Lymphadenectomy after neoadjuvant chemoradiation for esophageal adenocarcinoma: what is the optimal lymph nodes dissection?


 Background

The optimal lymph nodes dissection (LND) for esophageal adenocarcinoma (EAC) patients who underwent neoadjuvant chemoradiotherapy (NCRT) is controversial.
Methods

Patients were selected from Surveillance Epidemiology and End Results database. Multivariable Cox analysis was used to identify predictors of overall survival (OS). Restricted Cubic Splines (RCS) was used to examine the relationship between the number of LND and OS.
Result

2,019 patients with non-metastatic EAC underwent NCRT were stratified into three groups according to LND using X-tile software: group 1: 1–8, group 2: 9–14, group 3: ≥15. In Multivariable Cox Regression analysis, the death risk was reduced by 22% (P = 0.001), 43% (P < 0.001) respectively, for patients in groups 2, 3 compared with those in group 1. The results were similar for patients with pathological lymph node-negative (ypN0) EAC patients. But for pathological lymph node-positive (ypN+) patients, a significantly reduced hazard was present only in group 3 (P < 0.001). RCS exhibited a nonlinear relationship between the number of lymph nodes removed and OS for ypN0 EAC (P = 0.002). The risk of death sharply dropped until around 24 nodes removed and then started to steadily increase afterward. However, for ypN + EAC, it showed a linear relationship between LND and OS (P = 0.205), with a better OS when an increase in the number of lymph nodes removed.
Conclusions

For ypN0 patients, the optimal LND was approximately 24 lymph nodes, with the number of lymph nodes removed beyond 24 nodes did not provide additional benefit. However, for ypN + patients, a more extensive lymphadenectomy could favor survival.


Introduction
Esophageal cancer is still the leading cause of cancer mortality in the world. It is reported that more than 572,000 are newly diagnosed per year, causing more than 508,000 deaths globally.(1) Neoadjuvant chemoradiation followed by surgery become a standard treatment for esophageal cancer, which dramatically improved the 5-year overall survival. (2)(3)(4)(5) However, it remains controversial whether increasing the number of lymph nodes removed could enhance patient survival for EAC patients after NCRT. Although it is recommended to dissect at least 15 lymph nodes during esophagectomy based on National Comprehensive Cancer Network guidelines,(2) the optimum number of lymph nodes harvested is unknown for patients after NCRT.
Previous studies had investigated the association of the number of lymph nodes removed and survival, but the results showed con icting. (6,7) Hanna et al. and Susanna et al. reported that patients with a higher number of lymph nodes removed had better survival. (8,9) However, the study of Talsma et al. enrolled 161 patients with surgery alone and 159 patients with NCRT followed by surgery, demonstrated that the number of lymph nodes removed was a prognostic impact for patients who underwent surgery alone, but not in patients who underwent NCRT.(10) Shridhar et at. identi ed 358 patients had also concluded that the number of lymph nodes removed did not improve survival after NCRT. (11) To date, the number of lymph nodes removed for EAC patients after NCRT has not yet been standardized in the guidelines. Therefore, the focus of this present study based on multi-institution data from the SEER database aimed to evaluate the correlation between lymph node numbers removed and survival in EAC patients suffering NCRT and explore the optimal extent of lymphadenectomy as well.

SEER database and Patients Selection
The SEER database collected cancer incidence data from population-based cancer registries covering approximately 34.6 percent of the United States (http://seer.cancer.gov/). The National Cancer Institute's SEER*Stat software (version 8.3.6) was used for data of patients. We included 2,019 patients with nonmetastatic EAC who underwent NCRT followed by surgery between 1988 and 2016. The covariates were collected as follows: age, gender, primary tumor site, grade, cT stage, cN stage, ypN stage, and LND. The 8th edition of the TNM staging system for esophageal cancer as a reference was used in the study. The exclusion criteria included (1) patients who did not underwent neoadjuvant chemoradiation and esophagectomy (2) patients who had metastatic disease (3) patients with a histological type other than adenocarcinoma. The ICD-O3 code for EAC patients were as followed: adenocarcinoma (8140-8144), adenocarcinoma in an adenomatous polyp (8210), tubular adenocarcinoma (8211), adenocarcinoma with mixed subtypes (8255), papillary adenocarcinoma (8260), mixed cell adenocarcinoma (8323), mucinous adenocarcinoma (8480-8481) (4) patients with unknown data such as age at diagnosis, gender, primary tumor location, cT stage, cN stage, ypN stage, and the number of lymph nodes dissection (5) patients with survival time less than one month after NCRT followed by surgery (Fig. 1). In addition, the endpoint of this study was overall survival (OS), which was calculated as the time from the date of the surgery to the date of death from any cause or the date of the last follow-up.

