MRI prognostication in syndromes associated with frontotemporal lobar degeneration

Background: People with syndromes associated with frontotemporal lobar degeneration (FTLD) have a very variable prognosis (1-15years). This is only partly explained by clinical phenotype. Structural brain MRI may support prognostication and decision making in clinical practice and clinical trial design.

Methods: Structural 3T MRI brain scans from PiPPiN study participants with diagnosis of Behavioural variant frontotemporal dementia, Progressive supranuclear palsy, Corticobasal syndrome, non-fluent or semantic variant of Primary progressive aphasia had T1w PRISMA MRI scans (n =122, of whom 89 have died). Using Freesurfer, 68 cortical regional thicknesses and 19 grey matter subcortical regional volumes were measured, adjusted for age and sex using data from n=92 healthy volunteers (and total intracranial volume, TIV for the subcortical volumes). Survival effects were analysed by mass univariate Cox proportional hazard models using (i) cortical thickness and subcortical volumes and (ii) components from principal component analysis of MRI regions (PCA). The components were then used as predictors in logistic regression for status 1-5 years from scan.

Results: 

For subcortical regions, the second principal component (PC2) was related to survival. PC2 was characterized by positive loadings in brainstem, ventral diencephalon, pallidum, cerebellum, thalamus, caudate and putamen; and associated protective effect (HR: 0.7, CI: 0.61-0.79, adjusted p: 1 e-07) in the Cox model. PC2 was associated with reduced odds of dying within 2 years (OR: 0.72, CI: 0.56-0.91, adjusted p: 0.03), 3 years (OR: 0.62, CI: 0.47-0.78, adjusted p: 0.0005 ), 4 years (OR: 0.48 CI: 0.34-0.63, adjusted p: 4 e-05) and 5 years (OR: 0.47, CI: 0.33-0.64, adjusted p: 6 e-05). There were no significant interactions with diagnosis. For the cortex, PC2 reflected frontoparietal thickness, but the influence on survival was not significant (HR: 0.92, CI: 0.85-1.00, adjusted p: 0.08). 

Conclusion: 

Subcortical atrophy (worse than expected for age, sex and TIV) is linked to higher risk of death. This risk is present irrespective of diagnosis within the FTLD spectrum and it accumulates over 5 years.