Highlights
-
Until recently, riluzole has been the only disease-modifying treatment for ALS.
-
Network meta-analysis (NMA) compares all ALS interventions at the same time.
-
None was better than riluzole in improving mortality and functional decline.
Introduction
Amyotrophic lateral sclerosis (ALS) is a slowly progressive, fatal, neurodegenerative disease that affects the motor neurons of the brain and spinal cord. The median survival is three years, but around 10% may survive up to 10 years. Reference Prado, Pedro and Adiao1–Reference Van Damme, Al-Chalabi and Andersen3 Its etiology is not yet concretely identified; nevertheless, modulation of glutamate transmission, oxidative stress, mitochondrial and endoplasmic reticulum dysfunction, and genetic predisposition are leading candidates. Reference Van Damme, Al-Chalabi and Andersen3,Reference Turalde, Moalong, Espiritu and Prado4 Survival is usually associated with age and site of onset, duration of weakness, degree of clinical disability and respiratory function. Patients with ALS commonly die of respiratory failure. Reference Bensimon, Lacomblez and Meininger5
Until recently, riluzole was the only disease-modifying treatment approved for ALS. Riluzole is a 2-amino-3-trifluoromethoxy-benzothiazole that decreases presynaptic glutamatergic transmission and induces postsynaptic inhibition. In randomized clinical trials (RCT), 100 mg of riluzole per day prolongs survival by up to three months when compared to placebo, with its benefits significantly increased among patients with mild ALS and those with bulbar onset. Moreover, it also slows down functional decline, prolonging necessity for mechanical ventilation. Reference Bensimon, Lacomblez and Meininger5
Other extensively reviewed drugs with phase 2–3 clinical trials for ALS include edaravone, sodium phenylbutyrate and taurursodiol (SPT), and tofersen. Reference Van Damme, Al-Chalabi and Andersen3 Edaravone is a free radical scavenger, originally approved in Japan for acute ischemic stroke but now also being given for ALS. It is hypothesized to decrease oxidative stress in motor neurons by limiting lipid peroxides and hydroxyl radicals as shown by reduced free radical marker, 3-nitrotyrosine. Reference Zheng and Chen6–Reference Yoshino and Kimura8 SPT limits endoplasmic reticulum and mitochondrial dysfunction in motor neurons, while tofersen, an antisense oligonucleotide that decreases SOD1protein, has shown benefits among SOD1-positive ALS patients. Reference Miller, Cudkowicz and Shaw9,Reference Paganoni, Macklin and Hendrix10 Despite having positive outcomes in their respective clinical trials, the recent European Association of Neurology (EAN) guidelines for the management of ALS doesn’t recommend edaravone and SPT until their ongoing phase 3 clinical trials have concluded while tofersen is marred by possible incidence of severe adverse drug reactions. Reference Van Damme, Al-Chalabi and Andersen3
Currently, there are no head-to-head studies to compare the efficacy and safety of the four interventions discussed. Moreover, in clinical trials for edaravone, SPT and tofersen, each intervention is usually combined with riluzole, and compared to placebo with riluzole, as denying the patient with standard treatment in countries where it is available is possibly unethical. Network meta-analysis (NMA) is a statistical tool to compare three or more interventions at the same time when such comparisons have not yet existed. Reference White11,Reference Rouse, Chaimani and Li12 Moreover, NMA may also rank these interventions in terms of efficacy and safety and may indirectly compare all interventions to “pure” placebo used in clinical trials for riluzole. This study aims to compare all possible interventions for ALS and rank these interventions in terms of efficacy and safety using NMA methods.
Methods
Eligibility criteria
Population
Studies that employed sporadic or familial, suspected, probable or definite ALS based on El Escorial or Airlie-House Criteria, regardless of location of onset, were included in this review.
Intervention and control
Studies that used FDA-approved drugs, riluzole, edaravone, tofersen and SPT as interventions or as controls, were included in the present study, where some of them are in combination with each other or as pure interventions. As riluzole has been the standard of care for ALS, there is a high probability that succeeding disease-modifying therapy clinical trials used interventions in combination with this medication. Similarly, controls may be pure placebo or placebo combined with riluzole.
Study design
To ensure similarity, consistency and less heterogeneity, only double-blind randomized controlled clinical trials were used in this paper. Moreover, only RCTs can be used in NMA. Post hoc analysis containing the same set of patients as the original study was also excluded.
Outcome
As RCTs for ALS were limited, there were no restrictions on the study outcomes for the study used. Nevertheless, it was expected that some of the studies’ primary, secondary and safety outcomes included changes in ALS Functional Rating Scale-Revised (ALSFRS-R), mortality, survival and incidence of adverse events, adverse drug reactions, severe adverse events and tolerability.
Exclusion criteria
Studies written in other language that can’t be translated into English; those which can’t be retrieved despite attempts to communicate with the primary author as the last resort; and those from predatory journals were excluded.
