Clinical features of early esophageal neoplastic lesions at different stages and efficacy and prognosis after endoscopic submucosal dissection

Abstract

BACKGROUND

Esophageal cancer is a common malignancy, and endoscopic submucosal dissection (ESD) is the treatment of choice for early esophageal cancer.

AIM

To analyze clinicopathological features, ESD efficacy, and prognostic factors of differentiated esophageal neoplasms to optimize management strategies.

METHODS

A total of 264 Lesions in 245 patients treated with ESD (2014-2022) were retrospectively analyzed. Lesions were categorized as low-grade intraepithelial neoplasia (LGIN), high-grade intraepithelial neoplasia, early-stage carcinoma, or superficial carcinoma. Clinicopathological data and outcomes were compared.

RESULTS

Early-stage cancers showed more surface vesicles than LGIN (P = 0.002). Intraepithelial papillary capillary loop (IPCL)-A predominated in LGIN, while IPCL-B2 was frequent in cancer (P < 0.001). Superficial carcinomas had higher vertical margin positivity (P < 0.001). Curative resection correlated with differentiation, body mass index (BMI), and symptoms; complications were linked to gender, BMI, and lesion size. The 3-, 5-, and 8-year disease-free survival rates were 96.8%, 91.7%, and 86.4%, respectively. Age [hazard ratio (HR) = 1.018] and prior esophageal cancer (HR = 3.050) predicted poorer survival.

CONCLUSION

Differentiated esophageal neoplasms exhibit distinct clinicopathological features. ESD provides durable efficacy, but high-risk patients (older age, prior cancer) require vigilant surveillance.

Key Words: Esophageal intraepithelial neoplasia; Early esophageal cancer; Superficial esophageal cancer; Endoscopic submucosal dissection; Long-term outcomes

Core Tip: This study examined 264 cases of esophageal mucosal lesions in 245 patients who underwent endoscopic submucosal dissection (ESD) treatment. The study found that esophageal tumors with different levels of differentiation exhibit differences in clinical and pathological characteristics. While ESD demonstrated sustained efficacy, elderly patients and those with a history of esophageal cancer had poorer prognoses and required close follow-up.

INTRODUCTION

Esophageal cancer is the eighth most common cancer and the sixth leading cause of cancer-related death worldwide[1]. China accounts for more than half of the global incidence and mortality of esophageal cancer, and esophageal squamous cell carcinoma (ESCC) is the most aggressive type and is particularly prevalent in Eastern Europe and Asia[2-4]. The lack of typical symptoms in the early stages results in a 5-year survival rate of only 10%-30%[1].

The pathological staging of esophageal cancer is based on the Vienna Classification and the Japanese Esophageal Cancer Classification Criteria[5]. Early-stage lesions include low-grade intraepithelial neoplasia (LGIN), high-grade intraepithelial neoplasia (HGIN), early-stage cancer (confined to the mucosal layer, with no lymph node or distant metastasis), and superficial cancer (confined to the mucosal layer and submucosal layer, unrelated to lymph node or distant metastasis). Advanced esophageal cancer is characterized by invasion extending into the muscularis propria or beyond.

In recent years, endoscopic submucosal dissection (ESD) has achieved remarkable efficacy as the treatment of choice for superficial infiltrative ESCC, with 5-year survival rates of up to 90%[6,7]. However, there is a paucity of long-term efficacy observations, and the factors affecting prognosis are still being explored. Additionally, the indications for ESD need to be improved. This study aimed to analyze the clinical features, endoscopic characteristics, long-term efficacy, and influencing factors of esophageal mucosal tumors (LGIN, HGIN, early-stage carcinoma, and superficial carcinoma) with different degrees of differentiation after ESD. The findings are expected to elucidate the pivotal factors influencing the pathogenesis and prognosis of esophageal cancer, thereby providing a foundation for the development of early diagnostic methods and personalized treatment strategies.

MATERIALS AND METHODS

Patient selection

This study retrospectively included patients with esophageal lesions who underwent ESD at our center between June 2014 and October 2022. Clinical data were obtained from the hospital's electronic medical record system.

Inclusion criteria: (1) Early-stage esophageal neoplasia histologically confirmed by endoscopy and biopsy; (2) No evidence of lymph node involvement or distant metastasis on imaging, meeting the established indications for ESD; and (3) Written informed consent obtained prior to the procedure.

Exclusion criteria: (1) Patients meeting ESD indications but not undergoing the procedure; (2) Incomplete follow-up data; (3) Prior anticancer treatment before ESD (such as radiotherapy and chemotherapy); or (4) Severe comorbidities such as cardiovascular, hematological, or neuropsychiatric disorders, or significant hepatic or renal dysfunction.

The study was approved by the Institutional Ethics Committee (Approval No. 2024 Ethics Review 032) and conducted in accordance with the Declaration of Helsinki.

ESD procedure and histopathological assessment

The lesion boundaries were identified by pigmented endoscopy and Lougol's iodine staining, and the lesion was marked by electrocoagulation 0.3 cm from the edge of the lesion and elevated by multiple submucosal injections of a mixture of indigo carmine, epinephrine, and saline. Submucosal dissection was then performed with a needle knife. ESD specimens were fixed in 10% formalin for 24 hours, sectioned at 2-mm intervals, and evaluated by an experienced pathologist. Pathological assessment included infiltration depth, lymphovascular and perineural invasion, histological subtype, differentiation grade, and vertical and horizontal margins (Figure 1).

Figure 1 Endoscopic submucosal dissection of an early esophageal squamous cell carcinoma. A: White light endoscopy to describe lesion; B: Virtual pigment description of the lesion; C: Pericyclic marker lesion; D: Submucosal dissection; E: Post-endoscopic resection of ulcer; F: Fixed specimen.

