Abstract
ABSTRACT
Background and Objective
For their acceptance as endpoints in clinical trials, biomarkers and clinical outcome assessment (COA) measures should show reliability by being reproducible (precise) and responsive to change. We evaluated precision and responsiveness of MRI-derived brain volume metrics and various COAs in ceroid-lipofuscinosis, neuronal-3 (CLN3) disease.
Methods
This longitudinal observational study (
NCT03307304
) included 44 individuals with CLN3-related disease (ages 0.5–29.1 years, 48% female). Brain volumetric measures were computed using SynthSeg-v2.0 from T1-weighted 3D MPRAGE images acquired at 3T (Median [Q1, Q3] total MRI follow-up duration from baseline: 2.04 [0.00, 3.69] years). Compartmental tissue volumes were normalized to a model-adjusted total intracranial volume and scaled to the tissue-specific volume fraction expected in healthy adolescent brains to extract measures of brain atrophy. COAs included the Vineland Adaptive Behavior Scales, verbal IQ, and the Unified Batten Disease Rating Scale (UBDRS). Analysis of linear mixed-effects models included only CLN3 participants with syndromic presentation, having three or more visits per measurement to estimate slope, its relative standard error (RSE), and the minimum inter-visit interval (
T
min
) for statistically significant progression for each participant.
T
min
to detect progression in 75% of the population (
T
min,75%
) was computed for each metric.
Results
Among brain volumetric measures, brain parenchyma volume (BPV) and gray matter volume (GMV) were the most precise and responsive metrics. BPV showed an overall
intra
-individual decline with age of −1.32%/year (95% CI: −1.64% to −0.99%), low RSE of 12.5%, and short
T
min,75%
of 0.85 years (95% CI: 0.47–1.24). Amongst COAs, the UBDRS Clinical Global Impression (CGI) score was the most precise and responsive with an overall
intra
-individual increase of 1.26 units/year (95% CI: 1.03–1.49), RSE of 24.10%, and
T
min,75%
of 2.56 years (95% CI: 1.73–3.46), though with lower precision (RSE higher) and lower responsiveness (
T
min,75%
longer) than BPV.
Discussion
MRI-derived metrics, in particular BPV and GMV, detected change in brain atrophy with high precision and responsiveness suggesting that they would be informative for evaluating potential neuroprotective therapies in CLN3 disease. These findings support incorporating MRI-derived volumetric measurements alongside clinical assessments in future clinical trials in CLN3.