Evaluation of Four Lymph Node Classifications for the Prediction of Survival in Hilar Cholangiocarcinoma

Background An important prognostic indicator of hilar cholangiocarcinoma (HCCA) in patients after surgery is metastasis of lymph nodes (LN). However, there are many types of LN staging systems to the issue of a better determination of the prognosis of patients through the lymphatic staging system which needs research. Based on the above, we tried to re-evaluate the staging system of HCCA LNs. We compared the American Joint Committee on Cancer (AJCC), number of metastatic LNs (MLN), ratio of LN (LNR), and log odds of MLNs (LODDS) in individuals undergoing curative resection to determine the best LN staging system. Methods In the current study, we retrospectively analyzed 229 patients undergoing curative resection. We evaluated the impact of the stage of AJCC pN, LNR, LODDS, and MLN on OS (overall survival) and RFS (recurrence-free survival). According to the curve of receiver operating characteristic (ROC), we compared the predictive capacity of different staging systems of LN for survival and recurrence. Results Multivariate analysis results revealed that LODDS >  − 0.45 (95% CI = 1.115–2.709, P = 0.015; 95% CI = 1.187–2.780, P = 0.006) are independent risk factors affecting OS and RFS, respectively. Compared with LN status, AJCC pN stage, MLN, and LNR, the variable having the highest area under the ROC curve (AUC) was LODDS when predicting 1-year, 3-year, and 5-year OS and RFS. Conclusion This study shows that metastasis of LNs is a key indicator for predicting patient death and recurrence. Among them, LODDS is the best LN staging system for the prognostic evaluation of HCCA patients after surgery. Clinicians can incorporate LODDS into HCCA patient lymphatic staging system for a more accurate prognosis of HCCA patients post-surgery. Supplementary Information The online version contains supplementary material available at 10.1007/s11605-021-05211-x.


Introduction
Hilar cholangiocarcinoma (HCCA) is a highly malignant tumor with an increasing incidence in recent years 1 . The prognosis of HCCA is usually poor, so an accurate prognosis is extremely important. Metastasis to lymph nodes (LN) is a key tumor prognosis indicator 2, 3 . The prognosis of most tumors is directly related to metastasis of the LN and their number. Studies have shown that more than 30% of patients with HCCA who underwent curative resection have LN metastasis [4][5][6] . However, it is still unknown which LN staging system is in use to judge patient prognosis more accurately.
To better describe the status of LNs, some scholars use different methods to classify LN metastases 7,8 . Among Zhi-Peng Liu and Qing-Yi Zhang contributed equal to this work. them, the pN stage as per the American Joint Committee on Cancer (AJCC) staging is based on incremental tumorinfiltrated positive LNs 9,10 . LN ratio (LNR) is defined as the number of positive lymph nodes divided by the total number of lymph nodes examined 8 . Studies have shown that the LNR is superior to the number of metastatic LNs (MLN) for staging in cholangiocarcinoma [11][12][13] . However, LN staging systems such as LNR, MLN, and AJCC pN stage have limitations in predicting the prognosis of patients 14 . In recent years, the log odds of MLNs (LODDS) has been included by clinicians as a new prognostic indicator 15 . LODDS were calculated as the log of [MLN ± 0.5/total number of LNs retrieved (TLN) − MLN + 0.5] 16 . Kwon et al. have confirmed that LODDS is the best LN staging system for judging the prognosis of patients with ampullary adenocarcinoma 17 . Zhou et al. used MLN, LNR, LODDS, and other LN staging systems to assess the prognosis of patients with distant metastatic cancer of the stomach and found that LODDS is the most accurate system of LN staging for assessing the prognosis of patients 18 . Conci et al. evaluated the LN analysis system of 99 HCCA patients and pointed out that LODDS is the best LN staging system for the prediction of the overall survival (OS) time of patients 19 . However, his study did not compare the performance of the LN staging system in assessing HCCA recurrence.
Thus, this study aimed to compare MLN, LNR, AJCC pN stage values, and LODDS in evaluating the OS and recurrence-free survival (RFS) of patients with HCCA who underwent curative resection and finally determine the most suitable LN staging system.

Patient Selection
We retrospectively evaluated the medical records of 229 newly diagnosed HCCA patients who underwent curative resection in Southwest Hospital of China from January 2006 to December 2019. Each patient was confirmed to be HCCA by postoperative pathology. Patients who were found to be unresectable at exploration, had received adjuvant chemotherapy and radiotherapy, and died within 30 days after surgery were excluded. Curative resection was defined as complete resection of all macroscopic and microscopic HCCA tumor with microscopically clear resection margins in the surgical specimens. Regardless of preoperative computed tomography (CT) or magnetic resonance imaging (MRI) or suspicion of lymph node metastasis, patients underwent loco-regional lymphadenectomy. All lymph node tissue underwent pathological biopsy. This research followed the ethical guidelines of the World Medical Association (WMA; Declaration of Helsinki). The approval for the present research was obtained from the Ethics Committee of Southwest Hospital (approval number KY2021129), and all patients obtained informed, written consent for the clinical study.

