaluation remained the only independent prognostic factor for DMFS. As a critical adjuvant treatment modality, RT has been identified to significantly reduce the risk of local recurrence [12]. Delayed initiation of RT may compromise its advantages in local control (LC). In EC, DM typically arises secondary to local recurrence, with patients experiencing local recurrence exhibiting a 4-fold higher risk of DM compared to those maintaining LC [13]. So, SRI may indirectly influence DM through its effect of LC. Our previous study found that the OS was worse in patients with relapse than in the patients without relapse [14], providing a theoretical foundation for understanding SRI’s impact on survival endpoints. For HIR and HR patients, initiating postoperative RT within 33 days was recommended. Although SRI emerged as an independent prognostic factor for LRFS, its effect on DM was limited. The observed marginal significance in OS suggested a survival benefit trend mediated by enhanced LC.

The optimal timing for adjuvant RT has been of great interest in recent years and has been studied in various cancers, including breast cancer [15,16,17], cervical cancer [18], and soft tissue sarcomas [19]. Most findings suggest that longer intervals between postoperative RT and surgery correlate with worse LC and poorer survival. However, it remains unclear in patients with EC due to inconsistent findings from previous retrospective studies and the ethical limitations of prospective studies. Ahmad et al. [3] found that while SIR as a continuous variable showed no association with LC or DSS. However, multivariate analysis revealed that SRI >6 weeks had a marginally significant impact on LC (p=0.060) and served as an independent prognostic factor for DSS (p<0.005). Suphasynth et al. [9] reported that SRI ≥6 weeks was associated with significantly lower RFS (81.2% vs. 94.4%, p=0.008) and OS (83.2% vs. 95.5%, p=0.012). In multivariate analysis, SRI ≥6 weeks emerged as the only independent risk factor for OS (p=0.020). Ghanem et al. [7] demonstrated that SRI >8 weeks correlated with worse 5-year OS (p=0.048). Independent predictors of reduced OS included older age, African-American ethnicity, higher comorbidity burden, advanced tumor grade, lymphovascular invasion, stage II disease, and number of resected LNs. Although SRI was a significant predictor of OS in univariate analysis, it lost independent significance in multivariate analysis (p=0.280). Zhu et al. [8] observed significantly reduced disease recurrence in patients with SRI ≤8 weeks (9% vs 18%; p=0.010), particularly isolated vaginal recurrence (0% vs 6%, p=0.040). The 5-year RFS differed significantly between SRI ≤8 and >8 weeks groups (89% vs. 80%, p=0.040), while no differences were found in 5-year DSS (p=0.990) or 5-year OS (p=0.880). Similarly, Fabrini et al. [4] using a 9-week cutoff, reported significantly higher local recurrence rates in patients with SRI >9 weeks (8.1% vs. 0%, p=0.046), with SRI remaining an independent prognostic factor for local recurrence in multivariate analysis (p=0.046). Cattaneo et al. [5] found that SRI ≥9 weeks was associated with higher recurrence rates (49% vs. 8.9%, p<0.001) and worse 5-year DSS (66% vs. 93%, p<0.001). Multivariate analysis confirmed SRI ≥9 weeks as an independent prognostic factor for both RFS (p=0.001) and DSS (p=0.001). American Brachytherapy Society recommends that adjuvant VBT is usually not performed until at least 4 weeks after surgery [20]. It is important to confirm that the vaginal cuff is completely healed before starting RT. The increasing use of robotic or laparoscopically assisted vaginal hysterectomies may increase the SRI [20]. National Comprehensive Cancer Network (NCCN) Guideline (version 1.2024) recommends that initiate RT as soon as the vaginal cuff is healed, preferably no later than 12 weeks after surgery [21]. According to our studies, it seems reasonable to follow the treatment guideline recommendations for total early-stage EC patients, but shorter SRI can bring better survival outcomes to the HIR and HR patients.