Information sources
Medline through its PubMed interface and Embase served as the main databases searched for this review. Other databases were also scrutinized for further articles including Cochrane, Clinical Trials. Org, CINAHL and different journal websites. Reference lists of trials obtained from database search were also scanned for possible additional articles.
Search strategy
Two neurophysiologists (MBP and KJA) did the search independently. The search terms were employed, which are shown in Table 1.
Search terms
The search terms were scanned using the titles and abstracts only. Moreover, the search was restricted only to clinical trials and those written only in English.
Selection process
Only papers with available abstracts were examined for eligibility. If the title or abstract was deemed irrelevant by any of the authors, the article was rejected. Duplicate copies were also excluded. Papers that passed the initial screening were checked for eligibility. If the paper matched the inclusion and exclusion criteria, the full article was obtained, if not, the title and abstract were excluded, and the reason for rejection was tabulated. If the full article can’t be obtained, an email was sent to the corresponding author of the paper. If there is no response in two weeks, the potential paper is rejected. In case of disagreement, a systematic review and meta-analysis expert serving as a referee was consulted. The remaining papers were extracted and analyzed by the authors. The process of selection is described in Prisma Flow Chart diagram (See Figure 1).
Prisma Flow Diagram.
Data collection process
All information necessary for the current review was extracted from the papers obtained. Each information extracted was scrutinized and reviewed by two independent authors (MBP and KJA) and was initially placed in an Excel file. Eventually, these were transferred to statistical software capable of conducting conventional (Revman 5.4) and network meta-analytic (STATA SE 18.0) methods. A table of included studies was made to summarize the result of data collection.
Data items
For the characteristics of the papers obtained, the author’s name, study name, year and country of publication were extracted. Moreover, the specific inclusion and exclusion criteria, methodology, intervention, controls and their doses, primary and secondary and safety outcomes, and the projected sample size for each paper were also obtained. Specifically for the population, the number of patients included in the study, the total number included in the intervention and “control,” the male/female ratio of ALS and the duration of ALS per group were summarized.
For the meta-analysis, the total sample size as well as the sample size per intervention group per paper were extracted. The following served as the main outcomes of this paper:
-
1. Mortality/Survival: the number of deaths per group at the end of study.
-
2. Mean Change in ALSFRS-R: mean end ALSFRS-R minus the baseline mean ALSFRS-R
The rest served as the secondary and safety outcomes:
-
1. Survival rate per group at end of the study
-
2. Incidence of tracheostomy per group at the end of the study
-
3. Incidence of Adverse Event (AE), Severe Adverse Event (SAE) and withdrawal per group
-
4. Mean Forced Vital Capacity (FVC) change per group from baseline to end
-
5. Mean change in grip and pinch grade
-
6. Change in modified Norris Scale
-
7. Mean change in ALSAQ4
Study risk of bias assessment
Risk of Bias 2 tool (ROB-2 tool) was utilized to assess the presence of bias per study. Specific explanations per study were included if bias was detected. Reference Sterne, Savović and Page13
Effect measures
For continuous outcomes, standard mean difference was used to summarize the effect measure. If categorical, relative risk or odds ratio was utilized.
Synthesis methods
Continuous and categorical variables were presented as medians (range) and proportions, respectively. For the systematic review, qualitative description and summary of each article included were presented in a table of included studies. A conventional meta-analysis was done if at least two studies with the same intervention, control and outcomes were present. Heterogeneity was assessed using I2, with at least 50% signifying heterogeneous outcomes. To assess the presence of publication bias, funnel plots will be created in case at least 10 studies were included in the current review. This is to ensure that funnel plot has enough power to differentiate chance from real bias. Reference Sterne, Sutton and Ioannidis14
Since all studies included were RCT, a NMA comparing five interventions (riluzole, edaravone, tofersen, SPT and placebo) was done, especially when similarity and consistency were assured. For consistency, global consistency test using the NMA program of STATA was done. A p-value greater than 0.05 assured consistency. If possible, all five interventions were compared and ranked in terms of efficacy, safety and tolerability. A p-value of 0.05 served as the cut-off for statistically significant result.
Operational definition of interventions
In clinical trials beyond riluzole, the disease-modifying therapies including their placebos are usually combined with riluzole. Hence, for edaravone, tofersen and SPT, the proper names for interventions are “edaravone plus riluzole,” “tofersen+ riluzole” and “SPT + riluzole,” respectively. In their respective studies, they are compared to “riluzole plus placebo.”
In this case, we did two analyses in some outcomes. In the initial analysis, we assumed that each intervention was independent of riluzole, while in the second analysis, we assumed that each intervention was combined with riluzole. This is important as there may be an interaction (additive or multiplicative) between intervention and riluzole, not similar to the result when the pure intervention was used.
