Data from EMPOWER and BENEFIT-ALS are from the placebo groups. All groups are described in the introduction.
Data are based on a Cox proportional hazards regression model of time to clinical events. The 3 percentages are slope of change in percent predicted slow vital capacity per month.
Customize your JAMA Network experience by selecting one or more topics from the list below.
Andrews JA, Meng L, Kulke SF, et al. Association Between Decline in Slow Vital Capacity and Respiratory Insufficiency, Use of Assisted Ventilation, Tracheostomy, or Death in Patients With Amyotrophic Lateral Sclerosis. JAMA Neurol. 2018;75(1):58–64. doi:10.1001/jamaneurol.2017.3339
What is the natural history of decline in respiratory function, as measured by percentage predicted slow vital capacity, and its relationship to major clinical events and death in amyotrophic lateral sclerosis?
In this study of 893 placebo-treated patients with amyotrophic lateral sclerosis who participated in 2 clinical trials and an amyotrophic lateral sclerosis clinical trials database, older age at disease onset and lower functional scores were associated with faster rate of slow vital capacity decline. Slow vital capacity change over time was associated with meaningful clinical events, including time to respiratory insufficiency or death.
Findings suggest that decline in respiratory function measured by slow vital capacity is an important indicator of clinical progression and could be a useful end point in future amyotrophic lateral sclerosis clinical trials.
The prognostic value of slow vital capacity (SVC) in relation to respiratory function decline and disease progression in patients with amyotrophic lateral sclerosis (ALS) is not well understood.
To investigate the rate of decline in percentage predicted SVC and its association with respiratory-related clinical events and mortality in patients with ALS.
Design, Setting, and Participants
This retrospective study included 893 placebo-treated patients from 2 large clinical trials (EMPOWER and BENEFIT-ALS, conducted from March 28, 2011, to November 1, 2012, and from October 23, 2012, to March 21, 2014, respectively) and an ALS trial database (PRO-ACT, containing studies completed between 1990 and 2010) to investigate the rate of decline in SVC. Data from the EMPOWER trial (which enrolled adults with possible, probable, or definite ALS; symptom onset within 24 months before screening; and upright SVC at least 65% of predicted value for age, height, and sex) were used to assess the relationship of SVC to respiratory-related clinical events; 456 patients randomized to placebo were used in this analysis. The 2 clinical trials included patients from North America, Australia, and Europe.
Main Outcomes and Measures
Clinical events included the earlier of time to death or time to decline in the Amyotrophic Lateral Sclerosis Functional Rating Scale–Revised (ALSFRS-R) respiratory subdomain, time to onset of respiratory insufficiency, time to tracheostomy, and all-cause mortality.
Among 893 placebo-treated patients with ALS, the mean (SD) patient age was 56.7 (11.2) years, and the mean (SD) SVC was 90.5% (17.1%) at baseline; 65.5% (585 of 893) were male, and 20.5% (183 of 893) had bulbar-onset ALS. In EMPOWER, average decline of SVC from baseline through 1.5-year follow-up was −2.7 percentage points per month. Steeper declines were found in patients older than 65 years (−3.6 percentage points per month [P = .005 vs <50 years and P = .007 vs 50-65 years) and in patients with an ALSFRS-R total score of 39 or less at baseline (−3.1 percentage points per month [P < .001 vs >39]). When the rate of decline of SVC was slower by 1.5 percentage points per month in the first 6 months, risk reductions for events after 6 months were 19% for decline in the ALSFRS-R respiratory subdomain or death after 6 months, 22% for first onset of respiratory insufficiency or death after 6 months, 23% for first occurrence of tracheostomy or death after 6 months, and 23% for death at any time after 6 months (P < .001 for all).
Conclusions and Relevance
The rate of decline in SVC is associated with meaningful clinical events in ALS, including respiratory failure, tracheostomy, or death, suggesting that it is an important indicator of clinical progression.
