Risk factors and prognosis of postoperative early relapse after neoadjuvant therapy for locally advanced esophageal squamous cell carcinoma: a retrospective study

Abstract

Background

This study aimed to identify risk factors associated with early postoperative relapse and evaluate their impact on survival outcomes after neoadjuvant therapy for locally advanced esophageal squamous cell carcinoma (LA-ESCC), thereby informing strategies to optimize postoperative clinical management.

Methods

Patients with LA-ESCC who underwent neoadjuvant therapy followed by surgical resection at West China Hospital between January 2018 and December 2023 were screened and enrolled. Patients were categorized into early relapse (≤ 6 months) and late relapse (> 6 months) groups based on the interval from surgery to relapse.

Results

A total of 183 LA-ESCC patients who received neoadjuvant therapy followed by surgical resection were included, with 79 experiencing early relapse and 104 experiencing late relapse. Logistic regression analysis showed that postoperative TNM stage III-IV, ypN + stage, R1 resection, lymphovascular invasion (LVI), perineural invasion (PI), and absence of postoperative adjuvant therapy were risk factors for early relapse (p < 0.05). Multivariate regression further identified R1 resection as an independent predictor (p < 0.05). Cox regression analysis demonstrated that LI was a prognostic factor for overall survival (OS) in patients with early relapse (p = 0.045). Kaplan-Meier analysis revealed significantly reduced OS (12.5 months vs. 26.9 months, HR = 2.96, p < 0.001) and survival after relapse (SAR) (5.9 months vs. 11.0 months, HR = 1.81, p < 0.001) in the early relapse group compared to the late relapse group. However, relapse patterns did not differ significantly between groups (p > 0.05).

Conclusion

R1 resection is an independent risk factor for rapid postoperative relapse in LA-ESCC patients following neoadjuvant therapy. Furthermore, LVI significantly affects patients survival outcomes, and early relapse was strongly associated with reduced overall survival (OS).

Introduction

Esophageal cancer (EC) is one of the most common gastrointestinal malignancies worldwide [1]. China accounts for 53.7% of newly diagnosed cases and 55.3% of deaths, with nearly 90% of cases being esophageal squamous cell carcinoma (ESCC) [2, 3]. About 50% of patients are diagnosed at a locally advanced stage [4]. Radical surgery remains the cornerstone of ESCC treatment [5]. In recent years, neoadjuvant therapy has become the standard treatment strategy for resectable locally advanced esophageal squamous cell carcinoma (LA-ESCC), significantly improving patients’ overall survival (OS) rates [6–8]. However, the overall risk of postoperative relapse remains high, with several international studies reporting relapse rates between 27% and 52.4% [9–11]. This leads to a poor prognosis, with the 5-year overall survival rate remaining at only 30–50% [12, 13].

Among ESCC patients who experience postoperative relapse, early relapse—defined as occurring within 6 months after surgery—is of particular concern. These patients tend to have more aggressive biological behavior [14], and studies have shown that survival after early relapse is significantly shorter than after late relapse, indicating a worse prognosis [15]. Nevertheless, there is a lack of clear risk prediction for early relapse, which hinders clinicians from effectively identifying high-risk patient groups for individualized treatment and follow-up strategies.

Therefore, thoroughly investigating the risk factors and survival determinants associated with rapid postoperative relapse in patients with LA-ESCC, as well as establishing effective prediction or risk assessment models, has become a critical issue that urgently needs to be addressed in the clinical management of LA-ESCC. In this retrospective study, we analyzed the clinicopathological characteristics and prognostic factors related to early relapse following neoadjuvant therapy in LA-ESCC patients, aiming to provide evidence for optimizing postoperative management and improving patient outcomes.

Methods

Study population

Patients with LA-ESCC who underwent surgery following neoadjuvant therapy were included in a single-centre cohort analysis. The study adhered to the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) guidelines and received approval from the Ethics Committee of West China Hospital, Sichuan University (2024–2390), in accordance with the Declaration of Helsinki.

