NIMG-40. NON-INVASIVE IN VIVO SIGNATURE OF IDH1 MUTATIONAL STATUS IN HIGH GRADE GLIOMA, FROM CLINICALLY-ACQUIRED MULTI-PARAMETRIC MAGNETIC RESONANCE IMAGING, USING MULTIVARIATE MACHINE LEARNING

Abstract

PURPOSE: Mutational status of isocitrate dehydrogenase (IDH1) is a defining feature of the World Health Organization classification scheme for high grade gliomas (HGGs). IDH-mutant HGGs confer significantly improved prognoses when compared with IDH-wildtype, which typically describe the most common malignant primary HGGs in adults, namely glioblastoma. HGGs are densely cellular, pleomorphic tumors with high mitotic activity, with glioblastoma having either microvascular proliferation, or necrosis, or both. We hypothesize that integrative analysis of multi-parametric magnetic resonance imaging (mpMRI) via multivariate machine learning (ML), will enhance subtle yet important radiographic HGG characteristics, and reveal imaging signatures determinant of IDH1 mutational status. METHODS: 86 HGG patients were retrospectively identified with available pre-operative clinically-acquired mpMRI data (T1, T1-Gd, T2, T2-FLAIR, DTI, DSC-MRI). Each HGG was delineated into subregions of enhancement, non-enhancement, and peritumoral edema/invasion. 342 quantitative imaging phenomic (QIP) features extracted across subregions from all mpMRI, comprising descriptors of size, morphology, texture, intensity, and biophysical growth modeling. Cross-validated sequential feature selection determined the most discriminative QIP features for our integrative ML predictor of IDH1 status. The predicted classifications, following a 10-fold cross-validation, were compared with the IDH1 status obtained by next generation sequencing, or immunohistochemistry. RESULTS: 61 QIP features, primarily descriptive of tumor texture, were determined as most important for an IDH1 imaging signature. Using this signature, our predictor classified IDH1 mutational status with an accuracy of 88.4% (sensitivity= 66.7%, specificity=92.9%). CONCLUSION: Quantitative analysis of clinically-acquired mpMRI reveals subtle/visually-imperceptible, yet informative features, which integrated via ML yield a non-invasive in vivo IDH1 imaging signature in HGG. Knowledge of IDH1 mutational status at initial presentation can influence therapeutic decision-making, which will have a significant impact on patient care. Particularly in this precision medicine era, as mutant-IDH enzyme inhibitors and immunotherapy targeting IDHmutant tumor cells are developed, imaging to diagnose and follow IDHmutant tumors can be invaluable.∗equal contribution.