Proposal of the optimal numbers of examined and positive lymph nodes to the 8th edition of American Joint Committee on Cancer (AJCC) staging for 758 patients with distal cholangiocarcinoma

Introduction The American Joint Committee on Cancer (AJCC) recommended retrieval of at least 12 lymph nodes and firstly classified N category by the number of positive lymph nodes (PLNs) for Distal Cholangiocarcinoma (DCC). Objective The end of this cohort study was to explore the optimal cut-off values of the number of examined lymph nodes (ELNs) and PLNs to better stratify patients by utilizing a population-based database. Methods A number of 758 patients with DCC from the Surveillance, Epidemiology, and End Results (SEER) database were enrolled in the study and comparing by the survival analysis. Results Survival analysis found that patients with ELNs < 5 had a lower 3-year disease-specific survival rate than ELNs ≥ 5 in N0M0 cohort (35.3% vs. 53.0%, P = 0.001) and in M0 cohort (42.7% vs. 32.8%, P = 0.006); survival curves between patients with ELNs < 12 and ELNs ≥ 12 were overlapped in N0M0 cohort (P = 0.256) and in M0 cohort (P = 0.233). Among patients with ELNs ≥ 5, using the optimal cut-off value of the number of PLNs (0, 2) could accurately stratify patients, but the recommendation of the number of PLNs (0, 3) by the AJCC could not. Conclusions This study recommended examining at least 5 lymph nodes and defining PLNs = 1–2 as the N1 category and PLNs ≥ 3 as the N2 category, which may better stratify distal cholangiocarcinoma patients and improve the accuracy of the eighth edition AJCC staging.


Introduction
Distal Cholangiocarcinoma (DCC) was an epithelial cell malignancy arising from the origin of the cystic duct to ampulla of vater [1], accounting for approximately 41.0-44.1% of cholangiocarcinoma [2][3][4]. And its incidence has gradually increased in recent year [5]. Although pancreaticoduodenectomy was the only potential treatment that can cure DCC patients with the resectable disease [6], the prognosis of patients with DCC after surgical resection was still poor, and the 5-year overall survival rates varied from .5% [7][8][9][10].
Adequate lymph node dissection could not only reduce the risk of tumor recurrence, but also improve the accuracy of staging. The number of lymph nodes retrieved could provide additional prognostic value [11]. Several articles reported lymph node status was one of the most important independent predictors for DCC patients with the resectable disease [2,12]. Thus, American Joint Committee on Cancer (AJCC) recommended that at least 12 lymph nodes be retrieved, whereas the minimum number of examined lymph nodes examined (ELNs) has not yet been determined [13]. And eighth edition AJCC cancer staging manual firstly defined positive lymph nodes (PLNs) = 1-3 as N1 category and PLNs � 4 as N2 category [13], but there was some controversy about the effect of the number of PLNs on the prognosis of DCC patients [10,[14][15][16]. The end of this study was to explore the optimal cut-off values of the number of ELNs and PLNs to better stratify patients by utilizing a population-based database.

Database and patients
The data in this study was obtained from the Surveillance, Epidemiology, and End Results (SEER) database, which was one of the most representative large tumor databases in North America.
A number of 817 DCC patients who underwent primary surgical treatment from January 2010 to December 2015 were identified from the SEER � Stat 8.3.6 Software (https://seer.cancer. gov/data/) (S1 Table). The selected database was the Incidence-SEER 18 Regs Custom Data (with additional treatment fields), Nov 2018 Sub (1975-2016 varying). A total of 59 patients whose TNM stage (7th AJCC) were unknown and those who died within 1 month were excluded. Eventually, 758 DCC patients were enrolled in this study.
The data was including the following: (i) general demographic information: gender, age, race; (ii) clinicopathologic information: tumor size, tumor grade, the 7th AJCC stage, lymphadenectomy (number of lymph nodes removed � 1), number of ELNs and PLNs, adjuvant radiotherapy, adjuvant chemotherapy, and follow-up data.
diagnosis to the death of DCC or the last follow-up (November 2018). Survival analysis was comparing by the Kaplan-Meier method (log-rank test). Univariate and multivariate analysis was comparing by Cox proportional hazard regression model, and all variables related to survival (P < 0.2) in univariate analysis were enrolled in multivariate analysis. The X-tile software (https://medicine.yale.edu/lab/rimm/research/software/) was used to determine the optimal cut-off value for the number of ELNs and PLNs. Survival analysis was used for comparing the prognostic stratification ability of the optimal cut-off values of the number of ELNs and PLNs and the recommendation of ELNs (12) and PLNs (0, 3) by the AJCC. A P < 0.05 was considered statistically significant. Statistical analysis was using STATA 16.0 software.

Prognostic factors
In univariate analysis, tumor size, tumor grade, 7th AJCC T stage, lymph node metastasis, and distant metastasis were associated with prognosis (all P < 0.005,

Exploration of the best cut-off value of number of examined lymph nodes
Among 682 patients with ELNs � 1 in M0 cohort, the best cut-off value of ELNs by X-tile software was 5. Survival analysis found that patients with ELNs < 5 had a lower 3-year DSS rate than those with ELNs � 5 in M0 cohort (32.8% vs. 42.7%, P = 0.006, Fig 1A); survival curves between patients with ELNs < 12 and ELNs � 12 were overlapped in M0 cohort (P = 0.233, Fig 1B). Besides, among 332 patients with ELNs � 1 in N0M0 cohort, the best cut-off value of ELNs by X-tile software was also 5. Survival analysis found that patients with ELNs < 5 had a lower 3-year DSS rate than those with ELNs � 5 in N0M0 cohort (35.3% vs. 53.0%, P = 0.001, Fig  1C); Survival curves between patients with ELNs < 12 and ELNs � 12 were overlapped in N0M0 cohort (P = 0.256, Fig 1D).

