Assessing network degeneration and phenotypic heterogeneity in genetic frontotemporal lobar degeneration by decoding FDG-PET

Abstract

Genetic mutations causative of frontotemporal lobar degeneration (FTLD) are highly predictive of a specific proteinopathy, but there exists substantial inter-individual variability in their patterns of network degeneration and clinical manifestations. We collected clinical and ¹⁸Fluorodeoxyglucose-positron emission tomography (FDG-PET) data from 39 patients with genetic FTLD, including 11 carrying the C9orf72 hexanucleotide expansion, 16 carrying a MAPT mutation and 12 carrying a GRN mutation. We performed a spectral covariance decomposition analysis between FDG-PET images to yield unbiased latent patterns reflective of whole brain patterns of metabolism ("eigenbrains" or EBs). We then conducted linear discriminant analyses (LDAs) to perform EB-based predictions of genetic mutation and predominant clinical phenotype (i.e., behavior/personality, language, asymptomatic). Five EBs were significant and explained 58.52 % of the covariance between FDG-PET images. EBs indicative of hypometabolism in left frontotemporal and temporo-parietal areas distinguished GRN mutation carriers from other genetic mutations and were associated with predominant language phenotypes. EBs indicative of hypometabolism in prefrontal and temporopolar areas with a right hemispheric predominance were mostly associated with predominant behavioral phenotypes and distinguished MAPT mutation carriers from other genetic mutations. The LDAs yielded accuracies of 79.5 % and 76.9 % in predicting genetic status and predominant clinical phenotype, respectively. A small number of EBs explained a high proportion of covariance in patterns of network degeneration across FTLD-related genetic mutations. These EBs contained biological information relevant to the variability in the pathophysiological and clinical aspects of genetic FTLD, and for offering valuable guidance in complex clinical decision-making, such as decisions related to genetic testing.

Keywords: Clinical neurology; FDG-PET; Frontotemporal dementia; Frontotemporal lobar degeneration; Machine learning.