Fig. 1

Three distinct gene sets are associated with HCC disease stage and immune environment. a Three distinct gene clusters whose expression was associated with different disease stages. Clusters were identified by analysis of variance hypothesis testing. Clustering was done for all patients based on all 15,524 genes (transcriptome), with each patient profile (column) labeled with tissue classification (cirrhosis or HCC tumor) and tumor stages (T1–T3). b The gene signature (average of z-scores of genes in each cluster) shows a clear association between stage and signature. Top panel: Cluster 1 genes show a monotonically decreasing trend with progressive disease stages. Middle panel: expression of genes in Cluster 2 was higher in T1 than in cirrhotic tissues but decreased with more advanced stages of HCC. Bottom panel: Cluster 3 genes show a steady increase in expression with progressive disease stages. c Geneset enrichment analysis (GSEA) using Hallmark Gene Sets in Molecular Signature Database was used to identify the top-scoring genes and pathways in Clusters 1 and 3. Cluster 1 was overrepresented by Hallmark gene sets describing immune response while Cluster 3 was dominated by Hallmark gene sets associated with WNT pathway activation. None of the Hallmark gene sets were significantly enriched in Cluster 2 (not shown). d Aggregated expression of genes describing T effector signature (T-eff), including GZMA, GZMB, PRF1, EOMES, and CD8A, was significantly correlated with Cluster 1, p-value < 0.001. e Correlation between T-eff gene signature (distilled from RNA-Seq dataset) and CD8A expression (measured by immune histopathology). Strong directional relationship can be detected between the two assessments