Unraveling Glioblastoma Diversity: Insights into Methylation Subtypes and Spatial Relationships
Poster No:
853
Submission Type:
Abstract Submission
Authors:
Martha Foltyn-Dumitru1, Haidar Alzaid1, Felix Sahm1, Wolfgang Wick1, Martin Bendszus1, Philipp Vollmuth1, Marianne Schell1
Institutions:
1Heidelberg University Hospital, Heidelberg, Germany
First Author:
Co-Author(s):
Introduction:
Gliomas, particularly Isocitrate Dehydrogenase (IDH) wild-type variants, represent a formidable challenge in neuro-oncology due to their intrinsic heterogeneity (Parker et al., 2015). As our understanding of glioma subtypes evolves, integrating advanced imaging modalities and molecular profiling becomes imperative for unraveling the complex interplay between tumor genetics and neuroanatomy. This retrospective study elucidates the intricate relationships between specific brain regions and molecular subtypes in glioblastomas.
Methods:
A cohort of 441 consecutive patients with IDH-wild-type glioma, diagnosed between 2009 and 2020, underwent preoperative MRI at Heidelberg University Hospital. The imaging protocol encompassed T1-weighted images, 2D FLAIR, and T2-weighted images. Molecular analysis determined IDH status and subclassification via DNA methylation profiling (Capper et al., 2018).
Initially, brain extraction and tumor segmentation utilized HD-GLIO, a deep learning-based approach, further validated by a neuroradiology resident (Isensee et al., 2019; Kickingereder et al., 2019). Subsequently, all T1-weighted images were registered to the FSL 1mm MNI template using a modified version of the fsl_anat function, employing both linear and non-linear registration with additional tumor area masking. The resulting transformation mask was applied to the tumor masks. For the third step, support vector regression-based lesion-symptom mapping (SVR-LSM) was employed to detect distinct brain regions associated with methylation subtypes (DeMarco & Turkeltaub, 2018; Zhang et al., 2014). Lesion maps were resampled to 2mm³, data were corrected for lesion volume, and beta maps were thresholded at p < 0.005 based on 10,000 permutations, with a minimum cluster size of >100 voxels.
Initially, brain extraction and tumor segmentation utilized HD-GLIO, a deep learning-based approach, further validated by a neuroradiology resident (Isensee et al., 2019; Kickingereder et al., 2019). Subsequently, all T1-weighted images were registered to the FSL 1mm MNI template using a modified version of the fsl_anat function, employing both linear and non-linear registration with additional tumor area masking. The resulting transformation mask was applied to the tumor masks. For the third step, support vector regression-based lesion-symptom mapping (SVR-LSM) was employed to detect distinct brain regions associated with methylation subtypes (DeMarco & Turkeltaub, 2018; Zhang et al., 2014). Lesion maps were resampled to 2mm³, data were corrected for lesion volume, and beta maps were thresholded at p < 0.005 based on 10,000 permutations, with a minimum cluster size of >100 voxels.
Results:
Out of the initially screened 441 patients, 423 (95.9%) met inclusion criteria, with six patients lacking voxels meeting the minimum lesion cutoff (n=10), and the rest excluded due to missing high-resolution T1w images. Following DNA methylation profiling, patients were classified into methylation subclasses. Predominantly, RTK II was observed (n=172, 40.7%), followed by MES (n=142, 33.6%), RTK I (n=75, 17.7%), and others (n=34, 8%). SVR-LSM unveiled distinct brain regions for different methylation subtypes. MES revealed a left-hemispheric cluster (superior temporal gyrus, posterior temporal lobe, insula cortex, posterior limb of internal capsule). RTK I showed two clusters in the right superior and middle frontal brain parenchyma. RTK II showed three clusters in the left hemisphere, two in the frontal lobe (including the inferior frontal gyrus) and one in the parietal lobe (supramarginal and angular gyrus). See Figures 1 and Table 1.
·Figure1. A) Overlay of all tumor masks. Unthresholded beta maps for the distinct methylation subclasses, B) MES, C) RTK I, and D) RTK II. E) Final clusters: red MES, blue RTK I, and green RTK II.
·Table 1. Overview of methylation subclasses.
Conclusions:
This study bridged the gap between molecular heterogeneity and spatial characteristics in glioblastomas through SVR-LSM, elucidating complex relationships. It unveiled associations between molecular subtypes of glioblastomas and their spatial characteristics. The significance of the work was underscored by the integration of multi-modal approaches, offering a comprehensive understanding of glioma heterogeneity. These insights not only unraveled the intricate landscape of glioma biology but also held promise for guiding future research and informing clinical applications.
Disorders of the Nervous System:
Neurodegenerative/ Late Life (eg. Parkinson’s, Alzheimer’s) 2
Genetics:
Genetic Association Studies 1
Modeling and Analysis Methods:
Multivariate Approaches
Keywords:
ADULTS
Data analysis
Modeling
MRI
Multivariate
Neoplastic Disease
Neurological
STRUCTURAL MRI
1|2Indicates the priority used for review
Provide references using author date format
DeMarco, A. T., & Turkeltaub, P. E. (2018). A multivariate lesion symptom mapping toolbox and examination of lesion-volume biases and correction methods in lesion-symptom mapping. Hum Brain Mapp, 39(11), 4169-4182. https://doi.org/10.1002/hbm.24289
Isensee, F., Schell, M., Pflueger, I., Brugnara, G., Bonekamp, D., Neuberger, U., Wick, A., Schlemmer, H. P., Heiland, S., Wick, W., Bendszus, M., Maier-Hein, K. H., & Kickingereder, P. (2019). Automated brain extraction of multisequence MRI using artificial neural networks. Hum Brain Mapp, 40(17), 4952-4964. https://doi.org/10.1002/hbm.24750
Kickingereder, P., Isensee, F., Tursunova, I., Petersen, J., Neuberger, U., Bonekamp, D., Brugnara, G., Schell, M., Kessler, T., Foltyn, M., Harting, I., Sahm, F., Prager, M., Nowosielski, M., Wick, A., Nolden, M., Radbruch, A., Debus, J., Schlemmer, H. P., . . . Maier-Hein, K. H. (2019). Automated quantitative tumour response assessment of MRI in neuro-oncology with artificial neural networks: a multicentre, retrospective study. Lancet Oncol, 20(5), 728-740. https://doi.org/10.1016/s1470-2045(19)30098-1
Parker, N. R., Khong, P., Parkinson, J. F., Howell, V. M., & Wheeler, H. R. (2015). Molecular heterogeneity in glioblastoma: potential clinical implications. Front Oncol, 5, 55. https://doi.org/10.3389/fonc.2015.00055
Zhang, Y., Kimberg, D. Y., Coslett, H. B., Schwartz, M. F., & Wang, Z. (2014). Multivariate lesion-symptom mapping using support vector regression. Hum Brain Mapp, 35(12), 5861-5876. https://doi.org/10.1002/hbm.22590