Clinical data from ESCC patients who underwent surgical resection at the Cancer Centre of West China Hospital, Sichuan University, between January 2018 and December 2023 were collected. Patients with indications for neoadjuvant therapy (cT2N + M0 or cT3-4aN0-3M0), as assessed according to the guidelines of the National Comprehensive Cancer Network (NCCN) and the Chinese Society of Clinical Oncology (CSCO), were included in the study. Inclusion criteria were as follows: (1) age 18–80 years; (2) Eastern Cooperative Oncology Group (ECOG) Performance Status (PS) score of 0–1; (3) pathological biopsy confirming ESCC; (4) receipt of 2–4 cycles of neoadjuvant therapy; (5) underwent radical esophagectomy for ESCC. Exclusion criteria were: (1) coexistence of other tumor histologies, including small cell carcinoma or adenocarcinoma; (2) death without relapse within 3 months; (3) incomplete clinical follow-up data; (4) no relapse observed. Incomplete data were defined as loss to follow-up without an event (relapse or death) within 6 months or incomplete medical records of key data.

Treatment

Treatment decisions included neoadjuvant therapeutic strategies, surgical approaches, and adjuvant therapies, all of which were discussed by the multidisciplinary team to develop individualized treatment plans. Notably, the selection of these treatment options was independent of the study’s design and conduct. The neoadjuvant regimen comprised concurrent radiotherapy: platinum-based chemotherapy combined with radiotherapy (following the CROSS and NEOCRTEC5010 study protocols) [7]; chemotherapy combined with immunotherapy based on the Keystone-001 and ESCORT-1st study protocols [8]. Radical surgery was performed 4–8 weeks after neoadjuvant treatment, with the surgical approach—primarily Ivor-Lewis or McKeown procedures—determined by surgical specialists according to therapeutic requirements. Given that this study includes cases of upper thoracic esophageal carcinoma, it is important to note that while the NCCN recommends radical concurrent chemoradiotherapy for these patients, the Chinese Guidelines on Perioperative Management of Resectable Esophageal Cancer (2023 edition) suggest that surgery may be considered as a potential option for patients whose tumor margin is ≤ 5 cm from the cricopharyngeus, provided multidisciplinary treatment is ensured [16]. Therefore, the assessment of surgical indications should be determined through multidisciplinary discussion. Postoperative adjuvant therapy indications were evaluated, and adjuvant therapy was initiated within 4 weeks after surgery based on surgical pathological findings.

Assessment of treatment response and staging diagnosis is performed using pre- and post-neoadjuvant therapy imaging, including computed tomography (CT), upper gastrointestinal barium swallow, upper gastrointestinal endoscopy, magnetic resonance imaging (MRI), and positron emission tomography-computed tomography (PET-CT). Clinical T staging primarily relies on enhanced CT to evaluate the cervical, thoracic, and abdominal regions. Chest and abdominal MRI serve as supplementary evaluation methods. For N staging, further assessment is conducted with endoscopic ultrasound-guided fine-needle aspiration (EUS-FNA). M staging diagnosis often utilizes PET-CT for additional support. Analysis of surgical specimens includes evaluation of tumor size, surgical margins, degree of differentiation, lymphovascular invasion, perineural invasion, lymph node response, the extent and number of lymphadenectomies, the number and percentage of regionally positive lymph nodes, and tumor regression scores (TRS).

Follow-up

Imaging examinations and upper gastrointestinal endoscopy are routinely performed postoperatively every three months for two years, followed by every six months until five years post-surgery. In some cases, the frequency of follow-up may be increased in real-time based on clinical symptoms. The site of recurrence is determined based on CT or MRI results, endoscopic examination (with endoscopic ultrasound performed if necessary), and tissue biopsy. When clinical findings are suggestive of distant metastasis or when CT evaluation is challenging, PET-CT may be used in conjunction. Follow-up data is recorded in the clinical case electronic system for review, and any missing or uncertain information may be obtained through telephone interviews with patients, their families, or attending physicians.

