A novel nomogram for individualized preoperative prediction of lymph node metastasis in patients with intrahepatic cholangiocarcinoma

Abstract

Objective: Constructing and validating a nomogram model for preoperative prediction of intrahepatic cholangiocarcinoma (ICC) lymph node metastasis to assist decision making during surgery. Methods: Retrospectively collecting the clinical and pathological data of 1 031 ICC patients who underwent partial hepatectomy at Eastern Hepatobiliary Surgery Hospital of Naval Military Medical University,General Hospital of Eastern Theater Command,or Zhongda Hospital Southeast University from January 2003 to January 2014. There were 682 males and 349 females; mean age was 54.7 years(range:18 to 82 years). There were 562 patients who underwent lymph node dissection and 469 patients who did not. Among the patients in the dissection group,Lasso regression method was used to filtrate preoperative variables related to lymph node metastasis and establish a nomogram. Bootstrap method was used to internally validate the discrimination of the nomogram,and the accuracy of the nomogram was assessed by using calibration curves. Patients were divided into low-moderate and high-risk groups based on model prediction probability. Propensity score matching(PSM) was used to analyze the overall survival (OS) and recurrence-free survival (RFS) of patients with and without lymph node dissection in the two groups,and to judge the importance of lymph node dissection in the two groups. Results: Six factors related to ICC lymph node metastasis were determined by Lasso regression,including hepatitis B surface antigen,CA19-9,age,lymphadenopathy,carcinoembryo antigen and maximum tumor diameter. These factors were integrated into a nomogram to predict ICC lymph node metastasis. The aera under curve value was 0.764,and the C-index was 0.754. Stratified analysis showed that OS and RFS in the high-risk group of lymph node metastasis were significantly lower than those in the low-medium risk group(median OS:14.6 months vs. 27.0 months,P<0.01; median RFS:9.1 months vs. 15.5 months,P<0.01). In the high-risk group,the median OS was 16.7 months and 6.3 months(Log-rank test: P=0.187;Wilcoxon test:P=0.046),and the median RFS was 11.0 months and 4.8 months(P=0.403),respectively in the lymph node dissection group and undissected group after PSM. In the low-medium-risk group,the median OS was 22.7 months and 26.7 months(P=0.288),and the median RFS was 13.0 months and 14.5 months(P=0.306),respectively in the lymph node dissection group and undissected group after PSM. Conclusions: The nomogram could be used for preoperative prediction of lymph node metastasis and prognostic stratification in patients with ICC. For patients with high risk of lymph node metastasis predicted by the model,active dissection should be performed. For patients predicted to be at low-moderate risk,lymph node dissection might be optional in some specific cases.