Esophageal cancer ranks seventh in incidence rate and is the sixth most fatal cancer globally.¹ The incidence of squamous cell carcinoma varies geographically, with Asia accounting for 90% of esophageal cancer cases.² With the advancements in endoscopic equipment and techniques, endoscopic submucosal dissection (ESD) has emerged as the mainstay treatment modality for superficial esophageal squamous cell neoplasia (ESCN), facilitating an accurate histopathological assessment, low local recurrence rate, and high en bloc resection rate.³ However, ESD is technically challenging and carries a risk of complications, especially for patients with widespread lesions (length ≥3 cm and extending ≥3/4 of the esophageal circumference),⁴ which can lead to intractable stenosis and requires multiple sessions of endoscopic dilatation in 88% to 100% of cases.^5-8 Thus, widespread ESCN remains a clinically troublesome condition.

Endoscopic radiofrequency ablation (ERFA) is technically convenient and less likely to result in esophageal stenosis, thus offering advantages in treating widespread and circumferential lesions.⁹ Recently, several studies have demonstrated its safety and efficacy in the treatment of superficial ESCN,^9-12 with complete response (CR) rates ranging from 84% to 97% and rates of stenosis from 0.0% to 28.6%.^9,13,14 Wang et al.¹⁵ compared the safety and efficacy of ERFA and ESD in the treatment of large early ESCN (length ≥3 cm and extending ≥1/2 of the esophageal circumference) and revealed that these two modalities are equally effective in the short term. However, there are no studies on the long-term outcomes of ESD and ERFA for ESCN. Therefore, this study aimed to describe and compare the long-term outcomes regarding ESD and ERFA for widespread superficial ESCN.

1. Patients

This retrospective cohort study enrolled consecutive patients from January 2015 and December 2021 in a tertiary referral hospital in China (Changhai Hospital, Department of Gastroenterology, Ward III). Following were the inclusion criteria: (1) at least one Lugol-unstained lesion 1/2 of the esophageal circumference and extended ≥3 cm; (2) preoperative diagnosis of squamous high-grade intraepithelial neoplasia (HGIN) or intramucosal squamous cell carcinoma; and (3) flat-type lesions (Paris type 0-IIb). Following were the exclusion criteria: (1) endoscopic ultrasound or computed tomography indicating submucosal invasion or lymph node or distal metastasis; (2) previous endoscopic treatment, radiation or surgery of the esophagus; (3) esophageal stenosis preventing endoscope passage; (4) uncontrolled coagulopathy (platelet count <75,000/μL or international normalized ratio >2); (5) initial treatment with surgical or chemoradiotherapy; and (6) unable to complete follow-up or determine whether death and time of death. Fig. 1 illustrates the patient enrollment with a flowchart. For each eligible patient, ESD and ERFA were offered as treatment options, and the benefits and potential risks of these two modalities were explained. The treatment was ultimately chosen by the patients based on necessary suggestions from clinicians. The study was conducted based on the Declaration of Helsinki, and the protocol was approved by the Ethics Review Board at Changhai Hospital, Naval Medical University (IRB number: CHEC2015-086). Prior to any procedure, written informed consent was obtained from all patients.

Figure 1. Flowchart showing the treatment process and outcomes of the study groups. CRT, chemoradiotherapy; ESD, endoscopic submucosal dissection; ERFA, endoscopic radiofrequency ablation; CR, complete response.

2. Performance of ESD

Endotracheal intubation and general anesthesia were used for all ESD procedures. The patients were placed in the left lateral decubitus position. The ESD procedure details and equipment were described in a previous report from our center.¹⁶

3. Performance of ERFA

The ERFA was delivered via the HALO360 (n=13) or HALO90 (n=20) ablation catheter. Before ERFA, lesions were located and measured with Lugol's solution (1.5%). Ablative energy (10 or 12 J/cm²) was delivered through the balloon catheter. Ablation catheter was initially performed on the treatment area. Thereafter, the ablation catheter was removed to allow cleaning of the electrodes before a second ablation procedure. The mucosal coagulum was removed using an endoscopic cap between ablations.