Outcome measures

Tumor node metastasis (TNM) staging was performed according to the American Joint Committee on Cancer (AJCC)/Union for International Cancer Control (UICC) 8th edition, assessing Primary Tumor, Regional Lymph Node, and Distant Metastatic Lesions. Response to treatment was evaluated based on the Response Evaluation Criteria in Soft Tissue Tumors (RECIST) definition of TRS and the American Society of Pathology (ASCP) TRG, which describe the pathological response of the tumor to neoadjuvant therapy [17, 18]. TRS0 indicates complete regression; TRS1, moderate regression; TRS2, mild regression; and TRS3, poor regression. R1 resection refers to the circumferential resection margin. Overall survival (OS) was defined as the interval from definitive diagnosis to death or final follow-up. Disease progression-free survival (DFS) was defined as the time from surgery to tumor relapse. Survival after relapse (SAR) was defined as the time from diagnosis of relapse to death or final follow-up. Relapse patterns were classified as primary site (anastomotic) relapse, regional lymph node relapse, local relapse (primary site + regional lymph nodes), non-regional lymph node metastasis, single or multiple organ metastasis, thoracoabdominal metastasis, and systemic relapse.

Statistical methods

Statistical analyses were conducted using R software version 4.1.2 (http://www.r-project.org). The chi-square test was used for comparisons of categorical data, the Wilcoxon rank sum test for continuous data, and the Kruskal-Wallis test for multiple comparisons of non-parametric data. Risk factors for relapse were analyzed using univariate and multivariate logistic regression, while risk factors affecting survival to early relapse were assessed using the Cox proportional hazards model. Survival analyses were performed using the log-rank test and the Kaplan-Meier method. A two-sided p value of less than 0.05 was considered statistically significant.

Result

Baseline characteristics

A total of 689 esophageal cancer patients were collected after neoadjuvant therapy for EC. Of these, 626 cases (90.9%) were identified as ESCC patients. Twenty-nine cases (4.6%) with incomplete data were excluded, resulting in 597 eligible cases (95.4%). Among these, 183 cases (30.7%) experienced recurrence. Of the recurrent cases, 79 (43.2%) were early relapses and 104 (56.8%) were late relapses. The enrollment flow chart is shown in Fig. 1.

Fig. 1.

Flow chart of patient screening. ESCC, esophageal squamous cell carcinoma

The demographic characteristics of the two groups (age, gender, smoking history, drinking history, weight) were balanced. Compared to the late relapse group, the early relapse group had a higher proportion of postoperative TNM stage III/IV, ypN+, R1 margins, lymphovascular invasion (LVI), and perineural invasion (PI) cases (p < 0.05). In terms of adjuvant therapy, the trend was the opposite. There were no statistically significant differences in the remaining clinical data between the two groups (Table 1).

Table 1.

Comparison of clinical data between the early relapse group (≤ 6 months)and the later relapse group (> 6 months)

