Fig. 6: Metrics for proliferation and stemness correlate with xenograft survival, with in vitro stemness predicting in vivo drug response.

a Kaplan–Meier plots for orthotopic xenografts of 10,000 G34 (solid lines) and 400,000 G528 (dashed lines) treated with chemotherapies at concentrations and regimens based on published literature. G528 was only tested with temozolomide, carboplatin, and irinotecan. Statistical comparisons are shown in Supplementary Fig. 10. b Proportional hazards model built using in vivo survival data and in vitro viability data in the TME model at the dose-below-spheroid-IC₅₀ value. c–e Collective in vitro responses at the dose-below-spheroid-IC₅₀ value of G34 (dark blue) and G528 (light blue) treated with drugs in the TME model for invasion (c), proliferation (d), and stemness (e). Each data point is n = 4, with 6 drug responses for G34 and 3 drug responses for G528. Pearson correlation coefficients and p values are shown on respective plots (N = 9). f–h Proportional hazards models built using in vivo survival data and data from (c–e), examining the ability to predict xenograft survival given the percent invasion (f), proliferation (g), and stemness (h). Only stemness was able to predict in vivo therapeutic response (by proportional hazards model with *p < 0.05). NS indicates not significant.