Registration
The proposal of this review was approved by the local REB (RGAO 2024-1633, RIDO 2024-336) and was registered in Prospero (CRD42024524672).
Results
Study selection/ result of the search
There were 132 hits noted in the database search (riluzole:13; edaravone:13; SPT: 4; tofersen: 3) while one paper was marked in citation scanning. No duplicates were rejected. Ninety-nine papers were immediately excluded since they were deemed irrelevant by the authors upon initial check of the titles. Unfortunately, one paper (from citation scanning) was not retrieved despite extensive search. Of the 33 papers left, only 8 papers passed the inclusion and exclusion criteria (see Figure 1). Reference Page, McKenzie and Bossuyt15 Twenty-five journals were excluded for several reasons (see supplementary material 1).
Study characteristics
The eight studies included in this review were published from 1994 to 2022. Both riluzole and edaravone had three clinical trials included, while SPT and tofersen each had one. The three studies about riluzole were done in France, while the three papers for edaravone were done in Japan. The clinical trials for tofersen and SPT were both done in the USA.
Methodology
All papers included were multicenter, phase 2–3, randomized, prospective, double-blind, placebo-controlled clinical trials. The median number of centers that participated was 16.5 (range: 5 to 32).
Population
A total of 1,601 ALS patients were included in this review, 1,185 in the intervention group and 416 in the control group. Six papers used probable or definite ALS based on Airlie House El Escorial Criteria while one used definite ALS only. The clinical trial for tofersen used SOD1-positive ALS patients. Most studies used patients who were less than 75 years old (5/6), with median duration of at most three years (range: 1.5 to more than 5 years). All studies excluded tracheostomized patients.
Intervention
Riluzole was given as 50 mg tablet per orem twice a day until the end of the study, while edaravone was administered as 60 mg IV to run for one hour for 14 days in the first cycle, then for 10 days every 2 weeks until the end of the study. In one study, riluzole was also given as 50 and 200 mg per day. Reference Lacomblez, Bensimon, Meininger, Leigh and Guillet16 Patients received tofersen intrathecally as bolus injection of 15 ml solution (100 mg), three doses once every two weeks then five doses once every four weeks, while SPT was received as sachet (3 g sodium phenyl butyrate and taurursodiol), taken as one sachet per day for the first three weeks and two sachets per day thereafter. Reference Miller, Cudkowicz and Shaw9,Reference Miller, Cudkowicz and Genge17 Around 86%, 64% and 68% of patients also received riluzole aside from edaravone, tofersen and SPT, respectively. Reference Paganoni, Macklin and Hendrix10,Reference Abe, Itoyama and Sobue18 In the sole study for SPT, around 25% also received edaravone while 22% received riluzole and edaravone. Reference Paganoni, Macklin and Hendrix10
Placebo
All studies used placebo as controls. Nevertheless, in the edaravone, tofersen and SPT studies, placebos were usually given with riluzole in 90.4%, 64% and 77%, respectively.
Outcome
Primary outcome
The three studies about riluzole used survival, while the rest of the studies used change in ALSFRS-R from baseline to end of the study as their primary outcome.
Secondary outcomes
The secondary outcomes were variable but mostly consist of outcomes about change in muscle strength (4), change in FVC (7), subjective assessment by the patients (3), global clinical impression scale (3), grip strength (4), pinch strength (3), change in modified Norris scale (5), change in ALSAQ40 (4), average time to death (4) and changes in SOD1 protein (1) or NFL(2).
Safety and tolerability outcomes
Most studies include adverse events, presence of adverse drug reactions, severe adverse events and tolerability as outcomes.
Risk of bias
Aside from the study of Panagoni, which has a high risk of bias, and Miller, which has some concerns, both in randomization, all papers included had low risk for bias. The paper in the Panagoni study had high risk for bias in random sequence, as there was an error in kit distribution in 17 patients. In the Miller study, there was a significant difference in terms of baseline concentration of neurofilament (Nfl) in the cerebrospinal fluid (CSF) between tofersen and placebo groups. In the result, although statistically non-significant, the higher reduction of Nfl in the intervention group compared to placebo may be clinically significant.
