Automated structural imaging analysis detects premanifest Huntington's disease neurodegeneration within 1 year

Abstract

Intense efforts are underway to evaluate neuroimaging measures as biomarkers for neurodegeneration in premanifest Huntington's disease (preHD). We used a completely automated longitudinal analysis method to compare structural scans in preHD individuals and controls. Using a 1-year longitudinal design, we analyzed T(1) -weighted structural scans in 35 preHD individuals and 22 age-matched controls. We used the SIENA (Structural Image Evaluation, using Normalization, of Atrophy) software tool to yield overall percentage brain volume change (PBVC) and voxel-level changes in atrophy. We calculated sample sizes for a hypothetical disease-modifying (neuroprotection) study. We found significantly greater yearly atrophy in preHD individuals versus controls (mean PBVC controls, -0.149%; preHD, -0.388%; P = .031, Cohen's d = .617). For a preHD subgroup closest to disease onset, yearly atrophy was more than 3 times that of controls (mean PBVC close-to-onset preHD, -0.510%; P = .019, Cohen's d = .920). This atrophy was evident at the voxel level in periventricular regions, consistent with well-established preHD basal ganglia atrophy. We estimated that a neuroprotection study using SIENA would only need 74 close-to-onset individuals in each arm (treatment vs placebo) to detect a 50% slowing in yearly atrophy with 80% power. Automated whole-brain analysis of structural MRI can reliably detect preHD disease progression in 1 year. These results were attained with a readily available imaging analysis tool, SIENA, which is observer independent, automated, and robust with respect to image quality, slice thickness, and different pulse sequences. This MRI biomarker approach could be used to evaluate neuroprotection in preHD.