Characteristics	Patients, No. (%)		P value
	early relapse group (n = 79)	late relapse group (n = 104)
Age			0.145
≤ 60	39 (49.4)	59 (37.5)
> 60	40 (50.6)	65 (62.5)
Gender			0.797
Male	68 (86.1)	92 (88.5)
Female	11(13.9)	12 (11.5)
Smoking			0.891
No	29(36.7)	46(34.6)
Yes	50 (63.3)	68 (65.4)
Alcohol			0.682
No	29 (36.7)	24 (22.7)
Yes	50 (63.3)	70 (67.3)
BMI			0.439
normal	43 (54.4)	61 (58.7)
abnormal	18 (45.6)	16 (41.3)
Primary site			0.502
Upper	11 (13.9)	10 (9.6)
Middle/Lower	68 (86.1)	94 (90.4)
Tumor length (mean (SD))	5.80 (2.22)	5.21 (1.98)	0.061
cTNM stage			0.666
II	6 (7.6)	11 (10.6)
III/IV	73 (92.4)	93 (89.4)
cT stage			0.271
T2	7 (8.9)	4 (3.8)
T3/4	72 (91.1)	100 (96.2)
N stage			0.274
N0	7 (8.9)	16 (15.4)
N+	72 (91.1)	88 (84.6)
Type of neoadjuvant treatment			0.469
nCRT	49 (62.0)	71 (68.3)
nICT	30 (38.0)	33 (31.7)
Number of lymph node dissections (mean (SD))	23.65 (10.28)	23.08 (9.50)	0.699
ypTNM stage			0.038
Ⅰ/II	28 (35.4)	54 (51.9)
III/IV	51 (64.6)	50 (48.1)
ypT stage			0.070
T0/1/2	31 (39.2)	56 (53.8)
T3/4	48 (60.8)	48 (46.2)
ypN stage			0.016
N0	26 (32.9)	54 (51.9)
N+	53 (67.1)	50 (48.1)
TRS			0.594
0	25 (31.6)	38 (36.5)
1/2/3	54 (68.4)	66 (63.5)
Lymph node reaction			0.906
No	67(84.8)	90(86.5)
Yes	12 (15.2)	14 (13.5)
Differentiation level			1.000
well	27 (34.2)	36 (34.6)
worse	52 (65.8)	68 (65.4)
Resection Margin			< 0.001
R0	58 (73.4)	97 (93.3)
R1	21(26.6)	7(6.7)
Lymphovascular invasion			0.017
negative	46 (58.2)	79 (76.0)
positive	33 (41.8)	25 (24.0)
Perineural invasion			0.002
negative	43 (54.4)	80 (76.9)
positive	36 (45.6)	24 (23.1)
Adjuvant treatment			0.025
yes	41(51.9)	72(69.2)
no	38 (48.1)	32 (30.8)
Locoregional rucurrence			0.847
no	39(49.4)	54(51.9)
yes	40 (50.6)	50 (48.1)
Distant metastasis			0.462
no	18(22.8)	18(17.3)
yes	61 (77.2)	86 (82.7)
Statu			0.048
death	65 (82.3)	71 (68.3)
live	14(17.7)	23(31.7)

nICT neoadjuvant immunochemotherapy, nCRT neoadjuvant chemoradiotherapy, TNM tumor, node, and metastasis, TRS tumor regression score

Risk factors for early relapse

Univariable logistic regression analysis showed that postoperative pathological TNM stage III/IV, ypN+, R1 resection, LVI, PI, and lack of adjuvant therapy were risk factors for early relapse in patients. Based on the results of univariable logistic regression analysis, variables related to outcome (p < 0.1) were selected for multivariable logistic regression analysis, which demonstrated that R1 resection was an independent risk factor for early relapse (P = 0.008). The results of univariable and multivariable logistic regression analyses for predictors of early relapse are presented in Table 2.

Table 2.

Logistic regression analysis for predictors of early relapse group

Variable	Univariable-logistic regression analysis		Multivariate-logistic regression analysis
	Odds ratio (95% CI)	p-Value	Odds ratio (95% CI)	p-Value
Age
≤ 60	Reference
> 60	0.62(0.34–1.11)	0.109
Gender
Male	Reference
Female	0.81(0.34–1.94)	0.630
Smoking
No	Reference
Yes	0.91(0.50–1.68)	0.769
Alcohol
No	Reference
Yes	0.84(0.45–1.55)	0.571
BMI
normal	Reference
abnormal	1.19(0.66–2.14)	0.568
Primary site
Upper	Reference
Middle/Lower	0.66(0.26–1.64)	0.367
cTNM stage
II	Reference
III/IV	1.44(0.51–4.08)	0.493
cT stage
T2	Reference
T3/4	0.41(0.12–1.46)	0.169
cN stage
N0	Reference
N+	1.87(0.73–4.79)	0.192
Type of neoadjuvant treatment
nCRT	Reference
nICT	1.32(0.71–2.43)	0.379
ypTNM stage
I/II	Reference
III/IV	1.97(1.089–3.59)	0.027
ypT stage
T0/1/2	Reference
T3/4	1.81(1.00-3.27)	0.051	0.76(0.35–1.66)	0.497
ypN stage
N0	Reference
N+	2.20(1.20–4.04)	0.011	1.48(0.73–2.99)	0.272
TRS
0	Reference
1/2/3	1.24(0.67–2.31)	0.490
Lymph node reaction
Yes	Reference
No	1.15(0.50–2.65)	0.740
Differentiation level
well	Reference
worse	1.02(0.55–1.89)	0.951
Resection margin
R0	Reference
R1	5.02(2.01–12.53)	0.001	3.80(1.42–10.17)	0.008
Lymphovascular invasion
negative	Reference
positive	2.26(1.20–4.27)	0.011	1.34(0.64–2.82)	0.441
Perineural invasion
negative	Reference
positive	2.79(1.48–5.27)	0.002	2.01(0.91–4.44)	0.084
Adjuvant treatment
yes	Reference
no	2.09(1.14–3.83)	0.018	1.72(0.90–3.31)	0.103

