Identification of CCL20 and LCN2 as Efficient Serological Tools for Detection of Hepatocellular Carcinoma

Abstract

Objectives: To discover a more powerful diagnostic tool for the detection of hepatocellular carcinoma (HCC).

Methods: 16 extracellularly located candidates were selected by analyzing the expression array datasets in GEO. 10 of them were validated in clinical samples by ELISA. Differences of each variable were compared by one-way ANOVA or Kruskal-Wallis test. CCL20 and LCN2 were determined in all samples (HCC, 167; liver cirrhosis, 106; and healthy control, 106) and finally chosen for the construction of the combination model by binary logistic regression. The models were first built using a comprehensive control, including both liver cirrhosis (LC) and healthy donors. Then, the models were rebuilt by using the LC group alone as a control. ROC analysis was performed to compare the diagnostic efficiency of each indicator.

Results: Levels of CCL20 and LCN2 in HCC sera were significantly higher than those in all controls. Using the comprehensive control, ROC curves showed that the optimum diagnostic cutoff of the CCL20 and LCN2 combination was 0.443 (area under curve (AUC) of 0.927 (95% CI 0.896-0.951), sensitivity of 0.808, specificity of 0.892, and accuracy of 0.859). For detection of HCC from LC control, the optimum diagnostic cutoff was 0.590 (AUC of 0.919 (95% CI 0.880-0.948), sensitivity of 0.814, specificity of 0.868, and accuracy of 0.834). Furthermore, the model maintained diagnostic accuracy for patients with HCC in the early stage, with the sensitivity and specificity of 0.75 and 0.77 from LC control, yet the AFP only reached 0.5 and 0.67, respectively.

Conclusion: A combination model composed of CCL20 and LCN2 may serve as a more efficient tool for distinguishing HCC from nonmalignant liver diseases.