Glioblastoma multiforme (GBM), the most common malignant primary adult brain tumor, is highly refractory to treatment due to its invasive nature, resulting in a median survival time of ~ 1 year after diagnosis. The primary treatment challenge is the insidious propensity of glioma cells to aggressively infiltrate normal brain (A), making surgical resection palliative rather than curative.  Much attention has been focused on the role of proteases in tumor invasion.  Although the human “degradome”, the repertoire of proteases produced by cells, consists of at least 569 proteases (B), most studies restrict their attention to matrix metalloproteinases (MMPs).  However, the poor outcomes of the MMP inhibitor clinical trials suggest a significant contribution of other protease families to glioma invasion.

     In contrast to microarrays, which are based on a hybridization methodology, TaqMan low density arrays (TLDA) utilize PCR amplification of small products (C) and thus can detect differential expression of low copy number genes and is less sensitive to RNA degradation which is common in human surgical samples. In collaboration with Prof. Dylan Edwards and Dr. Caroline J. Pennington (School of Biological Sciences, University of East Anglia, Norwich UK), microfluidic TLDA cards containing primer/probe sets to 380 of the most widely expressed degradome genes were used to establish the protease signature of normal brain (i.e., from patients undergoing surgery for epilepsy) compared to metastatic and primary brain tumors (e.g., meningioma, grade II, grade III and recurrent oligodendroglioma, pilocytic astrocytoma, and GBM).  Surgical samples were obtained through collaboration with Drs. Helen L. Fillmore and William C. Broaddus (Dept. of Neurosurgery, Virginia Commonwealth University Medical Center, Richmond VA) and Drs. Paul L. Penar (UVM Division of Neurosurgery) and William W. Pendlebury (UVM Department of Pathology).  We hypothesized that, given the unique invasiveness of GBM, a distinct subset of proteases would be exclusively up-regulated in GBM relative to other tumor types.

     TLDA based profiling confirmed dysregulated MMP expression previously identified by microarray based analysis. Although TLDA profiling did not identify a GBM-specific protease signature, patient-specific protease signatures were apparent (D). Thus, TLDA screening may prove a useful pre-surgical tool to establish tailor-made anti-protease cocktail therapy that may be more efficacious than previous single MMP inhibitor therapies.  Furthermore, TLDA could better define resection boundaries.  As a proof of principle, MRI-defined stereotactic samples of tumor core, tumor-stromal interface, and “normal” adjacent brain were analyzed from one patient.  While the tumor core (102 genes) and tumor-stromal interface (139 genes) had significantly increased protease expression, the adjacent tissue classified as pathology-free still had 84 genes increased at least 4-fold relative to control samples, indicating the need for greater resection.  In contrast to microarray-based approaches, TLDA profiling, including data analysis, can be completed within 24 hours. Thus, TLDA screening may prove a useful pre-surgical tool to better establish surgical borders.