Abstract Clinical data indicate that certain regions within glioblastoma (GBM) have a relatively intact blood-brain barrier (BBB), but controversy surrounds whether associated heterogeneous delivery may limit efficacy for otherwise highly active drugs with poor brain distribution. In this study, the efficacy of molecularly targeted therapies was compared in relevant GBM patient-derived xenograft (PDX) models grown either as heterotopic or orthotopic tumors. The impact of further disrupting BBB integrity in non-responsive orthotopic tumors was evaluated by over-expressing vascular endothelial growth factor (VEGF) and measuring drug distribution by matrix-assisted laser desorption ionization mass spectrometry imaging (MALDI-MSI). The brain:plasma ratio for five targeted agents was determined by LC-MSMS and treatment efficacy was evaluated in target-relevant GBM PDX lines: erlotinib (brain:plasma = 0.02 ± 0.02; efficacy tested in GBM6), palbociclib (brain:plasma = 0.06; efficacy tested in GBM22), AZD1775 (brain:plasma = 0.05; efficacy tested in GBM22), SAR405838 (brain:plasma = 0.01 ± 0.003; efficacy tested in GBM108), and AZD4547 (brain:plasma = 0.045 ± 0.02; efficacy tested in GBM150). Each of these drugs was effective in relevant GBM PDX flank tumor models with a 27% to 218% prolongation in the median time to exceed their pre-specified endpoint compared to placebo treatment (p≤0.02 for each drug). In contrast, none of these drugs were effective in prolonging survival in the same target-relevant orthotopic tumor models (-6% to 8% extension in median survival; p>0.05 for each drug). Consistent with partial and heterogeneous disruption of the BBB, MALDI-MSI of erlotinib, AZD1775 or SAR405838 distribution within orthotopic tumors demonstrated highly heterogeneous drug levels with large regions within each tumor approaching the low drug levels observed within surrounding normal brain. To further evaluate whether heterogenous drug distribution in orthotopic tumors could account for poor treatment efficacy, GBM108 was transduced with lentiviral constructs encoding for VEGF or empty vector (EV). While both GBM108-sublines were readily detectable with gadolinium-enhanced magnetic resonance imaging (MRI), the GBM108-VEGF tumors had a significantly more disrupted BBB, as evidenced by a more uniform and intense distribution of a brain-impenetrant TexasRed-dextran vascular marker. Similarly, the distribution of SAR405838, measured by MALDI-MSI, was markedly higher and more uniform in the GBM108-VEGF tumors when compared to the GBM108-EV tumors. Enhanced delivery of SAR405838 into orthotopic GBM108-VEGF models translated into a marked enhancement in treatment efficacy in comparison to GBM108-EV with a median survival prolongation of 37 vs. 4 days, respectively (p=0.0055). Collectively, these data highlight the importance of testing novel therapeutic agents in orthotopic tumor models and suggest that limited brain penetration for many molecules may significantly limit their efficacy in brain tumors that have a partially intact BBB. In the face of continued failure to develop effective targeted agents for GBM, these in vivo results highlight the importance of re-evaluating the dogma in neuro-oncology that the BBB is fully disrupted in GBM and, therefore, drug delivery across the BBB is not a major factor limiting treatment efficacy. Citation Format: Jann N. Sarkaria, David Calligaris, Daniel Ma, Karen Parrish, Ann C. Mladek, Janice Laramy, Minjee Kim, Shuangling Zhang, Mark Schroeder, Brett L. Carlson, Katrina Bakken, Aaron Johnson, Nathalie Agar, William Elmquist. The critical importance of the blood-brain barrier in modulating the response to otherwise highly effective targeted therapies in patient-derived orthotopic glioblastoma xenografts. [abstract]. In: Proceedings of the AACR Special Conference: Patient-Derived Cancer Models: Present and Future Applications from Basic Science to the Clinic; Feb 11-14, 2016; New Orleans, LA. Philadelphia (PA): AACR; Clin Cancer Res 2016;22(16_Suppl):Abstract nr B25.