Outcome Measures
The distribution of tumors along the bile duct during our operation is classified as per Bismuth-Corlette 20 . Classification of the LN location is based on the English version of the 3rd edition of the Japanese Cholangiocarcinoma Classification 21 . As per the AJCC staging system (8th edition), the LN metastasis involved in the common bile duct, cystic duct, hepatic artery, and portal vein is N1; the metastatic LN involved in the mesentery, abdominal cavity, and para-abdominal aorta is N2 9 . LNR is the ratio of the MLN/TLN 11 . Based on the outcomes of earlier studies, patients were divided into intervals of LNR as LNR0, LNR = 0; LNR1, 0.01 < LNR ≤ 0.25; and LNR2, LNR > 0.25 11 . The influence of the number of MLN on the prognosis is ascertained by values for cut-off as 0, 1 ~ 3, and > 3 19 . LODDS is defined as log [(MLN + 0.5) divided by (TLN-MLN + 0.5)] 22 . We defined the intervals of the LODDS LN staging system for HCCA patients according to previous studies but found that the proportion of patients in each interval of LODDS varied remarkably 19 . Therefore, we performed three equal categorizations according to the numerical order of LODDS. That is, all patients were categorized into the following three LODDS intervals: LODDS1, LODDS ≤ − 0.85; LODDS2, LODDS − 0.85 < LODDS ≤ − 0.45; and LODDS3, LODDS > − 0.45.
The focus of this study was OS and RFS. OS was defined from the surgery date to the last follow-up or death due to any cause. The definition of RFS was the time from the surgery date to the diagnosis of recurrence or no disease found at the last follow-up. If the patient did not relapse, the definition of RFS was from the surgery date to the last follow-up or the date of death.

Statistical Analysis
For continuous variables, median and IQR (interquartile range) or the mean and standard deviation were used. Fisher's exact test or the chi-square test was applied for categorical variables, and Mann-Whitney U test was implemented for numerical variables. We used 25 mm and 150 U/L as the cut-off value of tumor size and CA19-9 19,23 . The univariate survival rate was determined using the Kaplan-Meier method, and for comparison, the log-rank test was applied. We used the model of Cox regression for multivariate assessments to ascertain independent prognostic criteria. We analyzed the factors having P < 0.10 in the univariate analysis as covariates in multivariate analysis. In addition, we calculated risk ratios (HRs) and 95% confidence intervals (CIs), and HR < 1.0 indicates a survival benefit. For an accurate evaluation and comparison of the prognostic ability of different methods of LN staging, we used receiver operating characteristics curve (ROC) and area under the ROC curve (AUC) to compare LODDS, LNR, MLN, AJCC pN stage, LN status, and other indicators in predicting the difference in the mortality and recurrence rate of patients at each time point. Statistical analyses were performed using SPSS® version 26.0 (IBM, Armonk, New York, USA).
The CA19-9 level, vascular invasion, method of resection, tumor size, and degree of differentiation were included as independent variables into the multivariate Cox analysis, and different LN staging methods were used to perform multivariate survival analysis on OS and RFS.   (Table 5).  Table 1.

Comparing the Predictive Ability of the LN Staging System
In 1-year OS, the cut-off value of LODDS is − 0.667896. In 3-year OS, the cut-off value of LODDS was − 0.520696. In 5-year OS, the cut-off value of LODDS was − 0.823365. In 1-year RFS, the cut-off value of LODDS was − 0.520696. In 3-year RFS, the cut-off value of LODDS was − 0.667896. In 5-year RFS, the cut-off value of LODDS was − 0.823365.