4. Postoperative management and surveillance

The patients were hospitalized following ESD or ERFA and were monitored for hematemesis, dyspnea, chest pain, or signs of infection. After nil per os of 24 hours, patients were provided with a full liquid diet in the absence of chest pain, dyspnea, or other symptoms. Discharge occurred once patients demonstrated tolerance to a regular diet.

After ESD, resected specimens were evaluated for curability. The term “en bloc resection” refers to the removal of a tumor or tissue in one complete piece, as opposed to a piecemeal resection performed at multiple segments. R0 resection was characterized by both horizontal and vertical margins being completely free from cancerous tissues. Curative resection was defined as no positive margins or submucosal or lymphovascular invasion in specimen pathology.^17-19 Additional esophagectomy, radiotherapy, and/or chemotherapy were suggested for non-curative resection; however, the final decision depended on the patient’s physical condition, life expectancy, or patient’s discretion. The therapeutic response to ERFA was evaluated 3 months after the procedure using Lugol chromoendoscopy and biopsies. Residual lesions were treated with repeated ERFA.

All patients, including those who received additional treatments due to non-curative initial treatments, were followed up at 6 and 12 months, and then annually until death or December 31, 2023. Endoscopy with Lugol’s staining, targeted biopsies, and chest computed tomography were performed during each surveillance. During follow-up, patients with local recurrence underwent endoscopic treatment or esophagectomy based on the disease stages. Patients with lymph node metastasis received radiotherapy and/or chemotherapy.

Oral steroids, which have been recommended for the prevention of stenosis when involving ≥3/4 of the circumference after ESD (if no contraindication), were administered at a dosage and course in accordance with clinical guidelines.²⁰ In our study, 15 patients were administered oral steroids for postoperative stenosis prevention. None of the patients in the ERFA group utilized steroid therapy as a preventive measure against stenosis.

5. Outcome measures

The primary outcome measure was recurrence-free survival (RFS). The secondary outcomes included overall survival (OS), disease-specific survival (DSS), CR at 12 months, and complications. Procedure characteristics were also documented and compared.

RFS was defined as the time from ESD or ERFA until the first metastasis or recurrence. Recurrence referred to the presence of pathology-confirmed HGIN or carcinoma in the treatment area during the follow-up beginning at 6 months. Metastasis included lymph nodes and distal metastasis. OS was calculated from the date of ESD or ERFA until death, irrespective of the cause, while DSS was measured from the date of ESD or ERFA until death specifically attributed to esophageal cancer. CR was defined as the absence of HGIN or squamous cell carcinoma in any biopsy specimen obtained from the treatment area. Complications included bleeding events, perforations, stenoses and other related adverse effects. Procedure-related bleeding events were characterized as instances requiring hemostatic interventions, such as thermocoagulation or endoscopic clipping. Esophageal stenosis refers to postoperative benign stenosis and was defined as the inability to successfully pass through a standard 11 mm diameter endoscope. Stenosis was classified into five levels to the Atkinsons’ grade of dysphagia.²¹

6. Statistical analysis

Categorical variables were expressed as the number of cases and percentages and statistical comparisons were performed using either the chi-square test or Fisher exact test. The Shapiro-Wilk test was employed to assess the normality of continuous variables. Variables were reported as mean±standard deviation and compared using the Student t-test if followed a normal distribution, or as median with interquartile range (IQR) and compared using the Mann-Whitney U test if not normally distributed. Estimation and comparison of the OS, DSS, and RFS rates were conducted utilizing the Kaplan-Meier method and log-rank test. Univariate and multivariate Cox regression analyses were employed to identify predictors of RFS and to estimate the adjusted hazard ratios (HRs) of treatments. In the univariate analysis, variables with p-value <0.20 were included in the multivariate analysis. Statistical significance was set at 0.05 (two-sided). R software (version 4.1.3; R Foundation for Statistical Computing, Vienna, Austria) and SPSS (version 25.0; IBM Corp., Armonk, NY, USA) were utilized for date analysis.

