Investigational Scans for the Prediction of Genetic Mutations in Patients with Glioma, ReGIT Study

PRIMARY OBJECTIVES:

I. To improve voxel sensitivity and reduce the computational costs of statistical multiscale mapping (SMM), we will build new classification and FWER correction frameworks and compare with baseline SMM performance.

Ia. Acquire a minimum of 2 stereotactic biopsy samples on 20 glioma patients and map whole exome deoxyribonucleic acid (DNA) and ribonucleic acid (RNA) patterns to spatially-correlated pre-surgical image voxels;

Ib. Use leave-one-patient-out cross validation to estimate generalized performance of the new methods compared to baseline.

II. Determine the prognostic values of 15-oxygen (O)-water (H2O)-PET, Fluoro-ethyl-tyrosine (FET)-PET, and MRI.

IIa. Use logistic regression, generalized linear models, and non-parametric machine learning with cross validation to determine the individual and collective multiscale prediction effects of each imaging modality.

III. Determine the clinical translational potential of SMM.

IIIa. Evaluate SMM ability to predict overall clinical outcomes and local and tumor-wide treatment response in a realistic glioma population.

OUTLINE:

Patients receive 15O-H2O intravenously (IV) and then undergo PET/CT, then 18F-FET IV and PET, followed by gadobutrol IV and MRI at baseline. Patients undergo collection of tissue samples during stereotactic tumor biopsy per standard of care (SOC). Patients also undergo the collection of blood samples at baseline.