Result of individual studies (see Table 2)
Riluzole
-
Bensimon, 1994 – The first randomized, double-blind, placebo-controlled clinical trial about riluzole was done by Bensimon et al in 1994. In this study, they compared 77 probable or definite ALS patients taking 100 mg of riluzole per day against 78 patients taking placebo. Survival at 12 (74% vs 58%, p = 0.014) and 21 months (49% vs 37%, p = 0.046) was significantly better in the riluzole group. Overall, riluzole reduced mortality by around 39% at 12 months and 20% at 21 months leading them to conclude that riluzole may be a possible disease-modifying drug for ALS. Nevertheless, the issue of including 24 patients who did not satisfy the inclusion criteria may threaten its validity as exclusion of this cohort in post hoc analysis made the difference in survival non-significant. Reference Bensimon, Lacomblez and Meininger5
-
Lacomblez, 1996 – Together with Bensimon, Lacomblez compared the efficacy of different doses of riluzole against placebo in 1996. They compared 237 probable or definite ALS patients taking 50 mg of riluzole against 236 patients receiving 100 mg, 244 patients taking 200 mg of riluzole and 242 patients on placebo. Accordingly, adjusted survival at 12 and 18 months was significantly reduced by 24%, 35% and 39% vs placebo for 50, 100 mg and 200 mg of riluzole, respectively (p = 0.004). In this case, they have concluded that riluzole is a safe and effective disease-modifying medication for ALS. Reference Lacomblez, Bensimon, Meininger, Leigh and Guillet16
-
Bensimon, 2002 – To determine its effect among severe ALS patients, Bensimon conducted another trial comparing riluzole against placebo. In this study, 82 probable or definite ALS patients taking 100 mg riluzole with at least one of the following, >75 years old, <60% FVC or ALS of more than 5 years in duration, were compared against 86 patients on placebo. Unlike the first two clinical trials, the survival at 18 months was not significantly different between the two groups (26.8% vs 25.6%, p = 0.93). They have concluded that although safe, riluzole may not be effective in ALS patients with advanced disease. Many baseline factors, including sex, age, baseline muscle testing score and Visual Analog Scale (VAS) for stiffness and tiredness, were not equal between the two groups. Moreover, the study was underpowered, probably committing type 2 error. Reference Bensimon, Lacomblez, Delumeau, Bejuit, Truffinet and Meininger19
Table of included studies
ALS = Amyotrophic Lateral Sclerosis; FVC = Forced Vital Capacity; ADR = adverse drug reaction; VAS = Visual Analog Scale; ALSFRS-R = ALS Functional Rating Scale- Revised; R = Riluzole; P = Placebo; E = edaravone; AE = Adverse Events; SAE = Severe Adverse Events; SOD1 = Superoxide dismutase 1; T = Tofersen; Nfl = Neurofilament; CSF = cerebrospinal fluid; SPT = Sodium Phenylbutyrate and Taurursodiol; SVC = slow vital capacity.
Edaravone
-
Abe, 2014 – Abe compared 102 probable or definite ALS patients given edaravone 60 mg IV per day for 14 days at initial cycle and 10 days for succeeding cycles, against 104 patients on placebo. Around 89% and 88% of patients were also on riluzole, respectively. Accordingly, the mean change in ALSFRS-R from baseline to end of study was not significantly different between the two groups [-5.7 (0.85) vs -6.35 (0.84), p = 0.411]. Moreover, no significant differences were also noted on secondary outcomes. In this study, edaravone failed to confirm its efficacy in ALS. Reference Abe, Itoyama and Sobue18
-
Abe, 2017 – In 2017, another clinical trial was conducted by Abe et al, now using very mild cases of ALS. Accordingly, only probable or definite ALS patients with at least 2 score on all ALSFRS-R items, FVC >80%, ALS duration of <2 years, <75 years old, and those with Japan ALS severity score of only one or two were included. Ninety-one percent of patients in both groups were also taking riluzole when the study was conducted. Sixty-nine patients on edaravone were compared to 68 patients on placebo. This time, the mean difference in change in ALSFRS-R was significant at six months (-5.01 (0.64) vs -7.5 (0.66), p = 0.0013). Its application in very mild cases limits its external validity. Reference Abe, Aoki and Tsuji20
-
Abe-b, 2017 – Abe and his team also conducted a parallel and exploratory clinical trial on the use of edaravone among more severe probable or definite ALS patients in the same year. Only 13 patients on riluzole and 12 patients on placebo were compared. No statistical difference was noted in terms of primary [-6.52 (1.78) vs -6.0 (1.83), p = 0.84] and secondary outcomes. As there were only 25 patients, the study was underpowered. Reference Abe, Itoyama and Aoki21
Sodium phenylbutyrate and taurursodiol (SPT)
-
Panagoni, 2020 – Panagoni conducted a phase 2, randomized, placebo-controlled clinical trial among definite ALS patients with <1.5 years duration from onset and expected FVC of >60%. He compared 87 patients receiving SPT against 48 patients on placebo. In this study, the proportion of patients on riluzole was 68% and 77% for the SPT and placebo groups, respectively. In terms of mean change in ALSFRS-R per month, progression was significantly worse in the placebo group compared to the SPT group (-1.24 vs -1.66, p = 0.03). There was no significant difference in terms of all secondary outcomes, including mean change in ATLIS, change in SVC and hazard for death or hospitalization. The major issue in this study was the error in kit distribution in 17 patients. Reference Paganoni, Macklin and Hendrix10
Tofersen
-
Miller, 2022 – Miller et al compared 39 SOD1-positive ALS patients taking tofersen against 21 patients taking placebo. No significant differences in the primary outcomes (change in ALSFRS-R) and secondary endpoints were noted. However, reductions in total SOD1 protein concentration (29% vs 16%, p = ns) and Nfl (60% vs 20%, p = ns) were seen. Nevertheless, risk of myelitis, aseptic meningitis, lumbar radiculopathy, increased intrathecal pressure and papilledema were noted in the intervention group. In this study, the efficacy of tofersen was not established. The difference in baseline characteristics (Nfl higher in tofersen group, rate of decline higher in tofersen at baseline) and low power may have contributed to the result. Reference Miller, Cudkowicz and Genge17
Result of the synthesis
Efficacy
Primary outcome
Two primary outcomes were used in eight studies: survival and change in ALSFRS-R from baseline to end of the study. Of the three riluzole studies that used survival, two yielded statistically significant results. Reference Bensimon, Lacomblez and Meininger5,Reference Lacomblez, Bensimon, Meininger, Leigh and Guillet16 The same authors also conducted the same clinical trial in severely affected patients but failed to demonstrate its benefit. Reference Bensimon, Lacomblez, Delumeau, Bejuit, Truffinet and Meininger19 This led them to conclude that riluzole may not modify motor axons in advanced states.