Outcome definitions and follow-up protocol

Short-term outcomes: (1) Complications. Acute bleeding: Intraprocedural hemorrhage requiring hemostasis. Delayed bleeding: Clinically evident bleeding (> 48 hours post-ESD) confirmed by endoscopy. Perforation was defined as a state in which the esophageal wall was punctured, resulting in communication of the esophageal lumen with the thoracic or abdominal cavity, or the development of persistent postoperative fever and severe chest pain, assisted by the detection of free gas on thoracic and abdominal computed tomography (CT) or the detection of an esophageal wall defect on gastrointestinal pan-glucosamine angiography. Stricture: Symptomatic dysphagia plus inability to pass a standard adult gastroscope; and (2) Resection quality. R0 resection: En bloc removal with histologically negative horizontal/vertical margins. Curative resection (CR): Complete endoscopic resection of the lesion with no risk of lymph node metastasis. Noncurative resection (NCR): Any violation of CR criteria.

Long-term outcomes: (1) Local recurrence was defined as the development of early esophageal cancer at the ESD site of the primary lesion, whereas metastatic recurrence was defined as recurrence involving other organs or lymph nodes; and (2) Survival metrics: Overall survival (OS); disease-free survival (DFS); recurrence-metastasis-free survival (RMFS); disease-specific survival (DSS).

Follow-up schedule: Standardized protocol: Months 3/6/12: Endoscopy + CT/magnetic resonance imaging + tumor markers. Annually thereafter (or biannually for high-risk: SM2/NCR). Additional treatment after ESD was based on the European Society of Gastroenterology Guidelines for Histopathological Diagnostic Endoscopy[6]. Additional treatment (surgery or radiotherapy) was given for lymphovascular infiltrating lesions and lesions invading the deeper layers of the submucosa (> SM2). After multidisciplinary discussion, additional treatment was provided on a patient-by-patient basis, taking into account the patient's age, comorbidities, and preferences.

Statistical analysis

Statistical analyses were performed using SPSS 27.0 and GraphPad Prism 10.0. Continuous variables were tested for normality via the Shapiro-Wilk test, presented as mean ± SD (normal) or median (interquartile range) (non-normal), and compared using ANOVA/t-test or Kruskal-Wallis/Mann-Whitney U test, respectively. Categorical variables were expressed as n (%), compared by χ² test, with Fisher's exact test used for expected frequencies < 5. Ordinal variables were analyzed via Kruskal-Wallis test with Bonferroni correction for pairwise comparisons. Multivariate logistic regression included variables with P < 0.05 from univariate analysis, using stepwise backward elimination (removal probability = 0.10) and Hosmer-Lemeshow test for goodness-of-fit. Survival outcomes were estimated by Kaplan-Meier method with log-rank test for group comparisons. Cox proportional hazards regression was used for multivariable survival analysis (Schoenfeld residuals to verify proportional hazards assumption), with hazard ratio (HR) and 95%CI reported. Sensitivity analysis validated the robustness of the results by excluding cases of superficial cancer. Statistical significance was set at two-tailed P < 0.05.

RESULTS

A total of 245 patients with 264 Lesions were examined. The mean age of the patients was 62.3 ± 7.8 years (range 44-86 years), of which 69.8% (171/245) were male.

Clinical and white light endoscopic characteristics

We analyzed and counted the epidemiological and clinical characteristics of early esophageal cancer with different degrees of differentiation, and there were no significant differences.

Endoscopic lesions were predominantly erythema and mucosal roughness. Surface vesiculation was significantly higher in patients with early carcinoma than in those with LGIN (33.8% vs 8.5%, P = 0.002). Submucosal vascular blurring or disappearance was significantly less in patients with LGIN than in those with HGIN and early carcinoma (18.6% vs 1.4%, 13.8% vs 1.4%, P = 0.002; Table 1).

Table 1 Clinical and endoscopic features of differentially differentiated esophageal mucosal tumors, n (%).

