Nomogram for predicting the prognosis of patients with hepatocellular carcinoma presenting with pulmonary metastasis

Abstract

Background: Only a few studies have investigated the clinical features and outcomes of patients with pulmonary metastasis from hepatocellular carcinoma (HCC) at the initial diagnosis. This study aimed to evaluate the prevalence, risk factors and prognostic factors related to pulmonary metastasis and then construct a nomogram to predict the outcomes of patients with HCC presenting with pulmonary metastasis. Methods: The Surveillance, Epidemiology, and End Results (SEER) database was used to select patients. A total of 25,236 eligible patients diagnosed with HCC from 2010 to 2015 were selected. Then, 897 patients with HCC presenting with pulmonary metastasis at the initial diagnosis were included in the primary set (n=598) and validation set (n=299). Logistic and Cox regression analyses were used to determine the risk factors and prognostic factors for pulmonary metastasis. A nomogram predicting the prognosis of patients with HCC presenting with pulmonary metastasis was constructed based on independent prognostic factors identified in Cox regression analyses. Both internal and external validations of the nomogram were performed using discrimination and calibration plots. Results: The prevalence of pulmonary metastasis was 3.6% (897/25,236) in the entire cohort diagnosed with HCC as the initial diagnosis. Age, race, Edmonson-Steiner classification grade I/III, higher T stage, N stage, alpha fetoprotein(AFP) levels, brain metastasis, bone metastasis and intrahepatic metastasis were positively correlated with the development of HCC with pulmonary metastasis at the initial diagnosis. Prognostic factors incorporated in the nomogram were sex, T stage, bone metastasis, AFP levels, treatment, radiation and chemotherapy. The concordance index (C-index) of the nomogram in the primary set was 0.661 (95% CI: 0.633–0.688), indicating considerable predictive accuracy. The calibration curves showed consistency between the nomogram and the actual observations. When the nomogram was applied to the validation set, the results also remained reconcilable, and the C-index of the nomogram was 0.657 (95% CI: 0.626–0.698). Conclusion: A list of risk factors associated with pulmonary metastasis occurrence in patients with HCC was selected, and the nomogram accurately predicted the prognosis of patients with HCC presenting with pulmonary metastasis at the initial diagnosis.