Fig. 5: Deconvolution analysis of pHGG immune infiltration by subgroup.

a Barplots of the estimated median infiltration of specific cell types as a proportion of all non-cancer cell types (range scaled from 0 to 1) in 401 pHGG (pediatric High-Grade Glioma) by subgroup (as per Mackay et al.³⁸). b t-SNE plot representing the methylation profiles of 401 pHGG. The colors of dots in the central panel map to subgroup. Background shading represents the 2D spatial density estimation of the amount of tumor-infiltrating lymphocytes (TILs); red shading equals relatively greater than average infiltration and blue less than average. Exploded side panels represent enlarged areas of interest wherein both dot color and background shading represent the relative amount of the particular immune cell infiltration denoted. Red denotes relatively greater than average infiltration and blue less than average. c Kaplan–Meier plot showing significant difference in overall survival in WT-A (log-rank, p < 0.001, n = 80), WT-C (log-rank, p = 0.045, n = 80), and G34 subgroups (log-rank, p = 0.011, n = 42) by low (<median) or high (>median) levels of B-cell and NK infiltration. d Boxplot showing the proportion of monocytes and CD4T cells in pHGG by the presence/absence of a MAPK mutation (Wilcoxon = 3614, p < 0.001, n = 143). Box represents interquartile range, center line represents median, whiskers represent range of minima and maxima excluding outliers. e Boxplot showing TIL proportion as estimated by methylCIBERSORT for a subset of pHGG samples for which histopathology-based estimates of lymphocyte infiltration were available (F = 7.8, p = 0.007, n = 61). Patients are categorized as per Rutledge et al.⁴⁷. Box represents interquartile range, center line represents median, whiskers represent range of minima and maxima. Estimates of TILs were significantly greater in patients classified as Categories 1 (present) or 2 (abundant) than Category 0 (absent).