Prognostic impact of treatment modalities and examined lymph nodes count on survival in T3–4 esophageal cancer: a retrospective cohort and predictive modeling study from the Surveillance, Epidemiology, and End Results database

Introduction

In 2022, esophageal carcinoma (EC) was the 11th most commonly diagnosed cancer and the 7th leading cause of cancerrelated deaths worldwide, with approximately 511,000 new cases and 445,000 deaths (1). By 2050, global cases are projected to rise to about 923,000 and deaths to 826,000, representing increases of roughly 80.5% and 85.4%, respectively, from 2022 (2). These trends highlight a growing disease burden and a pressing public health challenge in the decades ahead (3). EC is primarily composed of two histologic subtypes, namely esophageal squamous cell carcinoma (ESCC) and esophageal adenocarcinoma (EAC). ESCC is the predominant form worldwide. While its overall incidence and mortality have been declining, it remains highly prevalent in regions such as Southeast Asia and Central Asia (4). In contrast, the incidence of EAC has risen substantially, and it has become the most common type of EC in several areas, including the United States and Western Europe (5,6).

The early symptoms of EC are often subtle and nonspecific. These symptoms are frequently overlooked, leading to delayed diagnosis. As a result, more than 90% of patients are diagnosed at an advanced stage (7). For advanced EC, multimodality therapy is the established standard, with neoadjuvant therapy followed by curative resection representing the most widely adopted approach (8-10). In clinical practice, however, a substantial number of patients still proceed directly to esophagectomy due to tumor biology, comorbidities, or limited access to neoadjuvant treatment, which raises important questions regarding the role of postoperative adjuvant therapy (11). Notably, even among patients who have undergone neoadjuvant therapy and curative resection, whether adjuvant therapy confers an additional survival benefit remains controversial, and its appropriate indications, optimal regimens, and extent of survival benefit require further validation through high-level evidence (12,13). In radical esophagectomy for EC, lymph nodes dissection is a critical determinant of long-term prognosis. The number of examined lymph nodes (ELNs) correlates with improved overall survival (OS) and provides the necessary basis for accurate pathological staging and informed adjuvant therapy decisions (14,15). However, neoadjuvant therapy can substantially reduce both primary tumor and nodal burden, sometimes achieving pathological complete response (pCR), thereby altering the underlying tumor biology (16,17). In this context, the survival benefit of extensive lymphadenectomy remains unconfirmed and must be carefully balanced against the associated risk of surgical complications (18).

Therefore, this study aims to compare the survival differences among T3–4 EC patients undergoing different treatment modalities, including surgery alone (SA), neoadjuvant therapy plus surgery (NS), adjuvant therapy plus surgery (AT), and perioperative therapy (PT). Furthermore, it seeks to analyze whether the number of ELNs reaches a prognostic threshold within each treatment subgroup and the impact of this threshold on survival. Finally, a nomogram incorporating treatment strategy and the number of ELNs is constructed to provide a more accurate predictive tool for OS in patients with T3–4 EC. We present this article in accordance with the TRIPOD reporting checklist (available at https://jtd.amegroups.com/article/view/10.21037/jtd-2025-1-2675/rc).

Methods

Study population

In this retrospective study, data were obtained from the Surveillance, Epidemiology, and End Results (SEER) database, a population-based cancer registry in the United States. Patients diagnosed with EC between 2007 and 2021 (n=58,851) were extracted from 17 SEER registries using SEER*Stat software. The analyzed variables included age, sex, race, marital status, histologic type, tumor location, grade, tumor (T) stage, node (N) stage, metastasis (M) stage, number of ELNs, surgical status, vital status, and survival months. The primary endpoint was OS, defined as the duration from diagnosis to death from any cause or the last known follow-up. The study was conducted in accordance with the Declaration of Helsinki and its subsequent amendments.

Inclusion criteria were as follows: (I) diagnosis of primary EC; (II) age ≥18 years; (III) diagnosed between 2007 and 2021; (IV) underwent radical esophagectomy; (V) postoperative pathological stage of pT3–4; and (VI) availability of complete postoperative follow-up data. Exclusion criteria were as follows: (I) preoperative diagnosis of metastatic disease or concurrent other primary malignancy; (II) OS <1 month; (III) number of ELNs =0; and (IV) missing key clinicopathological or follow-up information. Based on these criteria, 4,485 patients were included in the final analysis. According to treatment strategy, patients were categorized into four groups: the NS group (n=3,116, 69.5%), the AT group (n=328, 7.3%), the PT group (n=736, 16.4%), and the SA group (n=305, 6.8%) (Figure 1).

Figure 1 Flowchart of patient inclusion. AT, adjuvant therapy following surgery; M, metastasis; N, node; NS, neoadjuvant therapy plus surgery; PT, perioperative therapy; SA, surgery alone; SEER, Surveillance, Epidemiology, and End Results; T, tumor.

Statistical analysis

Categorical variables are presented as frequencies and percentages, while continuous variables are summarized as medians with interquartile ranges. Survival comparisons between groups were performed using Kaplan-Meier curves. To control for potential confounding factors, 1:1 propensity score matching (PSM) with a caliper width of 0.02 was applied to address baseline imbalances between different treatment groups and between subgroups stratified by the number of ELNs. Survival analyses were repeated in the matched cohorts. For prognostic model development and validation, all eligible patients were randomly divided into a training set and a validation set at a 7:3 ratio. In the training set, variables with P<0.05 in univariate Cox regression were further selected via least absolute shrinkage and selection operator (LASSO) regression for dimension reduction. The retained variables were then entered into a multivariate Cox proportional hazards model to identify independent prognostic factors and to construct a nomogram. Model performance was evaluated as follows: discrimination was assessed using Harrell’s concordance index (C-index) and receiver operating characteristic (ROC) curves; calibration was examined with calibration plots for 1-, 3-, and 5-year OS; and clinical utility was evaluated via decision curve analysis (DCA). X-tile software (version 2.0) was employed to determine the optimal prognostic cut-off values for ELNs under different treatment modalities, based on the maximum χ² principle derived from Kaplan-Meier survival analysis and the log-rank test. Patients were subsequently stratified into high- and low-ELNs subgroups accordingly. All statistical analyses were performed using R software (version 4.2.2; R Foundation for Statistical Computing, Vienna, Austria). A two-sided P<0.05 was considered statistically significant.

