FALS included 225 index cases of unrelated families with probable or definite ALS, 131 males and 94 females (M:F ratio=1.4:1) with mean age of onset at 55 years (SEM 1, median 55 years, range 21–85 years,) and mean disease duration of 51 months (SEM 4, median 33 months, range 3–336 months, including 32 censored data). SALS included 725 patients, 420 males and 305 females (M:F ratio=1.4:1), with mean age of onset at 56 years (SEM 1, median 58 years, range 21–87 years) and mean disease duration of 72 months (SEM 4, median 48 months, range 1–354 months, including 90 censored data). Data were censored at the last date of the patients visit. Control samples were age-matched Caucasian individuals of French background (n=580). Most DNA samples were collected over the past 15 years at the ALS National Referal Center of Pitié-Salpêtrière Hospital (Paris). During the same period, the number of ALS patients followed by the center was 7784. Some families were collected by other French ALS Centers belonging to the French ALS study group. Criterion for family inclusion was that at least two members were affected. In FALS the disease was transmitted in a dominant (n=164) or putatively recessive (all affected members belonged to a single sibling with none of the parent affected, n=47) manner. In 14 remaining families, the ALS patients were distant relatives (more than 2nd degree relatives). All participants signed a consent form allowing performing research. Protocols were approved by the Medical Research Ethics Committees of the 'Comité d'Ethique de la Pitié-Salpêtrière' and 'Assistance Publique-Hôpitaux de Paris'.
All FALS have previously been screened for SOD1, ANG, TARDBP, FUS, DAO and OPTN. FALS without male-to-male transmission (n=130) had also been analysed for the X linked UBQLN2 gene. Patients with mutation in SOD1 (n=26), ANG (n=1), TARDBP (n=8), FUS (n=13), DAO (n=1), OPTN (n=2), UBQLN2 (n=1) and with no previously identified genetic defect (n=173) were included in the C9ORF72 repeat analysis. For each patient, sequencing of ALS related genes and determination of C9ORF72 repeat length were performed on the same DNA sample.
The analysis of the C9ORF72 repeat was performed by a repeat-primed PCR amplification as previously described. This analysis was completed by a classical fluorescent fragment-length analysis allowing the detection of non-expanded C9ORF72 alleles. Both analyses were repeated twice for each patient sample to ensure reproducibility of the results, determine whether the repeat expansion was present at the heterozygous or homozygous state and because repeat primed assay efficiency highly depends on DNA concentration and quality. DNA of some homozygote individuals was sequenced to precisely correlate the number of base pairs in the fluorescent assay with the number of GGGGCC repeats. For clinical comparison, only patients with GGGGCC repeat numbers greater than 50 were included in the C9ORF72 group. Although no biological data is available to support that 50 repeats is a suitable cut-off to determine pathogenicity, this 50 repeat cut-off corresponds to the detection limit of the method we used.
Clinical data could be partially recovered for up to 334 relatives belonging to the 225 analysed families and 725 SALS cases. The exact number of patients included in each statistical analysis depended on the availability of the corresponding data and is summarised in Table 1 and Table 2. Clinical parameters including gender ratio (n=334 FALS, 719 SALS), age of onset (n=290 FALS, 512 SALS), site of onset (n=283 FALS, 703 SALS), disease duration (n=264 FALS, 512 SALS) and presence of FTD behavioural variant (n=162 FALS) were compared in five FALS groups represented by patients with long C9ORF72 repeat, mutations in SOD1, TARDBP, FUS or other FALS with still unidentified mutations (Table 1). In addition, 2 groups of SALS patients either positive or negative for expanded C9ORF72 repeats were included. Disease duration was also compared between C9ORF72 patients with bulbar (n=42) or spinal (n=70) onset and with (n=25) or without (n=28) FTD. FALS were arbitrarily separated according to age of onset (≤40, 41–50, 51–60, 61–70, ≥71) to study the frequency of ALS-related genes in each subgroup. Two consecutive subgroups were pooled when their distributions were similar.
Statistical analyses were performed as previously described. Briefly, the Cox proportional hazards regression model was used to compare age at onset and disease duration between the five groups of FALS patients. If the difference was significant (p<0.05), a log rank analysis compared groups by pairs. Proportions of patients classified according to gender, site of onset or presence of FTD were compared by pairs using Fisher's exact tests. Proportions of C9ORF72-positive subgroup of patients (n=70) with available cognitive status data (FTD or pure ALS) classified according to site of onset (bulbar or spinal) were also compared using Fisher's exact test. Statistical analyses were performed using the SPSS V.11.0 data analysis software (SPSS Inc.).