Use of a novel patient-specific model of glioma growth kinetics to elucidate underlying biology as measured by gene expression microarray.

71 Background: Gliomas are heterogeneous diseases with a wide distribution of growth kinetics that can be estimated prior to treatment and that are prognostic for patient outcome after treatment. Coherent molecular data sets have been made available through cooperative projects such as REMBRANDT and TCGA. We apply our novel patient specific method of measuring the net proliferation and diffusion rate from routinely available preoperative MRI sequences on patients included in these publicly available data sets to assess the underlying biology with imaging. Methods: The normalized microarray data from REMBRANDT (n=475) was used to discover a set of genes differentially expressed among GBM patients when compared with lower grade gliomas (n=853). 647 of these genes were also assessed with probesets in TCGA (n=466). Of these 466 patients, 84 also had preoperative MRI imaging available through The Cancer Imaging Archive (TCIA), for which net diffusion (D) proliferation (ρ) were estimated. Differential gene expression comparing these patients was performed. Results: 37 genes were differentially expressed with D. Genes implicated in cell adhesion, ECM maintenance and the production of focal adhesions are negatively correlated with D. Genes positively correlated with D are related to cell motility and pseudopodia formation. When considering ρ, 20 genes were found to be differentially expressed. A subset of these genes is related to hypoxia and therapy resistance. Some genes are also increased after radiation in cell-lines. Clustering on these genes revealed two classes; one with a survival advantage (p=0.0002). Conclusions: This work demonstrates the potential to assess underlying differences in biology in a heterogeneous disease through patient specific assessment of routinely available imaging. We find that more diffuse tumors will under-express genes involved in focal-adhesions and production of ECM, while they express genes in pathways related to motility and pseudopodia formation. Furthermore, tumors with high ρ are seen to express genes related to treatment resistance, which may explain worse survival in these patients. Future work will verify these markers in model organisms.

Duke Scholars

Author Erich Senin Huang Surgery