	LGIN (n = 71)	HGIN (n = 97)	Early cancer (n = 65)	Superficial cancer (n = 31)	P value
Gender					0.818
Male	53 (74.6)	71 (73.2)	44 (67.7)	22 (71.0)
Female	18 (25.4)	26 (26.8)	21 (32.3)	9 (29.0)
Age (years)	61.34 ± 6.94	62.8 ± 8.10	62.91 ± 7.66	61.84 ± 9.06	0.578
Comorbidities
Diabetes	8 (11.3)	10 (10.3)	6 (9.2)	1 (3.2)	0.676
Hypertension	20 (28.2)	24 (24.7)	20 (30.8)	6 (19.4)	0.642
Cardiovascular disease	10 (14.1)	14 (14.4)	7 (10.8)	3 (9.7)	0.873
Liver cirrhosis	8 (11.3)	3 (3.1)	5 (7.7)	1 (3.2)	0.167
Respiratory diseases	7 (9.9)	13 (13.4)	6 (9.2)	8 (25.8)	0.141
Gastroesophageal reflux disease	38 (31.7)	46 (38.3)	26 (40.0)	10 (32.3)	0.174
Barrett's esophagus	3 (4.2)	3 (3.1)	4 (6.2)	0 (0)	0.575
Atrophic gastritis	49 (69.0)	64 (66.0)	43 (66.2)	23 (74.2)	0.836
Hiatal hernia	5 (7.0)	2 (2.1)	2 (3.1)	0 (0)	0.315
Previous tumor history	7 (9.9)	12 (12.4)	8 (12.3)	4 (12.9)	0.943
Previous esophageal cancer history	1 (1.4)	3 (3.1)	0 (0)	2 (6.5)	0.195
Smoking history	28 (39.4)	48 (49.5)	31 (47.7)	19 (61.3)	0.225
Alcohol history	22 (31.0)	32 (33.0)	13 (20.0)	13 (41.9)	0.133
Family history	17 (23.9)	12 (12.4)	5 (7.7)	5 (16.1)	0.055
Accompanying symptoms
Dysphagia	9 (12.7)	13 (13.4)	9 (13.8)	8 (25.8)	0.392
Retrosternal pain	16 (22.5)	26 (26.8)	10 (15.4)	9 (29.0)	0.314
Weight loss	10 (14.1)	12 (12.4)	8 (12.3)	2 (6.5)	0.716
Other	40 (56.3)	51 (52.6)	41 (63.1)	13 (41.9)	0.248
White light endoscopic
Redness	48 (67.6)	72 (74.2)	48 (73.8)	21 (67.7)	0.735
Whitening	11 (15.5)	16 (16.5)	9 (13.8)	4 (12.9)	0.973
Roughness	52 (73.2)	82 (84.5)	47 (72.3)	25 (80.6)	0.194
Erosion	6 (8.5)c	16 (16.5)	22 (33.8)a	5 (16.1)	0.002
Nodule	2 (2.8)	7 (7.2)	10 (15.4)	5 (16.1)	0.024
White coating	7 (9.9)	13 (13.4)	11 (16.9)	3 (9.7)	0.640
Submucosal vascular blurring or loss	1 (1.4)b,c	18 (18.6)a	9 (13.8)a	3 (9.7)	0.002

^aP < 0.05 vs low-grade intraepithelial neoplasia.

^bP < 0.05 vs high-grade intraepithelial neoplasia.

^cP < 0.05 vs early cancer.

LGIN: Low-grade intraepithelial neoplasia; HGIN: High-grade intraepithelial neoplasia.

Magnifying endoscopy + narrow-band imaging characteristics

Intraepithelial papillary capillary loop (IPCL) staging is based on the AB staging criteria of the Japanese Society of Endoscopy[8]. Compared with patients with HGIN, early cancer, and superficial cancer, lesions in patients with LGIN were more likely to be IPCL-A type (9.3% vs 29.6%, 3.1% vs 29.6%, 0% vs 29.6%, P < 0.001), and lesions in early cancer and superficial cancer were more likely to be IPCL-B2 type (44.6% vs 4.2%, 45.2% vs 4.2%, P < 0.001; Table 2).

Table 2 Narrow-band spectral imaging performance of esophageal mucosal tumors of varying differentiation using magnifying endoscopy, n (%).

	LGIN (n = 64, 25.9%)	HGIN (n = 92, 37.3%)	Early cancer (n = 61, 24.6%)	Superficial cancer (n = 30, 12.2%)	P value
IPCL
A	21 (29.6)b,c,d	9 (9.3)a	2 (3.1)a	0 (0)a	< 0.001
B1	40 (56.3)	64 (66.0)	30 (46.2)	15 (48.4)	0.066
B2	3 (4.2)b,c,d	18 (18.6)a,c,d	29 (44.6)a,b	14 (45.2)a,b	< 0.001
B3	0 (0)	1 (1.0)	0 (0)	1 (3.2)	0.353

^aP < 0.05 vs low-grade intraepithelial neoplasia.

^bP < 0.05 vs high-grade intraepithelial neoplasia.

^cP < 0.05 vs early cancer.

^dP < 0.05 vs superficial cancer.

LGIN: Low-grade intraepithelial neoplasia; HGIN: High-grade intraepithelial neoplasia; IPCL: Intra-epithelial papillary capillary loop.

Pathological characteristics

Compared to patients with LGIN, patients with superficial carcinoma had a higher rate of positive horizontal (19.4% vs 0%, P = 0.001) and vertical (29.0% vs 0%, P < 0.001) margins. Patients with HGIN, early-stage carcinoma, and superficial carcinoma had longer longitudinal diameter of lesions (H = 39.511, P < 0.001) and wider circumferential extent (H = 25.813, P < 0.001). Early-stage and superficial carcinomas infiltrated deeper than LGIN (H = -118.338, P < 0.001; H = -160.645,P < 0.001), with fewer type 0-IIb lesions (64.5% vs 83.1%, 58.1% vs 83.1%, P < 0.001) and more combined lesions (10.8% vs 0%, 22.6% vs 0%, P < 0.001). Early stage and superficial carcinoma lesions were more deeply infiltrated than HGIN (H = -118.338, P < 0.001; H = -160.645, P < 0.001). Superficial carcinomas were more deeply infiltrated than early carcinomas (H = -42.307, P = 0.003) and had a higher rate of positive vertical margins than LGIN and HGIN (16.1% vs 0%, P < 0.001; Table 3).

Table 3 Pathological characteristics of esophageal mucosal tumors with different degrees of differentiation, n (%).