Results

Participant characteristics

The demographic and clinicopathological characteristics of the cohort are summarized in Table 1. The patients were predominantly male (3,767, 84.0%), White (4,043, 90.1%), and under 60 years of age (2,520, 56.2%). Tumors were most frequently located in the lower esophagus (3,714, 82.8%) and were histologically adenocarcinoma (3,286, 73.3%). The majority of patients were married (3,058, 68.2%). Pathologic staging distributions were as follows: T3 in 4,143 patients (92.4%), N1 in 2,772 (61.8%), and M0 in 4,152 (92.6%). Poorly differentiated (grade III) histology was present in 2,074 cases (46.2%). The median number of ELNs was 16. A total of 2,796 patients died, with a median follow-up time of 24 months.

Comparison of baseline characteristics before and after PSM across treatment groups

This study compared four primary treatment modalities for EC. To address confounding due to non-random treatment assignment, we performed PSM for each pairwise comparison. PSM balanced key baseline covariates between the matched cohorts, with all significant pre-matching differences eliminated (P>0.05). Adequate balance was confirmed, as standardized mean differences (SMD) for all variables approached zero. Tables S1-S6 and Figures S1-S6 present the baseline characteristics and covariate balance plots for all pairwise comparisons, respectively.

Comparison of prognosis between neoadjuvant therapy followed by surgery and SA

Prior to matching, Kaplan-Meier curve analysis demonstrated that OS was significantly higher in the NS group than in the SA group (P<0.001). Specifically, the 1-, 3-, and 5-year OS rates for the NS group were 80.4% [95% confidence interval (CI): 79.0–81.8%], 49.6% (95% CI: 47.8–51.5%), and 38.4% (95% CI: 36.6–40.4%), respectively. In contrast, the corresponding OS rates for the SA group were 54.1% (95% CI: 48.8–60.0%), 24.8% (95% CI: 20.4–30.3%), and 17.7% (95% CI: 13.8–22.7%). After PSM, the estimated 1-, 3-, and 5-year OS rates were 74.7% (95% CI: 69.9–79.8%), 44.6% (39.1–50.8%), and 35.7% (30.2–42.1%) for the NS group, versus 54.2% (48.8–60.1%), 24.7% (20.2–30.1%), and 17.5% (13.6–22.5%) for the SA group. The difference in survival curves remained statistically significant (P<0.001) (Figure S7).

Comparison of prognosis between neoadjuvant therapy followed by surgery and surgery followed by adjuvant therapy

Prior to matching, Kaplan-Meier curve analysis demonstrated that OS was significantly higher in the NS group than in the AT group (P<0.001). Specifically, the 1-, 3-, and 5-year OS rates for the NS group were 80.4% (95% CI: 79.0–81.8%), 49.6% (95% CI: 47.8–51.5%), and 38.4% (95% CI: 36.6–40.4%), respectively. In contrast, the corresponding OS rates for the AT group were 76.6% (95% CI: 72.1–81.3%), 36.6% (95% CI: 31.6–42.4%), and 24.8% (95% CI: 20.3–30.2%). After PSM, the estimated 1-, 3-, and 5-year OS rates were 78.7% (95% CI: 74.3–83.3%), 47.6% (95% CI: 42.2–53.7%), and 36.4% (95% CI: 31.0–42.8%) for the NS group, versus 76.5% (95% CI: 72.0–81.3%), 36.4% (95% CI: 31.4–42.2%), and 24.5% (95% CI: 20.0–30.0%) for the AT group. The difference in survival curves remained statistically significant (P=0.006) (Figure S8).

Comparison of prognosis between neoadjuvant therapy followed by surgery and PT

Prior to matching, Kaplan-Meier curve analysis demonstrated that OS was not significantly different between the NS and PT groups (P=0.90). Specifically, the 1-, 3-, and 5-year OS rates for the NS group were 80.4% (95% CI: 79.0–81.8%), 49.6% (95% CI: 47.8–51.5%), and 38.4% (95% CI: 36.6–40.4%), respectively. The corresponding OS rates for the PT group were 86.0% (95% CI: 83.4–88.7%), 46.3% (95% CI: 42.2–50.8%), and 33.2% (95% CI: 29.1–37.9%). After PSM, the estimated 1-, 3-, and 5-year OS rates were 78.6% (95% CI: 75.6–81.7%), 45.0% (95% CI: 41.2–49.0%), and 35.9% (95% CI: 32.2–40.0%) for the NS group, versus 86.1% (95% CI: 83.5–88.8%), 46.3% (95% CI: 42.2–50.8%), and 33.1% (95% CI: 29.0–37.8%) for the PT group. The difference in survival curves remained statistically non-significant (P=0.21) (Figure S9).

Comparison of prognosis between surgery followed by adjuvant therapy and SA

Prior to matching, Kaplan-Meier curve analysis demonstrated that OS was significantly higher in the AT group than in the SA group (P<0.001). Specifically, the 1-, 3-, and 5-year OS rates for the AT group were 76.6% (95% CI: 72.1–81.3%), 36.6% (95% CI: 31.6–42.4%), and 24.8% (95% CI: 20.3–30.2%), respectively. In contrast, the corresponding OS rates for the SA group were 54.1% (95% CI: 48.8–60.0%), 24.8% (95% CI: 20.4–30.3%), and 17.7% (95% CI: 13.8–22.7%). After PSM, the estimated 1-, 3-, and 5-year OS rates were 75.6% (95% CI: 69.9–81.9%), 38.3% (95% CI: 32.0–45.8%), and 27.1% (95% CI: 21.4–34.3%) for the AT group, versus 54.9% (95% CI: 48.3–62.3%), 22.4% (95% CI: 17.2–29.1%), and 15.8% (95% CI: 11.3–21.9%) for the SA group. The difference in survival curves remained statistically significant (P<0.001) (Figure S10).

