Prognostic marker candidates for cancer immunotherapy in the training data set of glioblastoma multiform (GBM).

(A) Random survival forest analysis for cancer immunotherapy-related genes in GBM. Top 20 of variable importance was shown. (B) Prognosis prediction formula for cancer immunotherapy-related genes in GBM. The Z1 score is calculated by the expression values of 16 immunosuppressive pathway genes. (C,D) Kaplan-Meier survival analysis using the Z1 score = 4.11. (C) Overall survival analysis (OS). (D) Disease free survival analysis. HR, hazard ratio. (E-H) Kaplan-Meier survival analysis for representative immunosuppressive pathway genes in GBM. (E) CD276/B7-H3. (F) TIGIT. (G) VTCN1. (H) CD86. Numbers in the parentheses indicate the threshold of gene expression. High and low indicate subgroups with over and under the threshold. OS, overall survival. HR, hazard ratio. Subgroups were divided by the median score of Z1 and the median expression of genes.