	LGIN (n = 71)	HGIN (n = 97)	Early cancer (n = 65)	Superficial cancer (n = 31)	P value
Lesion location					0.033
Upper 1/3	18 (25.4)b	9 (9.3)a	10 (15.4)	2 (6.5)
Middle 1/3	36 (50.7)	48 (49.5)	34 (52.3)	14 (45.2)
Lower 1/3	17 (23.9)	40 (41.2)	21 (32.3)	15 (48.4)
Lesion length (cm)					< 0.001
≤ 2	46 (64.8)b,c,d	25 (25.8)a	12 (18.5)a	9 (29.0)a
> 2 and < 5	22 (31.0)b,c	58 (59.8)a	37 (56.9)a	14 (45.2)
≥ 5	3 (4.2)c,d	14 (14.4)	16 (24.6)a	8 (25.8)a
Circumferential extent					< 0.001
< 1/2	59 (83.1)b,c,d	60 (61.9)a	31 (47.7)a	17 (54.8)a
≥ 1/2 and < 3/4	12 (16.9)	27 (27.8)	16 (24.6)	3 (9.7)
≥ 3/4	0 (0)b,c,d	10 (10.3)a,c,d	18 (27.7)a,b	11 (35.5)a,b
Invasion depth					< 0.001
M1	71 (100)c,d	97 (100)c,d	0 (0)a,b	0 (0)a,b
M2	0 (0)c	0 (0)c	36 (55.4)a,b,d	1 (3.2)c
M3	0 (0)c,d	0 (0)c,d	29 (44.6)a,b,d	5 (16.1)a,b,c
SM1	0 (0)d	0 (0)d	0 (0)d	23 (74.2)a,b,c
SM2	0 (0)	0 (0)	0 (0)	2 (6.5)
Vertical margin					< 0.001
Positive	0 (0)d	0 (0)d	4 (6.2)	5 (16.1)a,b
Negative	71 (100)d	97 (100) d	61 (93.8)	26 (83.9)a,b
Horizontal margin					0.001
Positive	0 (0)d	5 (5.2)	6 (9.2)	6 (19.4)a
Negative	71 (100)d	92 (94.8)	59 (90.8)	25 (80.6)a
Lymphatic invasion					0.364
Positive	0 (0)	0 (0)	1 (1.5)	0 (0)
Negative	71 (100)	97 (100)	64 (98.5)	31 (100)
Vascular invasion					< 0.001
Positive	0 (0)d	0 (0)d	0 (0)d	8 (25.8)a,b,c
Negative	71 (100)d	97 (100)d	65 (100)d	23 (74.2)a,b,c
Multiple lesions	17 (23.9)b	7 (7.2)a	8 (12.3)	6 (19.4)	0.015
Paris classification					0.001
0-I	3 (4.2)	0 (0)	1 (1.5)	0 (0)
0-IIa	4 (5.6)	11 (11.3)	10 (15.4 )	3 (9.7)
0-IIb	59 (83.1)c,d	74 (76.3)	42 (64.6)a	18 (58.1)a
0-IIc	5 (7.0)	9 (9.3)	6 (9.2)	3 (9.7)
Mixed type	0 (0)c,d	3 (3.1)d	7 (10.8)a	7 (22.6)a,b

^aP < 0.05 vs low-grade intraepithelial neoplasia.

^bP < 0.05 vs high-grade intraepithelial neoplasia.

^cP < 0.05 vs early cancer.

^dP < 0.05 vs superficial cancer.

0-I: Superficial and protruding; 0-IIa: Slightly elevated; 0-IIb: Flat; 0-IIc: Slightly depressed; 0-II: Superficial and excavated (type 0-III); Mixed type: In lesions with combined morphology (e.g., 0-IIa + 0-IIc); M1: Tumor limited to the mucosal epithelium; M2: Tumor limited to the lamina propria without involvement of the muscular mucosa; M3: Tumor involving the muscularis mucosa; SM1: Tumor infiltrated into the superficial submucosa (from the muscularis mucosa to the submucosa < 200 μm); SM2: Tumor involved the deep submucosa (≥ 200 μm in submucosa); LGIN: Low-grade intraepithelial neoplasia; HGIN: High-grade intraepithelial neoplasia.

ESD safety and efficacy

All patients with LGIN met the absolute indications for ESD, whereas all superficial cancers exceeded the absolute indications; superficial cancers were more likely to exceed the absolute indications for ESD compared to early-stage cancers (40.0% vs 0%, P < 0.001) The R0 rate of early-stage and superficial cancers was significantly lower compared to LGIN (86.2% vs 0%, 71.0% vs 0%, P < 0.001), and the NCR rate (15. 4% vs 0%, 45.2% vs 0%) and the incidence of postoperative stenosis (24.6% vs 4.2%, 35.5% vs 4.2%) were significantly higher (P < 0.001). Superficial carcinoma had a lower R0 rate (71.0% vs 94. 8%) and a higher rate of NCR than HGIN had (45.2% vs 5.2%, both P < 0.001). The proportion of intrinsic muscle layer injury was significantly higher in superficial carcinoma than in LGIN (4.2% vs 29.0%, P = 0.008).

Postoperative body mass index (BMI) was significantly lower in patients with superficial carcinoma than in LGIN (18.94 vs 24.22, H = 79.192), HGIN (18.94 vs 23.04, H = 51.243) and early-stage carcinoma (18.94 vs 23.53, H = 52.355) (all P < 0.05). Postoperative follow-up BMI was also lower in HGIN (H = 27.949) and early cancer (H = 26.838) than in LGIN (both P < 0.05). Patients with early-stage and superficial cancers had a higher incidence of long-term adverse outcomes (recurrence, metastasis, and death) (12.3% vs 2.8%, 19.4% vs 2.8%, P = 0.006) and an increased risk of recurrence (6.2% vs 0%, 12.9% vs 0%, P = 0.014; Table 4).

Table 4 Efficacy and safety of endoscopic mucosal dissection in esophageal tumors of varying differentiation, n (%).