1. Baseline characteristics

Of the 155 patients who fulfilled the inclusion criteria, 125 were ultimately enrolled. Among them, 92 patients with superficial ESCN underwent ESD, and 33 patients underwent ERFA (Fig. 1). The tumor characteristics and demographics are presented in Table 1. Baseline characteristics were evenly distributed between the two treatment groups.

Table 1 . Patient Demographics and Clinical Characteristics.

Variable	ESD (n=92)	ERFA (n=33)	p-value
Age, mean±SD, yr	68.4±8.5	66.9±10.4	0.427
Sex			0.824
Male	66 (71.7)	23 (69.7)
Female	26 (28.3)	10 (30.3)
Alcohol drinking	21 (22.8)	7 (21.2)	0.849
Cigarette smoking	20 (21.7)	9 (27.3)	0.518
Comorbidities
Hypertension	26 (28.3)	10 (30.3)	0.824
Diabetes	9 (9.8)	3 (9.1)	1.000
Cerebrovascular disease	3 (3.3)	1 (3.0)	1.000
Cardiovascular disease	2 (2.2)	3 (9.1)	0.222
Other primary malignancy	4 (4.3)	1 (3.0)	1.000
Family history	8 (8.7)	4 (12.1)	0.819
Lesion location*			0.170
Upper	14 (15.2)	7 (21.2)
Middle	51 (55.4)	12 (36.4)
Lower	27 (29.3)	14 (42.4)
Multiple lesions	14 (15.2)	9 (27.3)	0.125
Lesion length, median (IQR), cm	4.7 (3.9–6.0)	5.0 (4.0–9.0)	0.083
Preoperative diagnosis			0.839
HGIN	74 (80.4)	26 (78.8)
Intramucosal cancer	18 (19.6)	7 (21.2)

Data are presented as number (%) unless indicated otherwise..

ESD, endoscopic submucosal dissection; ERFA, endoscopic radiofrequency ablation; IQR, interquartile range; HGIN, high-grade intraepithelial neoplasia..

*Upper: proximal, more than 24 cm from the incisors; Lower: distal, more than 32 cm from the incisors; Middle: between the upper and lower regions..

2. Procedure characteristics

Procedural details are presented in Table 2. The procedure duration of ERFA was significantly shorter than that of ESD (median, 35.0 minutes vs 105.0 minutes, p<0.001). Compared with the ESD group, post-procedure hospital stay in the ERFA group was significantly shorter (median, 3.0 days vs 5.0 days, p<0.001). In the ESD group, the en bloc, R0 and curative resection rates were 100.0% (92/92), 90.2% (83/92), and 76.1% (70/92), respectively. In the ERFA group, the CR rate at 3 months was 78.8% (26/33).

Table 2 . Procedure Characteristics and Outcomes of ESD and ERFA.

Characteristic	ESD (n=92)	ERFA (n=33)	p-value
Procedure duration, median (IQR), min	105.0 (70.0–155.0)	35.0 (20.0–45.0)	<0.001
Post-procedure hospital stay, median (IQR)	5.0 (4.0–6.0)	3.0 (2.0–5.0)	<0.001
En-bloc resection	92 (100.0)	-	-
R0 resection	83 (90.2)	-	-
Curative resection rate	70 (76.1)	-	-
CR at 3 mo	-	26 (78.8)	-
CR at 12 mo	87 (94.6)	30 (90.9)	0.748

Data are presented as number (%) unless indicated otherwise..

ESD, endoscopic submucosal dissection; ERFA, endoscopic radiofrequency ablation; IQR, interquartile range; CR, complete response..

In the ESD group, additional treatments were recommended in 22 patients and were performed in nine (40.9%). Five patients underwent additional esophagectomy for submucosal invasion, and four patients received radiotherapy for positive margins. In the ERFA group, all seven patients with residuals at 3 months underwent repeated ERFA (Fig. 1).

