Figure 6

Real-life cohort evaluation of predictive panels and ROC curves for the 3- to 5-marker panels. Experimental validation of the predictive panels in the real-life cohort. Relative expression of candidate markers (a) RNASE2, (b) TNF, (c) MAP3K7, (d) CCNA1, (e) KCNJ2, and (f) DNAJB2 in PBMC samples of patients with HCC, CHB and healthy controls. Green violin plots represent the expression levels of genes in normal patients (n = 24), blue and red violin plots represent the expression levels of genes in patients with HCC (n = 39) and CHB (n = 15), respectively. Relative expression of genes represents as -ΔCt on y-axis. (g) The ROC curve depicts the performance of the three-marker panel comprising RNASE2, TNF-α, and DNAJB2 (red line), as well as the four-marker (yellow line) and five-marker (green line) panels, alongside the best introduced three-marker panel containing RNASE2, TNF-α, and MAP3K7 (blue line) for distinguishing HCC from healthy controls. (h) The ROC curve illustrates the ability to distinguish HCC from CHB, showing the performance of the three-marker panel containing RNASE2, TNF-α, and DNAJB2 (red line), the best three-marker panel introduced, RNASE2, TNF-α, and MAP3K7 (blue line), as well as the four-marker (yellow line) and five-marker (green line) panels. The AUC is displayed in the plot.