Spectrum and frequency of genetic variants in sporadic amyotrophic lateral sclerosis

Abstract

Therapy of motoneuron diseases entered a new phase with the use of intrathecal antisense oligonucleotide therapies treating patients with specific gene mutations predominantly in the context of familial amyotrophic lateral sclerosis. With the majority of cases being sporadic, we conducted a cohort study to describe the mutational landscape of sporadic amyotrophic lateral sclerosis. We analysed genetic variants in amyotrophic lateral sclerosis-associated genes to assess and potentially increase the number of patients eligible for gene-specific therapies. We screened 2340 sporadic amyotrophic lateral sclerosis patients from the German Network for motor neuron diseases for variants in 36 amyotrophic lateral sclerosis-associated genes using targeted next-generation sequencing and for the C9orf72 hexanucleotide repeat expansion. The genetic analysis could be completed on 2267 patients. Clinical data included age at onset, disease progression rate and survival. In this study, we found 79 likely pathogenic Class 4 variants and 10 pathogenic Class 5 variants (without the C9orf72 hexanucleotide repeat expansion) according to the American College of Medical Genetics and Genomics guidelines, of which 31 variants are novel. Thus, including C9orf72 hexanucleotide repeat expansion, Class 4, and Class 5 variants, 296 patients, corresponding to ∼13% of our cohort, could be genetically resolved. We detected 437 variants of unknown significance of which 103 are novel. Corroborating the theory of oligogenic causation in amyotrophic lateral sclerosis, we found a co-occurrence of pathogenic variants in 10 patients (0.4%) with 7 being C9orf72 hexanucleotide repeat expansion carriers. In a gene-wise survival analysis, we found a higher hazard ratio of 1.47 (95% confidence interval 1.02-2.1) for death from any cause for patients with the C9orf72 hexanucleotide repeat expansion and a lower hazard ratio of 0.33 (95% confidence interval 0.12-0.9) for patients with pathogenic SOD1 variants than for patients without a causal gene mutation. In summary, the high yield of 296 patients (∼13%) harbouring a pathogenic variant and oncoming gene-specific therapies for SOD1/FUS/C9orf72, which would apply to 227 patients (∼10%) in this cohort, corroborates that genetic testing should be made available to all sporadic amyotrophic lateral sclerosis patients after respective counselling.

Keywords: amyotrophic lateral sclerosis; genetics; motor neuron disease.

Graphical Abstract

Figure 1

Study flow chart. Two thousand three hundred and forty ALS patients of 16 centres of the German MND-NET with a negative family history of ALS were enrolled in the study. Forty-seven patients dropped out due to a change in the family history of ALS or a change in the final diagnosis. Nineteen patients dropped out because quality criteria in sequencing were not met. Seven patients were excluded as no C9orf72 HRE data were available. The final analysis data set contained 2267 patients. All variants were classified according to the ACMG guidelines. C, ACMG variants class; HRE, hexanucleotide repeat expansion; NGS, next-generation sequencing; wt, wild-type.

Figure 2

Pie charts of the abundance of gene variants in all investigated genes. Patients with multiple Class 4 variants (n = 3), multiple Class 3 variants (n = 30) and patients harbouring the C9orf72 HRE and an additional pathogenic variant (n = 7) were excluded from the graphical visualization as they could not be assigned to a single gene. The left pie plot presents Class 3/variant of uncertain significance variants as a separate group with 15.4% of all patients falling into this group. The right pie plot shows only Class 4 variants, the most abundant pathogenic variants are in the C9orf72 gene (C9orf72 HRE), followed by pathogenic variants in the SOD1, NEK1, TARDBP, SQSTM1, TBK1, OPTN, FUS, FIG4, SETX, MAPT, ARHGEF28, DCTN1, ERBB4, CHMP2B, GRN, VCP, HNRNPA2B1 and NEFH genes. n, total number of patients included in the respective analysis, numbers in chart slices: percentage of all patients in the respective analysis.

Figure 3

Graphical visualization of pathogenic variants in selected genes of our cohort. Variants are depicted in the lollipop plot overlying the respective gene structure. Lollipops with brown filling represent novel pathogenic variants, lollipops with purple filling represent known pathogenic variants found in our cohort and lollipops with black filling represent known variants in the literature. The horizontal bars represent functional protein domains. Arabic numerals correspond to amino acid numbers. Disord, disordered protein region according to MobiDB-lite rules; fs, frameshift; AA, amino acid; AZI2, 5-azacytidine induced 2; TANK, TRAF family member-associated NF-kappa-B activator; TBKBP1, TBK1 binding protein 1; MYO6, myosin VI; HD domain, histidine-aspartate domain.

Figure 4

Point diagram of the distribution of demographic and clinical patient data in different groups. (A) Distribution of age at onset shows a significantly younger age at onset for patients with a single pathogenic variant [without the C9orf72 hexanucleotide repeat expansion (HRE), n = 96] compared with the reference group (n = 1599) and younger age at onset for patients with combinations of pathogenic variants (n = 3) compared with patients with a single pathogenic variant. (B) A higher Δ ALSFRS-R/m for patients harbouring the C9orf72 HRE and an additional pathogenic variant (n = 7) than for patients with the C9orf72 HRE alone (n = 141) are shown (two-sided t-test followed by Benjamini–Hochberg corrections *P < 0.05, *P < 0.005).

Figure 5

Forrest diagram of gene-wise HRs for death from any cause. Significant HRs and their corresponding P-values are visualized in bold. In brackets, the 95% confidence interval is given. Patients without a causal variant serve as references (P-value: P-values from the Wald statistics * P < 0.05). N, number of individuals with survival data harbouring pathogenic variants in the respective gene; # events, total number of reported deaths from any cause within the observation period (global P-value: overall significance of the model using the likelihood ratio test).