Comparison of prognosis between surgery followed by adjuvant therapy and PT

Prior to matching, Kaplan-Meier curve analysis showed a significant survival disadvantage for the AT group relative to the PT group (P<0.001). Specifically, the 1-, 3-, and 5-year OS rates for the AT group were 76.6% (95% CI: 72.1–81.3%), 36.6% (95% CI: 31.6–42.4%), and 24.8% (95% CI: 20.3–30.2%), respectively. In contrast, the corresponding OS rates for the PT group were 86.0% (95% CI: 83.4–88.7%), 46.3% (95% CI: 42.2–50.8%), and 33.2% (95% CI: 29.1–37.9%). After PSM, the AT group continued to exhibit inferior survival outcomes. The estimated 1-, 3-, and 5-year OS rates for the matched AT cohort were 76.1% (95% CI: 71.4–81.0%), 37.5% (95% CI: 32.3–43.4%), and 25.0% (95% CI: 20.4–30.6%), versus 85.8% (95% CI: 81.8–90.0%), 42.8% (95% CI: 36.9–49.6%), and 30.9% (95% CI: 25.3–37.7%) for the matched PT cohort. The survival difference remained statistically significant (P=0.02) (Figure S11).

Comparison of prognosis between PT and SA

Prior to matching, Kaplan-Meier curve analysis demonstrated that OS was significantly higher in the PT group than in the SA group (P<0.001). Specifically, the 1-, 3-, and 5-year OS rates for the PT group were 86.0% (95% CI: 83.4–88.7%), 46.3% (95% CI: 42.2–50.8%), and 33.2% (95% CI: 29.1–37.9%), respectively. In contrast, the corresponding OS rates for the SA group were 54.1% (95% CI: 48.8–60.0%), 24.8% (95% CI: 20.4–30.3%), and 17.7% (95% CI: 13.8–22.7%). After PSM, the estimated 1-, 3-, and 5-year OS rates were 83.1% (95% CI: 78.2–88.4%), 44.0% (95% CI: 37.2–52.1%), and 34.5% (95% CI: 27.6–42.6%) for the PT group, versus 56.2% (95% CI: 50.2–62.9%), 25.7% (95% CI: 20.6–32.0%), and 18.6% (95% CI: 14.1–24.5%) for the SA group. The difference in survival curves remained statistically significant (P<0.001) (Figure S12).

Subgroup analysis of the relationship between elns and survival outcomes across different treatment modalities

This study examined the prognostic impact of different treatment modalities in EC. To account for differences in requirements for lymphadenectomy extent across various treatment modalities and their potential impact on the number of ELNs, we first determined treatment-specific optimal thresholds for the number of ELNs using X-tile software. The identified thresholds were 9 for the NS group, 16 for the AT group, 17 for the SA group, and 19 for the PT group. Based on these values, patients within each treatment cohort were classified into a low-ELNs subgroup (≤ threshold) and a high-ELNs subgroup (> threshold) (Figure S13).

To mitigate potential confounding between these subgroups, PSM was applied separately to the low- and high-ELNs subgroups within each treatment cohort. All significant differences were eliminated after PSM (P>0.05). The effectiveness of matching was further confirmed by SMD analysis, with post-matching values for all variables approaching zero, indicating well-balanced cohorts. Detailed pre- and post-matching baseline characteristics and SMD plots are provided in Tables S7-S10 and Figures S14-S17.

Prior to PSM, OS was significantly higher in the high-ELNs subgroup than in the low-ELNs subgroup within the NS cohort (P<0.001). The 1-, 3-, and 5-year OS rates were 81.7% (95% CI: 80.1–83.3%), 51.5% (95% CI: 49.4–53.7%), and 40.8% (95% CI: 38.7–43.1%) for the high-ELNs subgroup, compared to 76.3% (95% CI: 73.2–79.4%), 43.8% (95% CI: 40.3–47.7%), and 31.4% (95% CI: 28.1–35.2%) for the low-ELNs subgroup. After matching, the survival advantage of the high-ELNs subgroup remained significant (P<0.001). The matched high-ELNs subgroup had 1-, 3-, and 5-year OS rates of 83.2% (95% CI: 80.5–86.0%), 54.7% (95% CI: 51.0–58.7%), and 42.7% (95% CI: 38.9–46.8%), whereas the matched low-ELNs subgroup had corresponding rates of 76.3% (95% CI: 73.2–79.4%), 43.8% (95% CI: 40.3–47.7%), and 31.4% (95% CI: 28.1–35.2%) (Figure S18).

In the AT cohort, a significant survival advantage was observed for the high-ELNs subgroup over the low-ELNs subgroup prior to matching (P=0.02), but this difference was no longer statistically significant after PSM (P=0.09). Before matching, the 1-, 3-, and 5-year OS rates were 85.4% (95% CI: 79.9–91.2%), 43.3% (95% CI: 35.8–52.2%), and 30.8% (95% CI: 23.9–39.6%) for the high-ELNs subgroup, compared with 68.9% (95% CI: 62.4–76.2%), 30.9% (95% CI: 24.6–38.8%), and 19.6% (95% CI: 14.3–26.9%) for the low-ELNs subgroup. After PSM, the matched high-ELNs subgroup exhibited 1-, 3-, and 5-year OS rates of 85.5% (95% CI: 79.7–91.7%), 44.9% (95% CI: 37.0–54.5%), and 31.3% (95% CI: 24.1–40.8%), while the matched low-ELNs subgroup had corresponding rates of 70.6% (95% CI: 63.2–78.9%), 34.0% (95% CI: 26.6–43.4%), and 22.4% (95% CI: 16.0–31.3%) (Figure S19).

