Objective The lack of prognostic biomarkers in patients with amyotrophic lateral sclerosis (ALS) induced researchers to develop clinical evaluation tools for stratification and survival prediction. We assessed the correlation between patterns of functional involvement, considered as a cumulative number of body regions involved, and overall survival in a population-based series of patients with ALS (PARALS).
Methods We derived the functional involvement of four body regions at diagnosis using ALSFRS-R subscores for bulbar, upper limbs, lower limbs and respiratory/thoracic regions. We analysed the effect of number of body regions involved (NBRI) at diagnosis on overall survival, adjusting for age at onset, sex, site of onset, diagnostic delay, forced vital capacity, body mass index, mutational status, cognition and comparing it with King’s staging system.
Results The NBRI was strongly related to survival, with a progressive increase of death/tracheostomy risk among groups (two body regions HR=1.24, 95% CI 1.06 to 1.45, p=0007; three body regions HR=1.65, 95% CI 1.38 to 1.98, p<0.001; four body regions HR=2.68, 95% CI 2.11 to 3.39, p<0.001). Using ALSFRS-R score, the consistency between the number of regions involved and King’s clinical stage at diagnosis was very high (81%). The evaluation of respiratory/thoracic region and cognition allowed to subdivide patients into different prognostic categories. Regional spreading of the disease is associated with survival, independently from the initial region involved.
Conclusions The evaluation of NBRI, with the inclusion of initial respiratory/thoracic involvement and cognition, can be useful in many research fields, improving the stratification of patients. Our findings highlight the importance of the spatial spreading of functional impairment in the prediction of ALS outcome.
Statistics from Altmetric.com
Amyotrophic lateral sclerosis (ALS) is a neurodegenerative disorder characterised by the involvement of upper and lower motor neurons in different body regions. Disease onset occurs in bulbar or spinal regions, but spreading patterns remain largely unpredictable in a single patient. ALS phenotypic heterogeneity is widely recognised1 and its impact on disease progression and overall survival has been confirmed also in population-based series.2 The underlying mechanisms of this variability are still poorly understood: different genetic or environmental factors have been analysed, but genotype–phenotype correlations and environmental risk factor–phenotype correlations remain limited and inconclusive3
Older age, bulbar onset and diagnostic delay are generally considered as negative prognostic factors for overall survival, whereas gender and El Escorial criteria have shown inconsistent results.4 Also, nutritional status,5 cognition6 and respiratory function7 at diagnosis have been considered as prognostic markers. Among genetic mutations, c9orf72 expansion leads to a shorter survival compared with patients with non-mutated ALS.8
Recently, innovative prognostic biomarkers, such as neurofilaments,9 showed consistent association in disease progression prediction. Considering phenotypic heterogeneity, a multimodal approach was used to develop the ENCALS Survival Prediction Model,10 a validated model able to stratify patients in prognostic categories according to different candidate predictors.
In clinical practice, the ALS Functional Rating Scale revised (ALSFRS-R) is the most widely used surrogate marker of disease progression.11 Total ALSFRS-R score at diagnosis is a strong survival predictor, even when adjusted for forced vital capacity (FVC), age, sex and symptom duration12 and its progression rate provides an additional predictive index beyond ALSFRS-R alone.13 Two main ALS clinical staging systems, the King’s College staging system and the Milano-Torino staging (MITOS) system, can be calculated from the ALSFRS-R score.14 15 Recently, the metric quality of ALSFRS-R has been questioned, demonstrating that the total score lacks unidimensionality and does not fulfil fundamental measurement requirements.16 17
The aim of the present study was to assess the patterns of functional involvement in four body regions (upper limbs, lower limbs, bulbar and respiratory/thoracic regions) at diagnosis and their correlation to overall survival, using ALSFRS-R score as a multidimensional scale in a population-based series of patients with ALS.
All patients (n=1105) meeting the revised El Escorial Criteria for defined, probable and probable-laboratory supported ALS18 diagnosed in the period 2007–2014 in Piemonte and Valle d’Aosta regions, Italy, were included. Thirteen patients were excluded from the analyses because of missing information on clinical evolution and further 30 cases were excluded for their long disease course (survival from diagnosis >10 years, see the online supplementary material).