	LGIN (n = 71, 26.9%)	HGIN (n = 97, 36.7%)	Early cancer (n = 65, 24.6%)	Superficial cancer (n = 31, 11.7%)	P value
Short-term outcomes
Absolute indication	71 (100)b,c,d	88 (90.7)a,c,d	26 (40.0)a,b,d	0 (0)a,b,c	< 0.001
R0 resection	71 (100)c,d	92 (94.8)d	56 (86.2)a	22 (71.0)a,b	< 0.001
CR	71 (100)c,d	92 (94.8)d	55 (84.6)a,d	16 (51.6)a,b,c	< 0.001
Muscle layer injury	3 (4.2)d	13 (13.4)	8 (12.3)	9 (29.0)a	0.008
Complications	4 (5.6)c,d	17 (17.5)	16 (24.6)a	12 (38.7)a	< 0.001
Bleeding	0 (0)	3 (3.1)	0 (0)	0 (0)	0.363
Perforation	1 (0)	0 (0)	0 (0)	1 (3.2)	0.198
Stricture	3 (4.2)c,d	14 (14.4)	16 (24.6)a	11 (35.5)a	< 0.001
Long-term outcomes
BMI (kg/m2)	24.22 (22.76-25.60)b,c,d	23.04 (20.95-24.95)a	23.53 (20.57-24.73)a	21.97 (18.94-23.00)a	< 0.001
Adverse outcomes	1 (2.8)c,d	6 (6.2)d	8 (12.3)a	6 (19.4)a,b	0.006
Recurrence	0 (0)c,d	3 (3.1)d	4 (6.2)a	4 (12.9)a,b	0.014
Metastasis	0 (0)	1 (1.0)	1 (1.5)	0 (0)	0.802
Mortality	1 (1.4)	2 (2.1)	5 (7.7)	2 (6.5)	0.133
Multiple primary cancers	11 (15.5)	13 (13.4)	11 (16.9)	5 (16.1)	0.933

^aP < 0.05 vs low-grade intraepithelial neoplasia.

^bP < 0.05 vs high-grade intraepithelial neoplasia.

^cP < 0.05 vs early cancer.

^dP < 0.05 vs superficial cancer.

LGIN: Low-grade intraepithelial neoplasia; HGIN: High-grade intraepithelial neoplasia; CR: Curative resection; BMI: Body mass index.

Analysis of factors affecting the effectiveness of diagnosis and treatment

Degree of differentiation [HGIN odds ratio (OR): 7.527, early-stage cancer OR: 2.545] was significantly associated with R0 resection. Degree of differentiation (early-stage cancer OR: 3.722, HGIN OR: 11.460), BMI (OR: 1.205), and preoperative symptoms (retrosternal pain OR: 0.340) were considered independent risk factors for CR (Table 5). Sex (female OR 2.768), BMI (OR 0.837), lesion length (≥ 5 cm: OR 4.411), > 3/4 of the annulus circumference (OR: 5.820) and R0 resection (OR 0.288) were independent risk factors for complications. Sex (female OR: 2.450), lesion length > 5 cm (OR: 5.833), > 3/4 of the esophageal circumference (OR: 3.472), and postoperative stenosis prophylaxis (OR: 4.161) were independent risk factors for esophageal stenosis after ESD (Table 6). Age (HR: 1.018) and prior esophageal cancer (HR: 3.050) were significantly associated with poor prognosis. Prior esophageal cancer (HR: 18.230) and depth of lesion invasion (SM2 HR: 30.213) were significantly associated with recurrent metastasis (Table 7 and Figure 2).

Figure 2 Kaplan-Meier curves show overall survival, disease-free survival, and disease-specific survival during follow-up. A: Overall survival; B: Disease-free survival (DFS); C: Disease-specific survival; D: Differentiation-based DFS; E: DFS based on depth of invasion; F: DFS based on curative resection; G: Recurrence-metastasis-free survival (RMFS) based on curative resection; H: RMFS based on noncurative resection combined with adjuvant therapy; I: DFS based on prior esophageal cancer; J: RMFS based on the prior esophageal cancer. LGIN: Low-grade intraepithelial neoplasia; HGIN: High-grade intraepithelial neoplasia.

Table 5 Multifactorial analysis of factors associated with curative vs complete resection.

Variable	Category	P value	Odds ratio	95%CI
CR
Differentiation degree	HGIN	< 0.001	11.460	3.404-38.582
	Early cancer	0.014	3.722	1.312-10.564
	Superficial cancer		1
Retrosternal pain	No		1
	Yes	0.036	0.340	0.124-0.931
BMI		0.032	1.205	1.016-1.430
R0 resection
Differentiation degree	HGIN	< 0.001	7.527	2.294-24.695
	Early cancer	0.080	2.545	0.893-7.255
	Superficial cancer		1

HGIN: High-grade intraepithelial neoplasia; CR: Curative resection; BMI: Body mass index.

Table 6 Multifactorial analysis of complications and factors associated with esophageal stricture.

Variable	Category	P value	Odds ratio	95%CI
Complications
Gender	Male		1
	Female	0.009a	2.320	0.838-6.423
BMI	-	0.008a	0.837	0.734-0.956
Lesion length (cm)	≤ 2		1
	≥ 5	0.021a	4.411	1.251-15.548
Circumferential extent	< 1/2		1
	≥ 3/4	< 0.001a	5.820	2.180-15.537
R0 resection	No		1
	Yes	0.023a	0.288	0.098-0.841
Esophageal stenosis
Gender	Male		1
	Female	0.032a	2.450	1.078-5.571
Lesion length (cm)	≤ 2		1
	≥ 5	0.026a	5.833	1.232-27.626
Circumferential extent	< 1/2		1
	≥ 3/4	0.033a	3.472	1.104-10.922
Stenosis prevention	No		1
	Yes	0.006a	4.161	1.490-11.615

^aP < 0.05.

BMI: Body mass index.