In five studies that used change in ALSFRS-R as primary outcome, only two (edaravone and SPT) showed benefits of intervention over placebo. Reference Paganoni, Macklin and Hendrix10,Reference Abe, Aoki and Tsuji20 Nevertheless, in the initial study of edaravone Reference Abe, Itoyama and Sobue18 and in an exploratory trial using severe ALS cohort, Reference Abe, Itoyama and Aoki21 the efficacy of edaravone was not established. Moreover, the trial with positive results only used mild cases of ALS, limiting its external validity. Likewise, no significant difference was noted among SOD1-positive ALS patients when tofersen was compared to placebo by Miller et al. Reference Miller, Cudkowicz and Genge17
Secondary outcomes
Several secondary outcomes were used in different studies. In one study, age, duration, FVC, bulbar function score, tiredness and stiffness score were established as risk factors for death. Reference Bensimon, Lacomblez and Meininger5 In one riluzole and one edaravone study, modified Norris score was noted to be better in the intervention group than placebo group, Reference Bensimon, Lacomblez, Delumeau, Bejuit, Truffinet and Meininger19,Reference Abe, Aoki and Tsuji20 but not in other riluzole, edaravone or other interventions. Aside from change in ALSAQ score in the study by Abe about edaravone, Reference Abe, Aoki and Tsuji20 no other secondary outcomes were noted to be statistically different from placebo in all studies included. Remarkably, tofersen reduced SOD1 and Nfl concentration more than the placebo among SOD1-positive patients, but the difference was not significant. Reference Miller, Cudkowicz and Genge17
Safety outcomes
Although adverse events, discontinuation, adverse drug events and severe adverse events were slightly higher in the intervention groups in most of the studies, the rates were still comparable. Accordingly, in riluzole studies, patients experienced worsening of asthenia, spasticity, elevation of AST and ALT, mild increase in BP and GI disturbances. Occasionally, impaired respiratory function and elevated liver enzyme were the causes of discontinuation of ALS patients in the riluzole group. Similarly, the more common adverse events reported in the edaravone group were GI disturbances, headache, stomatitis and headache. Reference Abe, Itoyama and Aoki21 As the medication was given intrathecally, procedural pain, headache, pain in the arms or legs, and back pain were common complaints of patients receiving tofersen. Notably, patients were warned of possible myelitis, aseptic meningitis and increased intrathecal pressure due to this drug. Reference Miller, Cudkowicz and Genge17
Reporting bias
In terms of allocation concealment, blinding of participants and outcome assessor, selective reporting, incomplete outcome data, and low risk of bias in all studies were noted. In the study by Paganoni and Miller, high risk and some concerns about bias were seen. Accordingly, there was an error in kit distribution in the SPT study, while there was a big difference in the baseline Nfl concentration between the two groups in the tofersen study.
Meta analysis
The three studies for riluzole can be combined in terms of survival. Likewise, the change in ALSFRS-R can be combined in three studies about edaravone. No conventional meta-analysis was done for both SPT and tofersen. Since all of the studies in the riluzole and edaravone trials were written by almost the same authors (Bensimon and Lacomblez for riluzole and Abe for edaravone), fixed effect model was used. Nevertheless, one paper in both interventions used another set of ALS patients, hence a random effects model (REML) meta-analysis with sensitivity analysis was also done.
Riluzole 100 mg vs placebo
Survival
Compared to placebo, ALS patients taking riluzole had 36% higher probability of surviving (OR: 1.36, I2 = 4%, p = 0.03) using the fixed effects model (FEML). In REML (OR:1.37, p = 0.04), and in sensitivity analysis removing the study by Bensimon in 2002 (OR: 1.44, p = 0.02), the differences remained significant. The heterogeneity was low in all three analyses (See Figure 2).
