Clinico-genetic findings in 509 frontotemporal dementia patients

Abstract

Frontotemporal dementia (FTD) is a clinically and genetically heterogeneous disorder. To which extent genetic aberrations dictate clinical presentation remains elusive. We investigated the spectrum of genetic causes and assessed the genotype-driven differences in biomarker profiles, disease severity and clinical manifestation by recruiting 509 FTD patients from different centers of the German FTLD consortium where individuals were clinically assessed including biomarker analysis. Exome sequencing as well as C9orf72 repeat analysis were performed in all patients. These genetic analyses resulted in a diagnostic yield of 18.1%. Pathogenic variants in C9orf72 (n = 47), GRN (n = 26), MAPT (n = 11), TBK1 (n = 5), FUS (n = 1), TARDBP (n = 1), and CTSF (n = 1) were identified across all clinical subtypes of FTD. TBK1-associated FTD was frequent accounting for 5.4% of solved cases. Detection of a homozygous missense variant verified CTSF as an FTD gene. ABCA7 was identified as a candidate gene for monogenic FTD. The distribution of APOE alleles did not differ significantly between FTD patients and the average population. Male sex was weakly associated with clinical manifestation of the behavioral variant of FTD. Age of onset was lowest in MAPT patients. Further, high CSF neurofilament light chain levels were found to be related to GRN-associated FTD. Our study provides large-scale retrospective clinico-genetic data such as on disease manifestation and progression of FTD. These data will be relevant for counseling patients and their families.

Fig. 1. Clinical distribution for genetic subgroups of FTD.

The figure shows the distribution of clinical subtypes of FTD according to their genetic diagnosis (see a for the distribution within the full cohort and b within the solved cases). There was no significant difference in the distribution of clinical diagnoses between the genetic subgroups.

Fig. 2. Age of onset and disease severity compared between the genetic subgroups.

Individuals without a definite genetic diagnosis are labeled as “unsolved” (a) The median age of onset was compared between the genetic subgroups with MAPT patients having the earliest age of onset (P = 0.007, post-hoc test). b The severity of FTD at the first presentation as assessed using the FTLD-CDR score was highest (most severe) in patients with the pathogenic variants in MAPT whereas cases with pathogenic TBK1 variants were least severely affected, however, differences were not significant Outliers are depicted as separate dots with their cohort ID.

Fig. 3. Comparison of neurochemical findings between the genetic subgroups.

The figures depict the mean levels of biomarkers and the 95% confidence intervals. Outliers are plotted as separate dots. The figures show the differences between serum NfL (a), CSF NfL (b), pNfH (c), CSF p-Tau (d), CSF Tau (e), CSF A-beta 1-42 (f), serum progranulin (g) and CSF progranulin (h) levels between C9orf72-, GRN-, MAPT-, TBK1-, TARDBP1- and genetically unsolved patients. Only CSF NfL as well as serum and CSF progranulin levels were significantly different between subgroups. Significantly different levels between subgroups are highlighted using asterisks (post-hoc test, *P < 0.5, **P = 0.007, ***P < 0.001). Note, that patients with the clinical diagnosis of FTD/MND were omitted in figures depicting NfL levels as MND would cause elevated levels independent from the genetic diagnosis.

Fig. 4. ROC analysis of CSF progranulin and CSF NfL to predict the presence of a pathogenic GRN variant.

a depicts the ROC analysis of CSF progranulin to predict GRN-associated FTD with a calculated optimum cutoff at 2.04 ng/mL (Sensitivity = 0.886, 1-Specificity = 0.000, Youden’s J = 0.886). b used serum NfL as a marker to predict the presence of a pathogenic variant in GRN. The AUC was 0.78 [0.62–0.94], P = 0.001 and the calculated optimum cutoff was 3258 ng/mL (Sensitivity = 0.889, 1-Specificity = 0.297, Youden’s J = 0.592).