Locoregional Radiogenomic Models Capture Gene Expression Heterogeneity in Glioblastoma

Abstract

Radiogenomics mapping noninvasively determines important relationships between the molecular genotype and imaging phenotype of various tumors, allowing advances in both clinical care and cancer research. While early work has shown its technical feasibility, here we extend 1 radiogenomic mapping to a locoregional level that can account for the molecular heterogeneity of tumors. To achieve this, our data processing pipeline relies on three main steps: 1) the use of multi-omics data fusion to generate a set of 100 interpretable gene modules, 2) the use of patch-based image analysis (specifically of contrast-enhanced T1-weighted weighted MR images) combined with Generalized Linear Models (GLM) to establish potential links between module expressions and local MR signal, and 3) the use of expression heatmaps based on GLMs decision values to explore visualization of tumor molecular heterogeneity. The performance of the proposed approach was evaluated using a leave-one-patient-out crossvalidation method as well as a separate validation data set. The top performing models were based on a small set of 20 features and yielded Area Under the receiver operating characteristic Curve (AUC) above 0.65 on the validation cohort for eight modules. Next, we demonstrate the clinical and biological interpretation of four modules using molecular expression heatmaps superimposed on clinical radiographic images, showing the potential for assessing tumor molecular heterogeneity and the utility of this method for precision treatment in clinical decision making and imaging surveillance.

ImaGene: a web-based software platform for tumor radiogenomic evaluation and reporting

“…The field of imaging genomics or radiogenomics focuses on finding associations between radiomic characteristics of tissue ROIs and molecular characteristics such as genomic, transcriptomic and proteomic profiles of tumor cells. Using data from The Cancer Imaging Archive (TCIA), a group recently conducted a locoregional study using MRI images of glioblastoma tissues by generating heat maps corresponding to both core and boundary regions of specific tumor ROIs, revealing substantial genetic heterogeneity ( Depeursinge et al , 2018 ). The gene expression profiles of each of the sampled regions were correlated with known information regarding the genes involved in multiple pathways potentially leading to oncogenesis ( Depeursinge et al , 2018 ).…”

“…Using data from The Cancer Imaging Archive (TCIA), a group recently conducted a locoregional study using MRI images of glioblastoma tissues by generating heat maps corresponding to both core and boundary regions of specific tumor ROIs, revealing substantial genetic heterogeneity ( Depeursinge et al , 2018 ). The gene expression profiles of each of the sampled regions were correlated with known information regarding the genes involved in multiple pathways potentially leading to oncogenesis ( Depeursinge et al , 2018 ). Another recent study investigated gene expression profiles of pancreatic ductal adenocarcinoma (PDAC) ROIs using CT images by comparing the radiomic features of these ROIs with their genotypes and stromal content ( Attiyeh et al , 2019 ).…”

ImaGene: A web-based software platform for tumor radiogenomic evaluation and reporting

“…The field of imaging genomics or radiogenomics focuses on finding associations between radiomic characteristics of tissue ROIs and molecular characteristics such as genomic, transcriptomic, and proteomic profiles of tumor cells. The Cancer Imaging Archive (TCIA) recently conducted a locoregional study using MRI images of glioblastoma tissues by generating heat maps corresponding to both core and boundary regions of specific tumor ROIs, revealing substantial genetic heterogeneity (Depeursinge et al 2018). The gene expression profiles of each of the sampled regions were correlated with known information regarding the genes involved in multiple pathways potentially leading to oncogenesis (Depeursinge et al 2018).…”

“…The Cancer Imaging Archive (TCIA) recently conducted a locoregional study using MRI images of glioblastoma tissues by generating heat maps corresponding to both core and boundary regions of specific tumor ROIs, revealing substantial genetic heterogeneity (Depeursinge et al 2018). The gene expression profiles of each of the sampled regions were correlated with known information regarding the genes involved in multiple pathways potentially leading to oncogenesis (Depeursinge et al 2018). Another recent study investigated gene expression profiles of pancreatic ductal adenocarcinoma (PDAC) ROIs using CT images by comparing the radiomic features of these ROIs with their genotypes and stromal content (Attiyeh et al 2019).…”

“…), and if the value of the decision threshold is set to 0.5 (as an example), all the expression values falling below 0.5 will be set to 0 and the ones landing above 0.5 will be set to 1, for both y observed and y predict for every gene label ( y ). Such a binarization allows ImaGene to calculate AUCs using binarized y observed and y predict values for each label (gene) at various measurement (aka normalized gene-expression) cut-offs, which aids users evaluate the model’s ability to classify labels (genes) into groups of high and low measures (at definite thresholds) indicating them the measures at which the labels (genes) are more reliable to be predicted using imaging features as the dataset (Depeursinge et al 2018). A plot of the AUC values versus decision thresholds is generated for all gene labels providing users with a visualized overview (along with the respective supplementary data files) indicative of a regressionmodel’s classification performance.…”