Exploration of the best cut-off value of the number of positive lymph nodes
The optimal cut-off values of the number of PLNs obtained by X-tile software were all 0 and 2 in three cohorts (682 M0 patients with ELNs � 1, 558 patients with ELNs � 5 and 371 patients

PLOS ONE
The optimal numbers of examined and positive lymph nodes for patients with distal cholangiocarcinoma with ELNs � 12). The three cohorts of patients were grouped according to the optimal cut-off value of PLNs (0, 2) and the recommendation of PLNs by the AJCC (0, 3), respectively. The DSS rates of the grouped patients were compared by the survival analysis.

Discussion
Adequate lymph node dissection could not only improve the prognosis of patients with DCC, but also better stratify the patients. The AJCC recommended that at least 12 lymph nodes be retrieved, whereas the minimum ELNs have not yet been determined [13]. In addition, the 8th AJCC cancer staging manual firstly proposed optimal cut-off value for the number of PLNs of 0 and 3, and it defined N stage as N0, N1, and N2 categories [13]. However, there was some controversy about the effect of number of PLNs on the prognosis for postoperative patients with DCC [10,[14][15][16]. The aim of this study was to investigate the impact of ELNs and PLNs on the prognosis of DCC patients with resectable disease.
This study recommended examining at least 5 lymph nodes. Although the lymph node dissection was a routine procedure in DCC patients in the European Society for Medical Oncology (ESMO) Clinical Practice Guidelines [17], there was currently some controversy regarding the minimum ELNs [12,13,18,19]. The results of this study found that whether in the N0M0 cohort or the M0 cohort, examining at least 5 lymph nodes could improve the prognosis of patients with DCC (Fig 1A and 1C), but whether the number of ELNs was greater than 12 had nothing to do with the prognosis (Fig 1B and 1D). Furthermore, in patients with ELNs � 12, neither the optimal cut-off value of the number of PLNs (0, 2), nor the recommendation of PLNs by AJCC (0, 3) failed to efficiently stratify patients (Fig 2E and 2F). In contrast, among patients with ELNs � 5, using the optimal cut-off value of the number of PLNs (0, 2) could accurately stratify patients (Fig 2C). Several studies suggested a minimum ELNs of 12 for DCC patients, but they were small in sample size [12,20]. Moreover, a large sample study from the

PLOS ONE
The optimal numbers of examined and positive lymph nodes for patients with distal cholangiocarcinoma SEER database found that the prognosis of DCC patients with ELNs = 4-9 was better than other patients [11]. But these studies did not consider the effect of the number of ELNs on the number of PLNs [11,12,20]. In summary, this study recommended examining at least 5 lymph nodes to better stratify patients and improve the accuracy of the N staging.
This study suggested that defining PLNs = 1-2 as N1 category may improve the prognostic stratification of patients in the eighth edition of AJCC staging. Nearly half of the patients in this study (49.7%) had lymph node metastasis, which was within the range previously found (44.4-58%) [15,16]. Lymph node metastasis was significantly associated with poor prognosis, consistent with previous findings [2,18,21]. The DCC patients with PLNs were prone to intrahepatic metastases and relapse [22,23], which will further lead to poor prognosis. Kiriyama M et al. [14] found that the number of PLNs was a strong predictor of survival for patients with DCC. Thus, the 8th AJCC staging defined PLNs = 1-3 as N1 category and � 4 as N2 category [13]. Nonetheless, there was some controversy about the effect of the number of PLNs on the prognosis of patients. Firstly, in Kiriyama M et al.'s paper [14], patients with PLNs � 4 were only 10% (37/370), and the patients from Japan was selected, which may not be suitable for European and American countries. Secondly, two large sample studies found no difference in prognosis between patients with PLNs � 4 and 1-3 [10,24], which was consistent with the results of this study. Thirdly, we found that among patients with ELNs � 5, using the optimal cut-off value of PLNs (0, 2) could accurately stratify patients (Fig 2C), but the recommendation of the number of PLNs by the AJCC (0, 3) could not ( Fig 2D). Fourthly, previous articles also partly supported the results of this study, and reported that patients with PLNs � 3 were associated with poor prognosis [15,16]. Hence, this study supported defining PLNs = 1-2 as N1 category.
This study also had several limits. Firstly, this study was limited due to its retrospective nature, and it was necessary to validate our results in large sample prospective studies. Secondly, several clinical features that have a significant impact on prognosis were lacking due to the SEER database, such as surgical margins, tumor recurrence, detailed plans for chemotherapy and radiotherapy, and complications. Thirdly, the number of lymph nodes retrieved may depend on the lymph node dissection technique of each institution, and the hospital type was not clear. Fourthly, there may be some errors in the data from the SEER database. But the SEER database, as a large national database, provided extensive universality for the results of the study.

Conclusions
This study recommended examining at least 5 lymph nodes and defining PLNs = 1-2 as the N1 category and PLNs � 3 as the N2 category,which may better stratify DCC patients and improve the accuracy of the eighth edition AJCC staging.
Supporting information S1