The abilities of quantitative analyses of 7 genes hypermethylation in serum DNA, α-fetoprotein (AFP) and prothrombin-induced vitamin K absence II (PIVKA-II), and various combinations to detect HCC were evaluated in a training cohort of 164 HCV-infected patients (108 HCCs; 56 non-HCCs). An optimal hybrid detector, built using data for 2 methylated genes (SPINT2 and SRD5A2), AFP, and PIVKA-II, achieved the most satisfactory ability to detect HCC in the training cohort. We evaluated the ability of the optimal hybrid detector to detect HCC in an independent validation cohort of 258 consecutive HCV-infected patients (112 HCCs; 146 non-HCCs) who were newly enrolled in 4 distinct institutes.
In the validation cohort of 258 patients, accuracy, sensitivity, and specificity of the hybrid detector for detection of HCC were 81.4 % , 73.2 % , and 87.7 % , respectively. Notably, even when detecting HCC ≤ 2 cm in diameter, the hybrid detector maintained markedly high abilities (84.6 % accuracy, 72.2 % sensitivity, 87.7 % specificity). Youden's index (sensitivity + specificity − 1) for HCC ≤ 2 cm was 0.60, vastly much superior to the 0.39 for AFP at a cut-off value of 20 ng/ml and the 0.28 for PIVKA-II at a cut-off value of 40 mAU/ml.
These results show that the optimal hybrid blood detector can detect HCV-related HCC more accurately.