Fig. 2: Survival outcomes according to the microenvironment-based classification (TSM versus HM).

a, b RFS (a) and OS (b) of the study cohort. c OS in the TCGA colon cancer dataset. The tumor-supportive signature genes and tumor-supportive signature score cutoff from the study cohort were applied to the TCGA dataset for validation. d–f Upper bar plots indicate the simulation results of prognostic difference (Cox HR between the two groups (TSM versus. HM)). The groups were divided according to the mean scores of the tumor-supportive signature in the NBT. Tumor-supportive signature gene sets were selected on the basis of the highest HR; when HRs were identical, the set with the fewest genes was chosen. The x axis indicates the number of DEGs (mostly overexpressed in tumors) used for calculation, and the y axis indicates the HR (top) and −log P value (bottom) between the two groups (TSM versus HM). Lower graphs indicate the OS of patients with head and neck squamous cell carcinoma (d), renal cell carcinoma (e) and lung squamous cell carcinoma (f). g, h GSEA for NBT (g) and tumors (h). P value was adjusted using the Benjamini–Hochberg procedure.