Figure 5.

The value of the IGLoS score in predicting the response to ICI in pan‐cancer immunotherapy cohorts. A) Pan‐cancer immunotherapy cohorts with survival information were accessed from the TIGER portal. B) Correlation between IGLoS scores and the GSVA scores of tumor cells' intrinsic pathways causing resistance to ICI, including the IFNG signature, the MHC I signature, the neoantigen signature, the PI3K pathway, the PTEN pathway, and the Wnt pathway. C) The bar plots show that the IGLoS score can predict the response to immunotherapy in two melanoma cohorts but limites predictive efficacy was observed in other immunotherapy cohorts. D) The forest plot shows that the IGLoS score is associated with a trend toward a worse prognosis in more than half of the immunotherapy cohorts. E) The optimal predictor varied from cohort to cohort in the random forest model for predicting responses to ICI. The IGLoS score was among the top predictors in the IMvigor (muscle‐invasive urothelial carcinoma), GSE91061 (melanoma), and PRJNA482620 (glioblastoma) cohorts. The height of the bars represented the VIMP, with a large positive value indicating a more important predictor. The rank of the predictors indicates the minimal depth, with the minimal depth increasing from left to right. A lower minimal depth represents a more important predictor in the model. F) Representative IHC staining images of PDL1 from part of GBM patients in the Gusu in‐house dataset with higher IGLoS score and lower IGLoS score.