Another finding in our study was that LN evaluation was an independent prognostic factor for OS, CSS, DFS, LRFS and DMFS. In piecewise multivariate analyses, it consistently remained an independent predictor for DMFS. The impact of LN evaluation methods on patient survival remains controversial. Ahmad et al. [3] found an association between surgical LN evaluation and higher LC rate, whereas two prospective studies showed that lymphadenectomy failed to improve survival outcomes while increasing adverse events [22,23]. These discrepant findings may stem from population heterogeneity (the prospective studies included more early-stage LR patients), inconsistent LN dissection quality, and imbalances in adjuvant therapies. LN evaluation can improve staging accuracy and guide subsequent adjuvant treatment decisions. Imaging cannot completely identify the occult LN metastasis, pelvic LN metastases occur in about 10% of patients with clinical stage I EC [24]. Insufficient adjuvant therapy intensity in such cases may adversely affect prognosis. Therefore, based on our real-world study and surgical advancements like sentinel LN biopsy, further research is needed to better define the clinical value of LN evaluation.

The major strength of this study was the use of subgroup analyses. The previous results were based on the data of the whole group of patients with early-stage EC, but no study analyzed the effect of SRI on survival outcomes in different risk groups. However, there are great differences among patients with different risk factors [25]. A consistent relative treatment effect may be observed across all patient types, but certain HR subgroups may have greater absolute benefits or should be treated with more positive treatment attitudes [26]. Patients with more risk factors may benefit more from initiating RT earlier. Our study found that SRI within 33 days was recommended for HIR and HR group. The study had the largest institutional sample size to determine the optimal RT timing for early-stage EC. All patients received postoperative RT at our institution using similar treatment principles and techniques to ensure minimum errors. Our study excluded patients who received adjuvant chemotherapy, which makes the results more reliable for patients who receive RT only. Notably, we used RCS model to evaluate the relationship between SRI and LRFS. Although the nonlinearity test did not reach statistical significance, the RCS curve suggested a clinically meaningful threshold effect. Additionally, the MSRS was used to determine the optimal SRI cutoff for the entire cohort, the result was consistent with the RCS-derived threshold. This method provides robust statistical support for SRI threshold analysis.

However, this study has some limitations. First is its retrospective design, which can be inherently biased. Selection bias and differential loss to follow-up are among the obvious limitations of this study. Multivariate Cox regression analysis was used to analyze the factors affecting patient outcomes. But this statistical method requires the data to be unbiased and balanced, which weakens the reliability of the results for retrospective studies. We attempted to use propensity score matching and prognostic model but limited by the small sample size after matching and the small number of positive events. Due to database limitations, comorbidity indices were unavailable; however, we assessed KPS, with 99% of patients scoring ≥80, justifying its exclusion from further analysis. Future database iterations will incorporate comprehensive comorbidity assessments. Another concern was the significant difference in median follow-up between groups after applying the optimal SRI cutoff. Because 48.7% of patients with SRI ≥42 days underwent RT within 5 years, compared to only 9.4% in the SRI <42-day group, potentially overestimating survival in the SRI ≥42 days group. Extended follow-up is needed to obtain more stable long-term outcomes.

Conventional staging modalities have limitations in accurate prognostic predictions. With continuous development in precision medicine, EC was classified into four types by Proactive Molecular Risk Classifier for Endometrial Cancer (ProMisE) based on somatic mutation burden and somatic copy number alternations. Retrospective analysis indicated that these four molecular groups responded differently to therapy and were associated with different clinical prognoses [27,28,29], but SRI in different molecular subtypes is still an unexplored area. Our cohort lacks molecular typing results, we will collect more and complete information to define the roles of SRI in each molecular subgroup clearly.

In conclusion, in patients with early-stage EC who did not receive chemotherapy, delaying adjuvant RT after hysterectomy resulted in worse survival outcomes. Based on our findings, we recommend that all patients receive RT as soon as the vaginal cuff is healed. And initiating RT within 33 days can bring better survival outcomes to the HIR and HR patients.

SUPPLEMENTARY MATERIALS

Table S1

Risk stratification defined by the European Society for Medical Oncology EC Clinical Practice Guidelines (2016)

Table S2

Univariate analysis of risk factors for survival outcomes in the entire cohort

Table S3

Univariate analysis of risk factors for survival outcomes in the in the high intermediate-risk and high-risk group

Fig. S1

A comparison of DMFS curves among the SRI of <42 days and ≥42 days in the entire cohort with landmark analysis.

Fig. S2

A comparison of RRFS curves among the EBRT±VBT and VBT group.

Fig. S3

Recurrence patterns of EBRT±VBT and VBT group.