Statistical Analysis
All statistical calculations were analyzed using SPSS (version 25.0, IBM, Armonk, NY, USA) and R software (version 4.0.2, R Foundation for Statistical Computing, Vienna, Austria). The best cutoff values for LND were calculated using X-tile software (version 3.6.1), which is a valuable tool to generate the optimal cut-point with minimum p values.(12) First, we divided the patients into three groups according to the optimal cut-off value generating by X-tile program (Fig. 2). Then Chi-Square test was utilized to identify differences among these groups. Survival curves were estimated using the Kaplan-Meier method and compared by the log-rank test. In Univariate analysis, all factors with p value < 0.10 were entered into Multivariate Cox Regression analysis to determine independent prognostic factors. Finally, the relationship between LND and OS was evaluated using restricted cubic splines (RCS), which could reveal the true nature and complexity of relationships between continuous variables. (13) In addition, RCS could recognize the hazard function in exion point (ie, threshold). (14) The threshold was considered as the optimal LND with the lowest risk of death. All statistical analyses were two-sided analysis, and signi cance is de ned as P < 0.05.

Patients Characteristics
A total of 2,019 EAC patients who underwent NCRT followed by surgery met inclusion criteria. The median LND was 12 (range 1 to 83). As showed in Table 1, the majority of patients were male (89.1%), cT3-4 stage (72.0%), and cN + stage (64.8%). 93.3% of tumors were located in the lower third of esophagus. Patients were strati ed by LND using X-tile software (group 1: 1-8, group 2: 9-14, group 3: ≥15). There were 696 patients in group 1, 544 patients in group 2, and 779 patients in group 3. The signi cant difference among these groups were tumor site, cT and cN stage: patients with ≥ 15 LNs were more likely to have cT stage (P = 0.03) and cN + stage (P < 0.001).  Figure 3 showed that patients in group 3 had a better OS than in other groups. The 5-year OS were 33.9% in group 1, 42.3% in group 2, and 52.0% in group 3 respectively (P < 0.001).  (Table 3), while only grade and LND (P < 0.05 for both) were the prognostic factors for ypN + EAC (Table 4). In Multivariable Cox Regression analysis, the number of lymph nodes removed was an independent prognosticator of OS for both ypN0 and ypN + patients. For ypN0 EAC patients, the death hazard was reduced by 30% (P < 0.001), 44% (P < 0.001) respectively, for patients in groups 2, 3 compared with those in group 1. Figure 4A showed the 5-year OS were 39.2% in group 1, 51.1% in group 2, 58.6% in group 3 respectively (group1 vs group 2: P < 0.001; group1 vs group 3: P < 0.001; group2 vs group 3: P = 0.094) for ypN0 patients. There was no signi cant difference between group 2 and group 3 on OS. (P = 0.094). However, for ypN + patients, a signi cantly reduced hazard was present only in group 3 (HR = 0.589, P < 0.001). Figure 4B showed the 5-year OS were 18.3% in group 1, 23.0% in group 2, 34.3% in group 3 respectively (group1 vs group 2: P = 0.282; group1 vs group 3: P < 0.001; group2 vs group 3: P < 0.001) for ypN + patients.

The relationship between LND and survival
Restricted cubic spline (RCS) analysis with adjusted for age, gender, tumor site, grade, cT stage was used to classify the association between the number of LND and survival. Figure 5A demonstrated a nonlinear relationship between LND and OS for patients with ypN0 EAC. The death hazard sharply decreased until around 24 lymph nodes removed and then started to steadily increase afterward (P = 0.002 for nonlinearity). The lowest risk of death was approximately 24 lymph nodes removed. However, for ypN + patients, it presented a linear relationship between LND and OS (P = 0.205 for non-linearity), with a better OS when an increase in the number of lymph nodes removed (Fig. 5B). But the incremental reduction in the probability of death was not evident when LND was over 24 nodes.