Riluzole 100 mg vs Placebo. 2A) fixed effects model; 2B) random effects model; 2C) Without Bensimon 2002. M-H = Mantel- Haenszel; CI = confidence interval.
Safety Outcomes (see supplemental material 2)
No statistical difference was noted between the intervention and control group in terms of adverse events in both FEML (OR: 1.19, p = 0.47) and REML (OR: 1.19, p = 0.47). Nevertheless, patients in the intervention group were 2.31 times more likely to withdraw or discontinue their medications in the clinical trial in both FEML (OR: 2.31, p < 0.001) and REML (OR: 2.31, p < 0.001).
Edaravone 60 mg vs placebo
Efficacy Outcomes (See supplementary material 3)
Using FEML, all outcome parameters that can be combined in the three studies were significantly better in the edaravone group. Specifically, those in edaravone group had a) 1.22 points lower ALSFRS-R score (standard mean difference (SMD): 1.22, p < 0.001, I2 = 98%; b) 1.3% points lower decline in FVC (SMD: 1.3%, p < 0.001, I2 = 94%); 0.79 points lower decline in grip strength (SMD: 0.79, p < 0.001, I2 = 91%); 0.92 points lower decline in pinch score (SMD: 0.92, p < 0.001, I2 = 93%); 1.29 points lower modified Norris score (SMD: 1.29, p < 0.001, I2 = 95%; and 0.61 points lower ALSAQ40 (SMD: -0.61, p < 0.001, I2 = 97%) compared to placebo at the end of the study. In Abe-b in 2017, even when patients with severe ALS were removed, the difference was still significant (SMD: 1.37, p < 0.001, I2 = 99%). Nevertheless, all comparisons were heterogeneous.
When REML was used, although there was a tendency towards better outcomes for the edaravone group, these became non-significant. Moreover, the effect measures still had considerable variability even when REML was used.
Safety outcomes
In terms of safety, no significant differences were noted between edaravone and control. Specifically, the incidence of mortality, adverse events, severe adverse events and withdrawal was almost the same for both groups (see supplemental material 3).
Network meta-analysis
Efficacy outcomes
-
A. Death at the End of the Study (placebo vs riluzole 50 vs riluzole 100 vs riluzole 200 vs edaravone vs tofersen vs SPT)
-
a. Placebo (with or without riluzole 100 mg)
-
Different interventions, doses of riluzole and placebo were compared in terms of mortality. The comparison was consistent (chi2 = 0.18, pp = 0.67) based on global test for inconsistency. Accordingly, in terms of the incidence of death at the end of the study, all interventions were not significantly different from placebo. Moreover, compared to one another, no statistical differences were also noted (see Figure 3).
Interval plot with mortality as outcome (A = edaravone, B = placebo, C = R100, D = R50, E = R200, F = SPT, G = tofersen).
-
b. Placebo (no riluzole 100 mg added)
As riluzole was added to the placebo in most patients of clinical trials involving interventions aside from riluzole, a sensitivity analysis where the placebo in these trials was considered as riluzole 100 mg was done. Accordingly, the placebo group (taken from the riluzole clinical trials) was compared to riluzole 100 mg (from riluzole 100 mg and “placebo” of trials for edaravone, SPT and tofersen), riluzole 50 mg, riluzole 200 mg, edaravone (appropriately named edaravone + riluzole), tofersen (tofersen + riluzole) and SPT (SPT + riluzole). The test for inconsistency was insignificant. Likewise, no significant difference in terms of mortality between different interventions was noted. (see supplemental material 4A)
-
B. Rate of Tracheostomy [placebo(with or without 100 mg riluzole) vs tofersen vs edaravone vs riluzole 50 vs riluzole 100 vs riluzole 200 ]
In this outcome, all interventions, as before, were compared to each other, except SPT. Using the test for inconsistency, no inconsistent findings were found (chi2 = 0.06, p = 0.81). At the end of the study, no significant differences between each intervention and placebo, and between each other, were noted (see supplemental material 4B).
-
C. Mean ALSFRS-R Change [placebo (with riluzole 100 mg) vs edaravone vs SPT vs tofersen]
Unfortunately, mean ALSFRS-R change was not an outcome for any riluzole studies, hence we can’t compare this intervention with other interventions using this outcome. Using the global test for inconsistency by STATA, no inconsistency was noted (chi2, p = 0.42). Although there was a tendency towards better score at the end of the study for edaravone (SMD: 1.23, CI: -0.4, 2.86) and SPT (SMD: 2.52, CI: -0.17, 5.21), all comparisons to placebo and to other interventions were not significant (see Figure 4).
Comparison of interventions, in terms of slowing functional decline or ALSFRS-R (A = edaravone, B = placebo, C = SPT, D = tofersen).
