Widespread structural brain involvement in ALS is not limited to the C9orf72 repeat expansion

Abstract

Background: In patients with a C9orf72 repeat expansion (C9+), a neuroimaging phenotype with widespread structural cerebral changes has been found. We aimed to investigate the specificity of this neuroimaging phenotype in patients with amyotrophic lateral sclerosis (ALS).

Methods: 156 C9- and 14 C9+ patients with ALS underwent high-resolution T1-weighted MRI; a subset (n=126) underwent diffusion-weighted imaging. Cortical thickness, subcortical volumes and white matter integrity were compared between C9+ and C9- patients. Using elastic net logistic regression, a model defining the neuroimaging phenotype of C9+ was determined and applied to C9- patients with ALS.

Results: C9+ patients showed cortical thinning outside the precentral gyrus, extending to the bilateral pars opercularis, fusiform, lingual, isthmus-cingulate and superior parietal cortex, and smaller volumes of the right hippocampus and bilateral thalamus, and reduced white matter integrity of the inferior and superior longitudinal fasciculus compared with C9- patients (p<0.05). Among 128 C9- patients, we detected a subgroup of 27 (21%) with a neuroimaging phenotype congruent to C9+ patients, while 101 (79%) C9- patients showed cortical thinning restricted to the primary motor cortex. C9- patients with a 'C9+' neuroimaging phenotype had lower performance on the frontal assessment battery, compared with other C9- patients with ALS (p=0.004).

Conclusions: This study shows that widespread structural brain involvement is not limited to C9+ patients, but also presents in a subgroup of C9- patients with ALS and relates to cognitive deficits. Our neuroimaging findings reveal an intermediate phenotype that may provide insight into the complex relationship between genetic factors and clinical characteristics.

Figure 1

Region-wise and vertex-wise comparison of cortical thickness in C9− and C9+ patients with ALS. (A) Region-wise comparison of C9− and C9+ patients with ALS, showing widespread involvement of the cortex in C9+ patients with ALS. Blue-coloured areas show significant cortical thinning in C9+ patients with ALS compared with C9− patients with ALS (p<0.05, FDR-corrected). (B) Vertex-wise comparison between C9− and C9+ patients with ALS. Coloured clusters were significantly thinner in C9+ patients with ALS compared with C9− patients with ALS (cluster-wise corrected using a Monte-Carlo simulation with 10 000 iterations). 1=pars opercularis; 2=insula; 3=superior parietal (in the left hemisphere extending to the temporal lobe); 4=lateral occipital; 5=isthmus cingulate; 6=fusiform; 7=rostral middle frontal. ALS-FTD, amyotrophic lateral sclerosis with comorbid frontotemporal dementia; C9+, C9orf72 repeat expansion present; C9−, C9orf72 repeat expansion absent.

Figure 2

Comparison of white matter integrity in C9− and C9+ patients with ALS. The ILF is bilaterally affected in C9+ patients with ALS, characterised by an increased RD and lower FA, compared with C9− patients with ALS. SLFP and SLFT showed increased RD of the SLFP and SLFT in the right hemisphere of C9+ patients. Dots are estimated marginal means (ie, means corrected for effects of age and gender) and whiskers represent 95% CIs. Comparisons between groups and corresponding p values are shown in online supplementary table S4. *p value <0.05 after FDR correction. ALS-FTD, amyotrophic lateral sclerosis with comorbid frontotemporal dementia; FA, fractional anisotropy; ILF, inferior longitudinal fasciculus; RD, radial diffusivity; SLFP, parietal parts of the superior longitudinal fasciculus; SLFT, temporal parts of the superior longitudinal fasciculus.

Figure 3

Flow chart of model development and application for classifying patients as C9+ or C9− based on neuroimaging phenotype. First, the model was developed in 14 C9+ patients with ALS and 28 age-matched and gender-matched C9− patients with ALS. A set of 11 imaging variables constituted the prediction model. Second, the model was applied to the remaining 128 C9− patients with ALS who were not used for model development. According to the model, 27 patients were classified with a ‘C9+’ neuroimaging phenotype and 101 patients with a C9− imaging phenotype. ACL, amyotrophic lateral sclerosis; C9+, C9orf72 repeat expansion present; C9−, C9orf72 repeat expansion absent.

Figure 4

Identification of patients with a C9orf72 neuroimaging phenotype in a group of C9− patients with ALS. (A) Elastic net logistic regression identified a characteristic pattern of brain involvement distinguishing C9+ from C9− patients with ALS based on 11 imaging variables, the ‘C9+’ neuroimaging phenotype. Patients with ALS were subsequently classified into patients with and without this characteristic neuroimaging phenotype, and cortical thickness was compared in a region-wise analysis with healthy controls (n=92). Blue-coloured areas show significant cortical thinning in patients with ALS compared with healthy controls (p<0.05, FDR-corrected). (B) Classification of a subgroup of 128 C9− patients with ALS (156 minus 28 C9− patients with ALS of the training set) revealed 27 C9− patients with ALS with a ‘C9+’ neuroimaging phenotype with widespread cortical involvement congruent to C9+ patients with ALS (n=14), whereas 101 C9− patients with ALS had a phenotype of cortical thinning restricted to the primary motor areas. Clinical characteristics are shown per study group. ^aFAB in a subset of patients (maximum score 18): C9+ ALS n=9, C9− ALS with ‘C9+’ neuroimaging phenotype n=18, C9− ALS without ‘C9+’ neuroimaging phenotype n=67. ^bVFI in a subset of patients: C9+ ALS n=8, C9− ALS with ‘C9+’ neuroimaging phenotype n=14, C9− ALS without ‘C9+’ neuroimaging phenotype n=64. ^cALS-FTD according to the Neary criteria. ^dp<0.01 (Mann-Whitney U). ALS-FRS R, ALS-Functional Rating Scale Revised; ALS-FTD, amyotrophic lateral sclerosis with comorbid frontal temporal dementia; FAB, Frontal Assessment Battery; NA, not available; VFI, Verbal Fluency Index.