Changes in Apparent Diffusion Coefficient Across the Macroscopic Tumour Margin Correlate with Novel Tissue Measures of Infiltration in a Preclinical Glioblastoma Model. (31st January 2018)

• Record Type: Journal Article
• Authors: Thompson, Gerard; Vallatos, Antoine; Birch, Joanna; Al-Mubarak, Haitham; Gallagher, Lindsay; Gilmour, Lesley; Waldman, Adam; Holmes, William; Chalmers, Anthony
• Is Part Of: Neuro-oncology. Volume 20(2018)Supplement 1
• Page Start: i15
• Page End: i15
• Publication Date: 2018-01-31
• Subjects: Brain Neoplasms -- Periodicals; Brain -- Tumors -- Periodicals; Brain -- Cancer -- Periodicals; Nervous system -- Cancer -- Periodicals; 616.99481
• DOI: 10.1093/neuonc/nox238.066
• Languages: English
• ISSNs: 1522-8517

Abstract

Abstract: INTRODUCTION: A growing literature links diffusion-weighted imaging (DWI) characteristics with glioma molecular signatures, histology, and outcomes. Recent studies have explored 'transition coefficients' – changes in diffusion properties at the interface between macroscopic tumour and adjacent brain parenchyma – as markers of invasive or progressive phenotype, and have related these to survival, possibly reflecting the role of microinvasive disease in progression. Correspondence between ADC and cellularity is, however, confounded by necrosis, extracellular matrix composition, and vascularity. Moreover, validation of DWI markers of tumour burden has been limited to comparison of macroscopic, voxelwise ADC values with microscopic fields, at vastly different scales confounded by sampling bias. Here we examine transition coefficients by exploring relationships between a novel tissue-based quantification of tumour burden and ADC values at the infiltrating edge of a clinically-relevant, invasive, murine glioblastoma (GBM) model. METHODS: Anatomical imaging and DWI of G7 orthotopic GBM xenografts in CD-1 nude mice was performed at 7T immediately preceding cull. Brains were excised, frozen, sectioned, and stained for HLA (Human Leukocyte Antigen) to determine the cellular burden of human-derived tumour at the macroscopic scale. Several slices were co-registered to imaging data and their relative HLA content was normalised and averaged. Macroscopic tumour boundary was delineated using ADC and T2w images without reference to HLA. For each tumour boundary voxel, a perpendicular profile allowed simultaneous centrifugal ADC and HLA measurement. RESULTS: Significant inverse correlations between ADC and HLA were demonstrated across the infiltrating tumour edge. Areas of localised non-correlation corresponded to non-parenchymal boundaries limiting infiltration e.g. ventricular margin. DISCUSSION: This work is the first relating ADC values to infiltrating glioblastoma tissue burden using a novel approach at comparable spatial resolutions. It provides novel direct validation of imaging against GBM invasion at a cellular level, and a platform for future non-invasive imaging characterisation of glioma endophenotype.