Discussion
HCCA is the most common type of cholangiocarcinoma. Currently, surgical resection is the only curable method. LN metastasis has been shown to be a key factor in evaluating the prognosis of patients with cholangiocarcinoma undergoing radical surgery 24,25 . The AJCC LN staging, 8th edition, is defined according to the number of LN metastasis areas, without considering the LN metastases number. Many studies believe that to assess cholangiocarcinoma prognosis based on the number of LN metastasis areas is controversial 26,27 . Studies have pointed out that to evaluating the location of LN metastasis is not sufficient. The number of LN metastases also has a direct impact on the prognosis of patients 28 . In this study, we also got similar results. We found that    this study believe that the location and number of LN metastases can affect patients' prognosis. To further explore the effect of the number of LN metastases on the prognosis, we tried to use more methods according to the LN metastases number to evaluate the patient's OS and RFS, such as LNR and LODDS, to estimate the best LN staging system for evaluating the patient's prognosis. Many scholars believe that the sufficient LN examination number is the prerequisite for the accurate number of positive LNs in gastrointestinal tumors [29][30][31][32][33] . A large-scale study collected the clinicopathological data of 20,068 patients with gallbladder cancer, ampullary cancer, and extrahepatic bile ducts from the SEER Cancer Registry. More than 10 LN examinations are needed to accurately stage the LNs 34 . However, it is controversial that expanding the scope of lymph node dissection is beneficial to patients with HCCA. Hakeem et al. indicated that the prognosis of HCCA patients with more than 20 lymphatic examinations was worse 6 . They believed that more lymph node examinations may not improve the prognosis of patients with hilar cholangiocarcinoma after surgery. The more LNs checked, the more traumatic the operation was. Our team believes that the most suitable system for LN staging should be found in the limited number of LN examinations. Moreover, it should be accurate to assess the prognosis of patients and avoid deliberately taking more LNs for examination. The median of the number of LN examinations in this study was only 3 (range, 1-9).
The number of LNs that were positive increases as the number of LN examinations increases. Therefore, it is inaccurate to estimate the prognosis based alone on the number of positive LNs. LODDS and LNR are the systems of LN staging according to the number of positive LNs proposed in recent years. As per various studies, LODDS and LNR are more reliable in assessing the prognosis of gastrointestinal tumors 11,13,15,16,35-37 . In the current study, the number of positive LNs, AJCC LN staging, LODDS, and LNR were used to assess the OS of HCCA patients who underwent radical surgery. Our results indicate that the AUC of LODDS was significantly higher than other LN staging systems when evaluating patients with 1-, 3-, and 5-year OS. This also suggests that LODDS was found to be the best LN staging system for predicting patient mortality.
At present, studies have pointed out that LODDS has better predictive performance in the assessment of gastrointestinal cancer. Zhou et al. analyzed the clinicopathological data of 1999 patients retrospectively with distant metastatic gastric cancer and pointed out that when predicting 1-year cancer-specific survival (CSS) and 2-year CSS, compared with MLN and LNR, LODDS had the highest AUC 18 . In addition, there are also studies comparing different LN analysis systems in evaluating the prognosis of HCCA patients undergoing radical surgery. Conci et al. compared the predictive capacity of LODDS, MLN, and LNR on OS in 99 HCCA patients. They believe that LODDS has the highest AUC when predicting 3-year OS 19 . Therefore, Conci believes that LODDS is the best LN staging system when predicting patient mortality. After analyzing the clinicopathological data of 437 HCCA patients, Bagante et al. concluded that a minimum of four LNs should be assessed for HCCA patients undergoing radical surgery 38 . When the number of ELNs is greater than 4, LODDS shows better prediction performance than the AJCC pN stage for evaluating OS 38 . However, none of the above reports on the prognostic relationship between LODDS and HCCA mentioned the relationship between the system of LN staging and tumor recurrence. We hypothesize that if a different LN staging system is used to evaluate the tumor recurrence of HCCA patients, that is, to evaluate the patient's RFS, whether the same conclusions as the evaluation of OS can be obtained, that is, LODDS is the best LN staging system for preciting the mortality rate of HCCA. Based on this, we used LODDS, LNR, the number of positive LNs, and AJCC LN staging to evaluate the 1-, 3-, and 5-years RFS of HCCA patients and found that the AUC of LODDS was the highest. Therefore, we conclude that LODDS is the best LN staging system for predicting the recurrence rate of HCCA patients.
This study has the following limitations. Due to the retrospective nature, all collected data were biased. This study was like other earlier reported studies conducted at single centers. In addition, the pathological results of each case in this study were examined by the different hepatobiliary pathology team while using the same pathological examination process. At the same time, our study lacked external verification. Some variables that are considered critical to predicting prognosis were missing, such as comorbidity, microvascular invasion, or patients' performance status. However, the results of the current study can improve the debatable issue of the LN staging system in HCCA.

Conclusions
In summary, LN metastasis is a key factor in judging prognosis. LODDS, on the basis of number of positive LNs, seems to be the best LN staging system for predicting the mortality and recurrence rate of HCCA patients after radical resection. In addition, proper LN dissection is of great significance for clarifying tumor staging and selecting treatment methods.