Radiogenomics in Newly Diagnosed Glioblastoma

Abstract

  Predicting the biological characteristics of diseases through non-invasive means can be considered as the ultimate goal of radiology. Magnetic resonance imaging (MRI) can help characterize the tissues of interest from a variety of perspectives. Indeed, MRI is expected to provide profound information pertaining to glioblastoma beyond mere anatomical and geometrical extension of the disease within the brain. Recent advances in computational radiology and bioinformatics have facilitated the novel concept of “radiogenomics” in the field of radiology. This rapidly expanding research field entails construction of mathematical algorithms that help predict the genetic characteristic of neoplasms by combining hundreds or thousands of quantitatively evaluated radiographical textures. Radiogenomics involves four procedures, i.e., image normalization, lesion segmentation, texture extraction, and mathematical modeling. Each procedure requires profound knowledge of image analysis. However, the lack of a commercially-available analytic system is a barrier to the validation of observations from different research groups. Despite these constraints, it is still possible to identify the trend and the potential applications of radiogenomics in the context of glioma. For example, although prediction of IDH mutation in lower grade glioma seems to be feasible, prediction of MGMT promoter methylation status in newly diagnosed glioblastoma (nGBM) is a challenge. On the other hand, several reports have shown the promising potential of radiogenomics in prognostic assessment of nGBM. In this review article, the author aimed to address the basics as well as the most promising applications of radiogenomics in the context of nGBM while carefully respecting the limitations of this novel technique.