Prior to matching, OS was significantly higher in the high-ELNs subgroup than in the low-ELNs subgroup within the SA cohort (P<0.001). The 1-, 3-, and 5‑year OS rates were 69.5% (95% CI: 60.9–79.2%), 35.1% (95% CI: 26.8–46.1%), and 27.5% (95% CI: 19.8–38.1%) for the high-ELNs subgroup, compared with 46.8% (95% CI: 40.4–54.1%), 19.9% (95% CI: 15.1–26.3%), and 12.9% (95% CI: 8.9–18.6%) for the low-ELNs subgroup. After PSM, the survival benefit of the high-ELNs subgroup remained statistically significant (P=0.02). The matched high-ELNs patients had 1-, 3-, and 5year OS rates of 68.6% (95% CI: 59.6–78.9%), 34.7% (95% CI: 26.1–46.2%), and 27.6% (95% CI: 19.6–38.7%), while the matched low-ELNs patients exhibited corresponding rates of 47.6% (95% CI: 38.3–59.2%), 23.8% (95% CI: 16.4–34.6%), and 16.9% (95% CI: 10.7–26.9%) (Figure S20).

Prior to matching, the high-ELNs subgroup demonstrated a significant survival advantage over the low-ELNs subgroup within the PT cohort (P=0.001). The 1-, 3-, and 5-year OS rates were 90.6% (95% CI: 87.0–94.3%), 56.4% (95% CI: 49.8–64.0%), and 42.8% (95% CI: 35.4–51.8%) for the high-ELNs subgroup, compared with 83.4% (95% CI: 79.9–87.1%), 40.8% (95% CI: 35.8–46.4%), and 28.4% (95% CI: 23.8–33.9%) for the low-ELNs subgroup. After PSM, the survival advantage of the high-ELNs subgroup remained statistically significant (P<0.001). The matched high-ELNs patients exhibited 1-, 3-, and 5-year OS rates of 90.4% (95% CI: 86.7–94.3%), 57.3% (95% CI: 50.5–65.0%), and 43.1% (95% CI: 35.5–52.4%), whereas the matched low-ELNs patients showed corresponding rates of 80.5% (95% CI: 75.5–85.8%), 37.2% (95% CI: 30.8–44.9%), and 23.2% (95% CI: 17.6–30.6%) (Figure S21).

To evaluate the robustness of the treatment-specific ELNs thresholds and to minimize the potential confounding effect of inadequate lymph node dissection, we performed a sensitivity analysis. After excluding patients with ELNs <5 from each treatment cohort, we recompared survival differences between high- and low-ELNs subgroups before and after matching. The results demonstrated that in the NS, AT, SA, and PT cohorts, the high-ELNs subgroups consistently exhibited a significant survival advantage both before and after matching (all P<0.05), indicating that the thresholds identified in this study remained stable (Figures S22-S29 and Tables S11-S14).

Variable selection and prognostic model construction for survival prediction

To develop and validate the prediction model, the overall cohort was divided into a training set and an internal validation set at an approximate ratio of 7:3 using stratified random sampling. The baseline demographic and clinical characteristics of patients in the training and validation sets are presented in Table 2. No statistically significant differences were observed in any baseline variables between the training and validation sets (P>0.05), confirming that the two sets were well-balanced.

We applied a structured variable selection process to develop the prognostic model. Candidate variables were first evaluated using univariate Cox regression in the training set. Those with a P value below 0.05, specifically gender, age, T stage, N stage, M stage, Grade, ELNs, and treatment modality, were retained for further analysis. To mitigate overfitting and perform feature selection, LASSO regression was employed. The optimal regularization parameter λ was chosen by minimizing the mean squared error. The trajectory of coefficient shrinkage and the corresponding mean squared error across λ values are presented in Figure 2. LASSO regression selected five variables with nonzero coefficients, namely age, N stage, M stage, ELNs, and treatment modality. These were entered into a multivariate Cox proportional hazards model, where all remained independent predictors with P<0.05 (Table 3). Based on the variable selection results, a total of five independent prognostic factors, including age, N stage, M stage, the number of ELNs, and treatment modality, were incorporated to construct a nomogram for the visual prediction of 1-, 3-, and 5-year OS in patients with EC (Figure 3).

Figure 2 Screening of variables via LASSO regression. (A) Characteristics of coefficient variations across variables. (B) Selection of the optimal regularization parameter λ in the LASSO model using cross-validation. LASSO, least absolute shrinkage and selection operator.

Figure 3 Nomogram for predicting OS in patients with T3–4 EC. AT, adjuvant therapy following surgery; EC, esophageal cancer; ELN, examined lymph node; M, metastasis; N, node; NS, neoadjuvant therapy plus surgery; OS, overall survival; PT, perioperative therapy; SA, surgery alone; T, tumor.

The predictive performance of the nomogram was comprehensively evaluated. Regarding discrimination, the model yielded a C-index of 0.608 in the training set and 0.616 in the validation set. The calibration curves for 1-, 3-, and 5-year survival demonstrated good agreement between the nomogram-predicted probabilities and the actual observed outcomes in both sets (Figure 4). Time-dependent ROC analysis further confirmed its discriminatory ability, with area under the curve (AUC) values for predicting 1-, 3-, and 5-year OS of 0.645, 0.620, and 0.618 in the training set, and 0.658, 0.619, and 0.631 in the validation set, respectively (Figure 5). DCA further supported the favorable clinical utility of the model (Figure 6).

Figure 4 Calibration curves of the nomogram for predicting 1‑, 3‑, and 5‑year OS in the training (A-C) and validation sets (D-F) of T3–4 EC patients. The red dots indicate the concordance between predicted and observed outcomes. The closer the red line is to the gray reference line, the better the calibration performance of the nomogram. EC, esophageal cancer; OS, overall survival.

Figure 5 ROC analysis of the nomogram for predicting 1-, 3-, and 5-year OS in the training (A-C) and validation (D-F) sets of T3–4 EC patients. AUC, area under the curve; CI, confidence interval; EC, esophageal cancer; OS, overall survival; ROC, receiver operating characteristic.

Figure 6 DCA of the nomogram for predicting 1‑, 3‑, and 5‑year OS in the training (A-C) and validation (D-F) sets of T3–4 EC patients. DCA, decision curve analysis; EC, esophageal cancer; OS, overall survival.