Amyotrophic lateral sclerosis (ALS) is a rapidly progressive disease of the upper and lower motor neurons that results in weakness of skeletal muscles, including those responsible for breathing. The progressive decline of skeletal muscle function results in disability and death. The median survival time from onset of symptoms is 2 to 4 years.1,2 Death is usually caused by respiratory failure owing to loss of motor neurons supplying innervation to the diaphragm and chest wall muscles.3
Respiratory muscle function is commonly assessed in the clinic and in ALS clinical trials by measuring vital capacity (VC), the maximal volume displaced from the lung (often reported as percentage predicted instead of absolute volume),4 using either a forced VC (FVC) or a slow VC (SVC) maneuver. In addition to respiratory muscle strength, there are other contributors to VC, including elastic recoil of the lung, airway patency, and chest wall anatomy.5 The rate of decline of FVC predicts survival of patients with ALS.6,7 The importance of respiratory dysfunction in ALS has long been recognized; as one indication, the commonly used Amyotrophic Lateral Sclerosis Functional Rating Scale (ALSFRS) was revised in 1999 to increase the number of respiratory questions in the scale from 1 question to 3 questions that evaluate dyspnea, orthopnea, and respiratory insufficiency.8 Despite this change, the ALSFRS-Revised (ALSFRS-R) is still insensitive to specific changes in respiratory function over time.8,9 Therefore, direct measurements of respiratory function using measures like VC are also important.1 While FVC has been the most widely used method for respiratory assessment in ALS,10 the patient must expel air quickly and forcefully, which may cause fatigue and induce bronchospasm and result in an underestimation of actual lung capacity.11 Although FVC and SVC have been shown to be tightly correlated and can be used interchangeably,12 SVC is easier for the patient with ALS to perform even in the presence of orofacial paresis because it involves exhalation of air in a slow, gentle manner after a maximal inspiration.11,12 This makes SVC ideal to use in a clinical trial setting because it can be obtained from patients with advancing disease, which can minimize missing data and may reduce any potential underestimation of actual lung capacity due to effort.
Riluzole, which was approved in 1995 by the US Food and Drug Administration for the treatment of ALS, extends survival and time to tracheostomy by several months but has not demonstrated a significant positive influence on skeletal muscle strength.13,14 Edaravone was recently approved in the United States (May 2017) based on significantly less decline over 6 months in the ALSFRS-R total score compared with placebo, although patients with ALS who were enrolled had normal respiratory function.15 Measures of respiratory function reflecting skeletal muscle strength, such as VC, can be sensitive measures of a treatment effect. In other studies16,17 in which experimental treatments (topiramate and minocycline, respectively) were found to be deleterious, a negative influence on VC was observed with active treatment vs placebo. However, the extent to which demographic factors influence change in VC has not been completely studied. In addition, the ability of change in VC over time to predict clinically important outcomes in ALS is still unclear. Many aspects of the pathophysiology of ALS remain unknown. While functional assessments, such as the ALSFRS-R and FVC, are established markers of disease progression, the prognostic value of SVC is less clear. Given the increasing use of SVC rather than FVC in clinical trials of new treatment for ALS, we sought to investigate determinants of the rate of decline in SVC and its value as a prognostic biomarker of ALS disease progression.
Herein, we used placebo data from 2 large, randomized, double-blind clinical trials to investigate the natural history of respiratory function decline in patients with ALS, as measured by SVC. These trials included (1) the phase 3 EMPOWER trial (NCT01281189) of dexpramipexole in ALS18 and (2) the phase 2 BENEFIT-ALS (Blinded Evaluation of Neuromuscular Effects and Functional Improvement With Tirasemtiv in ALS) trial (NCT01709149).19 We also used the PRO-ACT (Pooled Resource Open-Access ALS Clinical Trials) database.20 In addition, data from EMPOWER were used to examine the association of several clinical variables with the rate of decline of SVC and to investigate how changes in SVC were associated with other clinically meaningful events in patients with ALS.
The placebo group from EMPOWER,21 a large, randomized, double-blind phase 3 trial (conducted from March 28, 2011, to November 1, 2012) designed to evaluate safety and efficacy of dexpramipexole vs placebo in patients with ALS,18 was analyzed in the present study. All participants provided written informed consent for the study and institutional review board approvals were received at all sites before enrollment. In brief, the trial enrolled adults in the United States, Canada, Australia, and Europe with possible, laboratory-supported probable, probable, or definite ALS according to the revised El Escorial criteria, with symptom onset within 24 months before screening and with upright SVC at least 65% of predicted value for age, height, and sex. Patients were assessed for at least 1 year, with maximal follow-up of 1.5 years. Assessments included SVC, ALSFRS-R,8 sniff nasal inspiratory pressure (SNIP),22 time to respiratory failure (defined as tracheostomy with permanent assisted ventilation or use of noninvasive ventilation for at least 22 hours per day for at least 10 consecutive days), and survival. Of 468 patients randomized to the placebo group, 456 had at least 1 postbaseline measurement of SVC and were included in this analysis. Therefore, all patients evaluated had between 2 and 18 observations from which slopes were calculated.
