NIMG-05IDENTIFICATION OF IMAGING SIGNATURES OF THE EPIDERMAL GROWTH FACTOR RECEPTOR VARIANT III (EGFRvIII) IN GLIOBLASTOMA

Abstract

Glioblastoma (GBM) is the most common and aggressive adult brain tumor, with 14 months average survival. Epidermal growth factor receptor variant III (EGFRvIII) mutation is an important factor in driving tumor progression and defining prognosis in GBM patients, hence evidence of its presence can affect treatment decisions. The aim of this study is to identify quantitative imaging signatures of EGFRvIII. We used preoperative multi-parametric (T1-Gad, T2-FLAIR, Dynamic-Susceptibility-Contrast) magnetic resonance imaging data from a retrospective cohort of 64 patients (42 EGFRvIII-negative) with de novo GBM. We hypothesized that EGFRvIII-positive tumors, the more aggressive subtype, have a uniformly dense distribution of tumor cells throughout the peritumoral edematous region, as opposed to EGFRvIII-negative tumors, where tumor cell burden decreases farther from the tumor. To assess this peritumoral heterogeneity, we defined one region of interest (ROI) adjacent to the tumor and another at the farthest from the tumor but still within the edematous tissue. Perfusion temporal dynamics of each ROI were summarized via principal component analysis. The Bhattacharyya coefficient was used as a measure of separability (range [0,1]) between the dynamics of the two ROIs, for each patient. Values close to 0 indicate similar perfusion dynamics between the ROIs, which is consistent with uniformly and aggressively infiltrating tumors. Conversely, values close to 1 indicate substantial difference between the two ROIs, which would be consistent with less infiltrative tumors. The distributions of these separability measurements between EGFRvIII-negative and EGFRvIII-positive patients were very highly separable, with median values of 0.48 (Interquartile range: 0.251-0.647) and 0.209 (Interquartile range: 0.064-0.309), respectively. A two-tailed paired t-test confirmed the statistical significance of the results (p-value = 0.00007). These results suggest that discrimination of the EGFRvIII mutation status, which is critical for personalized treatment decisions and response evaluation, can be achieved based solely on assessing the peritumoral heterogeneity on in vivo imaging data.