Characterization of the Tumor Microenvironment in Osteosarcoma Identifies Prognostic- and Immunotherapy-Relevant Gene Signatures

Figure 5

Construction and validation of TMEscore for predicting clinical outcome. (a) Heatmap of consensus matrix (consensus matrix ). Unsupervised clustering based on 7 genes divided the samples into two categories. (b) Alluvial representation of the relationship between TME subtypes, TMEscore subgroups, and clinical outcomes. (c) KM analysis of high- and low-TMEscore subgroups in TCGA cohort (d) and external validation cohort (GSE21257). (e) KM analysis of high- and low-TMEscore subgroups in all OS cohorts (TCGA and GSE21257). (f) GSVA enrichment analysis showed the activation status of signaling pathways in various TMEscore subgroups. Heatmap visualization of these biological processes. Red indicates activated pathways. Blue indicates suppressed pathways.

(a)

(b)

(c)

(d)

(e)

(f)