Figure 4 from T-Cell Receptor and Immune Gene Expression Pharmacodynamics for Durvalumab Alone and with Tremelimumab or Bevacizumab in Unresectable Hepatocellular Carcinoma

Gene expression. Peripheral blood gene expression changes across the durvalumab plus bevacizumab, durvalumab monotherapy, T75 + D, STRIDE, and tremelimumab monotherapy arms (A) and changes in gene expression signatures in the Hallmark set (B), the consensus tumor microenvironment set (C), and the Bagaev immune set (D) across the durvalumab plus bevacizumab, durvalumab monotherapy, and STRIDE arms. In panel A, to identify differential gene expression in each trial arm, the DESeq2 capability was applied in R to compare the end of cycle 1 with day 1, while adjusting for baseline participant effects. A Venn diagram comparing differential gene expression per trial arm was produced according to the following criteria: |log2 fold change| ≥ 0.4 and P ≤ 0.05 (P values are unadjusted; no multiple hypothesis testing applied at this stage). In panels B–D, a mixed effects model accounting for random patient effects was applied to the cycle 1 vs. baseline change in signature expression, calculated independently per arm. The values shown in the legends of panels B–D are the estimated slope of the linear regression, comparing the end of cycle 1 with the baseline for each gene expression signature tested. The resulting P values of the test were used to select the signatures illustrated in the heatmap. *, P ≤ 0.05. D, durvalumab monotherapy; T, tremelimumab monotherapy; Treg, regulatory T cell.