For each patient, we collected age at onset, sex, site of onset, date of diagnosis, diagnostic delay and death/tracheostomy. Survival was assessed from 1 January 2007, until 31 December 2017, or until death/tracheostomy. In a subgroup of patients (58.5%), we obtained also FVC and body mass index (BMI). Mutational status for c9orf72, SOD1, TARDBP and FUS was available in 84.1% of patients. Neuropsychological evaluation was performed at diagnosis (diagnosis-evaluation interval ≤3 months) in 622 patients (58.6%). Patients’ cognitive status was classified according to the revised amyotrophic lateral sclerosis-frontotemporal spectrum disorder (ALS-FTD) Consensus Criteria19: the neuropsychological batteries used for classification were described in our previous work.20 To evaluate the effect of cognitive impairment, we subdivided patients into two categories: cognitively impaired patients (corresponding to ALS-FTD, ALSbi, ALSci, ALScbi categories in the revised ALS-FTD Consensus Criteria) and cognitively normal patients (corresponding to ALS-CN category).
Functional involvement assessment using the ALSFRS-R score
We derived the number of body regions involved (NBRI) at diagnosis using ALSFRS-R subscores for bulbar, upper limbs, lower limbs and respiratory/thoracic regions. We considered ALSFRS-R items 1, 2 and 3 for the bulbar region, items 4 and 5 for the upper limb region, items 8 and 9 for the lower limb region, and items 10, 11 and 12 for the thoracic/respiratory region. A region was considered to be affected when at least 1 point was lost from the maximum total subscore. Item 6 (dressing/hygiene) and item 7 (turning over in bed) were excluded from the analysis, being not specific for upper or lower limb involvement. In our series, as in a previous paper,21 these items never occurred alone in patients with upper and/or lower limb involvement.
Functional involvement assessment using ALSFRS-R score and neuropsychological evaluation
In the subset of 622 patients with neuropsychological evaluation at diagnosis, we considered the presence of cognitive impairment as another body region involved, stratifying patients in a further classification based on the number of motor body regions plus cognitive (NBRI-C). Using this classification, for example, a patient with bulbar and upper limb motor involvement and cognitive impairment has been considered as having three body regions involved (NBRI-C=3).
Differences of discrete and continuous variables of interest between numbers of regions involved were analysed using the χ2 test, Student t-test or Kruskal-Wallis test by ranks and Mann-Whitney U test, as appropriate. A p value <0.05 was considered significant.
The association between numbers of regions involved and survival was assessed using Cox proportional hazards models, adjusted for sex, age groups (15–44, 45–59, 60–74 and >75 years), diagnostic delay, ALSFRS-R total score and site of onset. Kaplan-Meier curves and log-rank test were calculated. We also performed sensitivity analysis considering the effect of FVC (below and above 80%) and BMI categories (BMI≤18.5 ‘underweight’, BMI 18.5–25 ‘normal weight’, BMI 25–30 'overweight or preobesity’, BMI≥30 ‘obesity’) according to the WHO classification,23 and excluding genetic mutation (c9orf72, SOD-1, TARBDP and FUS) carriers.
Data were analysed using Stata V.13.1.
Standard protocol approvals, registrations and patient consents
The study design was approved by the Ethical Committee of the Azienda Ospedaliero-Universitaria Città della Salute (Prot. No. 0036344). Patients signed written informed consent.
Of the 1062 patients included, 937 (88.2%) deceased or underwent tracheostomy before the end of follow-up. Five-hundred and sixty-five were men (53.2%), and the mean age at onset was 66.3 years (SD=11.0). At diagnosis, one region was functionally affected in 492 patients (46.3%), two regions in 288 patients (27.1%), three in 196 patients (18.5%) and four in 86 patients (8.1%). The number of affected regions at diagnosis progressively increased with the increase of the age at onset (p=0.001). Patients’ characteristics are shown in table 1.
NBRI was related to the site of onset (p<0.001) with more widespread disease at diagnosis in patients with respiratory onset compared with other groups. Mutational status did not affect the spreading of the disease at diagnosis (p=0.337).
