Fig. 6

Prediction of multiple treatment responses based on pyroptosis features. A PyroScore increased dependently among the RAI treatment condition groups. B ROC curves of the PyroScore to predict the responsiveness of PTC-RAI patients to RAI and ICI treatments. C, D Distribution differences in objective RAI treatment responses (C) and putative ICI treatment responses derived from TIDE (D) between PyroScore subgroups. E Sankey diagram shows the case flow to various sources or sinks between two types of PRG signatures and RAI and ICI treatment responses. F Comparisons of the scores of TIDE, interferon-gamma (IFNG), microsatellite instability (MSI), Merck18 (T-cell-inflamed signature), CD274, CD8, T-cell dysfunction, T-cell exclusion, myeloid-derived suppressor cells (MDSCs), cancer-associated fibroblasts (CAFs), and M2 tumor-associated macrophages (TAMs) in PyroScore subgroups. G TCIA algorithm predicted the probability of the response to PD1/PDL1 and CTLA4 blockers in the PyroScore subgroups. H, I The IC50 of MTT (H) and CTx agents (I) between PyroScore subgroups. J Top 30 of compounds (from the CTRP v2. database) with the most significant differences in predict IC50 between PyroScore subgroups