Abstract
PURPOSE: Liver metastasis is the final stage of cancer progression and is associated with poor prognosis. Although numerous indicators have been identified as having prognostic value for lung cancer and liver metastasis, liver metastases are still not diagnosed by imaging in many patients. To provide a more accurate method for clinical prediction of liver metastasis, we analyzed multiple factors to identify potential predictive factors for liver metastasis of lung cancer. METHODS: Patients first diagnosed with lung cancer between 2002 and 2016 (n = 1746) were divided into two groups, with and without liver metastasis. Serum concentrations of calcium, carcinoembryonic antigen (CEA), cancer antigen-125 (CA125), cancer antigen-153 (CA153), carbohydrate antigen-199 (CA199), cytokeratin fraction 21-1 (CYFRA21-1), total prostate-specific antigen (TPSA), and neuron-specific enolase (NSE) were analyzed in both patient groups. RESULTS: There was no significant difference in age or sex between the two groups. CA125 and NSE were significantly associated with liver metastasis. Compared with CA125, NSE was more specific, while it was less sensitive (P < 0.001). Further analysis of NSE concentrations was conducted in patients with non-small-cell lung cancer and indicated that NSE concentration differed significantly between those with and without liver metastasis (P = 0.023). We conducted analysis with NSE and CA125 combined, resulting in acceptable sensitivity (51.2%), specificity (72.6%), and area under the curve (0.64) values; sensitivity and area under the curve values were higher than those for individual factors, while specificity was higher than that for CA125. CONCLUSIONS: The combination of CA125 and NSE can assist prediction of liver metastasis of lung cancer, providing improved diagnostic accuracy.