The overall mean diagnostic delay was 0.98 years. The intervals between onset and diagnosis were marginally different in relation to the NBRI (one region: 0.67 years, IQR=0.41–1.03; two regions: 0.75 years, IQR=0.50–1.08; three regions: 0.75 years, IQR=0.50–1.08; four regions: 0.81 years, IQR=0.50–1.16; p=0.04), not showing a significant increase trend among groups (see Mann-Whitney U test in the online supplementary table). The distribution of ALSFRS-R total scores at diagnosis was significantly different (one region: median score 45, IQR=44–46; two regions: 42, IQR=39–44; three regions: 37, IQR=33–40; four regions: 29, IQR=24–36; p<0.001), as well as the rate of disease progression (ΔALSFRS-R=48 – ALSFRS-R at diagnosis/diagnostic delay in months, p<0.001) and cognitive impairment (p=0.003).
A more widespread functional involvement at diagnosis in the elderly could be explained postulating a greater diagnostic delay due to the existence of several clinical conditions that could mimic an initial motor neurons impairment (such as sarcopenia, brain vascular ageing and so on): in our series, we do not confirm this hypothesis, finding no difference in the diagnostic delay among different age classes (Kruskal-Wallis test, p=0.376). Another possible explanation is the presence of a more aggressive disease in the elderly, which is confirmed considering the overall mean survival reduction across different age classes (online supplementary material).
Different combinations of NBRI are shown in table 2.
Survival analysis according to the number of regions involved at diagnosis
Median survival time was inversely related to NBRI at diagnosis (one region: 2.9 years, IQR=2.1–5.6; two regions: 2.6 years, IQR=1.7–4.2; three regions: 1.8 years, IQR=1.2–3.1; four regions: 1.5 years, IQR=0.8–2.4; p=0.001).
Table 3 shows the Cox proportional hazard models for different adjustments. Adjusting for age and sex, the NBRI was strongly related to survival, with a progressive increase of death/tracheostomy risk among groups (two body regions HR=1.24, 95% CI 1.06 to 1.45, p=007; three body regions HR=1.65, 95% CI 1.38 to 1.98, p<0.001; four body regions HR=2.68, 95% CI 2.11 to 3.39, p<0.001) (table 3).
The association resulted to be even more significant adjusting also for diagnostic delay and site of onset (two body regions HR=1.53, 95% CI 1.30 to 1.79, p<0001; three body regions HR=2.09, 95% CI 1.75 to 2.50, p<0.001; four body regions HR=4.63, 95% CI 3.61 to 5.93, p<0.001).
To assess the effect of global functional impairment, we adjusted the analysis also for ALSFRS-R total score at diagnosis. Also, in this case, the association with the number of regions involved remained significant (table 3).
Kaplan-Meier curves for each group of NBRI are reported in figure 1. The number of involved regions proportionally increased the risk of death/tracheostomy (log-rank test, p≤0.001).
Sensitivity analysis: wild-type patients, nutritional status and respiratory function
We further performed a sensitivity analysis considering only wild-type patients (800 patients) and the association remained significant.
In a second sensitivity analysis in 621 patients (58.5%), which considered also measures of both FVC and BMI at diagnosis, the NBRI remained significantly associated with a cumulative increased risk of death/tracheostomy.
Sensitivity analysis: cognitive impairment spreading
In this sensitivity analysis, we considered cognition as a further body region involved. As motor impairment derived by ALSFRS-r reflects neuronal damage, in the same way, cognitive impairment assessed by neuropsychological evaluation could show disease spreading through non-motor areas.
We performed Kaplan-Meier curves (log-rank test, p≤0.001, see the online supplementary material) and Cox proportional hazard models adjusted for age, sex and diagnostic delay, showing that also NBRI-C classification is a strong survival determinant (table 3).
Number of regions and King’s stage comparison at diagnosis
The study population was also stratified according to King’s stage at diagnosis (table 4).
Using ALSFRS-R score, the consistency between the NBRI and King’s clinical stage at diagnosis was 81%. The inconsistency between the two staging systems was mainly related to an under-representation of initial respiratory/thoracic involvement in King’s system.14
Survival models using the King’s staging system were also assessed and their HRs were reported and compared with NBRI. King’s staging, adjusted only for age and sex, showed statistically cumulative and significant association with overall survival (table 3). Nevertheless, adjusting for diagnostic delay and site of onset the trend of increasing risk of death/tracheostomy resulted to be inverted in stages 3 and 4. Adjusting also for ALSFRS-R total score, both stage 3 (HR 1.26, 95% CI 0.96 to 1.63) and stage 4 (HR 0.85, 95% CI 0.59 to 0.24) lose their significance.
