How the Precise Habitats Can Predict the IDH Mutation Status and Prognosis of the Patients With High-grade Gliomas

Overview

High-grade glioma is the most common primary malignant tumor in central nervous system, and its high tumor heterogeneity is the main cause of tumor progression, treatment resistance and recurrence. Habitat imaging is a segmentation technique by dividing tumor regions to characterize tumor heterogeneity based on tumor pathology, blood perfusion, molecular characteristics and other tumor biological features. In some studies, the Hemodynamic Multiparametric Tissue Signature (HTS) method has been proven to be feasible. The Hemodynamic Multiparametric Tissue Signature (HTS) consists of a set of vascular habitats obtained by Dynamic Susceptibility Weighted Contrast Enhanced Magnetic Resonance Imaging (DSC-MRI) of high-grade gliomas using a multiparametric unsupervised analysis method. This allowed them to automatically draw 4 reproducible vascular habitats (High-angiogenic enhancing tumor; Low-angiogenic enhancing tumor; Potentially tumor infiltrated peripheral edema; Vasogenic peripheral edema) which enable to describe the tumor vascular heterogeneity robustly. In other studies, contrast-enhancing mass can divided into spatial habitats by K-means clustering of voxel-wise apparent diffusion coefficient (ADC) and cerebral blood volume (CBV) values to observe the changes of voxels in spatial habitat on the time line. Using this so-called spatiotemporal habitat to identify progression or pseudoprogression in cancer therapy. Above all, we have sufficient and firm reasons to deem that habitat imaging based on multiparametric MRI is more conducive to reflect the potential biological information inside the tumor and realize individualized diagnosis and treatment. To sum up, the assumption of this experiment is that the Habitats Created by preoperative or postoperative Multiparametric MRI ,such as conventional MRI sequences, Dynamic Susceptibility Weighted Contrast Enhanced Magnetic Resonance Imaging (DSC-MRI), Dynamic Contrast Enhanced Magnetic Resonance Imaging (DCE-MRI), Diffusion Weighted Magnetic Resonance Imaging(DWI) ,Vessel Size Imaging (VSI) ,or Magnetic Resonance Spectroscopy (MRS) can predict the molecular mutation status, prognosis, treatment residence, progression, pseudoprogression, and even recurrence and distant intracranial recurrence in patients with high-grade gliomas.

This is a single center experiment. The subjects of this study were patients diagnosed as high-grade glioma by multiparametric magnetic resonance imaging and pathological biopsy from January 1, 2008 to December 31, 2021(or at some interval within this period). Patients meeting the inclusion criteria will enter the next experimental stage. 1. Patients selection: The patients will be clustered according to the preoperative and postoperative examination methods performed by each patient, such as only DSC sequence, both DSC and DWI sequences, or a full set of functional imaging sequences at the same time; 2. Images segmentation: To select the patients who meet the experimental design, and use deep learning-based biomedical image segmentation methods, such as Brain Tumor Segmentation (BraTS) challenge, to segment more accurate and reproducible habitats as much as possible; 3. Construction of clinical model: We may be able to obtain the parameter values of the habitat, such as CBV, ADC, Ktrans, etc. Combining these data with the basic situation of patients can build a clinical model. 4. Construction of radiomics model: The radiomics analysis will probably be structured into four parts: habitats segmentation, feature extraction, feature selection and model construction. 5. We try to analyze that habitat imaging based on Multiparametric MRI is indeed better than the conventional rough and simple ROI analysis; We try to analyze the relation between the habitats and the IDH mutation status or MGMT promoter methylation status; We try to analyze the relation between the habitats and the overall survival (OS) of the patient; We try to analyze the habitats is conducive to differentiate recurrence from distant intracranial recurrence. Finally, statistical methods and survival analysis were used to determine whether the habitat was statistically significant for IDH mutation status and prognosis. For example, receiver operating characteristic curve (ROC) analysis evaluated the potential of the spatial habitats in IDH mutation prediction. The Kaplan-Meier curve evaluates the validation of the diagnosis in OS prediction in high-grade glioma.