Current methods of preclinical testing of potential therapeutics have, for the most part, failed to yield a clinical impact. This is particularly true or Glioblastoma multiforme (GBM) where prognosis has increased only by 2-3 months over the last 75 years and is coupled with a 5-year survival of less than 4%. Many promising preclinical studies have failed to live up to expectations when tested clinically. This problem is likely due t: a) limitations of the preclinical model system to predict efficacy and b) lack of reliable biomarkers for proper patient selection. We believe that comprehensively molecularly characterized patient-derived xenograft tumors (xenolines) will provide solutions to both problems. We have characterized a panel of 27 human GBM xenolines at the genomic, transcriptomic, and kinomic (global kinase activity assessment through a peptide substrate microarray) level to generate a "proband" model of clinical disease. We have found these xenolines can recapitulate all four molecular subtypes of GBM identified in The Cancer Genome Atlas while reproducing the key hallmarks of GBM when implanted orthotopically in immunocompromised mice. As a high-throughput extension to our orthotopic pre-clinical model, we have shown that we can grow disaggregated xenoline tumors in a novel three-dimensional (3D) culture system. This system incorporates many cells of the tumor microenvironment to produce "MicroTumors" suitable for higher throughput drug testing. Our comprehensive profiling of these xenolines will allow us to build potential clinical biomarkers of therapeutic response to drugs, particularly for small molecule kinase inhibitors (SMI's). To demonstrate this approach, we have identified novel "kinomic" clusters among these xenolines and have recently presented preliminary evidence that baseline kinomic profiles can differentiate treatment sensitive versus resistant tumors. Therefore, we hypothesize that a two-stage drug testing system utilizing 3D GBM xenoline MicroTumors to screen and profile drug response followed by validation in orthotopic xenoline tumors, will generate a more reliable preclinical therapeutic testing approach translating to more effective drug treatment in patients. A corollary is that xenoline testing will allow for molecular profile integration with drug response information producing a diverse set of probands to which patients with molecularly similar GBM tumors can be assigned for more rational clinical trial design. To demonstrate the feasibility of this system, as proof of concept e will select four drugs and 12 xenolines and: 1) Perform drug screening on MicroTumors grown in a unique 3D environment;2) Perform secondary drug testing in vivo with selected xenolines based on kinomic profiles and MicroTumor response data;and 3) Integrate chemovulnerability data with molecular profiles to begin to build a drug response prediction system. Relevance: The improved preclinical modeling of GBM not only will provide more clinically reliable drug sensitivity data but also patient selection criteria for smarter clinical trial design.
The cancer literature is filled with promising preclinical studies demonstrating impressive efficacy for new therapeutics, yet, the clinical success rate for these approaches have been minimal indicating that current methods used to predict efficacy are inaccurate. This disconnect between oversimplified laboratory models and human patients necessitates the development of models that more faithfully represent the disease. This project seeks to produce a suitable model with Glioblastoma multiforme serving as a prototypical system that addresses key factors often missing in preclinical models, notably the tumor microenvironment and molecular diversity/heterogeneity of human disease. This robust model will permit improved go-no-go decision making in drug development, provide early predictive response biomarkers, and superior research models of disease.