nICT neoadjuvant immunochemotherapy, nCRT neoadjuvant chemoradiotherapy, TNM tumor, node, and metastasis, TRS tumor regression score

Prognostic risk factors for early relapse

On univariate Cox analysis, LVI was identified as a risk factor affecting OS in patients with early relapse (HR=1.66, 95% CI: 1.01–2.71, p=0.045). Further screening of variables related to outcome (p<0.1) in multivariate Cox analysis revealed no independent risk factors affecting OS (p>0.05). The results of univariate and multivariate Cox analyses for prognostic risk factors in the early relapse group are presented in Table 3. The Cox proportional hazards model did not show that SAR was affected in patients with early relapse risk factors (p>0.05) (Table 4).

Table 3.

Cox analysis in OS

Variable	Univariable Cox analysis		Multivariate Cox analysis
	Hazard ratio (95% CI)	p-Value	Hazard ratio (95% CI)	p-Value
Age
≤ 60	Reference
> 60	1.19(0.73–1.93)	0.490
Gender
Male	Reference
Female	0.81(0.41–1.59)	0.537
Smoking
No	Reference
Yes	1.15(0.69–1.93)	0.593
Alcohol
No	Reference
Yes	1.48(0.88–2.49)	0.145
BMI
normal	Reference
abnormal	1.44(0.88–2.36)	0.146
Primary site
Upper	Reference
Middle/Lower	1.21(0.58–2.55)	0.607
cTNM stage
II	Reference
III/IV	2.72(0.85–8.70)	0.092	2.44(0.76–7.86)	0.136
cT stage
T2	Reference
T3/4	1.02(0.44–2.38)	0.958
cN stage
N0	Reference
N+	1.90(0.69–5.26)	0.216
Type of neoadjuvant treatment
nCRT	Reference
nICT	0.88(0.53–1.46)	0.628
ypTNM stage
I/II	Reference
III/IV	1.51(0.90–2.56)	0.121
ypT stage
T0/1/2	Reference
T3/4	1.18(0.71–1.95)	0.523
ypN stage
N0	Reference
N+	1.34(0.79–2.28)	0.277
TRS
0	Reference
1/2/3	1.03(0.61–1.74)	0.902
Lymph node reaction
Yes	Reference
No	1.09(0.54–2.22)	0.804
Differentiation level
well	Reference
worse	1.06(0.63–1.76)	0.837
Resection margin
R0	Reference
R1	0.79(0.45–1.42)	0.436
Lymphovascular invasion
negative	Reference
positive	1.66(1.01–2.71)	0.045	1.55(0.94–2.54)	0.084
Perineural invasion
negative	Reference
positive	0.95(0.58–1.55)	0.830
Adjuvant treatment
yes	Reference
no	1.44(0.88–2.36)	0.145

OS Overall Survival, nICT neoadjuvant immunochemotherapy, nCRT neoadjuvant chemoradiotherapy, TNM tumor, node, and metastasis, TRS tumor regression score

Table 4.