In this population-based study, which included 96% of patients with ALS incident in Piemonte and Valle d’Aosta Italian regions during the 2007–2014 period, we evaluated the importance of assessing the diffusion of motor involvement at diagnosis as a prognostic biomarker. First, we observed that the spatial spreading at diagnosis, considered as NBRI, was strongly related to overall survival, independently from the specific body region involved. Second, we found that the inclusion of the involvement of the respiratory/thoracic region provides useful information for the classification in more or less widespread disease. Third, we pointed out that the evaluation of cognitive dysfunction leads to a better stratification of patients’ outcomes.
The phenotypical heterogeneity in ALS is determined by various combinations of clinical traits, such as site of onset, the predominant involvement of upper or lower motor neurons and the rate of disease progression.24 Such heterogeneity has profound implications in determining clinical outcomes and in designing clinical trials.25 26
Recent neuropathological studies have proposed an explanation to the focality of onset and to upper and lower motor neurons involvement as a corticofugal spreading of TDP43 pathology,27 28 whereas the determinants of progression rate are far to be elucidated. They can reflect individual differences in disease-modifying genes, environmental modifiers exposure or both.29
The pattern of disease spreading has been related to survival30; for example, in lower limb-onset, the time to involvement of the following limb has been shown to be a prognostic factor regardless of the initial direction of spread.31 Besides, the time to bulbar involvement in spinal onset patients,32 the time to generalisation (ie, the time of spreading of the clinical signs from spinal or bulbar localisation to both)33 and the interval from onset to involvement of the second region have been reported to correlate with survival.21
In the present study, we have shown that the regional spreading of ALS symptoms can be easily derived from the ALSFRS-R scale, which is characterised by high internal consistency, reliability and responsiveness to change.34 The use of a validated scale minimises the risk of misclassifications of patients due to variability in neurological evaluation, allowing its administration in different clinical settings. Moreover, considering the worldwide use of ALSFRS-R score, our findings can be easily replicated using data from other population-based or clinical trial datasets. The use of ALSFRS-R regional subscores considered the multidimensionality of the scale, allowing to derive information on the spatial spreading of lesions in a way that is otherwise impossible to obtain with the total raw score. The NBRI at diagnosis resulted to have a remarkable prognostic value, independent from age, sex, site of onset, diagnostic delay, ALSFRS-R total score, FVC and nutritional status.
C9orf72 gene expansion is generally considered a negative prognostic factor, leading to an early age of onset, increased risk of dementia35 and shorter survival.36 Considerable phenotypic heterogeneity occurs across the other ALS-related genetic mutations.37 We found that the regional spreading of functional impairment at diagnosis is not significantly different in patients with c9orf72 expansion and other genetic mutations compared with wild-type patients, suggesting that the negative effect of c9orf72 mutations on overall survival is probably related to other factors, such as cognitive impairment and to its negative influence on mechanical ventilation and enteral nutrition adherence.38
We compared the prognostic yield of NBRI to King’s staging system, a validated clinical score that summarises the anatomical spread of disease.39 The King’s staging system considers as milestones, the progressive involvement of bulbar, upper and lower limb regions to classify stages 1, 2 and 3, but it does not consider the initial respiratory/thoracic involvement.22 In our population-based series, this difference resulted to be the major determinant of the lack of consistency between NBRI and King’s stage.
The involvement of the thoracic region is frequently underestimated, especially in the early phases of the disease: this can be due to many factors, such as the difficulty of assessing upper motor neuron (UMN) and lower motor neuron (LMN) signs at thoracic level, the limited number of neurophysiological studies of paravertebral muscles performed and the need of pulmonary function tests to detect early respiratory impairment, which can be less accurate in patients with bulbar symptoms. Despite these limitations, we have found that the evaluation of functional respiratory involvement at diagnosis derived from ALSFRS-R scale can provide reliable information about the involvement of the thoracic region. According to our findings, as indicated by the sensitivity analysis adjusted for FVC, thoracic region involvement attains a prognostic value as a sign of more widespread disease.
