NIMG-12GENE SIGNATURES ASSCOCIATED WITH DIFFUSION MRI ADC MAPPING OF GLIOBLASTOMA EDEMA/TUMOR INVASION

Abstract

BACKGROUND: Diffusion MR imaging (dMR) is widely used in clinical practice and captures a distinct molecular/cellular features of tumor tissue. The aim of this study is to correlate pretreatment tumor Apparent Diffusion Coefficient (ADC), measured by diffusion magnetic resonance (dMR) imaging with overall survival in patients with glioblastoma and determine its association with gene signatures. METHODS: We retrospectively identified 37 treatment-naive glioblastoma patients from The Cancer Genome Atlas (TCGA) whom had gene expression profiles and corresponding dMR imaging available in The Cancer Imaging Archive (TCIA). ADC mean were measured in edema/tumor invasion of GBM lesions and contralateral normal brain tissue. To normalize ADC values, ADC mean of edema/tumor invasion regions were divided by the contralateral ADC means. Using decision tree clustering method, patients categorized based on normalized ADC mean into high vs low groups. Kaplan-Meyer survival analysis was used to determine the difference of overall survival between the two groups of patients. Imaging genomic analysis was subsequently performed using GenePattern Comparative Marker Selection module (Broad Institute). To identify the associated gene network, the top 100 most positively and the top 100 most negatively correlated genes in the high group versus the low group were then analyzed with Ingenuity Pathway Analysis. RESULTS: Based on decision tree analysis, normalized ADC mean cutoff of 19.48 was used to categorize patients to high (n = 6) versus low (n = 31) groups. Kaplan-Meier analysis showed the patients with high normalized ADC mean had significantly (p = 0.02) better overall survival than the patients with low normalized ADC mean. Median survival in high and low group was 747 and 394 days, respectively. There was a significant genomic signature, associated with high vs low groups. The validation of these analyses was done in a separate group of glioblastoma patients. CONCLUSIONS: dMR characteristics can identify highly significant survival differences and specific genomic signature.