Discussion

This study systematically evaluated the impact of different treatment modalities on the prognosis of patients with T3–4 EC and identified subgroup-specific prognostic thresholds for the number of ELNs. The key findings demonstrate that treatment modalities significantly influence survival. NS group showed a clear survival advantage, with OS significantly superior to SA or AT group, while no statistically significant difference was observed compared with PT group. Among all treatment modalities, SA was associated with the worst prognosis. Furthermore, this study is the first to quantify prognostic thresholds of ELNs for each treatment subgroup, with corresponding thresholds of 9, 16, 17, and 19 ELNs for the NS, AT, SA, and PT groups, respectively. Exceeding these thresholds was significantly associated with improved patient survival. Furthermore, a sensitivity analysis was performed by excluding patients with ELNs <5 to account for potential inadequate lymph node dissection, which further confirmed the stability of the ELN thresholds across all treatment subgroups. We further developed and validated an individualized prognostic prediction model that integrates age, N stage, M stage, ELNs, and treatment modality.

Surgical intervention constitutes the fundamental component of managing resectable EC (19). For advanced stages of the disease, given the diversity of available treatment options, systematic evaluation and comparison of their efficacy are crucial for advancing personalized and precision medicine. Our study further demonstrates that patients treated with neoadjuvant, adjuvant, or PT achieve significantly improved OS compared with those undergoing SA. This is consistent with evidence showing that SA offers limited benefit in advanced EC, notably a median survival of only 15–18 months and a 5-year survival rate of 20–25% (20). Furthermore, when surgery is employed as the sole therapeutic approach, rates of both local and systemic recurrence are high, reaching 35% to 50% (21). Currently, neoadjuvant therapy has become the preferred standard strategy for advanced EC (8,9,22). Multiple meta-analyses have demonstrated that neoadjuvant therapy achieves a 5-year OS of approximately 45–50%, which is significantly higher than the 20–30% observed with SA or with surgery followed by adjuvant therapy (23). This is further supported by a network meta-analysis of 33 randomized controlled trials, which identified neoadjuvant chemoradiotherapy followed by surgery as the most effective strategy for improving survival in patients with resectable EC (24). Neoadjuvant therapy contributes to reducing tumor volume, thereby enhancing surgical feasibility (25). Additionally, research by Zhang et al. confirmed that neoadjuvant chemoradiotherapy offers significant advantages in terms of pCR rate, R0 resection rate, and OS (26). Given that neoadjuvant therapy can involve multiple organ systems and patients frequently have comorbidities with risks of nutritional decline, a comprehensive preoperative risk assessment is therefore essential (27). Research by Xiao et al. suggests that neoadjuvant therapy may be associated with higher perioperative mortality following esophagectomy (28). In contrast, a nationwide population-based study from Finland indicated no significantly increased risk of postoperative complications in patients who underwent this treatment (29). This discrepancy in findings likely stems from heterogeneity across studies in patient baseline characteristics, specific treatment protocols, levels of supportive care, and surgical team expertise. The present study indicates that PT did not demonstrate a significant OS advantage compared to neoadjuvant therapy. However, the randomized controlled trial employed a highly standardized FLOT regimen (fluorouracil, leucovorin, oxaliplatin, and docetaxel) as perioperative chemotherapy and demonstrated a survival advantage compared to preoperative chemoradiotherapy (30). In contrast, the SEER database analysis in this study was limited to identifying treatment modalities, without data on the specific protocols, agents, doses, or cycle completion. Furthermore, it lacks in-depth individual-level clinical data, including patient performance status, treatment tolerance, and the contextual background of clinical decision-making. Additionally, while the randomized trial enrolled a broader patient population spanning stages cT1 to cT4a, this study focused specifically on patients with a higher tumor burden in the T3–T4 subgroup. These patients with more advanced disease may have relatively limited tolerance for adjuvant therapy, potentially affecting treatment completion rates (31). The heterogeneity in specific treatment regimens, the absence of key clinical data, and the differences in study population characteristics may be potential reasons for the discrepancy between the findings of this observational study and the conclusions of the referenced randomized controlled trial. Nevertheless, our findings are consistent with other relevant research. A retrospective study reported comparable 2-year recurrence-free survival rates between neoadjuvant and PT groups, at 30.0% and 28.8%, respectively, with no significant difference (32). Additionally, a meta-analysis indicated that adding postoperative radiotherapy following preoperative radiotherapy significantly increases patient mortality (33).

Based on the present study and current evidence, we propose that neoadjuvant therapy combined with surgery should be established as the standard approach for advanced resectable EC, given its significant improvement in long-term survival. In clinical practice, however, treatment decisions should be individualized through multidisciplinary collaboration, with comprehensive assessment of tumor stage, histologic type, and patient performance status, along with enhanced supportive care throughout treatment (34). Furthermore, the decision to add adjuvant therapy should be carefully evaluated based on postoperative pathology, treatment tolerance, and recurrence risk (35). Additional studies are needed to refine patient selection and standardize treatment protocols.

Although the number of ELNs is a significant prognostic factor in EC (14,15), evidence suggests that more extensive lymph node dissection may not consistently improve patient outcomes (36). Our study defined treatment-specific minimum adequate ELNs yields, whereby meeting these thresholds was associated with a significantly improved prognosis. Neoadjuvant therapy not only achieves effective tumor downstaging but also induces significant biological and morphological alterations in regional lymph nodes. Consistent with existing literature and our findings, the lymph node yield is generally significantly lower in patients who undergo neoadjuvant therapy compared to those treated with SA (37,38). The reduction in the threshold for adequate ELNs yield following neoadjuvant therapy is primarily attributable to treatment-induced pathological biological effects. The underlying mechanism involves two interrelated aspects. On one hand, when treatment is effective, the clearance of micrometastatic tumor cells within the lymph nodes leads to fundamental alterations in their biological function and architecture. Consequently, the lymph nodes often undergo atrophic or regressive pathological changes, resulting in an objective decrease in their detectable number. On the other hand, therapy-induced extensive tissue fibrosis affects both the lymph nodes and the surrounding stroma, causing the nodes to adhere to adjacent tissues, harden in texture, and exhibit blurred margins. These changes significantly increase the technical difficulty of intraoperative visual and tactile identification, as well as postoperative pathological dissection of individual lymph nodes (39,40). Notably, studies have shown that the absence or nearabsence of detectable lymph nodes following neoadjuvant therapy may be associated with improved disease-free survival (41). Therefore, we suggest that in the context of neoadjuvant therapy, the evaluation of surgical quality and prognosis should focus more on the pathological response status of lymph nodes, such as the ypN stage, rather than adhering rigidly to the same absolute ELNs count required for SA patients (42). Accordingly, in clinical assessment, if the number of ELNs meets the lower threshold proposed in this study, it may be considered as meeting the adequacy criteria for dissection under this treatment approach. The surgical objective should consequently be to achieve an anatomically standardized lymphadenectomy, while avoiding non-therapeutic extended dissections performed solely to increase ELNs yield. This strategy contributes to optimizing the balance between ensuring radical oncologic efficacy and maintaining surgical safety.

