Around a third of frontotemporal dementia (FTD) is caused by mutations in three main genes: progranulin (GRN), microtubule associated protein tau (MAPT) and chromosome 9 open reading frame 72 (C9orf72). Pathophysiological processes induced by these mutations may differ and some FTD subtypes have been associated with damage to the white matter (WM). This damage is visible as hyperintense signal on T2-weighted magnetic resonance (MR) imaging. The Genetic FTD Initiative (GENFI) is a longitudinal cohort study aimed at furthering understanding of disease in individuals with these three mutations.


      T1 and T2-weighted MR sequences were acquired for 180 subjects within the GENFI cohort, divided into 76 non-carriers, 61 presymptomatic mutation carriers (25 GRN, 8 MAPT and 28 C9orf72) and 43 symptomatic carriers (7 GRN, 13 MAPT and 23 C9orf72). WM hyperintensities (WMH) were automatically segmented using an algorithm based on outlier modelling in a multivariate Gaussian mixture model. Location in the WM was coded according to 1) a relative distance between ventricles and the cortex, divided into four equidistant layers (1st layer periventricular, 4th layer juxtacortical), and 2) to the closest cortical lobe. WMH in the basal ganglia were also investigated. Infratentorial regions were excluded from the analysis. Log-transformed WMH volumes were adjusted for age, gender, total intracranial volume, scanner type and years before expected onset.


      Symptomatic GRN carriers had significantly more WMH than all other groups, but no differences could be detected between other subgroups (table 1). Symptomatic GRN carriers appeared to have a larger volume of WMH in the frontal and occipital regions compared with other symptomatic groups and presymptomatic GRN cases. Differences were most noticeable in periventricular layers, with higher WMH volumes in GRN cases compared with other groups.


      WMH patterns differ across FTD genetic subtypes, with symptomatic GRN carriers displaying particularly predominant fronto-occipital periventricular WMH. Future research should explore the pathophysiological mechanisms of WMH within genetic FTD.
      Figure thumbnail fx1
      Figure 1Effect sizes of observed differences between groups at local (layers and lobes) and global scales; PS - Presymptomatic ; S - Symptomatic.
      Table 1Comparison of global WMH volumes (mL) across mutation groups corrected for age, TIV, sex, scanner type, year before expected onset
      C9orf720.50[0.35 0.71]0.44[0.28 0.67]NC vs PS 0.99NC vs S 0.60PS vs S 0.87
      GRN0.50[0.40 0.62]0.56[0.39 0.81]1.51[0.77 2.94]NC vs PS 0.60NC vs S 0.0030PS vs S 0.012
      MAPT0.53[0.27 1.06]0.59[0.34 1.02]NC vs PS 0.87NC vs S 0.60PS vs S 0.87
      p-valuesNAC9orf72 vs GRN 0.86

      C9orf72 vs MAPT 0.99

      GRN vs MAPT 0.86
      C9orf72 vs GRN 0.0048

      C9orf72 vs MAPT 0.67

      GRN vs MAPT 0.012
      Acronyms: NC - Non-carriers; PS - Presymptomatic; S - Symptomatic; NA - Non applicable; CI - Confidence interval.