Sixty-seven consecutive patients with ALS followed in our unit were included in this study. Inclusion criteria were probable or definitive disease according to the revised El Escorial criteria (11), with neurophysiological changes supporting diagnosis (12), informed consent, and riluzole medication over, at least, 1 month. Exclusion criteria were pregnancy, smoking, other medical conditions (e.g., hematological diseases, diabetes, lung disorders, heart diseases, previous stroke, or peripheral neuropathy), severe dementia precluding appropriate informed consent, and marked symptoms of respiratory insufficiency, to avoid patient discomfort. No individual in the patient group or in the control group had any reported co-morbidity. The mean body mass index (BMI) of the ALS patients was calculated, as well as the mean BMI among men and women. During the same period, we assessed 82 healthy controls, who were either workers of the academic center or spouses of the patients. This study was approved by the Ethics Committee of Centro Académico de Medicina de Lisboa.

Patients were clinically evaluated at the time of blood sampling and on follow-up (every 3 months for the time of follow-up until death) by applying the functional scale revised—ALS Functional Rating Scale (13), considering its total score and the respiratory subscore (score from questions 10, 11, and 12), and through the assessment of the predicted percentage of forced vital capacity (%FVC). From this group, 78% of ALS cases are called “sporadic,” meaning the cause or causes of the disease are unknown. Nine percent of ALS cases are due to genetic mutations and are inherited from a family member (familial cases).

Furthermore, age, gender, diagnostic delay, region of onset, and presence (or absence) of cognitive changes (by clinical evaluation) were considered, since those variables have been reported as prognostic factors in ALS (14).

Human venous blood samples from ALS patients and controls were collected into K₃EDTA tubes, at the same period of the day (between 8 and 10 a.m.), and immediately centrifuged at 1,040 × g for 10 min, as previously described (9). Plasma was collected and stored at −80°C until analyzed.

Human plasma γ' fibrinogen concentration was quantified using a specific two-site sandwich kit enzyme-linked immunosorbent assay (ELISA; Gamma Couer ELISA kit, Gamma Therapeutics, Portland, OR, USA) as used by other authors (9, 15). The γ' fibrinogen assay was performed using human venous plasma samples isolated from blood collected into K₃EDTA anticoagulant tubes. A monoclonal anti-γ' fibrinogen antibody, specific for the unique carboxyl terminal sequence of the γ' fibrinogen chain, was used as the capture antibody. After incubation with the antibody, unbound non-γ' fibrinogen, such as γAγA fibrinogen, and others plasma components were removed by sequential washing. Secondly, an enzyme-labeled polyclonal anti-fibrinogen detection antibody was added, which binds to the captured γ' fibrinogen, forming a detection antibody sandwich. After this, several sequential washings removed excess detection antibody and a substrate for the horse radish peroxidase (HRP)-labeled detector was added to generate a colored product. Each sample (from either ALS patients or controls) was run in duplicate, and the mean γ' fibrinogen concentration was calculated. Absorbance of the end-product was read at 450 nm on an Infinite M200 microplate reader (Tecan, Männedorf, Switzerland). The γ' fibrinogen concentration was determined using a standard calibration curve prepared with solutions of different concentrations of this protein. The coefficient of variation (CV) between kits was quantified. The intra-assay CV is calculated as the average of the CVs of all samples in a given assay or set of assays. The inter-assay CV is calculated from the mean of the means ± standard deviation (SD) of a low and high γ' fibrinogen plasma control sample included in each kit. Intra- and inter-assay CVs below 10% are considered as good. Inter-assays values for the high control γ' fibrinogen concentration results in a value of 45.86 ± 0.61 mg/dL (mean of means ± SD) with a CV = 1.32% and in 32.54 ± 1.53 mg/dL (mean of means ± SD), CV = 4.70% for the low concentration control γ' fibrinogen in plasma. The intra-assays CV for healthy donors (n = 82) was 5.05 ± 3.72%, and for ALS patients (n = 67) it was 4.67 ± 3.04%.

Forced vital capacity (FVC) was determined with the patients in sitting position, using a computer-based USB spirometer (microQuark, Cosmed, Chicago, IL, USA). The best of three consistent tests was used to define the %FVC.

