BACKGROUND: Insufficient drug delivery across an intact blood-brain barrier (BBB) may limit the efficacy of therapeutic strategies. This study tested whether penetration across the BBB in glioblastoma could affect the efficacy of promising therapies. METHODS: The efficacy of a CDK4 inhibitor (PD0332991) and an MDM2 inhibitor (SAR405838), both with limited brain penetration, were tested singly in flank and orthotopic xenograft models in Mayo GBM22 and GBM108 lines, respectively. Structurally similar PI3K/mTOR inhibitors with limited (GDC-0980) and robust (GNE-317) brain penetration were tested in combination with bevacizumab in an orthotopic model of GBM10. Integrity of the BBB was assessed by injection of TexasRed-dextran conjugate 10 minutes prior to sacrifice. Drug concentrations were assessed by mass spectrometry. RESULTS: PD0332991 was effective in GBM22 flank tumors with a median time to reach the 1500mm3 endpoint of 116 days compared to 32 days for placebo (p < 0.05), while the same drug was ineffective in an orthotopic model (median survival 43 days vs. 44 days for placebo). Brain-to-plasma ratios for PD0332991 were 0.12 in wild-type mice compared to 10.3 in BBB-deficient mice. Similar results were obtained with the MDM2 inhibitor in GBM108 with profound effects on flank tumor growth (median time to 1500mm3 62 days for placebo versus 132 days with SAR405838; p < 0.05) but provided no survival benefit in an orthotopic model (median survival 38 days vs. 39 days). The effects of disrupting the BBB on the efficacy of SAR405838 in GBM108 orthotopic tumors will be reported. The clinical use of bevacizumab in GBM results in reduced contrast enhancement on MRI. Similarly, bevacizumab treatment of GBM10 orthotopic tumors results in a marked reduction in TexasRed-dextran accumulation within orthotopic tumors. Using this model, we compared the time to reach a moribund state following treatment with bevacizumab combined with either GDC-0980 or GNE-317. Compared to placebo (median survival 26 days), treatment with either GDC-0980 (29 days) or GNE-317 (33 days) did not significantly extend survival. However, when compared to bevacizumab alone (55 days), the combination of GNE-317 + bevacizumab significantly extended survival (71 days; p < 0.05) while GDC-0980 + bevacizumab did not (50 days). The brain-to-plasma AUC ratio for GNE-317 was 0.75 while GDC-080 was 0.07 in wild-type mice. The impact of bevacizumab therapy on tumor drug delivery will be reported. CONCLUSIONS: The efficacy of promising therapeutic strategies against relevant targets may be significantly compromised if tested using drugs that have limited penetration across the BBB. SECONDARY CATEGORY: Tumor Biology.