The results of this study indicate that the prognostic thresholds for ELNs in the adjuvant and PT groups were significantly higher than those in the neoadjuvant therapy group. This discrepancy may be attributed to variations in patient baseline characteristics and treatment response. Patients selected for adjuvant or PT often present with a greater baseline tumor burden and higher risk of lymph node metastasis, necessitating more extensive lymph node dissection to achieve accurate pathological staging and guide subsequent treatment (43). Notably, in the PT group, a suboptimal pathological response following neoadjuvant therapy often indicates more aggressive tumor biology, requiring surgery to further intensify local control on top of systemic treatment to achieve maximal tumor cytoreduction (44). Therefore, we argue that the differences in ELNs thresholds between treatment groups reflect the clinical application of individualized surgical strategies, guided by distinct disease stages and therapeutic goals. For patients with a high baseline tumor burden and elevated nodal metastasis risk, surgical efforts should aim to achieve the higher ELNs thresholds established in this study. This approach enables more accurate pathological staging, minimizes the risk of understaging, and provides a valuable quantitative benchmark to inform subsequent adjuvant therapy decisions. The prognostic model combining treatment strategy and ELNs count offers a quantifiable instrument for personalized clinical decision-making in T3–4 EC. By facilitating accurate outcome prediction, it assists in transitioning from experience-based to evidence-based clinical practice, thereby enabling more tailored treatment approaches and refined prognostic assessment.

There are several limitations in this study. First, as a retrospective analysis, although PSM and other methods were employed, they cannot fully eliminate selection bias arising from unmeasured confounding factors, which imposes constraints on the strength of causal inference. Second, the SEER database lacks detailed records of treatment parameters and key pathological response data, such as tumor regression grade (TRG). This limits our ability to conduct an in-depth exploration of the biological relationship between neoadjuvant therapy response and ELNs yield, and it also prevents direct validation of the clinical insights proposed in this study. It is particularly important to note that the ELNs yield thresholds identified through our data-driven approach are exploratory and hypothetical in nature. Although they demonstrated statistical significance within the current cohort, their biological plausibility and suitability as surgical quality metrics require independent validation through well-designed prospective or multicenter studies before they can be cautiously applied in clinical practice. Furthermore, the discriminative performance of the prognostic model developed in this study has room for improvement. Future refinements could incorporate additional prognostic dimensions. Finally, the demographic and histopathological composition of the study cohort may affect the generalizability of the conclusions to other populations, highlighting the need for validation in different epidemiological contexts in the future. To address these limitations, future prospective studies incorporating detailed treatment records, data on TRG and pCR, as well as comprehensive clinicopathological variables are warranted to validate the proposed ELNs yield thresholds and elucidate their underlying biological mechanisms.

Conclusions

This study not only confirms that NS provides significant survival benefits for patients with advanced EC, but also, through systematic comparison, clarifies the survival benefit hierarchy among different treatment strategies. For patients ineligible for NS, AT represents an important alternative strategy superior to SA. The selection of PT should be approached with greater deliberation. Furthermore, by establishing treatment-specific thresholds for ELNs, the study provides an objective and quantifiable benchmark for evaluating surgical quality across different treatment modalities, enabling a more precise balance between therapeutic benefit and surgical risk for individualized surgical planning. The prognostic model integrating treatment strategy and ELNs count provides a quantitative tool for personalized clinical decision-making, facilitating more precise management of EC.

Acknowledgments

None.

Footnote

Reporting Checklist: The authors have completed the TRIPOD reporting checklist. Available at https://jtd.amegroups.com/article/view/10.21037/jtd-2025-1-2675/rc

Peer Review File: Available at https://jtd.amegroups.com/article/view/10.21037/jtd-2025-1-2675/prf

Funding: This study was supported by the Science and Technology Program of the Joint Fund of Scientific Research for the Public Hospitals of the Inner Mongolia Academy of Medical Sciences (Nos. 2023GLLH0123 and 2024GLLH0384), the Inner Mongolia Medical University Joint Project (No. YKD2024LH021), the Beijing University Cancer Hospital Inner Mongolia Hospital “Qingmiao” Talent Plan (Nos. QM202325 and QM202312), the Inner Mongolia Hospital of Peking University Cancer Hospital-Demonstration Project for the Reform and High-quality Development of Public Hospitals (Gastrointestinal Tumors + Thoracic Tumors) Construction of Clinical Key Specialties (No. 12024YNQN013), the Natural Science Foundation of Inner Mongolia Autonomous Region (Nos. 2025LHMS08067 and 2022MS08054), the Peking University Cancer Hospital Inner Mongolia Hospital Public Hospital Reform and High-Quality Development Demonstration Project (Digestive Tract Cancer) Research Fund (Nos. 2023pp004, 2023SGGZ070, 2023SGGZ070-02, and 2023SGGZ070-05).

Conflicts of Interest: All authors have completed the ICMJE uniform disclosure form (available at https://jtd.amegroups.com/article/view/10.21037/jtd-2025-1-2675/coif). The authors have no conflicts of interest to declare.