Table 7 Cox regression multifactorial analysis of factors associated with time to recurrence, metastasis, death, and recurrence metastasis.

Variable	Category	P value	Hazard ratio	95%CI
Recurrence, metastasis, death survival
Age	-	0.031a	1.018	1.002-1.034
Prior esophageal cancer	No		1
	Yes	0.029a	3.050	1.120-8.303
Recurrence metastasis survival
Prior esophageal cancer	No		1
	Yes	0.001a	18.230	3.064-108.461
Depth of invasion	M1		1
	SM2	0.003a	30.213	3.207-284.650

^aP < 0.05.

DISCUSSION

The pathogenesis of ESCC is influenced by a variety of factors, including genetic susceptibility and environmental stimuli[2,9,10]. The precursor lesion of ESCC is dysplasia of the esophageal epithelium, ranging from LGIN to HGIN, which may eventually progress to invasive carcinoma. It has been reported that about 23% of LGINs progress to HGINs or invasive carcinoma within 5 years, while about 70% of HGINs progress to invasive carcinoma within 5 years. However, about 37% of precancerous lesions regress spontaneously within 5 years, and not all precancerous lesions progress, which may be closely related to genetic and environmental factors[11].

This study found no significant differences in demographics, comorbidities, or symptoms among patients with ESCC of varying differentiation. However, atrophic gastritis was more common and is strongly associated with increased ESCC and gastric cancer risk[10]. Smoking and alcohol consumption significantly raised the incidence of superficial carcinoma, aligning with previous findings[12]. LGIN detection was higher in individuals with a family history of esophageal cancer; likely due to active screening. The most common symptom was nonspecific upper gastrointestinal discomfort. Therefore, early endoscopic examination with targeted biopsy is essential for patients with risk factors, such as atrophic gastritis, gastrointestinal symptoms, family history, smoking, and alcohol use, to enable early detection and reduce mortality.

With increasing lesion differentiation, the longitudinal and circumferential extent, invasion depth, muscular involvement, and vascular invasion rates also increased. Early-stage or superficial carcinoma presented with larger lesions, deeper infiltration, reduced Paris 0-IIb staging, predominantly IPCL-B2 patterns, and more frequent surface vesiculation than LGIN under white light endoscopy. Patients with superficial carcinoma had higher rates of atrophic gastritis and smoking, which both contributed to esophageal cancer progression. Additionally, these patients showed higher positive horizontal margins and vascular infiltration in ESD specimens and were more prone to myotomal injury during ESD procedures.

ESD is an efficient and minimally invasive resection technique with a high resection rate. En bloc resection rate, R0 resection rate, and CR rate were 100%, 91.3%, and 88.6%, respectively, The CR rate was significantly higher than in previous studies; likely due to accurate preoperative evaluation and advancing ESD techniques in China, further validating the method's effectiveness. There were 30 patients with NCR (11.4%). Fourteen of those patients received additional treatment, including esophagectomy in five and radical chemoradiotherapy in 10. One patient underwent esophagectomy followed by additional chemoradiotherapy. Of the patients who received additional chemoradiotherapy, five experienced recurrence. One patient achieved remission after repeated ESD, one achieved remission after continued radical chemoradiotherapy, and three achieved a cure after esophagectomy. The remaining 16 patients underwent endoscopic monitoring. Some did not meet the CR criteria in the Japanese Esophageal Guidelines but had pathological findings indicating R0 resection. Others were older, had severe comorbidities, or had other reasons. Of the patients who did not receive additional treatment, two had recurrence (one with neck lymph node metastasis who subsequently died, and one who was relieved after undergoing ESD again), and 14 showed no abnormalities during follow-up. There were no ESCC-related deaths post-CR surgery, and the RMFS was significantly higher than in NCR patients. Follow-up observations are ongoing for these patients. The possibility of recurrence in NCR patients despite additional radiotherapy and chemotherapy suggests that postoperative adjuvant radiotherapy and chemotherapy cannot completely prevent disease recurrence, which is consistent with previous studies[13]. Further large-scale studies are needed to explore whether adjuvant radiotherapy can reduce the recurrence rate in NCR patients after ESD. Previous studies have demonstrated that patients with NCR who received additional treatment achieved better survival outcomes than those who did not. In contrast, our study found no significant difference in DFS between CR and NCR groups (P = 0.42); consistent with Beaufort et al[14], who reported no DFS difference during follow-up. However, RMFS was significantly higher in the CR group, indicating that the observed DFS similarity may be due to non-esophageal-cancer-related deaths. According to current evidence, additional treatment is recommended for NCR cases following ESD, particularly in patients with lymphovascular invasion, submucosal invasion, or positive margins, as it is considered both safe and beneficial for long-term prognosis[15-17].

This study found that the degree of tumor differentiation, retrosternal pain, and BMI were independent risk factors for NCR. Compared with superficial cancer, the CR rate was significantly higher in LGIN, HGIN, and early-stage cancer. The incidence of NCR and non-R0 resection increased with increasing lesion differentiation, deeper invasion, and higher risk of lymphatic metastasis. In particular, the presence of retrosternal pain may indicate that the tumor has invaded deeper tissues or nerves, and the extent of the lesion may be beyond the expected resection range of ESD. In addition, patients with a high BMI may impair the operator's ability to accurately assess the lesion boundaries due to limited exposure of the surgical field, thus increasing the risk of NCR. However, the differences between the results of this and some previous studies may be related to the different pathological stratification of the study population and the types of confounding factors. Based on these findings, it is recommended that in patients with retrosternal pain, high preoperative pathological differentiation and obesity, the indications for ESD should be evaluated more cautiously, and the extent of resection should be carefully assessed to avoid overtreatment or underestimation of the extent of tumor infiltration.

