Prognostic factors in patients with first diagnosis of hepatocellular carcinoma presenting with pulmonary metastasis and construction of a clinical prediction model.

ABSTRACT

In some areas where routine screening for hepatocellular carcinoma is not available, 30% of HCC patients present with extra-hepatic metastases at the first visit. The most common metastatic organ among them is the lung. The factors influencing the prognosis of this particular subgroup are questions that deserve to be explored. We screened the patients using the SEER database. After exclusion, 989 patients with first diagnosis of hepatocellular carcinoma with lung metastasis were included in this study. Based on Cox regression, the random forest and stepwise methods were applied to screen out risk factors that independently affect the overall survival or disease-specific survival of HCCPM patients and construct prognostic models, respectively. The data were set as training and validation sets, and the reliability and accuracy of the models were verified in different data sets using time-dependent ROC curves with decision curves. We found that the clinical factors affecting the overall survival of HCCPM patients were age grouping, chemotherapy, AJCC T-stage, pathologic grading, and surgery. The clinical factors affecting disease-specific survival in patients with hepatocellular carcinoma pulmonary metastases were age grouping, marital status, AJCC T-stage, pathological grading, and surgery. For the OS model for the training cohort, the 6-month AUC = 0.695, 12-month AUC = 0.692, and 18-month AUC = 0.72. While the DSS model for the training cohort resulted in a 6-month AUC = 0.671, 12-month AUC = 0.671, and 18-month AUC = 0.635. In this study, we developed and validated a model of prognostic risk factors for patients with lung metastases from hepatocellular carcinoma. Our prognostic model can prospectively predict the prognostic status of patients and improve clinical efficiency.