Cox analysis in SAR

Variable	Univariable Cox analysis		Multivariate Cox analysis
	Hazard ratio (95% CI)	p-Value	Hazard ratio (95% CI)	p-Value
Age
≤ 60	Reference
> 60	1.13(0.70–1.85)	0.613
Gender
Male	Reference
Female	0.82(0.42–1.61)	0.565
Smoking
No	Reference
Yes	1.23(0.73–2.05)	0.438
Alcohol
No	Reference
Yes	1.51(0.89–2.54)	0.124
BMI
normal	Reference
abnormal	1.40(0.86–2.29)	0.177
Primary site
Upper	Reference
Middle/Lower	1.22(0.58–2.55)	0.607
cTNM stage
II	Reference
III/IV	3.01(0.94–9.66)	0.064	2.24 (0.66–7.62)	0.198
cT stage
T2	Reference
T3/4	1.08(0.47–2.52)	0.855
cN stage
N0	Reference
N+	1.98(0.71–5.48)	0.190
Type of neoadjuvant treatment
nCRT	Reference
nICT	0.92(0.55–1.52)	0.746
ypTNM stage
I/II	Reference
III/IV	1.65(0.97–2.79)	0.062	1.15 (0.65–2.03)	0.633
ypT stage
T0/1/2	Reference
T3/4	1.14(0.69–1.88)	0.615
ypN stage
N0	Reference
N+	1.49(0.88–2.55)	0.140
TRS
0	Reference
1/2/3	1.12(0.66–1.89)	0.670
Lymph node reaction
Yes	Reference
No	1.15(0.57–2.33)	0.705
Differentiation level
well	Reference
worse	1.00(0.60–1.68)	0.991
Resection margin
R0	Reference
R1	0.83(0.47–1.48)	0.523
Lymphovascular invasion
negative	Reference
positive	1.56(0.95–2.56)	0.081	1.50(0.90–2.50)	0.122
Perineural invasion
negative	Reference
positive	0.83(0.51–1.36)	0.465
Adjuvant treatment
yes	Reference
no	1.40(0.85–2.29)	0.185

SAR Survival After Relapse, nICT neoadjuvant immunochemotherapy, nCRT neoadjuvant chemoradiotherapy, TNM tumor, node, and metastasis, TRS tumor regression score

Survival analysis

The median follow-up for relapsed patients was 22.5 months. In the early relapse group, the median follow-up was 16.1 months, with 65 events (82.2%) occurring by the follow-up cut-off. In the later relapse group, the median follow-up was 29.2 months, with 71 events (68.2%) by the cut-off, meeting the study endpoint requirements for event rates. A log-rank test for OS was performed comparing the early and later relapse groups. OS (12.5 months vs. 26.9 months, HR=2.96, 95% CI: 2.09-4.19, p<0.001) and SAR (5.9 months vs. 11.0 months, 95% CI: 1.29-2.55, HR=1.81, p<0.001) were both significantly shorter in the early relapse group compared to the later relapse group, and the difference between the Kaplan-Meier survival curves was significant (Figure 2).

Fig. 2.

Survival analysis stratified by relapse timing in patients. A Kaplan–Meier curves for OS, defined as time from initial diagnosis to death. Patients with early relapse (≤6 months) exhibited shorter OS compared to those with late relapse (>6 months). B Kaplan–Meier curves for survival after relapse (SAR), defined as time from first documented recurrence to death. A shorter SAR was observed in the early relapse group compared to the late relapse group. Survival curves were compared using the log-rank test. Shaded areas represent 95% confidence intervals. The number at risk is shown below each time point

Within the early relapse group, comparing LVI positive and negative patients, the median OS in the positive group was significantly shorter than in the negative group (11.2 months vs. 13.3 months, HR=1.66, 95% CI: 1.01-2.71, p=0.043) (Figure 3A). However, there was no statistically significant difference in SAR between the two groups (4.6 months vs. 6.6 months, HR=1.56, 95% CI: 0.95-2.56, p=0.079) (Figure 3B). Meanwhile, a comparison was made between the cTNM stage III/IV and cTNM stage I/II groups in terms of OS and SAR in the early relapse group. The OS comparison showed 12.4 months versus 14.3 months (HR=2.72, 95% CI: 0.85–8.70, p=0.080) (Figure 4A), while the SAR comparison was 9.3 months versus 5.5 months (HR=3.01, 95% CI: 0.94–9.66, p=0.050) (Figure 4B). The cTNM stage III/IV group exhibited numerically shorter OS and SAR than the cTNM stage I/II group, though neither reached statistical significance.

