Participants
The sample included in this study is part of a longitudinally followed-up cohort. From the 26 DM1 patients and 57 healthy controls (HC) assessed at baseline, 11 DM1 patients from the outpatient Neurology Service of the Donostia University Hospital and 11 HC recruited from healthy volunteers and patients’ relatives were re-scanned. The final sample included for the analyses was composed of 8 DM1 patients and 10 HC. A flow-chart showing the recruitment process and detailing the reasons for each excluded participant is shown in Figure 3. From the healthy volunteers recruited at baseline, only those whose age, gender and years of education were equal or closely similar (±5 years at most) to any re-scanned patient were invited to participate in order to form demographically equivalent groups. The DM1 sample was distributed as follows regarding disease onset and inheritance pattern: 4 adult onset (age of onset: 20-40) (50%) and 4 late onset patients (50%) (age of onset >40); 3 patients with maternal inheritance (37.5%), 4 with paternal inheritance (50%) and 1 inherited from both parents (12.5%).
Patients were excluded from the study if they had congenital (onset at birth), childhood (age of onset: 1-10) or juvenile (age of onset: 10-20) forms, history of major psychiatric or somatic disorder, acquired brain damage or alcohol or drug abuse, presence of corporal paramagnetic body devices that could impede an MRI study, and the presence of cerebral anomalies that could affect the MRI analysis. HC participants were required to satisfy the same inclusion criteria, except for the clinical diagnosis. These criteria were applied both at baseline and at follow-up.
This study was reviewed and approved by the Clinical Research Ethics Committee of the Gipuzkoa Health Area (DMRM-2017-01), in accordance with the principles of the Declaration of Helsinki and informed consent was obtained from all participants.
Clinical and neuropsychological assessment
Clinical and neuropsychological data were obtained at both baseline and follow-up by the research team’s responsible neurologist and neuropsychologist, respectively.
Regarding clinical data, muscular impairment was assessed with the Muscular Impairment Rating Scale (MIRS) 38, and CTG expansion size was determined from medical registries. Genetic assessment – PCR in DMPK alleles up to approximately 100 CTG and Southern blot analysis for larger expansions – was conducted in patients with no recent data available (within the last 5 years). The remaining clinical data concerning clinical form and maternal or paternal inheritance pattern were obtained from the patients’ medical records.
For all DM1 patients, neuropsychological assessment was conducted by an experienced neuropsychologist in the hospital facilities. The examiner was blind to the clinical condition (i.e., disease form, CTG repeats, MIRS or inheritance pattern) and to the MRI results. The administered battery of neuropsychological tests included several tests and subtests, and standardized T values were obtained according to Spanish population-based normative data. The employed assessment tools were: Wechsler Adult Intelligence Scale – Third Edition (WAIS III) 39, Rey-Osterrieth Complex Figure test (ROCF) 40, Rey Auditory Verbal Learning Test (RAVLT) 41, Stroop color and word test 42, Raven’s Progressive Matrices 43, verbal fluency (semantic and phonemic) 44,44 and California Computerized Assessment Package (CALCAP) 45.
The converted scores were employed to calculate six cognitive-domains: visuo-construction (Block design from WAIS-III and ROCF copy), verbal memory (RAVLT immediate recall, RAVLT delayed recall, Total RAVLT), attention/ processing speed (Digit span from WAIS III, STROOP word, STROOP color and the following CALCAP subtests: Simple Reaction Time (RT), election RT, Sequential 1 RT, Sequential 2 RT), executive functioning (Total RAVEN, semantic fluency, phonemic fluency, STROOP color-word, STROOP interference) and visual memory (ROCF delayed recall), and intellectual functioning. The intellectual functioning domain was estimated from vocabulary and block design subtests of the WAIS III based on Sattler and Ryan (reliability rxx=.93; validity r=.87).
