Holistic MRI-based Prognostic Model Development Using Genomic and Histopathologic Features in Glioblastoma Patients

Glioblastoma multiforme (GBM) is one of the malignant tumors with a high mortality rate. We aimed to predict the overall survival (OS) and progression-free survival (PFS) status of GBM via MRI-based features, named Visually AcceSAble Rembrandt Images (VASARI), genomics, and histopathologic information. The aforementioned data along with MRI of 188 patients were obtained, including the results of tumor biopsies and the mutation status of ATRX, CIC, EGFR, TP53, PDGFRA, IDH1, and 1p19q deletion genes. The dataset was randomly split into 80% and 20% for training and test datasets, respectively. LASSO was utilized to perform feature selection and classifier building. The area under the ROC curve (AUC) along with 95% confidence intervals were reported as indices showing model performance. For OS and PFS prediction, 7 and 5 distinct features were selected, respectively. Among the selected features, there were tumor histologic grade, age, and some of the VASARI features, including f3 and f13 (p-values <0.05). The combination of these features yielded an AUC of 0.90 (95%CI: 0.79–0.98) for OS prediction and 0.66 (95%CI: 0.59- 0.74) for PFS prediction. This study showed that a combination of VASARI, genomics, and histopathologic features can predict the OS and PFS of patients with glioblastoma and thus, enable physicians to achieve better management and treatment of patients.