Disease-modifying effects of TMEM106B in genetic frontotemporal dementia: a longitudinal GENFI study

Abstract

Common variants within TMEM106B are associated with risk for frontotemporal lobar degeneration with TDP-43 pathology (FTLD-TDP). The G allele of the top single nucleotide polymorphism, rs1990622, confers protection against FTLD-TDP, including genetic cases due to GRN mutations or C9orf72 hexanucleotide repeat expansions. However, the effects of interaction between TMEM106B-rs1990622 and frontotemporal dementia (FTD) mutations on disease endophenotypes in genetic FTD are unknown. This longitudinal cohort study was embedded within the GENetic Frontotemporal dementia Initiative (GENFI). We included 518 participants from 222 families [209 non-carriers; 222 presymptomatic carriers (C9orf72 = 79; GRN = 101, MAPT = 42); 87 symptomatic carriers (C9orf72 = 45; GRN = 29; MAPT = 13)] followed for up to 7 years. Using linear mixed-effects models, we examined the effects of a triple interaction between TMEM106B-rs1990622G allele dosage (additive model: 0, 1 or 2 alleles) and autosomal dominant FTD mutations with clinical status, and time from baseline on (i) grey matter volume using a voxel-based analysis; (ii) serum neurofilament light chain (NfL) levels; and (iii) cognitive and behavioural measures. Mean age of participants was 47.9 ± 13.8 years, 58.1% were female and 61% had at least one G allele. C9orf72: rs1990622G allele dosage was associated with less atrophy within the right occipital region in presymptomatic carriers at baseline, and reduced atrophy rate within putamen and caudate nucleus, right frontotemporal regions, left cingulate and bilateral insular cortices in symptomatic carriers over time; lower NfL levels in presymptomatic carriers at baseline; better executive functions and language abilities in presymptomatic carriers; and maintained overall cognitive functions and behaviour in symptomatic carriers over time. GRN: rs1990622G allele dosage was associated with reduced grey matter atrophy rate within the right temporal and occipital regions in presymptomatic carriers, and within the right frontal cortex and insula over time in symptomatic carriers; lower serum NfL levels over time in presymptomatic carriers and lower NfL levels at both baseline and over time in symptomatic carriers; and better global cognitive performance at baseline and higher attention/processing speed scores over time in symptomatic carriers. MAPT: rs1990622G allele dosage was associated with reduced grey matter atrophy rate within the right inferior frontal gyrus in symptomatic carriers, but no effects on serum NfL or cognitive/behavioural measures. TMEM106B-rs1990622G allele dosage showed protective effects on multiple endophenotypes predominantly in GRN and C9orf72 groups. Therefore, TMEM106B genotype should be assessed in clinical trials, particularly of GRN- and C9orf72-related genetic FTD, due to its modifying effects on biomarker, imaging, cognitive and clinical outcomes.

Keywords: TMEM106B; FTLD-TDP; additive model; blood-based biomarkers; cognition; structural MRI.

Figure 1

Longitudinal associations of TMEM106B-rs1990622^G allele dosage and brain volumes. The figure shows clusters of grey matter showing less atrophy rate associated with TMEM106B-rs1990622^G allele dosage over time. Results are derived from voxel-based morphology (VBM) linear mixed-effects models. Images are in neurological display convention. ΔpGM = difference in partial grey matter volume.

Figure 2

Associations of TMEM106B-rs1990622^G allele dosage and serum neurofilament light chain levels. Figure shows predicted serum neurofilament light chain (NfL) levels over time by TMEM106B-rs1990622^G allele dosage. Comparisons are derived from a single linear mixed effects model. Spaghetti plots in background show raw data. Significant differences: Baseline: (i) presymptomatic C9orf72 [β^int: −7.57; 95% confidence interval (CI): −14.57, −0.57; P^int: 0.03]; (ii) symptomatic GRN (β^int: −27.86; 95% CI: −37.44, −18.30; P^int: 2.53 × 10⁻⁸). Slope difference: (i) presymptomatic GRN (β^int: −3.10; 95% CI: −4.75, −1.39; P^int: 0.0003); (ii) symptomatic GRN (β^int: −6.43; 95% CI: −10.43, −2.43; P^int: 0.002). β^int and P^int at baseline: β and P-value for an interaction between TMEM106B-rs1990622^G allele dosage and gene-clinical status at time from baseline = 0. β^int and P^int over time: β and P-value for a triple interaction between TMEM106B-rs1990622^G allele dosage, gene-clinical status and time from baseline. See Supplementary Table 5 for complete outputs of the linear mixed-effects model including test statistics and degrees of freedom.

Figure 3

Associations of TMEM106B-rs1990622^G allele dosage with cognition and behaviour. Figure shows predicted cognitive composite, global cognition and Cambridge Behaviour Inventory-Revised (CBI-R) scores by TMEM106B-rs1990622^G allele dosage. Comparisons in each domain are derived from a single linear mixed-effects model per domain. Spaghetti plots in background show raw data. Refer to Supplementary Table 6 for complete outputs of the linear mixed-effects model, including test statistics and degrees of freedom.