Investigators from our team recently discovered that patients with glioma having mutation of the isocitrate dehydrogenase-1 (IDH1) gene tended to have larger, nonenhancing tumors localized in the frontal lobe (Lai et al, J Clin Oncol, 2011; 29:4482-90). Using an approach similar to lesion-symptom mapping, we developed a technique where MR images of the brains of patients with glioma were registered to stereotactic space, their tumors contoured, and the frequency of tumor occurrence spatially mapped and statistically compared between IDH1 mutant and wild type tumors (Analysis of Differential Involvement, ADIFFI maps). Results clearly suggested these tumors likely arise from a very specific region in the frontal lobe. Based on this interesting observation we decided to examine whether tumors with O(6)-methylguanine-DNA methyltransferase (MGMT) promoter methylation tended to occur more frequently in a specific location in the brain (Ellingson et al, Neuroimage, 2012; 59:908-16). Interestingly, we noticed that most tumors tended to be contiguous with the subventricular zone, known to harbor adult stem cells, and MGMT promoter methylated tumors tended to be lateralized to the left temporal lobe. Additionally, we noticed that if patients had tumors near this specific region in the left temporal lobe they were more likely to live longer, independent of their methylation status. At this time the University of California Cancer Research Coordinating Committee (UC CRCC) decided to help us build the first set of Probabilistic Radiographic Atlases for various tumor genotypes and phenotypes, comparing tumor volumes and

locations for various demographic, “-omic”, and interventional phenotypes.

Neuroradiologists and neuro-oncologists have known for a long time that tumor location is an important prognostic factor for malignant gliomas. Recent evidence in other tumor types (eg, medulloblastoma, IDH1 mutant gliomas, etc) suggests that this observation may reflect the location of tumor-specific cells of origin. Our current study provides additional insight into these rather complex phenomena, tying tumor location and volumetrics with various phenotypes and tumor behaviors (ie, radiogenomics).

In ways, our study has created many new questions regarding the potential mechanisms and reasons for the various observed tumor features for different phenotypes. Also, our approach has sparked interest in evaluating these same trends in other, independent datasets such as the The Cancer Genome Atlas or the Ivy Foundation.

With the help of the UC CRCC, we are currently constructing Version 2 of the UC Probabilistic Radiographic Atlases for Glioblastoma, focusing on more advanced imaging features (eg, texture analysis, shape features) and advanced imaging techniques (perfusion MRI, diffusion MRI, MR spectroscopy, and PET imaging features).

These findings were presented at the 2012 ASNR Annual Meeting, the 2011 Society of Neuro-Oncology Annual Meeting, and the 2012 International Society for Magnetic Resonance in Medicine Annual Meeting. We plan on potentially presenting new updates on our work at the 2013 ASNR Annual Meeting.

Read this article at AJNR.org . . .

 

If you liked this article, you may also like . . .

Differential Gene Expression in Glioblastoma Defined by ADC Histogram Analysis: Relationship to Extracellular Matrix Molecules and Survival

Relationship between Tumor Enhancement, Edema, IDH1 Mutational Status, MGMT Promoter Methylation, and Survival in Glioblastoma