Abstract
Objectives: The clinical usefulness of tumor markers alpha-fetoprotein (AFP) and des-gamma carboxyprothrombin (DCP) in the early detection of hepatocellular carcinoma (HCC) in patients with liver cirrhosis (LC), including those with marker decline after antiviral therapy, is limited. MicroRNAs (miRNAs) are expected to complement detection; however, their details remain unknown. Our prospective pilot study aimed to improve the surveillance of HCC high-risk LC patients by propensity scoring with tumor markers and additional predictors. Methods: Tumor markers and plasma levels of cytokines and miRNAs were observationally measured and statistically evaluated with propensity scoring in 85 eligible patients: 43 with current HCC (cHCC) including 8 with early-HCC, 22 with previous HCC cured (pHCC), and 20 with intact LC (iLC). Results: The analysis of the area under the receiver operating characteristic curve (AUC) showed that the best single predictor was AFP (0.794 for cHCC-discrimination and 0.771 for pHCC-discrimination). AFP-DCP integrated with miR-21-5p for cHCC-discrimination was 0.896; with IL-10 for pHCC-discrimination was 0.872, these were significantly better than those of AFP alone, independently (P < .01). The best single predictor for iLC-discrimination was IL-17 level (0.756). IL-17 integrated with AFP-DCP was 0.882, which was significantly better than that of IL-17 alone (P < .01). The positive likelihood ratio (pLR) for cHCC-discrimination by integration of AFP-DCP and miR-21-5p was 32.2. Preliminary validation analysis of early-HCCs compared to conventional AFP and DCP showed the combinations of AFP-DCP and 3 integrated predictors, miR-21-5p for cHCC-discrimination, IL-10 for pHCC-discrimination, and IL-17 for iLC-discrimination, sensitivity, specificity, and pLR, improved from 37.5% to 62.5%, 55.8% to 83.1%, and 0.85 to 3.70, respectively. Conclusion: The predictors of AFP-DCP combined with iR-21-5p, IL-10, and IL-17 by propensity scoring achieved higher discrimination of cHCCs, pHCCs, and iLCs, may be beneficial for the surveillance of early-HCCs, improving prediction of early-HCCs over conventional methods. However, further validation is required.