Fixel-Based Analysis of Diffusion Imaging as a Quantitative Marker of Disease State in Spinocerebellar Ataxia

Abstract

Objective: Spinocerebellar ataxias (SCAs) are a group of genetically heterogeneous neurodegenerative diseases causing progressive deterioration and reduced quality of life. Therapeutic advances have been limited by a lack of sensitive anatomic, functional, or diffusion imaging-based biomarkers. This study aimed to identify white matter differences in the brains of preataxic and early-stage SCA1 and SCA3 mutation carriers using diffusion magnetic resonance imaging data from a multisite trial setting.

Methods: Fixel-based analysis was used to estimate microscopic fiber density, macroscopic fiber-bundle cross-section, and a combined fiber density and fiber-bundle cross-section measure within 45 cerebral and cerebellar tracts. Multivariate ANOVAs compared controls (n = 16), pre-ataxic (n = 10 SCA1, n = 24 SCA3), and ataxic patients (n = 14 SCA1, n = 36 SCA3). Clinical variables were correlated with fixel metrics and receiver operating characteristic analyses identified white matter tracts sensitive to distinguishing controls from pre-ataxic SCA1 and SCA3.

Results: We found widespread white matter deficits in pre-ataxic and ataxic patients compared to controls with regard to fiber density, fiber-bundle cross-section, and combined measures, all of which were associated with clinical measures of ataxia severity. We also found the combined fiber density and fiber-bundle cross-section measure from cerebellar tracts distinguished controls from pre-ataxia with high sensitivity and specificity for both SCA1 (receiver operating characteristic area under the curve = 0.96) and SCA3 (area under the curve = 0.97). The receiver operating characteristic analyses revealed that cerebellar tracts resulted in greater area under the curve than cortico-spinal and transcallosal tracts.

Interpretation: These results demonstrate that fixel metrics offer sensitive disease-specific measures of early SCA disease state that correlate with standard clinical measures.

Trial registration: Clinical Trial Readiness for SCA1 and SCA3 (READISCA), NCT03487367. https://clinicaltrials.gov/ct2/show/NCT03487367.

Keywords: early‐stage SCA; neurodegeneration; white matter.

FIGURE 1

Tracts of interest included a subthalamo‐pallidal tract, nigrostriatal tract, corticostriatal tract, and cerebello‐thalamo‐cortical tract, as well as six sensorimotor tracts from the sensorimotor area tract template, 32 transcallosal tracts from the transcallosal tractography template, and three cerebellar tracts. GP, globus pallidus; ICP, inferior cerebellar peduncle; M1, primary motor cortex; MCP, middle cerebellar peduncle; PMd, dorsal premotor area; PMv, ventral premotor area; preSMA, pre‐supplemental motor area; S1, somatosensory cortex; SCP, superior cerebellar peduncle; SMA, supplemental motor area; STN, subthalamic nucleus.

FIGURE 2

SCA1 (top) and SCA3 (bottom) ROC curves distinguishing control from pre‐ataxic (left), control from ataxic (middle), and pre‐ataxic from ataxic (right). ROC curves were generated using FDC metrics from the 3 cerebellar tracts, the top 3 SMATT tracts, the top 3 TCATT tracts, a combination of the average of the 3 cerebellar, the average of the top 3 SMATT, and the average of the top 3 TCATT tracts, and the SARA and FARS ADL clinical scores. The line of identity is shown as a dotted line (black). The AUC is shown in parentheses within the legend with the 95% confidence interval (CI). Note, the pre‐ataxic vs. ataxic and the control vs. ataxic comparisons do not have ROC curves for the SARA and FARS ADL because the SARA was used to determine the ataxic group.