Fig. 3.

Survival analysis stratified by lymphovascular invasion (LVI) status in patients with early relapse. A Kaplan–Meier curves for OS, defined as time from initial diagnosis to death. Patients with LVI-positive exhibited shorter OS compared to those with LVI-negative. B Kaplan–Meier curves for survival after relapse (SAR), defined as time from first documented recurrence to death. A shorter SAR was observed in LVI-positive patients compared to LVI-negative patients. Survival curves were compared using the log-rank test. Shaded areas represent 95% confidence intervals. The number at risk is shown below each time point

Fig. 4.

Survival analysis stratified by clinical TNM stage in patients with early relapse. A Kaplan–Meier curves for OS, defined as time from initial diagnosis to death. Patients with cTNM stage III/IV exhibited shorter OS compared to those with stage I/II；B Kaplan–Meier curves for survival after relapse (SAR), defined as time from first documented recurrence to death. A shorter SAR was observed in cTNM stage III/IV compared to stage I/II. Survival curves were compared using the log-rank test. Shaded areas represent 95% confidence intervals. The number at risk is shown below each time point

Relapse pattern

In the early relapse group, the most common patterns, in descending order, were systemic relapse (32.9%), organ metastasis (29.2%), non-regional lymph node metastasis (12.7%), relapse at the primary site (11.4%), regional lymph node relapse (7.6%), thoracic and peritoneal membrane metastasis (3.7%), and local relapse (2.5%) (Fig. 5 A). In the late relapse group, the order was organ metastasis (37.5%), systemic relapse (27.8%), non-regional lymph node metastasis (15.3%), regional lymph node relapse (6.7%), relapse at the primary site (5.7%), local relapse (4.8%), and thoracic-abdominal membrane metastasis (1.9%) (Fig. 5B). Systemic relapse, organ metastasis, and non-regional lymph node metastasis were the top three recurrence patterns in both groups, indicating a predominance of distant metastatic relapse. There was no statistically significant difference in the relapse patterns between the two groups (Fig. 5C)

Fig. 5.

Relapse patterns in patients with early relapse and late relapse. A Distribution of relapse patterns in patients with early relapse. B Distribution of relapse patterns in patients with early relapse. C Statistical analysis of recurrence patterns showed no significant difference between early-relapse and late-relapse patients

