Lymph node sampling in resectable hepatocellular carcinoma: national practice patterns and predictors of positive lymph nodes

Nomogram predicting survival in patients with lymph node-negative hepatocellular carcinoma based on the SEER database and external validation

BACKGROUND

The relationship between lymph node (LN) status and survival outcome in hepatocellular carcinoma (HCC) is a highly controversial topic. The aim of this study was to investigate the prognostic factors in patients without LN metastasis (LNM) and to construct a nomogram to predict cancer-specific survival (CSS) in this group of patients.

METHODS

We screened 6840 eligible HCC patients in the Surveillance, Epidemiology and End Results（SEER）database between 2010 and 2019 and randomized them into a training cohort and an internal validation cohort, and recruited 160 patients from Zhongnan Hospital of Wuhan University as an external validation cohort. Independent prognostic factors obtained from univariate and multivariate analysis were used to construct a nomogram prediction model. The concordance index (C-index), area under curve (AUC), calibration plots and decision curve analysis (DCA) were used to assess the predictive power and clinical application of the model.

RESULTS

Univariate and multivariate analysis revealed age, gender, bone metastasis, lung metastasis, AFP, T stage, surgery and chemotherapy as independent prognostic factors. The C-index of the constructed nomogram for the training cohort, internal validation cohort and external validation cohort are 0.746, 0.740, and 0.777, respectively. In the training cohort, the AUC at 1-, 3-, and 5-year were 0.81, 0.800, and 0.800, respectively. Calibration curves showed great agreement between the actual observations and predictions for the three cohorts. The DCA results suggest that the nomogram model has more clinical application potential.

CONCLUSION

We constructed a nomogram to predict CSS in HCC patients without LNM. The model has been internally and externally validated to have excellent predictive performance and can help clinicians determine prognosis and make treatment decisions.

Development and Validation of a Novel Model to Predict Regional Lymph Node Metastasis in Patients With Hepatocellular Carcinoma

Background

The evaluation of the nodal status of hepatocellular carcinoma (HCC) is a classic but controversial topic. This study aimed to investigate the incidence of lymph node metastasis (LNM), explore the role of lymph node dissection (LND), and develop and validate a novel model to predict LNM in patients with HCC, not other specified (NOS).

Methods

The study cohort was taken from the Surveillance, Epidemiology, and End Results database. The annual percent change (APC) was calculated using the Joinpoint regression. Survival analyses adopted the competing risk model. The nomogram was constructed based on the least absolute shrinkage and selection operator (LASSO) logistic regression algorithm and validated by calibration curves. The area under the receiver operating characteristic curve (AUROC) was obtained to compare prognostic performance. Decision curve and clinical impact curve analyses were introduced to examine the clinical value of the models.

Results

A total of 8,829 patients were finally enrolled in this study, and 1,346 (15.2%) patients received LND. The LND rate showed no noticeable fluctuation in the last decade, with an APC of 0.5% (P=0.593). LNM was identified in 56 (4.2%) patients and confirmed an independent prognostic factor of HCC patients (P=0.005). There were 2,497 lymph nodes retrieved, and 93 (3.7%) of them were positive. After propensity score matching, LND indicated no direct oncologic benefit and did not worsen competing risks. Moreover, an increased number of lymph nodes retrieved could not improve prognoses. 1,346 patients with LND were further randomly divided into the training and validation sets with the ratio of 1:1. Race, tumor size, clinical T stage, extrahepatic bile duct invasion, and tumor grade were independent risk factors for LNM. The constructed model was well calibrated and showed good discrimination power and net benefits in clinical practice.

Conclusion

LNM is an independent prognostic factor in HCC, but routine LND seems to be unnecessary in HCC patients. The constructed model could predict the presence of LNM in HCC patients with good performance, which is meaningful to patient stratification and individual treatment strategies optimization.