3. Patient outcomes

The short- and long-term outcomes are presented in Table 2. At 12 months, the CR rates were comparable between the two groups (94.6% vs 90.9%, p=0.748). The overall median follow-up duration was 66 months (IQR, 41 to 86 months), and the median follow-up duration was 56 months (IQR, 41 to 82 months) in the ESD group and 81 months (IQR, 51 to 87 months) in the ERFA group. In the Kaplan-Meier analysis, RFS was significantly longer in the ESD group than in the ERFA group (p=0.004), while no significant differences were observed in OS (p=0.845) and DSS (p=0.494) (Fig. 2). The treatment methods for patients with local recurrence or metastasis are summarized in Supplementary Table 1. Most of recurrence cases (53.3%) were adequately managed with endoscopic therapy (5 ESD and 3 ERFA).

Figure 2. Kaplan-Meier survival curves for the ESD and ERFA groups. (A) Recurrence-free survival. (B) Overall survival. (C) Disease-specific survival. ESD, endoscopic submucosal dissection; ERFA, endoscopic radiofrequency ablation.

The univariate and multivariate analysis results for RFS are shown in Table 3. The crude HR of intramucosal cancer versus HGIN for recurrence was 6.45 (95% confidence interval, 2.29 to 18.17, p<0.001) in univariate analysis. In multivariate analysis, the HR of intramucosal cancer versus HGIN for recurrence was 5.55 (95% confidence interval, 1.83 to 16.81, p=0.002) after adjusting for patient age, lesion length, and treatment modalities. In subgroup analysis, RFS in the ERFA group was comparable to that in the ESD group in patients with a lesion involving ≥3/4 of the circumference (p=0.643) and length ≥7 cm (p=0.894) (Figs 3 and 4).

Figure 3. Kaplan-Meier survival curves for ESD and ERFA in different patient subgroups. (A) Recurrence-free survival for patients with a lesion involving ≥3/4 of the circumference. (B) Recurrence-free survival for patients with a lesion involving <3/4 of the circumference. ESD, endoscopic submucosal dissection; ERFA, endoscopic radiofrequency ablation.

Figure 4. Kaplan-Meier survival curves for ESD and ERFA in different patient subgroups. (A) Recurrence-free survival for patients with lesions ≥7 cm in length. (B) Recurrence-free survival for patients with lesions <7 cm in length. ESD, endoscopic submucosal dissection; ERFA, endoscopic radiofrequency ablation.

Table 3 . Univariate and Multivariate Cox Regression Analysis for Recurrence-Free Survival.

Characteristic	Univariate analysis			Multivariate analysis
	HR (95% CI)	p-value		HR (95% CI)	p-value
Treatment (ERFA vs ESD)	3.72 (1.32–10.52)	0.013		2.82 (0.95–8.41)	0.062
Age	1.07 (1.01–1.13)	0.038		1.04 (0.98–1.10)	0.233
Sex (male vs female)	0.83 (0.28–2.43)	0.736
Alcohol drinking (yes vs no)	0.23 (0.01–7.81)	0.224
Cigarette smoking (yes vs no)	0.57 (0.13–2.52)	0.454
Comorbidities (yes vs no)	0.67 (0.23–1.97)	0.472
Family history (yes vs no)	0.64 (0.08–4.89)	0.669
Lesion length (≥7 cm vs <7 cm)	2.09 (0.74–5.90)	0.162		1.93 (0.64–5.83)	0.247
Multiple lesions (yes vs no)	1.67 (0.53–5.26)	0.380
Involving ≥3/4 of the circumference (yes vs no)	1.28 (0.46–3.56)	0.631
Preoperative diagnosis (IMC vs HGIN)	6.45 (2.29–18.17)	<0.001		5.55 (1.83–16.81)	0.002

HR, hazard ratio; CI, confidence interval; ERFA, endoscopic radiofrequency ablation; ESD, endoscopic submucosal dissection; IMC, intramucosal cancer; HGIN, high-grade intraepithelial neoplasia..

4. Complications

Complications associated with ESD and ERFA are shown in Table 4. The ESD group showed a higher rate of perioperative complications than the ERFA group (7.6% vs 0.0%, p=0.234), including four perforations (4.3%) and three bleeding events (3.3%). All perforations were treated with antibiotics and endoclips. All bleeding events were managed endoscopically, none patient requiring transfusion.