BENEFIT-ALS23 was a randomized, double-blind, placebo-controlled, parallel-group phase 2b trial (conducted from October 23, 2012, to March 21, 2014) that evaluated tirasemtiv, a fast skeletal muscle troponin activator, vs placebo over 12 weeks.19 All participants provided written informed consent for the study and institutional review board approvals were received at all sites before enrollment. The trial included adults in the United States, Canada, and Europe with possible, laboratory-supported probable, probable, or definite ALS, with upright SVC at least 60% (later lowered by protocol amendment to >50%) of predicted value for age, height, and sex, at least 4 of 12 ALSFRS-R questions with scores of 2 or 3, and at least 1 moderately weak handgrip. Slow vital capacity was assessed as a secondary end point. A total of 210 placebo patients who had at least 1 postbaseline measurement of SVC were included in this analysis.
The PRO-ACT database20 includes 16 phase 2 and phase 3 trials and 1 observational study completed between 1990 and 2010. The present analysis used data from 8672 patients that were downloaded from the PRO-ACT database (https://nctu.partners.org/ProACT) on October 21, 2014, and were dated August 13, 2013. Of those 8672 patients, 227 placebo patients had SVC measured longitudinally and contributed to this analysis.
The slope of decline in percentage predicted SVC was calculated for the EMPOWER placebo group and in subgroups defined by age, sex, riluzole use, baseline percentage predicted SVC, site of ALS symptom onset, and baseline ALSFRS-R total score using a repeated-measures mixed model with numbers of days of SVC assessment from baseline and baseline percentage predicted SVC as independent variables and an unstructured covariance matrix. The models assumed a random slope effect and included the baseline time point with nonintercept specified. Slopes of decline in SVC were calculated for the BENEFIT-ALS placebo group and the PRO-ACT database in a similar fashion.
For the EMPOWER placebo group, the Spearman rank correlation coefficient (r) was used to evaluate strength of association between decline in SVC and continuous clinical variables of changes from baseline in SNIP and the total score of individual respiratory subdomain questions of the ALSFRS-R. The 95% CI was derived using Fisher transformation to stabilize variance.
A Cox proportional hazards regression model was used to estimate risk of clinical outcomes to end of the follow-up period based on the slope of decline in SVC from baseline to 6 months of each individual patient obtained from simple regression. The slope of SVC change from baseline to the month 6 visit was the explanatory variable, and adjustments were made for baseline riluzole use and the ALSFRS-R total score. Clinical outcomes included the earlier of time to death or any of the following events after 6 months: time to decline in any of the 3 items of the respiratory subdomain of the ALSFRS-R (items 10, 11, and 12), time to first onset of respiratory insufficiency (defined as use of tracheostomy or noninvasive ventilation for ≥22 hours per day for ≥10 consecutive days), or time to first occurrence of tracheostomy. All-cause mortality after 6 months was also evaluated.
All hypothesis tests were 2-sided with a significance level of .05. All analyses were performed using a statistical software program (SAS, version 9.4; SAS Institute Inc) on a Microsoft Windows operating system.
A total of 893 patients were included in this analysis (456 and 210 patients in EMPOWER and BENEFIT-ALS, respectively) who were randomized to placebo and had postrandomization SVC measures, as well as 227 from the PRO-ACT database. Baseline demographic and disease characteristics for these patient populations are summarized in Table 1. For EMPOWER, the mean patient age was 57.3 years, with 63.4% (289 of 456) male and 93.9% (428 of 456) of white race. Overall, characteristics of the EMPOWER patient population appear to be similar to those of the PRO-ACT database and the placebo group of BENEFIT-ALS, except for duration of symptoms and time from diagnosis to baseline. In BENEFIT-ALS, entry criteria were not based on time from symptom onset as was the case in EMPOWER, providing a rationale for this difference.
The 3 ALS clinical trial data sets were similar with regard to the rate of decline in SVC (Figure 1). The slope of SVC decline was −2.73 percentage points per month in EMPOWER, −2.74 percentage points per month in BENEFIT-ALS, and −2.90 percentage points per month in the PRO-ACT database.
Using the EMPOWER placebo group, we compared the slope of SVC decline in several potentially clinically relevant subgroups to investigate factors that might be associated with changes in respiratory function (Table 2). Age was an important determinant: the slope of change in SVC for the group older than 65 years was significantly steeper than for the group younger than 50 years (P = .005) and the group aged 50 to 65 years (P = .007), indicating greater decline in SVC in patients older than 65 years. In addition, patients with lower baseline ALSFRS-R total score (≤39, the median in EMPOWER) had a significantly steeper slope of change in SVC compared with patients with higher baseline ALSFRS-R total score (P < .001). No significant differences were seen in the slope of change of SVC as a function of sex (P = .21), riluzole use (P = .65), baseline percentage predicted SVC (P = .25 for <65% vs >75% and P = .88 for 65%-75% vs >75%), or site of ALS symptom onset (P = .06 for bulbar vs other).