Ethical Statement: The authors are accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved. The study was conducted in accordance with the Declaration of Helsinki and its subsequent amendments.

Open Access Statement: This is an Open Access article distributed in accordance with the Creative Commons Attribution-NonCommercial-NoDerivs 4.0 International License (CC BY-NC-ND 4.0), which permits the non-commercial replication and distribution of the article with the strict proviso that no changes or edits are made and the original work is properly cited (including links to both the formal publication through the relevant DOI and the license). See: https://creativecommons.org/licenses/by-nc-nd/4.0/.

References

1. Bray F, Laversanne M, Sung H, et al. Global cancer statistics 2022: GLOBOCAN estimates of incidence and mortality worldwide for 36 cancers in 185 countries. CA Cancer J Clin 2024;74:229-63. [Crossref] [PubMed]
2. Qi L, Sun M, Liu W, et al. Global esophageal cancer epidemiology in 2022 and predictions for 2050: A comprehensive analysis and projections based on GLOBOCAN data. Chin Med J (Engl) 2024;137:3108-16. [Crossref] [PubMed]
3. Li Z, Li J, Yu C, et al. The global, regional, and national burden of early, middle, and late-onset esophageal cancer and trends from 1990 to 2021: a systematic analysis of the Global Burden of Disease Study 2021. J Thorac Dis 2025;17:8477-96. [Crossref] [PubMed]
4. Zhao YX, Zhao HP, Zhao MY, et al. Latest insights into the global epidemiological features, screening, early diagnosis and prognosis prediction of esophageal squamous cell carcinoma. World J Gastroenterol 2024;30:2638-56. [Crossref] [PubMed]
5. Xie Z, Lin J, Li Z, et al. Esophageal cancer trends in the US from 1992 to 2019 with projections to 2044 using SEER data. Sci Rep 2025;15:22423. [Crossref] [PubMed]
6. Xie Y, Shi L, He X, et al. Gastrointestinal cancers in China, the USA, and Europe. Gastroenterol Rep (Oxf) 2021;9:91-104. [Crossref] [PubMed]
7. Wang J, Tang L, Lin L, et al. Imaging characteristics of esophageal cancer in multi-slice spiral CT and barium meal radiography and their early diagnostic value. J Gastrointest Oncol 2022;13:49-55. [Crossref] [PubMed]
8. Eyck BM, van Lanschot JJB, Hulshof MCCM, et al. Ten-Year Outcome of Neoadjuvant Chemoradiotherapy Plus Surgery for Esophageal Cancer: The Randomized Controlled CROSS Trial. J Clin Oncol 2021;39:1995-2004. [Crossref] [PubMed]
9. Shah MA, Kennedy EB, Catenacci DV, et al. Treatment of Locally Advanced Esophageal Carcinoma: ASCO Guideline. J Clin Oncol 2020;38:2677-94. [Crossref] [PubMed]
10. Williams KM, Banks KC, Velotta JB. Novel neoadjuvant immunotherapy treatment and surveillance strategies in resectable esophageal cancer: innovation leads to improved outcomes. J Thorac Dis 2025;17:1802-6. [Crossref] [PubMed]
11. Zeng Y, Su X, Zhou T, et al. Propensity-matched study on locally advanced esophageal cancer: surgery versus post-operative radiotherapy. Radiat Oncol 2024;19:130. [Crossref] [PubMed]
12. Lee Y, Samarasinghe Y, Lee MH, et al. Role of Adjuvant Therapy in Esophageal Cancer Patients After Neoadjuvant Therapy and Esophagectomy: A Systematic Review and Meta-analysis. Ann Surg 2022;275:91-8. [Crossref] [PubMed]
13. Jia W, Li C, Liu C, et al. Survival benefit of adjuvant therapy following neoadjuvant therapy in patients with resected esophageal cancer: A retrospective cohort study. PLoS One 2024;19:e0304937. [Crossref] [PubMed]
14. Altorki NK, Zhou XK, Stiles B, et al. Total number of resected lymph nodes predicts survival in esophageal cancer. Ann Surg 2008;248:221-6. [Crossref] [PubMed]
15. Liu Y, Yang H, Fu H, et al. Prognostic impact of examined lymph node count in pT1N0M0 esophageal cancer: A population-based study. Thorac Cancer 2019;10:1636-43. [Crossref] [PubMed]
16. Gaber CE, Sarker J, Abdelaziz AI, et al. Pathologic complete response in patients with esophageal cancer receiving neoadjuvant chemotherapy or chemoradiation: A systematic review and meta-analysis. Cancer Med 2024;13:e7076. [Crossref] [PubMed]
17. Simoni N, Pavarana M, Micera R, et al. Long-Term Outcomes of Induction Chemotherapy Followed by Chemo-Radiotherapy as Intensive Neoadjuvant Protocol in Patients with Esophageal Cancer. Cancers (Basel) 2020;12:3614. [Crossref] [PubMed]
18. Schuring N, van Berge Henegouwen MI, Gisbertz SS. History and evidence for state of the art of lymphadenectomy in esophageal cancer surgery. Dis Esophagus 2024;37:doad065. [Crossref] [PubMed]
19. Acharya R, Mahapatra A, Verma HK, et al. Unveiling Therapeutic Targets for Esophageal Cancer: A Comprehensive Review. Curr Oncol 2023;30:9542-68. [Crossref] [PubMed]
20. Garg PK, Sharma J, Jakhetiya A, et al. Preoperative therapy in locally advanced esophageal cancer. World J Gastroenterol 2016;22:8750-9. [Crossref] [PubMed]
21. Chen G, Wang Z, Liu XY, et al. Recurrence pattern of squamous cell carcinoma in the middle thoracic esophagus after modified Ivor-Lewis esophagectomy. World J Surg 2007;31:1107-14. [Crossref] [PubMed]