ESD perioperative complications such as bleeding (1.1%) and perforation (0.8%) can be effectively controlled with medication and endoscopy, and no deaths occurred. Due to the low event numbers, risk factor analysis for these complications was limited in statistical power. Nevertheless, these findings align with previous studies suggesting that ESD, when performed by experienced endoscopists, is a safe procedure with minimal life-threatening risks. The overall incidence of postoperative esophageal stricture was 16.7%, but there were significant differences between different stages of disease: Superficial cancer (35.5%) > early cancer (24.6%) > HGIN (14.4%) > LGIN (4.2%) (Table 4), suggesting that the risk of stricture increases with the depth of invasion and extent of the lesion. Among 44 patients with esophageal stricture, 21 (47.7%) did not undergo intervention due to mild symptoms (e.g., occasional difficulty swallowing), while 23 (52.3%) received endoscopic balloon dilation therapy. Among these, two developed refractory esophageal stricture (defined as persistent dysphagia after 5 balloon dilations), requiring multiple dilations or even stent placement: 1 male patient underwent 10 dilations and received one stent, and one patient required repeated interventions after repair surgery due to perforation during dilation, ultimately receiving a stent placement. However, neither experienced severe adverse events. Multivariate analysis showed that female sex (OR = 2.450), lesion length ≥ 5 cm (OR = 5.833), and circumferential extent ≥ 3/4 (OR = 3.472) were independent risk factors, while postoperative stenosis prevention measures reduced the risk of stenosis by 4.161 times (Table 6). In this study, women had a higher risk of esophageal stricture than men, which may be related to estrogen-mediated fibrosis enhancement[18]. The above results are consistent with previous studies[19,20]. Furthermore, this study found that the depth of lesion invasion is an important influencing factor: The stenosis rate of early-stage and superficial cancer was significantly higher than that of LGIN, suggesting that stenosis prevention should be tailored to the stage of the lesion. Based on the data from this study and evidence from the literature, prevention of stenosis should be implemented on an individualized basis according to risk level, with the core objective of reducing wound inflammation and fibrosis. This study focused on preventing postoperative stenosis through oral steroids, local steroid injections (such as triamcinolone), and balloon dilation. More specifically: (1) Drug intervention. Local steroid injections are used for moderate-risk patients, such as those with HGIN or early-stage cancer. Within 1 week postoperatively, glucocorticoids (e.g., triamcinolone) are injected at multiple sites on the wound surface to target and inhibit inflammatory responses. The most common preventive method is oral steroids, which are administered at a daily dose of 30-40 mg with a gradual reduction of 5 mg/week. However, in high-risk populations, the efficacy of this method is limited, with a stenosis incidence rate as high as 35%-50%[21]. Studies have shown that local steroid injections have a stenosis incidence rate of only 4.5% in high-risk populations[22]. Since few patients in this study used monotherapy, effective comparisons could not be made. Oral steroids (such as prednisone) may reduce fibrosis through systemic anti-inflammatory effects, but they have gastrointestinal side effects. Based on the research of Zhou et al[23], for high-risk patients, steroid injection combined with short-term oral steroids (1-2 weeks after surgery) may be superior to a single measure, which can be verified in the future; and (2) Mechanical intervention. Prophylactic endoscopic balloon dilation (EBD) is indicated for high-risk patients (circumference ≥ 3/4, length ≥ 5 cm), with EBD initiated 1-2 week postoperatively. This study demonstrated that although 70% of high-risk patients still developed strictures, dilation reduced the severity of strictures and delayed progression, with only one case of perforation (which healed after repair), suggesting that EBD can reduce the severity of strictures but cannot completely prevent them. This is associated with the characteristic of esophageal mucosal defects exceeding 3/4 circumference in this study. Severe defects leading to fibrosis are difficult to completely suppress through mechanical dilation alone and require combined pharmacological intervention. Biodegradable stents maintain the luminal morphology through physical support, and the combination of steroids can reduce the circumferential lesion stenosis rate to < 30%, but stent migration remains a problem[24]. Regenerative medicine uses human cells or tissues to promote esophageal mucosal regeneration and iatrogenic wound healing to prevent postoperative esophageal stricture. Cell injection, cell sheet transplantation, and extracellular matrix have been proven to be effective[25]. This study did not include the last two techniques, and their application in high-risk patients could be explored in the future. For high-risk groups (women, long lesions, and large circumferential lesions), individualized plans should be developed in advance, combining multiple preventive measures such as steroid injection and EBD. Postoperative follow-up should be intensified (endoscopic examination at 1, 3, 6, and 12 months) to detect and intervene in narrowing at an early stage. This single-center, retrospective study had a small sample size (e.g., n = 31 for superficial cancer) and did not compare the efficacy of different preventive regimens. Future multicenter studies are needed to validate the efficacy of combination therapy, particularly for preventive strategies targeting SM2-stage invasive lesions. Although ESD for early-stage esophageal tumors is safe, esophageal stricture remains a major challenge. Implementing risk stratification based on lesion stage and combining multiple preventive measures can significantly improve patients' long-term quality of life.

The cumulative OS at 3, 5, and 8 years in the study cohort was 98.4%, 96.6%, and 92.3%, respectively. DFS was 96.8%, 91.7% and 86.4%, and RMFS was 98.4%, 95.0% and 92.1%. The 10-year local recurrence metastasis rate with DSS was 4.5% and 99.6%, respectively. Five-year DFS rates for CR and NCR patients were 92.0% and 88.9%, respectively. A total of 10 patients died: One from ESCC, five from second primary tumors (3 Lung, 1 each esophageal and gastric) and four from lung cancer. Significant differences in DFS and RMFS were observed between lesions of varying differentiation invasion depth, with superficial carcinoma (especially invasive to SM2) showing the worst prognosis. Recurrent metastases increased with higher lesion differentiation and depth of invasion, consistent with previous studies[15,26]. Additionally, the study found a higher incidence of secondary lung cancer in esophageal cancer patients, reinforcing the need for regular monitoring for second primary malignancies post-treatment[16,17].