MRI acquisition and data preprocessing
Diffusion images were acquired using a 1.5 Tesla scanner (Achieva Nova, Philips) with a single-shot echo-planar diffusion imaging sequence (IVIM-EPI) with the following parameters: voxel size=1.75 × 1.75 × 2 mm3; 60 axial slices; TR = 9,967 ms; TE = 66 ms; matrix size = 128 × 128. A diffusion gradient was applied across 32 non-collinear directions with b-value = 800 s/mm2. Additionally, one set of images was acquired without diffusion weighting (b = 0 s/mm2). All participants’ MRI scans were acquired on the same scanner and with the same protocol at both timepoints.
White matter tract alterations were assessed with FSL v6.0.4 and Tract-Based Spatial Statistics (TBSS) (Smith et al., 2006) using Fractional Anisotropy (FA) images. First, all individual FA images were normalized non-linearly to the HCP1021-2mm template. Next, the mean image was computed across all participants and skeletonized to obtain the mean FA skeleton, which represents regions with high confidence bundles common to all participants and removing some of the subject-specific tract-based heterogeneities. FA images for each subject were then projected onto the mean skeletons.
WM lesion load was assessed according to the Wahlund scale (ARWM) 47. When lesions >5mm were identified, severity was rated from 0 (no lesions) to 3 (diffuse involvement). Lesion location was quantified separately in five different regions: (1) the frontal area; (2) the parieto-occipital area; (3) the temporal area; (4) the infratentorial area, including the brain stem and cerebellum; and (5) the basal ganglia, including the striatum, globus pallidus, thalamus, internal and external capsules, and insula.
Statistical analysis
The SPSS (IBM SPSS Statistics 24) statistical package was used for sample description. Data were analyzed through inter-group comparisons to compare DM1 patients and HC, using contingency analysis (Chi-square) for categorical data and parametric (t-test) or non-parametric (Mann-Whitney U) for interval data, when appropriate. Intra-group analysis of the longitudinal evolution of clinical and neuropsychological data was conducted using the Wilcoxon signed-rank test or dependent t-test, when appropriate. In order to control for selective attrition bias, intra-group differences of eligible DM1 patients who were lost to follow-up and those who were included for analyses were conducted through independent sample comparisons (t-test or Mann Whitney U, when appropriate).
For the main objectives of the present work, two analyses were conducted.
Intergroup and intragroup tract integrity differences: transversal and longitudinal analyses
For diffusion parameters, group comparisons at baseline and at follow-up were conducted using the randomise tool included in FSL, a nonparametric permutation test for finding cluster-based significant statistical differences between groups at the voxel level. Global FA values were obtained for each group at both timepoints. Major tracts over which alterations were assessed are shown in Supplementary Figure 1.
For multiple comparison correction, Threshold-Free Cluster Enhancement (TFCE) (Smith & Nichols, 2009) was used, with 5000 iterations and Family Wise Error (FWE) corrected at p=0.05, thus ensuring that the chance of false positives was no more than 5%, or equivalently, ensuring 95% confidence of no false positives. Group comparisons were conducted using two different contrasts: patient > control, and control > patients.
Global FA (the mean value across all voxels in the brain) and mean values of FA per tract were compared between groups and across longitudinal measures. First, a Kruskal-Wallis test was conducted between all the four possible values (HC baseline, DM1 baseline, HC follow-up, and DM1 follow-up), followed by additional Kruskal-Wallis tests as a post-hoc analysis for the following four comparisons: HC baseline vs HC follow-up, HC baseline vs DM1 baseline, DM1 baseline vs DM1 follow-up, HC follow-up vs DM1 follow-up). For multiple comparisons across the number of possible tracts, False Discovery Rate (FDR) was applied.
Association between tract integrity and clinical and neuropsychological data
For assessing the association between skeletonized FA images and both clinical and neuropsychological data, we only considered the intensity values belonging to the significant regions resulting from the group comparison at follow-up. These associations were analyzed at both baseline and at follow-up. For each subject, the mean value of FA within each significant tract was obtained and correlated with CTG expansion size, MIRS score, WM lesion load and the neuropsychological scores using the linear Pearson partial correlation index, controlling for the covariable of age.
In all statistical analyses, missing values were handled by listwise deletion.