22. Yang H, Liu H, Chen Y, et al. Neoadjuvant Chemoradiotherapy Followed by Surgery Versus Surgery Alone for Locally Advanced Squamous Cell Carcinoma of the Esophagus (NEOCRTEC5010): A Phase III Multicenter, Randomized, Open-Label Clinical Trial. J Clin Oncol 2018;36:2796-803. [Crossref] [PubMed]
23. Kamarajah SK, Phillips AW. ASO Author Reflections: Assessing the Impact of Neoadjuvant Therapy: A Real View Perspective. Ann Surg Oncol 2020;27:3193-4. [Crossref] [PubMed]
24. Pasquali S, Yim G, Vohra RS, et al. Survival After Neoadjuvant and Adjuvant Treatments Compared to Surgery Alone for Resectable Esophageal Carcinoma: A Network Meta-analysis. Ann Surg 2017;265:481-91. [Crossref] [PubMed]
25. Hsu PK, Chien LI, Wang LC, et al. Lymphovascular invasion and extracapsular invasion are risk factors for distant recurrence after preoperative chemoradiotherapy and oesophagectomy in patients with oesophageal squamous cell carcinoma. Eur J Cardiothorac Surg 2017;51:1188-94. [Crossref] [PubMed]
26. Zhang Z, Ye J, Li H, et al. Neoadjuvant sintilimab and chemotherapy in patients with resectable esophageal squamous cell carcinoma: A prospective, single-arm, phase 2 trial. Front Immunol 2022;13:1031171. [Crossref] [PubMed]
27. Groenewold MD, Olthof CG, Bosch DJ. Anaesthesia after neoadjuvant chemotherapy, immunotherapy or radiotherapy. BJA Educ 2022;22:12-9. [Crossref] [PubMed]
28. Xiao X, Hong HG, Zeng X, et al. The Efficacy of Neoadjuvant Versus Adjuvant Therapy for Resectable Esophageal Cancer Patients: A Systematic Review and Meta-Analysis. World J Surg 2020;44:4161-74. [Crossref] [PubMed]
29. Sirviö VEJ, Räsänen JV, Helminen O, et al. Neoadjuvant Treatment and Postoperative Complications After Surgery for Esophageal Cancer: A Population-Based, Nationwide Study in Finland. Ann Surg Oncol 2025;32:7914-28. [Crossref] [PubMed]
30. Hoeppner J, Brunner T, Schmoor C, et al. Perioperative Chemotherapy or Preoperative Chemoradiotherapy in Esophageal Cancer. N Engl J Med 2025;392:323-35. [Crossref] [PubMed]
31. Boyle JM, Kuryba A, Cowling TE, et al. Survival outcomes associated with completion of adjuvant oxaliplatin-based chemotherapy for stage III colon cancer: A national population-based study. Int J Cancer 2022;150:335-46. [Crossref] [PubMed]
32. Matsuura N, Yamasaki M, Yamashita K, et al. The role of adjuvant chemotherapy in esophageal cancer patients after neoadjuvant chemotherapy plus surgery. Esophagus 2021;18:559-65. [Crossref] [PubMed]
33. Malthaner RA, Wong RK, Rumble RB, et al. Neoadjuvant or adjuvant therapy for resectable esophageal cancer: a systematic review and meta-analysis. BMC Med 2004;2:35. [Crossref] [PubMed]
34. Villaflor VM, Allaix ME, Minsky B, et al. Multidisciplinary approach for patients with esophageal cancer. World J Gastroenterol 2012;18:6737-46. [Crossref] [PubMed]
35. Tian Q, Lin J, Hu L, et al. Enhancing survival in locally advanced esophageal cancer: a comparative analysis of neoadjuvant immunotherapy versus conventional neoadjuvant therapies using the SEER database. J Thorac Dis 2025;17:2778-801. [Crossref] [PubMed]
36. Lagergren J, Mattsson F, Zylstra J, et al. Extent of Lymphadenectomy and Prognosis After Esophageal Cancer Surgery. JAMA Surg 2016;151:32-9. [Crossref] [PubMed]
37. Kauppila JH, Wahlin K, Lagergren P, et al. Neoadjuvant therapy in relation to lymphadenectomy and resection margins during surgery for oesophageal cancer. Sci Rep 2018;8:446. [Crossref] [PubMed]
38. Bollschweiler E, Besch S, Drebber U, et al. Influence of neoadjuvant chemoradiation on the number and size of analyzed lymph nodes in esophageal cancer. Ann Surg Oncol 2010;17:3187-94. [Crossref] [PubMed]
39. Athauda A, Nankivell M, Langer R, et al. Pathological regression of primary tumour and metastatic lymph nodes following chemotherapy in resectable OG cancer: pooled analysis of two trials. Br J Cancer 2023;128:2036-43. [Crossref] [PubMed]
40. Chen P, Chen M, Bu Y, et al. Prognostic role of lymph node regression in patients with esophageal cancer undergoing neoadjuvant therapy. Pathol Oncol Res 2024;30:1611844. [Crossref] [PubMed]
41. Huang X, Jian Z, You R, et al. Positive Lymph Node Status Before and After Neoadjuvant Chemoradiotherapy Improves Prediction of Disease-Free Survival in Esophageal Squamous Cell Carcinoma Patients. Ann Surg Oncol 2025;32:3147-56. [Crossref] [PubMed]
42. Leone AG, Petrelli F, Klempner SJ, et al. Progression-free survival as a surrogate for overall survival in gastro-esophageal cancer trials with immunotherapy: A meta-analysis. Eur J Cancer 2026;232:116155. [Crossref] [PubMed]
43. Ando N, Kato H, Igaki H, et al. A randomized trial comparing postoperative adjuvant chemotherapy with cisplatin and 5-fluorouracil versus preoperative chemotherapy for localized advanced squamous cell carcinoma of the thoracic esophagus (JCOG9907). Ann Surg Oncol 2012;19:68-74. [Crossref] [PubMed]
44. Nagaki Y, Motoyama S, Sato Y, et al. Patterns and timing of recurrence in esophageal squamous cell carcinoma patients treated with neoadjuvant chemoradiotherapy plus esophagectomy. BMC Cancer 2021;21:1192. [Crossref] [PubMed]