-
D. Mean FVC Change (placebo vs edaravone vs tofersen vs SPT)
In limiting the decline of respiratory function, placebo, edaravone, tofersen and SPT were compared. All three interventions were better than placebo at the end of the study (edaravone: SMD: 1.85, CI: -2.51, 6.22) (SPT: SMD: 5.58, CI: -1.68, 12.84) (tofersen: SMD: 7.89, CI: 0.28, 15.5), but only tofersen reached statistical significance. Accordingly, those who took tofersen had 7.89-point lower decline when compared to placebo. Despite reaching statistical significance, the comparison was inconsistent (chi2 = 4.13, p = 0.04). Ranking all interventions, tofersen had the highest probability to be the best in halting the progression of respiratory failure while SPT had the highest probability to be the second best (see supplementary material 4C).
Safety outcomes
-
A. Adverse Events (placebo vs riluzole 50 vs riluzole 100 vs riluzole 200 vs edaravone vs tofersen vs SPT)
In terms of adverse events, no statistical difference in the incidence of adverse events was noted when different interventions (including placebo) were compared to one another (see supplemental material 4D).
Funnel plots
The funnel plots for both edaravone and riluzole in terms of primary outcomes showed possible symmetry. Nevertheless, due to low number of studies, the symmetry observed may be subjective (see supplemental material 5).
Discussion
Summary of main effects
In conventional meta-analysis, riluzole appeared to improve survival by 36 to 44% among ALS patients when compared to placebo in both fixed effects and REMLs. For edaravone, there was a tendency towards slower functional decline in the intervention group than placebo in the REML, which turned significant when FEML was used. Nevertheless, when all interventions, including placebo, were compared to each other using NMA, no significant differences in terms of mortality, functional decline and safety outcomes were found. Although tofersen appeared to slow vital capacity when compared to placebo in the NMA, the inconsistency of this effect measure may cast doubt on this finding. The younger demographics and more aggressive phenotype of SOD1-positive ALS patients receiving tofersen may have contributed to the significant inconsistency in this and other outcomes. Reference Huang, Liu, Yao, Qin and Su22,Reference Opie-Martin, Iacoangeli and Topp23
For the longest time, riluzole has only been the sole disease-modifying treatment for ALS. As such, it is difficult to conduct newer clinical trials without riluzole being part of the intervention or placebo as all clinical guidelines for ALS recommend its use. Because of this, the usual designs of clinical trials involved comparing mixture of intervention and riluzole against “riluzole”. This situation may explain why most clinical trials aside from riluzole had negative results since a new intervention is almost always compared to an “active intervention” proven to be effective. Those studies with significant results were only assumed to have additive effects to riluzole when, in fact, positive or negative interaction between intervention and riluzole is possible. For example, in the confirmatory trial for edaravone, almost 90% of patients in both intervention and control groups were in riluzole, while around 60 to 80% were in riluzole in the tofersen and SPT trials. Reference Paganoni, Macklin and Hendrix10,Reference Miller, Cudkowicz and Genge17,Reference Abe, Itoyama and Sobue18 Even when the remaining patients not on riluzole were analyzed, the result may still be not significant as doing this may significantly increase type 2 error in a condition where the difference in the effect measure between intervention and control is markedly minute. Although doing a clinical trial of these interventions in a country where riluzole is not available may circumvent this problem, this may not be recommended. Reference Prado, Hamoy-Jimenez and Adiao2,Reference Ditan and Turalde24
In all studies, the cohort of patients used was mostly mild to moderately severe ALS cases. This is understandable as most clinical trials discussed were aimed to prevent the progression of functional decline or death. Accordingly, in severe cases, there are only few motor neurons left, and given the lengthy duration of axonal recovery and almost nil probability of nerve regeneration, administering these medications at late stage may be useless as evidenced by the non-significant clinical trials which used this cohort of ALS patients as their population. Reference Bensimon, Lacomblez, Delumeau, Bejuit, Truffinet and Meininger19,Reference Abe, Itoyama and Aoki21 Unfortunately, in some studies, the benefit of interventions can only be applied in extremely early and mild cases of ALS, limiting their external validity. Reference Abe, Aoki and Tsuji20 Especially in low- to middle-income countries, ALS patients often come to the clinic when they can’t swallow or use their limbs anymore. Reference Prado, Hamoy-Jimenez and Adiao2,Reference Ditan and Turalde24 Nevertheless, despite seemingly ineffective at their stage, these patients still take these expensive interventions due to lack of option and desperation.
