NCMP-14. HIGH GLIOMA PROLIFERATION SIGNIFICANTLY CORRELATES WITH LANGUAGE DEFICITS REGARDLESS OF MRI LESION SUBTYPE

• Published in: Neuro-oncology (Charlottesville, Va.) Va.), 2023-11, Vol.25 (Supplement_5), p.v209-v210
• Main Authors: Dada, Abraham, Aabedi, Alexander, Kaur, Jasleen, Ahn, EunSeon, Brang, David, Hervey-Jumper, Shawn
• Format: Article
• Language: English
• Subjects: Neurological Complications of Cancer
• ISSN: 1522-8517; 1523-5866
• DOI: 10.1093/neuonc/noad179.0798

Abstract BACKGROUND We recently demonstrated that glioma-infiltrated cortex engages in task-specific computations but may encode less information while doing so. The mechanistic underpinnings of glioma burden and its behavioral impact remains unknown OBJECTIVE To investigate the relationship between glioma burden and degree of cognitive impairments. METHODS One-hundred-eight patients with newly diagnosed WHO 2-4 diffuse glioma were recruited from a UCSF prospective registry. MRI FLAIR and T1 post-gadolinium tumor volume masks were demarcated. A “Composite” mask was generated from the sum of T1 post-gadolinium and FLAIR volume masks. Participants completed language assessments using the Quick Aphasia Battery (QAB). Glioma burden was determined by tumor-specific marker expression (IDH-1 mutation and P53) and glioma proliferation (Ki-67 and MIB-1 proliferation index). For each mask type, support-vector regression voxel-lesion symptom mapping (SVR-LSM) and linear regression analyses were performed to identify anatomic regions of interest (ROI) and glioma-specific factors associated with lower QAB scores. RESULTS SVR-LSM analysis by mask type revealed significant clusters in the left superior temporal gyrus: composite lesions within a cluster of 41,762 voxels and T1 post-gadolinium lesions within a cluster of 42,570 voxels. Both were independently predictive of lower QAB scores. SVR-LSM analysis of FLAIR masks failed to produce any significant clusters. Multivariate linear regression analysis of composite lesions was utilized to control for composite lesion voxel volume, glioma grade, and tumor-specific marker expression. The presence of voxels overlapping with the composite lesion ROI scores (beta = -0.13; p < 0.001), and, relative to a low glioma proliferation group (< 2%), glioma proliferation between 31-50% (beta = -2.00, p = 0.044) were independently and significantly predictive of lower QAB scores. CONCLUSION Relative to a low glioma proliferation group (< 2%), glioma proliferation between 31-50% significantly predicts language impairments, with a linear relationship between voxel overlap with ROI and impairment severity.