Abstract
Ultrasound screening for hepatocellular carcinoma (HCC) in patients with liver cirrhosis has a poor sensitivity for small tumors. Circulating microRNAs (miRNAs) have been explored as HCC biomarkers, but results are diverging. Here, we evaluate if miRNAs up-regulated in HCC tissue can be detected in plasma and used as screening biomarkers for HCC. In this cross-sectional study, plasma, HCC tissue and surrounding non-tumorous liver tissue were collected from liver resections. Tissue miRNAs were identified and quantitated by RNA-sequencing analysis, and the fold-changes between HCC and surrounding liver tissue were calculated. The miRNAs up-regulated in HCCs were then re-analyzed in plasma from the same patients, and the miRNAs with the highest plasma levels were subsequently measured in plasma from an independent cohort of patients with cirrhosis or HCC. In tissues from 84 resected patients, RNA-sequencing detected 197 differentially expressed miRNAs, 40 of which had a raw count above 200 and were analyzed in plasma from the same cohort. Thirty-one miRNAs were selected for further analysis in 200 patients with HCC or cirrhosis. Of these, eleven miRNAs were significantly increased in HCC as compared to cirrhosis patients. Only miR-93-5p and miR-151a-3p were significantly associated with HCC, with an AUC of 0.662. In comparison, alpha-fetoprotein and des-gamma-carboxy prothrombin yielded an AUC of 0.816, which increased to 0.832 if miR-93-5p and miR-151a-3p were added. When including sex and age, the addition of miR-93-5p and miR-151a-3p did not further improve the AUC (from 0.910 to 0.911). In conclusion, micro-RNAs up-regulated in HCCs are detectable in plasma but have a poor performance as screening biomarkers of HCC.