Longitudinal stability of cortical grey matter measures varies across brain regions, imaging metrics, and testing sites in the ABCD study

Magnetic resonance imaging (MRI) is a vital tool for the study of brain structure and function. It is increasingly being used in individual differences research to examine brain-behaviour associations. Prior work has demonstrated low test-retest stability of functional MRI measures, highlighting the need to examine the longitudinal stability (test-retest reliability across long timespans) of MRI measures across brain regions and imaging metrics, particularly in adolescence. In this study, we examined the longitudinal stability of grey matter measures (cortical thickness, surface area, and volume) across brain regions, and testing sites in the Adolescent Brain Cognitive Development (ABCD) study release v4.0. Longitudinal stability ICC estimates ranged from 0 to .98, depending on the measure, parcellation, and brain region. We used Intra-Class Effect Decomposition (ICED) to estimate between-subjects variance and error variance, and assess the relative contribution of each across brain regions and testing sites on longitudinal stability. In further exploratory analyses we examined the influence of parcellation used (Desikan-Killiany-Tourville and Destrieux) on longitudinal stability. Our results highlight meaningful heterogeneity in longitudinal stability across brain regions, structural measures (cortical thickness in particular), parcellations, and ABCD testing sites. Differences in longitudinal stability across brain regions were largely driven by between-subjects variance, whereas differences in longitudinal stability across testing sites was largely driven by differences in error variance. We argue that investigations such as this are essential to capture patterns of longitudinal stability heterogeneity that would otherwise go undiagnosed. Such improved understanding allows the field to more accurately interpret results, compare effect sizes, and plan more powerful studies.