Our findings are in keeping with other studies focused on the symptoms burden in early disease stages.12 13 21 30 33 40 However, we have provided novel evidence on the prognostic value of the number of body regions functionally involved.
Cognitive impairment has been widely recognised as one of the major determinants of disease outcome.6 Recent findings suggested that ALS motor and cognitive components may worsen in parallel during disease progression.20 In our sensitivity analysis with NBRI-C, we found an exponential increase in risk across categories, confirming that the spread of the disease through cognitive areas is a cause of worse prognosis, independently from specific motor impairment.
Our study is not without limitations. First, we could not discriminate whether the functional involvement of a body region is mainly related to UMN or LMN dysfunction. We assume that functional involvement that leads to the loss of point in ALSFRS-R regional subscore is related to the presence of either UMN or LMN signs or both, considering that the body region with the highest UMN involvement at onset in general also had the highest frequency of LMN signs and vice versa.39 Second, we could not include in our analysis any specific biomarkers (such as neurofilaments – pNfH and NfL): recent findings showed a correlation between pNfH and NfL levels and the number of regions with both UMN and LMN involvement, demonstrating their higher correlation to the disease spreading than to rate of progression.9 Third, our study has not been designed to provide a personalised prediction model, like the ENCALS Survival Prediction Model, but to specifically untangle the role of the early spatial spread of functional impairment and the underlying pathology, from the mere progression rate. For this reason, the NBRI is not recommended to be used as a prognostic marker in a single patient. Fourth, due to their retrospective nature, our findings should be confirmed in prospective cohorts from different populations, to assess their validity and generalisability.
For this reason, we suggest that the evaluation of NBRI can be useful in many research fields (ie, neuropathology, prognostic biomarkers and neuroimaging), shedding new light on the importance of evaluating the spreading of functional impairment in the prediction of ALS outcome.
This work was in part supported by the Italian Ministry of Health (Ministero della Salute, Ricerca Sanitaria Finalizzata, grant RF-2016-02362405), the European Commission’s Health Seventh Framework Programme (FP7/2007-2013 under grant agreement 259867), the Italian Ministry of Research and University (Progetti di Rilevante Interesse Nazionale 2017, grant 2017SNW5MB), the Joint Programme - Neurodegenerative Disease Research (Strength, ALS-Care and Brain-Mend projects), granted by Italian Ministry of Education, University and Research, and Fondazione Mario e Anna Magnetto. This study was performed under the Department of Excellence grant of the Italian Ministry of Education, University and Research to the ‘Rita Levi Montalcini’ Department of Neuroscience, University of Torino, Italy. The authors give special thanks to their colleagues Barbara Iazzolino, Laura Peotta and Jean Pierre Zucchetti for sharing the data on neuropsychological evaluations
Contributors UM: study concept and design, drafting on the manuscript, critical revision of the manuscript for important intellectual content and study supervision. AndC, AdrC, FD: drafting on the manuscript, critical revision of the manuscript for important intellectual content and study supervision. GM and LM: drafting on the manuscript, critical revision of the manuscript for important intellectual content. CM, MARD, AntC, RV, MCT, MG, MB, SD, LC and FDM: drafting on the manuscript, critical revision of the manuscript for important intellectual content, administrative, technical and material support.
Funding The authors have not declared a specific grant for this research from any funding agency in the public, commercial or not-for-profit sectors.
Disclaimer The sponsor organizations had no role in data collections and analysis and did not participate to writing and approving the manuscript. The information reported in the manuscript has never been reported elsewhere.
Competing interests Adriano Chiò serves on scientific advisory boards for Mitsubishi Tanabe, Roche, Biogen, and Cytokinetics, and has received a research grant from Italfarmaco. Andrea Calvo has received research grant from Cytokinetics.Umberto Manera, Margherita Daviddi, Antonio Canosa, Rosario Vasta, Maria Claudia Torrieri, Maurizio Grassano, Maura Brunetti, Sandra D’Alfonso, Lucia Corrado, Fabiola De Marchi, Cristina Moglia, Fabrizio D’Ovidio, Letizia Mazzini, report no conflicts of interest.
Patient consent for publication Not required.
Provenance and peer review Not commissioned; externally peer reviewed.
Data availability statement Data are available upon reasonable request.
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.