Cortical microstructure is associated with disease severity and clinical progression in genetic frontotemporal dementia: a GENFI study

Abstract

The study of genetic frontotemporal dementia (FTD) allows investigating its earliest presymptomatic stages. Using cross-sectional T1-weighted and diffusion-weighted MRI, we test the hypothesis that cortical microstructural alterations, quantified as cortical mean diffusivity (cMD), are detectable earlier and are more strongly associated with clinical progression than cortical thickness (CTh). The sample comprised n = 710 individuals (47.8 ± 13.5 years, 56.6% female, 14.1 ± 3.3 years of education), including 118 symptomatic carriers and 305 presymptomatic carriers with mutations in C9orf72, GRN or MAPT genes, and 287 non-carriers, collected from 24 GENFI sites. A subset of n = 453 individuals (289 carriers, 164 non-carriers) were investigated across Clinical Dementia Rating (CDR) = 0, 0.5 and ≥1 stages. Two subsets had longitudinal clinical outcome measures, including n = 403 individuals (239 carriers, 164 non-carriers) with Cambridge Behavioural Inventory-Revised scores during 2.8 ± 1.6 years, and n = 261 individuals (164 carriers, 97 non-carriers) with CDR Sum-of-Boxes scores during 2.0 ± 0.8 years. Regional cMD and CTh were entered into linear mixed-effects models incorporating age, sex and education as covariates; site, and individual nested within site were random intercepts. The results demonstrated that cMD is more sensitive than CTh to track early cortical injury, with elevated cMD first observed at CDR = 0 in C9orf72 carriers, followed by MAPT carriers (from CDR = 0.5 stage), and by GRN carriers (beginning at CDR ≥ 1). At all stages, cortical microstructural injury had stronger effect size and was more widespread than cortical thinning. In all mutation carrier types, cMD was more strongly associated than CTh with subsequent clinical progression. Cortical microstructure is a promising biomarker to identify at-risk individuals before atrophy and clinical progression, with utility in therapeutic trials.

Fig. 1. Brain maps illustrating the interaction of mutation status (carrier vs non-carrier) with age on the regional cMD and CTh as dependent variables.

The top panel displays β coefficients from linear mixed-effects models indicating regions with a significant interaction effect, meaning that there was a significantly different trajectory of cMD vs age in mutation carriers compared with that in non-carriers (β coefficients depicted with red tones). The lower panel displays β coefficients from linear mixed-effects models for the respective interaction effect for CTh (blue tones). The scale represents β values ranging from 0.25 (light orange) to 0.85 (dark red) for cMD, and from −0.25 (light blue) to −0.85 (dark blue) for CTh. All analyses in mutation carriers are presented stratified by mutation type (C9orf72, GRN and MAPT). Only regions with P-values adjusted for multiple comparisons <0.05 (two-sided tests) are coloured. cMD cortical mean diffusivity; CTh cortical thickness.

Fig. 2. Brain maps illustrating the regional cMD and CTh topographical patterns in mutation carriers across disease stages defined by GENFI-CDR = 0, 0.5 and ≥ 1 vs non-carriers.

The top panel displays β coefficients from linear mixed-effects models indicating increased cMD (red tones) in mutation carriers vs non-carriers, and the lower panel displays β coefficients from linear mixed-effects models indicating regions of reduced CTh (blue tones). The scale represents β values ranging from 0.45 (light orange) to 3.0 (dark red) corresponding to elevated cMD in mutation carriers vs non-carriers, and from −0.45 (light blue) to −3.0 (dark blue) indicating reduced CTh in mutation carriers vs non-carriers. All analyses in mutation carriers are presented stratified by mutation type (C9orf72, GRN and MAPT). Only regions with P-values adjusted for multiple comparisons < 0.05 (two-sided tests) are coloured. cMD cortical mean diffusivity; CTh cortical thickness; GENFI-CDR GENetic Frontotemporal dementia Initiative Clinical Dementia Rating scores.

Fig. 3. Brain maps depicting associations of regional cMD and CTh in mutation carriers at baseline with longitudinal clinical outcome measures.

A Cambridge Behavioural Inventory-Revised (CBI-R), (B) CDR® plus NACC FTLD-NM SOB (GENFI-CDR-SOB). The brain maps display β coefficients from linear mixed-effects models indicating positive (red tones) or negative (blue tones) associations between baseline neuroimaging and longitudinal clinical data. For the associations involving baseline cMD, the scale represents β values ranging from 0.3 (light pink) to 0.7 (dark red) for CBI-R, and from 0.4 (light pink) to 0.95 (dark red) for GENFI-CDR-SOB. For the associations involving baseline CTh and longitudinal clinical data, the scale represents β values ranging from −0.3 (light blue) to −0.7 (dark blue) for CBI-R, and from −0.4 (light blue) to −0.95 (dark blue) for GENFI-CDR-SOB. All analyses are presented stratified by mutation type (C9orf72, GRN and MAPT). Only regions with P-values adjusted for multiple comparisons < 0.05 (two-sided tests) are coloured. CBI-R Cambridge Behavioural Inventory-Revised; cMD cortical mean diffusivity; CTh cortical thickness; GENFI-CDR-SOB GENetic Frontotemporal dementia Initiative Clinical Dementia Rating Sum-of-Boxes.

Fig. 4. Interaction plots illustrating the moderating effect of global cMD and of global CTh at baseline in mutation carriers on the longitudinal evolution of clinical outcome measures.

A Cambridge Behavioural Inventory-Revised (CBI-R), (B) CDR® plus NACC FTLD-NM SOB (GENFI-CDR-SOB). The regression lines represent model-predicted longitudinal clinical progression including 95% confidence bands, at three discrete levels of global cMD and of global CTh: mean, mean - 1 SD, and mean + 1 SD. All analyses are presented stratified by mutation type (C9orf72, GRN and MAPT). Individual dots represent repeated clinical visits. CBI-R Cambridge Behavioural Inventory-Revised; cMD cortical mean diffusivity; CTh cortical thickness; GENFI-CDR-SOB GENetic Frontotemporal dementia Initiative Clinical Dementia Rating Sum-of-Boxes; SD standard deviation.