Table 4 . Complications of ESD and ERFA Procedures.

	ESD (n=92)	ERFA (n=33)	p-value
Perioperative complications	7 (7.6)	0	0.234
Bleeding	3 (3.3)	0	0.565
Perforation	4 (4.3)	0	0.522
Perioperative mortality	0	0	-
Stenosis	35 (38.0)	5 (15.2)	0.016
Stenosis rate in lesions involving ≥3/4 of the circumference	22/31 (71.0)	5/15 (33.3)	0.015
Stenosis rate in lesion length ≥7 cm	8/18 (44.4)	4/14 (28.6)	0.358

Data are presented as number (%) or number/number (%)..

ESD, endoscopic submucosal dissection; ERFA, endoscopic radiofrequency ablation..

Stenosis was significantly more frequent in the ESD group than in the ERFA group (38.0% vs 15.2%, p=0.016). In patients with a lesion involving ≥3/4 of the circumference, the ESD group had a significantly frequent stenosis rate than the ERFA group (71.0% vs 33.3%, p=0.015) (Table 4). No significant differences in stenosis grade (median, 3.0 vs 3.0, p=0.843) were found between the ESD and ERFA groups. Stenoses were mainly treated with endoscopic balloon dilation (EBD) (Supplementary Table 2). In patients with a lesion involving ≥3/4 of the circumference, more sessions of EBD significantly were performed in the ESD group (median, 7.0 vs 4.0, p=0.016). After treatment, the stenosis grade decreased, while there were no significant differences between the two groups (median, 1.0 vs 1.0, p=0.843).

The management of widespread ESCN poses technical challenges in the context of ESD, which is accompanied by a higher risk of refractory stenosis and perioperative complications. Recent studies have shown that ERFA may offer advantages for widespread superficial esophageal lesions, with acceptable CR rates and a decreased risk of stenosis.^10-13 In the current study, we found that RFS was significantly longer in the ESD group than in the ERFA group (p=0.004), and ERFA (adjusted HR, 5.55; vs ESD) was an independent predictor of recurrence. However, no significant differences in OS (p=0.845) or DSS (p=0.494) were observed. Furthermore, a significantly lower rate of esophageal stenosis and fewer EBD sessions were observed in the ERFA group. To our knowledge, this is the first study directly comparing the long-term outcomes of ERFA and ESD for superficial ESCN.

The major concerns of ERFA are that the depth of treatment is limited to the mucosa and the entire lesion sample cannot be obtained for accurate pathological staging. In contrast, ESD enables en bloc resection, thereby minimizing residual lesions and enabling accurate postoperative pathological staging and suggestion of necessary additional treatments for non-curative patients. Despite the application of image-enhanced magnifying endoscopy, endoscopic ultrasound, and endoscopic biopsy, the accurate evaluation of the invasion depth of the ESCN before the procedure remains challenging. Wang et al.¹² reported 29.8% histological upstaging of ESCN in the final resected specimens compared with the pre-ESD biopsies. Furthermore, ESCN demonstrated frequent vertical colonization of the ducts and submucosal glands. Previously, submucosal ductal extension has been reported in 12% and 14% of patients with ESCN who were judged to be eligible for ERFA.^22,23 ESCN extending into ducts/submucosal glands was observed in 58% and 64% of ESD specimens.²⁴ ERFA specifically aims to ablate the mucosa and always results in incomplete ablation of submucosal glands, which may result in local or submucosal recurrence. Wang et al.¹² reported that glandular ductal involvement was found in 86% of resected specimens of local recurrence after ERFA treatment of ESCN. In the current study, submucosal invasion was found in 18.5% (17/92) of the ESD specimens, although all lesions were preoperatively judged to be intramucosal. This may contribute to the higher risk of recurrence in the ERFA group compared to the ESD group. Within a median follow-up time of 81 months, nine recurrences were identified among 33 patients in the ERFA group, indicating a higher frequency than previously reported outcomes from other studies.^9,12 This difference may be attributed to different duration of follow-up and judgment of recurrence rates. Yu et al.⁹ reported 11 in 78 recurrences or progressions during a 4-year follow-up. Wang et al.¹² reported six in 30 recurrences or progressions during a 40-month follow-up. Both studies reported recurrence rates in patients with successful ERFA who achieved CR. In the current study, recurrences were reported in all patients who underwent ERFA.