Discussion
Nowadays, neoadjuvant chemoradiotherapy followed by surgery has been regarded as a standard treatment for EAC. (3, 4, 15) It reported that approximately 20-51% of EAC patients could achieve pathological complete response (pCR) after NCRT. (16,17) Based on the frequency of pCR, we hypothesize that it is unnecessary to extend lymph node dissection for patients who underwent NCRT.
Although previous several studies have explored the effect of the number of lymph nodes removed for prognosis in patients with neodjuvant therapy. It remains controversial the optimal LND for patients after NCRT. (8)(9)(10)(11)(18)(19)(20)(21) Furthermore, few studies have used RCS to explore the linear relationship between LND and survival. Therefore, in this study, we aimed to investigate the survival effect of total number of lymph nodes removed and further determined the optimal LND in EAC patients receiving NCRT.
In our study, we based on a large sample analysis from the SEER database to examine the relationship between LND and survival. The median number of lymph nodes removed in our study was 12 (range 1 to 83). Patients were strati ed by LND using X-tile software ( reported that patients with a higher number of lymph nodes removed had better survival. (8,9) In addition, a larger cohort of 2698 patients from the Netherlands Cancer Registry, Visser et at. demonstrated that the number of lymph nodes removed was independently associated with survival following trimodality therapy.(21) These data and our study result, support that lower lymph nodes removed may result in worse survival. One possible explanation is that when the number of lymph nodes removed was not enough, potential metastatic lymph nodes may be undetected, resulting in poor prognosis. It reported that micro-metastases had been found in up to 50% of patients with histology-negative nodes. (22) Extensive the number of LND that may eliminate occult metastasis of lymph nodes, decrease local recurrence rates, and then improved survival. (23)(24)(25) Could a more extensive lymphadenectomy improve survival for patients with ypN0 stage after NCRT? At present, there is a lack of study to discuss this issue deeply. Therefore, in this study, strati cation analyses were conducted on OS according to the ypN stage. Interestingly, we found that for ypN0 EAC patients, the death hazard decreased by 30% (P < 0.001), 44% (P < 0.001) respectively, for patients in groups 2, 3 compared with those in group 1. The 5-year OS were 39.2%, 51.1%, 58.6% in group 1, 2, and 3 respectively (group1 vs group 2: P < 0.001; group1 vs group 3: P < 0.001; group2 vs group 3: P = 0.094). There was no signi cant difference between group 2 and group 3 on OS (P = 0.094 (21) Our results were similar to their research. In our study, for ypN + patients, a signi cantly reduced hazard was present in group 3 (HR = 0.589, P < 0.001). The 5-year OS were 18.3%, 23.0%, 34.3% in group 1, 2, and 3 respectively (group1 vs group 2: P < 0.001; group1 vs group 3: P < 0.001; group2 vs group 3: P < 0.001), which demonstrated less than 14 nodes removed were not enough, more lymph nodes removed were needed to improve survival.
What is the optimal lymph node dissection? National Comprehensive Cancer Network guidelines recommend at least 15 lymph nodes should be dissected during esophagectomy.(2) But for patients after NCRT, the optimal LND is unknown. In our study, we used the restricted cubic spline to determine the optimal number of lymph nodes removed. Interestingly, we found a nonlinear relationship between LND and OS for patients with ypN0 EAC. The death hazard sharply decreased until around 24 lymph nodes removed and then started to steadily increasing afterward (P = 0.002 for non-linearity The underlying mechanism is unknown. The ypN + was a signi cant adverse prognostic factor that has been con rmed in several studies. (28,29) Extensive lymph node dissection allows for more complete removal of positive lymph nodes and avoidance of potential metastatic lymph nodes. In our analysis, the optimal number of lymph nodes removed was not given a speci c value for ypN + patients, but our result con rmed that more extensive lymph node dissection was necessary and at least 24 lymph nodes could favor survival for ypN + patients.
Our study represents one of the largest analyses of EAC patients to evaluate the relationship between the number of LND and survival using the SEER database, which may reduce the potential for selection biases. Furthermore, our research only included adenocarcinomas to reduce the confounding effect of other histologies on our results. However, there are several limitations in our study. First of all, this study is a retrospective study with an underlying weakness. Secondly, the SEER database did not provide speci c description on radiation methods, chemotherapy regimen and esophagectomy skills. Further prospective studies are needed to validate these results.

Conclusion
In conclusion, the extent of lymph nodes resected was an independent prognostic factor for EAC patients after NCRT. The optimal LND is maybe approximately 24 lymph nodes for ypN0 patients, with the number of lymph nodes removed beyond 24 nodes did not provide additional bene t. However, for ypN + patients, a more extensive lymphadenectomy could favor survival. These results need con rmation in further research.

Declarations
Ethics statement This retrospective study analyzed data from the publicly accessible SEER database. Before the study, we obtained an o cial permit for the research purpose (ID: 19895-Nov2019). Informed consent or ethical approval was not required for this study. Patients included in and excluded from study.  The overall survival with strati cation for the number of lymph nodes dissection in whole cohort.

Figure 4
The overall survival with strati cation for the number of lymph nodes dissection in ypN0 (A) and ypN+(B) patients.