In the latest ALS guidelines published by EAN, recommendations for edaravone and SPT as Disease modifying therapies for ALS have been pending, while in Canada, only riluzole and edaravone were recommended. Reference Van Damme, Al-Chalabi and Andersen3,Reference Shoesmith, Abrahao and Benstead25 For tofersen, EAN guidelines emphasized its limitation to SOD1-positive ALS patients only and its possible multiple severe complications including myelitis. Reference Van Damme, Al-Chalabi and Andersen3 Specifically for edaravone and SPT, EAN recommendations will only be approved once the ongoing trials for oral edaravone and phase 3 confirmatory trial for SPT will show beneficial results. The oral edaravone trial had been terminated while the results of the phase 3 trial for SPT were negative, although published results have yet to be obtained for both. 26,27 As previously mentioned, primary clinical outcomes were not significantly better in the tofersen group, and its reduction of SOD1 and Nfl in the CSF were also not significant. Reference Miller, Cudkowicz and Genge17 It is as if we are back to square one where riluzole is the only medication beneficial for ALS. When another clinical trial is planned, especially for riluzole or edaravone, another set of authors should be tapped, as all RCTs using these interventions were written by the same authors. In the meantime, non-medication interventions such as modification in diet among others should be considered. Reference Prado, Pedro and Adiao1
The NMA section of this review aimed to compare all interventions at once. Unfortunately, riluzole trials used survival outcomes while others used change in ALSFRS-R. ALSFRS-R was developed in 1999 by Cedarbaum et al, hence it is not really the fault of riluzole trials why this functional rating scale was not included in all their three clinical trials which were written in 1994, 1999 and 2002. Reference Cedarbaum, Stambler, Malta, Fuller and Hilt28 In comparison of interventions that used change in ALSFRS-R, no significant difference was found when the interventions were compared to each other in our NMA. As mentioned earlier, majority of patients in the “placebo” group used riluzole, hence we can indirectly say that these interventions were compared to riluzole and that they were not significantly better than riluzole. Mortality was a common outcome in all studies; nevertheless, the differences were still not significant.
There are several possible reasons why riluzole may have benefited in the conventional meta-analysis, while none was observed in the NMA. In NMA, although we assume similarity and consistency between different studies of different interventions, more heterogeneity of outcome is still expected since the inclusion criteria between different interventions in NMA are more different than the inclusion criteria of different studies of the same intervention in conventional meta-analysis. As such, it is harder to detect statistically significant results in NMA. Moreover, with only eight studies available in the literature to be combined in NMA, this may result in significant type 2 error. In addition, the mixture of placebo only and placebo plus riluzole as control may further decrease the treatment effect difference between interventions, possibly decreasing the power of the study to detect any statistical significance. Furthermore, although all patients were diagnosed with ALS, they may be different in some characteristics. Some trials include different disease severities, while in the tofersen trial, they include SOD1-positive-only ALS patients, which commonly have earlier onset and more aggressive profile. This might lead to significant inconsistency, which was seen in one outcome (slowed vital capacity) and marked heterogeneity. High heterogeneity may result in imprecision which may also lead to wide confidence intervals and higher chance of including the null value of 1, which may be the case in our NMA. Similar to conventional meta-analysis, sensitivity analysis and meta-regression may be done to decrease heterogeneity, unfortunately, while we did sensitivity analysis by comparing “pure placebo” with other interventions, meta-regression requires that at least 10 studies be included for it to be properly done.
Recommendations for future studies
Ideally, if another RCT is to be done in any interventions above, a randomized, controlled, double-blind, clinical trial, comparing pure interventions against pure placebo, using survival AND change in ALSFRS-R as primary outcome measures is recommended. Moreover, while a head-to-head clinical trial comparing pure interventions is feasible, this may not be possible since 1) riluzole is a core recommendation in most if not all clinical guidelines for ALS and 2) the cost it will take to undertake such study will be massive since the treatment effect differences between interventions are small.
Conclusion
Despite the apparent benefit of riluzole in terms of survival in conventional meta-analysis, non-significant findings and the lack of a comparison of ALSFRS-R to placebo, edaravone, SPT and tofersen in NMA may preclude any strong recommendation for its use, especially in low- to middle-income country settings. Moreover, the difference in primary outcomes limits important comparison between interventions and while global consistency in NMA was satisfied, the heterogeneity of the patient population and low number of included studies limit the interpretability and reliability of our results. A head-to-head randomized, controlled, double-blind, comparison study, using survival and change in ALSFRS-R as primary outcome measures, may be needed in the future to confirm our results.
Supplementary material
The supplementary material for this article can be found at https://doi.org/10.1017/cjn.2025.10429.
Acknowledgments
None.
Author contributions
All authors contributed to the study conception and design. Material preparation, data collection and analysis were performed by MBP and KBP. The first draft of the manuscript was written by MBP and all authors commented on previous versions of the manuscript. All authors read and approved the final manuscript. Declaration of generative AI and AI-assisted technologies in writing process: During the preparation of this work the authors did not use AI in order to write this manuscript.
Funding statement
The authors did not receive support from any organization for the submitted work.
Competing interests
The authors declare that they have no conflict of interest.
Research involving human participants and/or animals
The manuscript does not contain clinical studies or patient data.
Informed consent
Not required.