Three types of statistical analyses were done for the obtained clinical and experimental data: (1) between-group comparisons to test whether γ' fibrinogen values differed between groups (ALS patients vs. controls) were compared using the parametric Student's t-test and the non-parametric Mann-Whitney U-test, and whether those two groups were well-matched demographically; (2) baseline assessments of the relation between γ' fibrinogen and either ALS diagnosis or the values of the clinical variables of interest (ALSFRS-R global and respiratory scores and %FVC) controlling for possible demographic and clinical confounders; and, (3) longitudinal assessments of the relation between γ' fibrinogen and ALS progression. Specifically, for (2) we performed separate multiple regressions on patients' ALSFRS-R and %FVC values (dependent variables), considering as independent variables γ' fibrinogen level, age, gender, presence of comorbid frontotemporal dementia (FTD), region of onset (spinal vs. bulbar vs. respiratory) and diagnosis delay. For (3) we underwent two approaches: firstly, we performed multiple regressions on the change of ALSFRS-R scores and %FVC values per year (dependent variables) that controlled for the aforementioned independent variables (14); secondly, we performed Cox regressions that controlled for the same independent variables, by categorizing ALS patients according to γ' fibrinogen levels (reference group, group 1: <30 mg/dL; group 2: 30–60 mg/dL; group 3: >60 mg/dL), and using continuous γ' fibrinogen levels. In both Cox regressions, the endpoint was set at 5 years after diagnosis (censoring point). Separate coefficients were used for spinal- and bulbar-onset patients in all regressions, as detailed above. Bonferroni correction was applied to the baseline and longitudinal multiple regressions (significant value: p < 0.05/3).

The sample of ALS patients consisted of 40 men (59.7%) and 27 women (40.3%), with a mean age of 63.3 ± 12.3 years (mean ± SD) and a mean disease duration of 24.4 ± 21.4 months (mean ± SD). The control group consisted of 46 men (56.1%) and 36 women (43.9%), with a mean age of 47.3 ± 11.6 years (mean ± SD). Gender distribution was similar across groups (p > 0.30), but controls were significantly younger (p < 0.001; Supplementary Table 1). No statistically significant differences were observed between BMI of healthy donors' (25.91 ± 3.85 kg/m², mean ± SD) and ALS patients' (25.06 ± 2.74 kg/m², mean ± SD; p = 0.27; Supplementary Table 1). γ' fibrinogen levels were significantly higher in ALS patients [51.6 ± 24.5 mg/dL, mean ± SD, lower 95% confidence interval (CI) = 45.6; upper 95% confidence interval (CI) = 57.6) than in controls (38.7 ± 16.6 mg/dL; mean ± SD, lower 95% confidence interval (CI) = 35.0; upper 95% confidence interval (CI) = 42.3); p = 0.0006; Figure 1].

Consistently with the results from the direct comparison of γ' fibrinogen values between patients and controls, multiple regression analysis showed that γ' fibrinogen concentration was increased in ALS patients when controlling for age and gender (p = 0.017; Supplementary Table 2). In ALS, γ' fibrinogen levels were not influenced by gender, age, diagnosis delay, region of onset, or clinical signs of dementia (all p > 0.25). Additionally, we did not detect any association between γ' fibrinogen levels and the clinical variables of interest (ALSFRS-R global score, ALSFRS-respiratory subscore, and %FVC; all p > 0.25; Supplementary Table 3).

Regression analysis disclosed that ALSFRS-R was positively associated with γ' fibrinogen levels, but it was not significant after Bonferroni correction (βγ′fibrinogen = 0.228; p = 0.034). There was no association with ALSFRS-R-respiratory subscore or FVC (p > 0.1).

The survival analysis (Cox proportional hazard regressions) provided strong evidence toward an independent association between higher γ' fibrinogen levels and longer survival (Figure 2; Supplementary Table 5). Both the concentrations of γ' fibrinogen between 30 and 60 mg/dL (group 2; B = −1.343; p = 0.011) and those above 60 mg/dL (group 3; B = −1.814; p = 0.003) were associated with longer survival, when compared to γ' fibrinogen concentrations below 30 mg/dL (group 1; reference group; Supplementary Table 5). Moreover, when coding γ' fibrinogen values continuously, higher γ' fibrinogen values was an independent predictor of longer survival (B = −0.014; p = 0.045; Supplementary Table 5). Survival analysis showed, as expected, a negative association between age and survival (B = 0.039; p = 0.036 for γ' fibrinogen values coded categorically; B = 0.040; p = 0.029 for γ' fibrinogen values coded continuously). The average survival in our population was similar to other ALS studies (14).