Abstract
PURPOSE: Hepatocellular carcinoma (HCC), a prevalent type of liver cancer, is mainly diagnosed in the advanced stage, leading to a high mortality rate. Recent advances have identified peripheral cytokines as a potential tool to predict disease outcomes and inform therapeutic decisions. Hence, in this study, we aim to build a predictive model for HCC based on serum levels of different cytokines. PATIENTS AND METHODS: We used immunoassay to quantify the concentrations of IL-27, MIP-1β, Perforin, sCD137, sFas, and TNF-α in the serum of 38 HCC patients and 15 healthy controls. Logistic regression was then used to construct classification models detecting HCC based on these cytokines. A nomogram of the best-performing model was generated to visualize HCC prediction. RESULTS: sFas and MIP-1β were found to be significantly higher in HCC patients compared to controls. Predictive models based on cytokine levels combining sFas, sCD137, and IL-27 performed the best in distinguishing HCC patients from healthy controls. This model has a bias-corrected area under the receiver operating characteristic (ROC) curve (AUC) of 0.948, a sensitivity of 92.11%, a specificity of 93.33%, and an accuracy of 0.925. CONCLUSION: Our findings suggest that serum cytokines have the potential to be utilized in HCC screening to improve detection rates.