Discussion

In this retrospective cohort study, we systematically examined the clinicopathological characteristics and prognostic factors linked to postoperative early relapse (≤ 6 months) in patients with LA-ESCC who underwent neoadjuvant therapy. Of the 597 patients included, early relapse occurred in 13.2% (79 cases), which is consistent with previously reported rates in international studies (approximately 10%−13%) [19] Consistent with previous studies, our findings reaffirm the critical role of R1 resection as a significant predictor of early postoperative relapse. R1 margins substantially increase the likelihood of residual microscopic tumor cells, which can promote early recurrence through mechanisms such as circulating tumor cells (CTCs) [20–22]. This effect may also be influenced by factors like treatment compliance, timeliness, and adherence to standardized protocols after surgery. Prior literature has substantiated the adverse impact of R1 resection on relapse risk in esophageal carcinoma, and our study further quantifies its specific contribution to early relapse Notably, however, some reports suggest that positive margins, in the absence of neoadjuvant therapy, may have a limited impact, highlighting the potential modifying effects of preoperative treatment on margin status [23]. Different neoadjuvant therapies yield varying results. Previous research indicates that pathological complete response (pCR) and lymph node tumor regression grading following neoadjuvant chemotherapy or chemoradiotherapy are independent risk factors for postoperative relapse [24]. With the advent of neoadjuvant immunotherapy, studies have identified postoperative positive lymph nodes, lack of response to neoadjuvant treatment, and resection of fewer than 20 lymph nodes as independent risk factors for early recurrence [25]. This contrasts with our findings, suggesting that the risk factors for early relapse after neoadjuvant immunotherapy in locally recurrent ESCC may differ Survival analyses in this study showed significant reductions in both OS and SAR among patients with early relapse, aligning with previous reports [26]. The link between LVI and poorer OS likely reflects its role in facilitating aggressive metastatic spread through mechanisms such as epithelial-mesenchymal transition (EMT), potentially mediated by transcription factors like ZEB1, as previously described [24, 27–29]. However, most studies have examined the effect of early relapse on OS, without assessing SAR after relapse. Our findings indicate that patients with early relapse also have shorter SAR, which may be due to the inherently poor prognosis of this group As LVI is a risk factor for survival in rapidly relapsing patients, we further compared OS and SAR between the LVI positive and LVI negative groups within this population. The OS in the LVI positive group was significantly shorter than in the LVI negative group. Although SAR showed a trend toward reduction, indicating poorer outcomes for LVI positive patients, the difference did not reach statistical significance. Compared to patients with advanced disease, those experiencing early relapse—especially those with LVI positivity—have poorer survival outcomes, presenting a significant challenge to current treatment strategies and necessitating special clinical attention. Notably, while R1 resection significantly affected early relapse, it did not independently predict OS in our Cox regression analysis, likely due to the complex interplay between surgical margins, tumor biology, and postoperative interventions [30]. Because of the substantial absolute difference in sample sizes between cTNM stage I/II and cTNM stage III/IV within the early relapse group, the survival analysis for OS and SAR in these two subgroups is of limited value Analysis of relapse patterns showed no significant differences between early and late relapse groups, consistent with literature reporting similar distributions of systemic, distant, and loco-regional relapse [31, 32]. However, a higher proportion of primary site recurrence, regional lymph node involvement, and thoracoabdominal metastases in the early relapse group suggests distinct clinical trajectories that warrant further investigation using standardized assessment protocols.

Research limitations

This study has several limitations. Firstly, as a retrospective, single-center study, despite the considerable overall sample size, the subset of patients experiencing early recurrence remained limited. This small subgroup affected statistical analyses, resulting in wide confidence intervals, which could introduce bias and limit the interpretability of results. To overcome this, future multicenter studies with larger patient cohorts and extended follow-up durations are warranted. Additionally, potential confounding factors, such as molecular biomarkers and the reliance on imaging modalities rather than pathological confirmation to detect recurrences, may affect the accuracy of recurrence identification. Future research incorporating advanced techniques like liquid biopsy (e.g., circulating tumor DNA) could enhance the accuracy of early relapse prediction and facilitate personalized postoperative management for high-risk LA-ESCC patients.

Conclusion

R1 resection is an independent risk factor for rapid postoperative relapse in LA-ESCC patients following neoadjuvant therapy. Furthermore, LVI significantly affects patients survival outcomes, and early relapse was strongly associated with reduced overall survival (OS).

Authors’ contributions

Ze-Gang Chen and Yue-yun Chen contributed equally to this article.Ze-Gang Chen and Yue-Yun Chen drafted the manuscript text. Ze-Gang Chen and Yue-Yun Chen performed statistical analyses. Zhen Lin, Pan-pan Lin, Qing Li and Yu-Shang Yang designed this study and revised the paper. Hua-Shan Shi gave critical comments. Zhen-Yu Ding gave critical comments and revised the paper. All authors reviewed the manuscript。.

Funding

Financial support and sponsorship: this work was supported by grants from the Sichuan Science and Technology Program (2023YFS0001) and Beijing Xisike Clinical Oncology Research Foundation (Year 2021) (Y-MSDPU2021-0183).

Data availability

No datasets were generated or analysed during the current study.

Declarations

Ethics approval and consent to participate

This study met the requirements of the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) guidelines, and the study was approved by the Ethics Committee of West China Hospital of Sichuan University (2024-2390) in accordance with the Declaration of Helsinki. All patients provided written informed consent.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.