In this study, the mean time to esophageal cancer recurrence was 41.0 ± 23.5 months (range: 8-85 months), with 81.8% of recurrences occurring within 5 years post-surgery. Recurrence rates significantly declined beyond 5 years[27], so a minimum of 5 years postoperative follow-up is recommended, as well as increased surveillance for 2-4 years. In Cox multivariate regression, each additional year of age was associated with a 1.8% increase in the risk of DFS (HR = 1.018), suggesting that advanced age may affect prognosis through immune aging or DNA repair capacity[28]. Patients with a history of esophageal cancer had a recurrence risk approximately three times higher than those without a history of the disease (HR = 3.05), and clinical follow-up strategies should be strengthened. To address potential bias from the small sample size of superficial carcinomas (n = 31), a sensitivity analysis excluding these cases was performed (Table 8). Notably, the HR for age (1.097 vs 1.018) and prior cancer history (10.429 vs 3.050) remained significant or even increased, confirming the robustness of these associations. These findings underscore the need for intensified surveillance in elderly patients and those with prior malignancies, regardless of lesion stage. To further eliminate the impact of esophageal cancer-specific mortality on the results, the analysis showed that the independent risk factors for RMFS were a history of esophageal tumors (HR = 18.230) and a lesion infiltration depth of SM2. Patients with a history of esophageal cancer have an 18.23-fold higher risk of recurrence and metastasis than those without a history of the disease (HR = 18.230), suggesting that residual micrometastases or chronic inflammation may drive recurrence[29]. Increased depth of invasion is significantly associated with a marked increase in lymph node metastasis rate, suggesting that SM2 Lesions may have a higher metastatic potential[30], with a significantly increased recurrence and metastasis rate. Two patients with stage SM2 invasive esophageal cancer developed choroidal invasion after ESD. One patient achieved long-term remission with adjuvant radiotherapy, while the other patient did not experience recurrence after esophagectomy. These cases suggest ESD combined with radiotherapy might be a viable option for SM2 Lesions, although larger studies are needed to confirm safety and efficacy. Consistent with previous studies[31], patients with NCR have a significantly increased risk of recurrence and metastasis. Such patients are recommended to undergo enhanced endoscopic monitoring 3-6 months after surgery and receive an individualized assessment of the need for adjuvant therapy (e.g., radiotherapy for patients not eligible for surgery). In addition, this study supports increased screening of people over 40 years of age or with a history of tumors, especially esophageal cancer, for early diagnosis and treatment to improve long-term survival and reduce the burden on society.

Table 8 Comparison of Cox regression results before and after excluding superficial cancer cases.

Variable	Model type	Hazard ratio (95%CI)	P value
Age	Full model	1.018 (1.002–1.034)	0.031a
	Sensitivity model	1.097 (1.032–1.165)	0.003a
Prior esophageal cancer	Full model	3.050 (1.120-8.303)	0.029a
	Sensitivity model	10.429 (1.227-88.634)	0.032a

^aP < 0.05.

Sensitivity model excludes superficial cancer cases (n = 31). Cox regression models were adjusted for age and prior cancer history. Results remained robust.

Our study revealed significant differences in the epidemiological and endoscopic characteristics of esophageal mucosal tumors with varying degrees of differentiation, highlighting the importance of accurate preoperative diagnosis in selecting appropriate ESD indications. ESD demonstrated high CR and R0 resection rates for early esophageal cancer, with excellent DSS and favorable long-term outcomes. Although bleeding and perforation risks are low, esophageal stricture remains a major challenge. Patients with NCRs had significantly lower DFS, necessitating further treatment strategies, as no clear consensus exists on whether surgery or radiotherapy is preferable. Additionally, esophageal cancer patients face a higher risk of death from other malignancies and recurrence within 5 years. Thus, close postoperative surveillance is essential for early detection of second primary tumors and recurrence.

This study had several limitations. First, as a single-center retrospective study, selection bias may be present despite strict inclusion criteria, limiting the generalization of the findings. Second, small subgroup sizes (e.g., superficial carcinoma, n = 31) may reduce statistical power, especially for rare events like bleeding and perforation. Finally, residual confounding cannot be fully excluded despite multivariate adjustment. Larger, multicenter prospective studies are needed to validate these results and improve risk stratification.

CONCLUSION

This study systematically analyzed the clinicopathological characteristics and long-term outcomes of early esophageal neoplastic lesions with different degrees of differentiation treated by ESD. ESD demonstrated favorable efficacy and safety, particularly in patients with LGIN and HGIN. Poor differentiation, retrosternal pain, and elevated BMI were identified as independent predictors of NCR, underscoring the importance of individualized preoperative assessment. Although adjuvant therapy may benefit patients with NCR, recurrence remains a concern, especially in those with SM2 invasion or a history of esophageal cancer. Post-ESD strictures were more common in female patients and those with long or ≥ 3/4 of the annulus circumference lesions. While prophylactic interventions such as steroid therapy and balloon dilation may reduce stricture severity, their efficacy requires further validation. Given the potential risk of recurrence and secondary malignancies, long-term and structured surveillance is essential. These findings support the refinement of risk-stratified treatment and follow-up strategies for early-stage esophageal cancer.