Despite the increased risk of recurrence in the ERFA group, the OS and DSS were similar between the two groups. Similar to previous studies, the majority of recurrences detected during scheduled follow-up could be managed with endoscopic therapy, and patient survival was not significantly affected. Moreover, the preoperative diagnosis of intramucosal cancer (adjusted HR, 5.55; vs HGIN) was identified as an independent predictor of recurrence, which was consistent with the results of Yu et al.,⁹ indicating that ERFA should be restricted to lesions within the epithelium. Therefore, we suggest that ERFA could be considered as a treatment choice for ESCN only in dedicated centers, with careful preoperative evaluation of invasion depth. After treatment, patients should be closely followed up with Lugol’s staining and chest imaging to detect and treat potential residuals and recurrences in time, so as not to affect patient survival.

Our study revealed several advantages of ERFA over ESD. One of the greatest concerns of widespread ESD is intractable postoperative stenosis, which occurs in 55% to 76% of patients even under different stenosis-prophylaxis measures.¹⁷ In the present study, we found a lower rate of stenosis in the ERFA group than in the ESD group (15.2% vs 38.0%, p=0.016), and fewer sessions of EBD (median, 4.0 vs 7.0, p=0.016) required for lesion involving ≥3/4 of the circumference. Other studies also demonstrated the esophageal stenosis rate ranging from 0% to 28.6% following ERFA. However, these could be relieved through a median range of 2.5 to 5.5 dilation sessions.^9,13,14 Compared to the ESD group, the procedure duration (median, 35.0 minutes vs 105.0 minutes, p<0.001) and incidence of perioperative complications (0.0% vs 7.6%, p=0.234) were lower in the ERFA group, suggesting that ERFA has a lower technical threshold and better perioperative safety. Thus, ERFA may offer advantages in treating widespread lesions, and for patients with comorbidities and of an older age.

It is important to acknowledge several limitations in our current study. Firstly, the study was retrospective and non-randomized, which may be affected by patient selection and recall biases. ERFA is a more recent modality option than ESD. Moreover, some details of the procedure or patient management may change, such as endoscopy systems, ESD instruments, and prophylactic oral steroids. Secondly, the total number of patients and events during follow-up in the ESD and ERFA groups was small; therefore, the statistical power for drawing robust conclusions is insufficient. Thirdly, although we performed multivariate Cox regression analysis to estimate the adjusted HRs, there were other potential confounders; thus, the results should be interpreted with caution. Finally, although we strictly followed the inclusion and exclusion criteria, the selection bias of operators in choosing treatment modalities was inevitable, which may have a decisive effect on the tumor prognosis.

In conclusion, ERFA was associated with a higher risk of recurrence for ESCN than ESD, while patient's survival was not affected. ERFA reduced the risk of esophageal stenosis compared to ESD. For ESCN limited to the mucosa, ERFA could be considered as a treatment choice for those who cannot tolerate other more invasive treatment methods or who are at an increased risk for esophageal stenosis, upon careful preoperative evaluation and close postoperative follow-up.

This research was supported by the Science and Technology Commission of Shanghai Municipality (21Y31900100).

No potential conflict of interest relevant to this article was reported.

Study concept and design: L.W.W., H.L., W.W., Q.Q.M. Data acquisition: X.T., L.X., Y.G. Data analysis and interpretation: C.T.Y., Y.B., J.F.X. Drafting of the manuscript: X.T., C.T.Y. Critical revision of the manuscript for important intellectual content: Y.G. Statistical analysis: Y.R.Z., J.F.X. Obtained funding: L.W.W. Administrative, technical, or material support; study supervision: L.W.W. Approval of final manuscript: all authors.

Supplementary materials can be accessed at https://doi.org/10.5009/gnl240308.