The relationship between SVC slope and other respiratory measures was also evaluated. Statistically significant correlations were observed between change from baseline in percentage predicted SVC and change from baseline in SNIP (r = 0.38; 95% CI, 0.35-0.42; P < .001), as well as change from baseline in the individual items of the ALSFRS-R, including dyspnea (r = 0.23; 95% CI, 0.20-0.27; P < .001), orthopnea (r = 0.23; 95% CI, 0.19-0.27; P < .001), and respiratory insufficiency (r = 0.26; 95% CI, 0.22-0.29; P < .001).
A Cox proportional hazards regression model was constructed to evaluate time to several clinically meaningful ALS events. Slowing the rate of decline of SVC by 1.5 percentage points per month reduced the risk in any component of the respiratory subdomain of the ALSFRS-R (items 10 [dyspnea], 11 [orthopnea], or 12 [respiratory insufficiency]) or death, first onset of respiratory insufficiency or death, first occurrence of tracheostomy or death, and death at any time after 6 months by 19%, 22%, 23%, and 23%, respectively (P < .001 for all) (Table 3). In this model, by setting baseline ALSFRS-R total score to its mean and the slope of change in SVC equal to −4.23 (−2.73 [observed in EMPOWER] and −1.23 percentage points per month respectively), the probabilities of respiratory failure–free survival were predicted for baseline riluzole use subgroup first and then were averaged out. The result is shown in Figure 2. An 80% probability of respiratory failure–free survival was seen at approximately 52 weeks for those with a slope of change of −4.23 percentage points per month in predicted SVC, at approximately 57 weeks for those with a slope of change of −2.73 percentage points per month, and at approximately 63 weeks for those with a slope of change of −1.23 percentage points per month.
Decline in respiratory skeletal muscle function results in disability and death in patients with ALS. In both clinical and experimental trial settings, this decline is commonly assessed by measuring VC using either a forced or slow maneuver. Given that VC is used frequently in a clinical setting to make important patient management decisions and is used often in clinical trials among patients with ALS, understanding the influence of other clinical variables on decline in VC and its relationship to clinically meaningful events is of clear importance. Vital capacity can be measured using an FVC or an SVC maneuver; under most circumstances, the values are almost identical, but SVC is slightly less variable in patients with very impaired breathing, spasticity, or significant bulbar dysfunction.12,24 For this reason, we think that SVC is a better measure than FVC, but we believe that our conclusions are equally applicable to either method of measuring VC.
In 3 independent data sets (the placebo groups of EMPOWER and BENEFIT-ALS and the population in the PRO-ACT database), average rates of decline in SVC in patients with ALS were −2.73 to −2.90 percentage points per month and were virtually identical, despite somewhat different inclusion criteria. These findings provide information on the natural decline of SVC and suggest that there is a consistent average rate of decline in SVC among patients with ALS who participate in clinical trials. Decline in SVC is also correlated with changes in the ALSFRS-R total score, with correlations ranging between 0.46 and 0.71 for ceftriaxone sodium and dexpramipexole studies.25 However, while the ALFRS-R is commonly accepted as the criterion standard for assessing changes in overall function in ALS, questions in the respiratory subdomain are insensitive to change,9 making it important to use direct measures of respiratory function, such as SVC.
Using the placebo group of EMPOWER, the largest relevant data set collected under a single clinical protocol, we found that younger age at disease onset and higher baseline ALSFRS-R total score, indicating better overall physical function at baseline, were both associated with slower declines in SVC. The observation that older patients had greater rates of decline in SVC is consistent with prior data that showed decreased survival times correlated with increased age at onset of ALS.1,26 It has been demonstrated that patients who are seen with higher initial ALSFRS-R total scores at their first ALS clinic visit have a better prognosis,27 and the findings presented herein that higher ALSFRS-R total scores at trial entry are associated with slower SVC progression may be a reflection of this. No significant influence on the rate of decline of SVC was seen in the present study for riluzole use. However, riluzole does not confer benefits on muscle strength or function, which may explain this observation.13
Decline in SVC was significantly correlated with questions in the respiratory subdomain of the ALSFRS-R. However, the correlations themselves were weak. These weak correlations of change in SVC with the ALSFRS-R suggest that the respiratory subdomain of the ALSFRS-R may not be a sensitive indicator of respiratory function in ALS. Some questions, including those in the respiratory subdomain, may be better suited to a 3-tiered response than the current 5 response options given the difficulty in distinguishing between the current choices of function.28
Change in SVC over time strongly predicted other clinically meaningful events, including respiratory failure or death in patients with ALS. These observations are consistent with other published reports demonstrating that change in FVC predicted survival1,6,7 and that higher SVC at baseline significantly reduced risk of death.29 Modeling a slower decline of SVC by 1.5 percentage points per month compared with the observed overall rate of decline of 2.7 percentage points per month in all 3 trial databases predicted a statistically significant risk reduction of respiratory failure by approximately 20%, also supporting the view that altering the rate of decline in SVC may reduce risk of respiratory failure or death and be clinically meaningful. It should be noted that the modeled slower decline is a substantial reduction in the rate of decline of SVC of more than 50%. This value was chosen because it was approximately the effect seen in the recent phase 2b study of tirasemtiv in ALS.19 A slower decline in SVC was associated with a reduction in risk of having respiratory symptoms that can interrupt activities of daily living. These findings suggest that slowing the rate of decline in SVC over 6 months may delay time to the development of respiratory symptoms or respiratory failure or death.
While there was a reduced rate of clinically meaningful events associated with the reduction in decline in SVC, it is somewhat surprising that such a marked change in SVC decline was associated only with moderate modifications of risk of other events predicted by the model. One reason for this may be that modeling this difference in decline of respiratory function does not include other potential beneficial influences of treatment with a muscle-directed therapy or disease-modifying agent. If this was the case, the influence on the clinically important outcomes that we modeled might well be greater under experimental circumstances, and reducing decline in respiratory function in patients with ALS may improve their prognosis to a greater extent than predicted herein.
A limitation of this study is that the data were analyzed retrospectively. In addition, owing to the limited duration of follow-up in the BENEFIT-ALS and PRO-ACT data sets, the main analyses were only performed for the EMPOWER data set.
Vital capacity is a commonly used clinical measure of respiratory muscle function that informs the clinical management of patients with ALS. Our analyses suggest that there is a consistent rate of decline in SVC in this patient population. Our study also suggests that the rate of decline in SVC is associated with the likelihood of clinically meaningful events, such as respiratory failure or death, in these patients. Consequently, we conclude that monitoring the rate of decline of SVC in patients with ALS can provide important prognostic information that is useful in their optimal clinical management, and observing changes in the rate of decline of SVC can be a clinically meaningful outcome measure for use in clinical trials.
Accepted for Publication: August 13, 2017.
Corresponding Author: Jinsy A. Andrews, MD, The Neurological Institute, Columbia University, 710 W 168th St, Bldg NI-3, New York, NY 10032 (firstname.lastname@example.org).
Published Online: November 27, 2017. doi:10.1001/jamaneurol.2017.3339
Open Access: This article is published under the JN-OA license and is free to read on the day of publication.
Author Contributions: Dr Andrews had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.
Study concept and design: Andrews, Meng, Kulke, Wolff, Bozik, Malik, Shefner.
Acquisition, analysis, or interpretation of data: Andrews, Meng, Kulke, Rudnicki, Bozik, Malik, Shefner.
Drafting of the manuscript: Andrews, Meng, Kulke, Wolff, Malik, Shefner.
Critical revision of the manuscript for important intellectual content: All authors.
Statistical analysis: Meng, Wolff.
Obtained funding: Malik.
Administrative, technical, or material support: Andrews, Kulke, Rudnicki, Wolff, Bozik.
Study supervision: Andrews, Kulke, Rudnicki, Wolff, Malik, Shefner.
Conflict of Interest Disclosures: Drs Meng, Kulke, Rudnicki, Wolff, and Malik reported being shareholders of Cytokinetics, Inc. Dr Shefner reported being a consultant to Biogen, Cytokinetics, Inc, and Neuraltus Pharmaceuticals, Inc and reported receiving grant funding from The ALS Association, Muscular Dystrophy Association, and ALS Finding a Cure.
Funding/Support: This analysis was designed by Cytokinetics, Inc, and Knopp Biosciences and was funded by Cytokinetics, Inc. Cytokinetics, Inc provided funding for writing and editorial support provided by Jennifer L. Giel, PhD, and Nicholas C. Stilwell, PhD, on behalf of Evidence Scientific Solutions, Philadelphia, Pennsylvania.
Role of the Funder/Sponsor: The study sponsor (Cytokinetics, Inc) was involved in the design and conduct of these analyses, including collection and analysis of data, generation of the statistical tables, and interpretation of this study.
Additional Contributions: Dafeng Chen, PhD (employed by Cytokinetics, Inc) provided statistical expertise (without additional compensation). We thank all study participants, their families and caregivers, study investigators, and site staff.
Create a personal account or sign in to: