Abstract
Objective: To determine if long-term topiramate therapy is safe and slows disease progression in patients with ALS.
Methods: A double-blind, placebo-controlled, multicenter randomized clinical trial was conducted. Participants with ALS (n = 296) were randomized (2:1) to receive topiramate (maximum tolerated dose up to 800 mg/day) or placebo for 12 months. The primary outcome measure was the rate of change in upper extremity motor function as measured by the maximum voluntary isometric contraction (MVIC) strength of eight arm muscle groups. Secondary endpoints included safety and the rate of decline of forced vital capacity (FVC), grip strength, ALS functional rating scale (ALSFRS), and survival.
Results: Patients treated with topiramate showed a faster decrease in arm strength (33.3%) during 12 months (0.0997 vs 0.0748 unit decline/month, p = 0.012). Topiramate did not significantly alter the decline in FVC and ALSFRS or affect survival. Topiramate was associated with an increased frequency of anorexia, depression, diarrhea, ecchymosis, nausea, kidney calculus, paresthesia, taste perversion, thinking abnormalities, weight loss, and abnormal blood clotting (pulmonary embolism and deep venous thrombosis).
Conclusions: At the dose studied, topiramate did not have a beneficial effect for patients with ALS. High-dose topiramate treatment was associated with a faster rate of decline in muscle strength as measured by MVIC and with an increased risk for several adverse events in patients with ALS. Given the lack of efficacy and large number of adverse effects, further studies of topiramate at a dose of 800 mg or maximum tolerated dose up to 800 mg/day are not warranted.
The course of ALS is relentless with a linear decline in strength with time during the active phase of the disease.1,2⇓ Survival is modestly prolonged by the glutamate inhibitor riluzole; there are no other known effective therapies. Recent studies implicate toxicity from excess excitation of the motor neuron by transmitters such as glutamate,3-7⇓⇓⇓⇓ mitochondrial dysfunction,8-11⇓⇓⇓ and free radical-mediated oxidative cytotoxicity12,13⇓ as possible causes of ALS. Several in vitro4,7,14⇓⇓ and in vivo15 studies suggest that the α-amino-3-hydroxy-5-methyl-isoxazole-4-propionic acid (AMPA)/kainate glutamate receptors may mediate the selective loss of motor neurons in patients with ALS. Pharmacologic agents that affect these pathways may be therapeutic for patients with ALS. Multiple lines of evidence support this assertion. First, riluzole, which remains the one Food and Drug Association (FDA)-approved treatment for patients with ALS, acts in a variety of ways that block glutamate transmission and release. Second, drugs that act as AMPA receptor antagonists have shown promise in a pilot study in a transgenic mouse model of ALS and in a small human trial. One such drug, GYKI-42466, was administered by miniosmotic pump to five transgenic ALS mice and prolonged survival by 25% (Personal communication, Rothstein). A randomized controlled trial of the same agent in 60 patients with ALS showed no statistically significant effects, but all outcome measures studied trended toward slower progression in treated patients.16 Another agent that has a similar mechanism of action is topiramate, an FDA-approved agent for epilepsy. Topiramate has been shown to have antiexcitotoxic properties because of its ability to diminish glutamate release from neurons and antagonize kainate activation of the AMPA glutamatergic excitatory amino acid receptor.17 Topiramate protects motor neurons in an in vitro model of chronic glutamate toxicity in a dose-dependent fashion.18 Based on the aforementioned data, we conducted a controlled clinical trial to determine whether topiramate slows disease progression and is safe and well tolerated in patients with ALS.
The average dose of topiramate as an add-on therapy for patients with epilepsy was 400 mg/day when this study was initiated. Eight hundred milligrams per day was on the high end of the dose used for patients with epilepsy but was well within the approved dose (up to 1,600 mg/day). The neurobehavioral effects of topiramate were known to be dose dependent. The dose for the current study was chosen based on preclinical studies. Results from an in vitro study of amelioration of glutamate-mediated toxicity by topiramate showed a dose-related effect between 10 and 100 μmol/L (Personal communication, Hulihan, Ortho-McNeil Pharmaceutical, Inc). We chose a dose of 800 mg/day, which should produce plasma levels of approximately 45 μmol/L, well within the therapeutic range defined by in vitro studies.
Methods.
General study design.
A randomized, double-blind, placebo-controlled trial design was used. The institutional review boards of each of the participating institutions approved the protocol and consent forms. Eligible patients were randomized in a 2:1 fashion to receive treatment for 12 months of 1) active topiramate (Ortho-McNeil Pharmaceutical, Inc., Raritan, NJ) or 2) placebo. A 2:1 (topiramate:placebo) randomization plan was chosen to enhance subject recruitment because this design would minimize the number of patients given placebo. This ratio also provides more safety information on the use of topiramate in patients with ALS for 12 months. An additional 30 patients were required with a 2:1 vs a 1:1 randomization. Patients were assigned to groups using computer-generated randomization (by Ortho-McNeil Pharmaceuticals) and were stratified by clinical site. The coordination center staff, all site investigators, coordinators, and clinical evaluators were blind to treatment assignment throughout the study. In addition, the clinical evaluators who performed the primary outcome measure were blind to adverse events and laboratory test results and did not review previous outcome measure scores. The site investigators had the ability to disclose treatment assignment in the event of a medical emergency if necessary for clinical care, but they did not have direct access to this information. An independent safety monitoring committee (SMC) reviewed the safety data every 4 months throughout the study, and an NIH Data and Safety Monitoring Board reviewed study progress and patient safety. Mortality and safety assessments were performed at each SMC meeting throughout the study. The SMC had the ability to recommend modifying the trial if they judged that patients’ safety would be jeopardized.
Patient selection criteria.
Eligible participants, aged 18 to 80 years, had a clinical diagnosis of definite or probable ALS,19 a forced vital capacity (FVC) ≥50% predicted, evidence of limb weakness in at least one extremity, and disease duration ≤3 years. Upper extremity strength had to be sufficient so that at least four of eight muscle groups could be evaluated using the quantitative isometric strength apparatus. Patients were excluded if they had taken other investigational drugs in the preceding 4 weeks. Patients could take riluzole if the dose was stable for at least 2 months before the baseline visit. Patients were not allowed to change the dosage of riluzole during the trial.
Study procedures.
At the screening visit, the study design was explained to all prospective participants, and informed consent was obtained. Screening procedures included assessment of eligibility criteria, complete medical history, general physical examination, safety laboratory tests (chemistry, liver function tests, and complete blood counts), FVC determination, and medication review. Screening procedures took place within 14 days of the baseline visit and randomization. Before initiation of study medication, study participants were evaluated using measurements of muscle strength in eight arm muscles (bilateral shoulder and elbow flexion and extension) by maximum voluntary isometric contraction (MVIC).20,21⇓ This is a standardized, validated measurement tool developed to characterize disease progression in patients with ALS that has proven useful in natural history studies and previous clinical trials in patients with ALS.20-23⇓⇓⇓ Raw scores for MVIC are standardized to published norms using Z score transformation (raw score minus norm average divided by norm SD).20,21⇓ An arm megascore is calculated as the average of all Z scores of the eight arm muscles tested at each visit. A Z score of 0 indicates average strength compared with a standard population of patients with ALS from a published database. A Z score of +1 or −1 indicates muscle strength 1 SD greater than or less than average. At the baseline visit, patients were randomized to treatment group assignment (topiramate or placebo). FVC, grip strength, and the ALS functional rating scale (ALSFRS)24 were also measured.
Before any site enrolled patients into this trial, formal training was provided on all outcome measures. For MVIC and FVC, variability of measurement on four healthy control subjects was assessed, with a criterion for acceptable intrarater variability set at <15%. Healthy subjects were chosen for variability testing because previous studies have demonstrated that test-retest variability is greater in healthy subjects than in patients;20 thus, the aforementioned criterion was established as more stringent than direct reliability testing on patients. Repeat variability assessments were also conducted at each site after 50% of patient enrollment had been accomplished. The average intrarater coefficient of variation was 7.3 ± 2.0% at all centers before study initiation and 6.9 ± 2.1% on repeat testing halfway through the study.
Subsequent visits occurred 1, 3, 6, 9, and 12 months after randomization. At each visit, MVIC, FVC, ALSFRS, vital signs, physical examination, reports of concomitant medications, and adverse events were obtained. Safety blood tests were performed 1 and 12 months after baseline visit. Deaths (or tracheostomy) were reported as they occurred: either event was considered a survival endpoint. Plasma samples for measurement of topiramate levels were obtained at the screening visit and after 1, 6, and 12 months, and were frozen. All samples were analyzed at study completion so as not to jeopardize the blinding of the study.
Study intervention.
The study medication consisted of tablets containing either topiramate 25 mg or 100 mg, or matching placebo, and was supplied by the manufacturer, Ortho-McNeil Pharmaceutical, Inc.
Study medication was titrated slowly during 16 weeks to a maximum tolerated dose or a maximum of 800 mg/day. Study medication was taken orally twice a day. At each site, investigators had the option to increase study medication dose or keep dosage stable, according to patient tolerance. If a patient experienced severe adverse effects, investigators also had the option of reducing dose. After 16 weeks, study protocol mandated that further dosage increases should not occur. At any time during the study, dose could be temporarily reduced or study medication stopped for adverse events. At the completion of the 12-month double-blind phase, participants were given the option to continue taking topiramate in an open-label study until data analysis was complete.
Given the known risk of weight loss with topiramate use, careful attention was given to educating the site personnel and the research participants to this risk. All research participants were informed of this known risk and advised to monitor their weight, to not skip meals if possible, and to contact their site investigator if they experienced weight loss or loss of appetite. Appetite stimulants could be administered if clinically indicated.
Statistical analysis.
Sample size.
The sample size was based on the expected average decline in arm megascore. Power analysis showed an 80% chance of detecting a 35% decrease in the rate of decline at a two-sided significance level of p = 0.05. Preliminary values for the rate of decline (−0.021/week) and its SD (0.022) were obtained from a previous study.23 Allowing for an approximate 20% withdrawal rate, the planned enrollment for the study was 288 patients: 192 in the topiramate group and 96 in the placebo group.
Analysis.
Based on the intent-to-treat principle, the data set for analysis included all randomized patients with the exception of two patients. The decision to exclude these patients from the data analysis was made before revealing the group randomization. One participant was randomized but never took a dose of study medication (topiramate group), and a second participant was found to have a different diagnosis (inclusion body myositis) and was withdrawn from the study (placebo group). In the intent-to-treat analysis, outcome measures that were past a subject’s early drug discontinuation date were used. All patients discontinuing therapy were encouraged to complete study visits for 12 months.
The primary statistical analyses were performed according to the intention-to-treat principle. Analysis of the primary outcome variable used a mixed model analysis of variance. The dependent variable was the patient’s arm megascore measured at baseline and months 1, 3, 6, 9, and 12. The independent variables were treatment, time, and a time-treatment interaction. The intercept and time variables were also modeled as random effects. Because the hypothesis was that the slope of the arm megascore was different for the different treatments, the test of treatment effect was based on the time-treatment interaction. If each patient had the same number of observations, this model would be equivalent to the following two-stage analysis. First, estimate the slope of each patient’s measurements by linear regression and, second, compare the slopes using a t-test. The use of a mixed model is an improvement to this analysis in the case where all patients do not have the same number of measurements. The analysis optimally weights each patient’s slope by a function of its precision. A sample of patient trajectories was examined without regard to treatment to see if the assumption of linearity is met. In addition, models with quadratic terms were fit to see if the results were sensitive to the model being fit. Tests for site time-treatment interaction were performed using the same model. We also tested for a dose effect among patients randomized to topiramate using the same model, with “dose” defined as the maximal dose achieved during the 16-week titration phase of the study.
The aforementioned analysis was repeated for the secondary outcome variables for efficacy. For the analysis of mortality, survival curves were plotted using the Kaplan–Meier method and tested for a treatment effect using a log-rank test. Patients who underwent tracheostomy were considered to have died on the date of their tracheotomy. The survival analysis was adjusted for baseline covariates (table 1) using a Cox proportional hazards model. Covariates were chosen using a forward selection procedure with a p = 0.05 criterion for inclusion. The analysis only included categorical covariates that included at least 5% of the population in each category.25 Site and postbaseline outcomes were not considered as covariates.
Table 1 Baseline patient characteristics
Fisher’s exact test was used to compare treatment groups with regard to occurrence of adverse events and abnormal laboratory and vital signs. Weight change was analyzed using the mixed model analyses of variance described previously. To determine whether change in patient weight influenced the MVIC, we repeated the primary analysis using patient weight as a time-varying covariate. Demographic and baseline variables were summarized descriptively for each treatment group. Discrete variables were compared using the Fisher’s exact test, and continuous variables were compared using t-test. The achieved dose and average dose were compared using t-test. The time to attrition was compared between the treatment and control groups using a Kaplan–Meier plot and compared using a log-rank test. Deaths that occurred within 30 days of drug discontinuation were considered as censored observations because the attrition of moribund patients was not considered as drug intolerance.
To explore the impact of missing data on all the outcome measures, two additional analyses were performed. The first alternative analysis imputed the data for patients who missed their 12-month visit and used the change from 12 months as the efficacy parameter. The most conservative imputation was to assume that patients missing the 12-month visit had the worst possible loss in muscle strength observed in their treatment group. Patients who died and those who withdrew from the study were treated the same way in this analysis. We also used a nonparametric analysis that orders outcomes so that patients who died were considered to have a worse outcome than patients who lived and patients who withdrew and lived.26 Patients who withdrew from the trial were censored at withdrawal at the last follow-up evaluation.
Results.
Enrollment and baseline characteristics.
Between July 1999 and August 2000, 296 patients with ALS were enrolled at 20 participating sites. One hundred ninety-eight patients were randomized to receive topiramate, and 98 patients were randomized to receive placebo. There were 13 inclusion criteria protocol violations: one subject was aged >80 years at entry (placebo group, age = 82 years); five patients had disease duration >3 years (four in topiramate group, one in placebo group; range, 3.1 to 4.5 years); one subject was misdiagnosed with ALS and subsequently diagnosed with inclusion body myositis (placebo group); and six patients had predicted FVC percentage <50% before randomization (five in topiramate group, one in placebo group; range, 40 to 46.2%). Site investigators responsible for the protocol violations were re-educated on protocol inclusion and exclusion criteria.
Demographic and clinical variables were comparable between the two groups at baseline (see table 1). There were no differences in the safety laboratory test values performed at the screening visit between the groups. Baseline values of primary and secondary outcome variables were also similar between the two groups (see table 1). During the course of the study, there were no differences between groups in the events of tracheostomy, feeding tube placement, and noninvasive positive pressure ventilation (NIPPV) use. Eight participants underwent tracheostomy (2.5% of patients in topiramate group, 3.1% in placebo group; p = 0.72), 66 underwent feeding tube placement (23.4% in topiramate group, 20.6% in placebo group; p = 0.66), and 62 began treatment with NIPPV (21.3% in topiramate group, 20.6% in placebo group; p = 1.00).
Tolerability of topiramate.
Treatment assignment did not affect the likelihood of study completion (p = 0.26; Fisher’s exact test). The flow of participants is shown in figure 1. One hundred sixty-two patients completed the 12-month study, and 42 patients died while taking study medication. One hundred fifteen participants elected to discontinue study medication (86 [43.7%] in the topiramate group, 29 [29.9%] in the placebo group), and 90 of these patients (64 [32.5%] in the topiramate group, 26 [26.8%] in the placebo group) withdrew from the trial. The most common causes of early drug discontinuation in the topiramate-treated participants were adverse experiences (61.6%), principally asthenia, thinking abnormalities, and diarrhea. The main reason for stopping drug in the placebo group was subject choice (41.4%). Eighty-nine patients (28.4%) from topiramate group and 33 (34.0%) from placebo group elected to enroll in the open-label study.
Figure 1. Flow chart for research participants. The number of participants completing the study and the number who died or withdrew participation is shown by treatment group assignment.
In general, topiramate was not as well tolerated as placebo. The probability of remaining on the study medication was greater for the placebo group than the topiramate group (figure 2; log rank p = 0.03). The maximum achieved dose after 16 weeks of titration and the average dose per day from the date study medication was started to the completion date was higher in the placebo group compared with the topiramate-treated group. The average maximum achieved dose after 16 weeks of titration was 408.8 ± 254.0 mg/day in the topiramate group compared with 523.2 ± 235.1 mg/day in the placebo group (t-test, p = 0.0002). The average dose/day was 282.7 ± 223.0 mg/day in the topiramate group compared with 402.0 ± 229.2 mg/day in the placebo group (t-test, p < 0.0001). In addition, only 65 of the 197 patients (33%) given topiramate achieved maximum dose of 800 mg/day compared with 52 of 97 patients (54%) given placebo (Fisher’s exact test, p = 0.0006). The mean compliance (assessed by pill counts) was slightly better in the placebo group than in the topiramate group: 91.3 ± 17.2% in the topiramate group and 94.7 +10.9% in the placebo group (t-test, p = 0.06). Plasma topiramate levels performed at screening and months 1, 6, and 12 revealed that four patients in the placebo group had at least on one occasion detectable plasma levels of topiramate and might have taken topiramate outside of the study.
Figure 2. The probability of staying on study medication for patients in the placebo group (closed squares) and in the topiramate treatment group (open squares) is shown.
Safety of topiramate.
The incidence of certain specific adverse effects was greater for patients taking topiramate. Table 2 lists the common adverse events by treatment group. The most commonly reported adverse experiences in both treatment groups included asthenia, anorexia, depression, nausea, and constipation. Of these events, anorexia, depression, diarrhea, ecchymosis, kidney stones, nausea, paresthesia, thinking abnormalities, and weight decrease were more likely to occur in the topiramate-treated group. All these adverse events were known side effects of topiramate. Kidney stones occurred more frequently (9.1%) than the 1.5% (US package insert) found in the epilepsy disease population. In addition, 6% of patients given topiramate compared with 1% of patients given placebo developed pulmonary embolism or deep venous thrombosis. The unexpected increased risk of thromboembolic events led to the Data and Safety Monitoring Board recommending early termination of the open-label phase of the clinical trial.
Table 2 Adverse events
The rate of change in weight and body mass index was approximately 2.5 times faster in the topiramate-treated group than in the placebo group (see figure E-1 on the Neurology Web site). The rate of weight loss was 0.93 (SEM = 0.07) kg per month in the topiramate-treated group compared with a loss of 0.37 (SEM = 0.09) kg per month in the placebo group (p < 0.0001). Four research participants required appetite stimulants; all were in the topiramate treatment group.
The occurrence of serious adverse events was similar between treatment groups. Forty-eight percent of participants given topiramate had at least one serious adverse experience compared with 42.2% of participants given placebo (Fisher’s exact test, p = 0.23). The most common serious adverse events were respiratory failure and hospitalization for percutaneous endoscopic gastrostomy. Seventy-six deaths occurred during the study: 28 in the placebo group and 48 in the topiramate group (29% vs 24%).
The percentage of patients who developed abnormalities in laboratory safety studies was similar in the two groups except for bicarbonate levels. Two percent of patients given placebo had abnormally low bicarbonate levels on at least one occasion after the baseline visit compared with 11.7% of patients given topiramate (Fisher’s exact test, p = 0.006). The mean decrease in bicarbonate level from screening was 3.1 ± 2.7 mmol/L at 1 month and 3.0 ± 4.6 mmol/L at 12 months in the topiramate group compared with 0.3 ± 2.9 mmol/L and 0.2 ± 3.7 mmol/L in the placebo group. There were no significant changes in vital signs over time for either treatment group.
Efficacy of topiramate.
The average decrease in arm megascore in the placebo group was similar to that projected in our sample size calculation and in other published studies.22,27⇓ The primary intent-to-treat analysis showed that topiramate-treated patients declined more rapidly than patients taking placebo (0.0997 ± 0.006 units/month vs 0.0748 ± 0.008 units/month, p = 0.012; table 3 and figure 3A). Topiramate use was also associated with a more rapid decrease in grip strength (p = 0.017, figure 3B), but there was no effect on the rate of decrease in FVC percentage predicted or ALSFRS score (figure 3, C and D) or tracheostomy-free survival (log-rank test, p = 0.59). Time to death for any cause was the same for both groups (data not shown).
Table 3 Treatment effects
Figure 3. The rate of decline in arm (A), grip (B) megascore, FVC percentage predicted (C), and ALSFRS (D) is shown. In all graphs, the data from the topiramate-treated group are represented by open squares; data from the placebo group are shown with closed squares.
The additional analysis with imputation of missing data resulted in no significant difference in arm strength or grip strength between the two groups. The nonparametric analysis that ordered patients first by survival and second by arm strength also failed to show any significant differences. However, the power to detect a difference of 35% was only 20% using this method for imputation.
There was a significant quadratic effect in the analysis of arm strength, grip strength and FVC. The rate of decline decreased as time increased. However, the analysis of the treatment effect measured by the linear time-treatment interaction was unchanged when the quadratic term was added to the model. A treatment effect was not masked by the nonlinear decrease in arm MVIC.
For the patients given topiramate, there was a tendency for arm strength to decline faster at higher achieved doses (p = 0.087). There was no evidence that the treatment effects varied significantly among the 20 centers (data not shown). When the patients’ weight, measured at each visit, was included in the model, there still was a significant difference in the rate of decline of MVIC between the topiramate and placebo treatment groups, indicating that the negative effect of topiramate was not directly mediated by weight loss. A weight-by-time interaction was not included because weight itself changed over time, and the idea was to see if the decrease in weight accounted for the treatment effect on MVIC.
Using a Cox proportional hazard model, baseline FVC percentage predicted value, ALSFRS, patient age, and time from symptom onset to screening were important predictors of survival (table 4). Higher baseline FVC percentage predicted value, greater baseline ALSFRS score, younger age, and longer time from symptom onset were predictors of longer survival.
Table 4 Survival: Cox proportional hazard model
Discussion.
In this controlled trial, high-dose topiramate accelerated the decline in arm strength for patients with ALS, as measured by MVIC, the prespecified primary outcome measure. There was a similar effect on grip strength, but there was no impact on the rate of decline of FVC, ALSFRS, or for overall or tracheostomy-free survival. Additional analyses using imputed data confirmed the lack of effect of topiramate on FVC, ALSFRS, and survival, but they failed to show significant differences between groups with respect to strength measurements. However, power was reduced to such an extent using these analyses that we believe that the primary analysis should be the basis for further discussion. The primary analysis is based on the assumption that the data from patients who left were “ missing at random.” This would allow the missingness to depend on the outcome measure made before the patient left but not on the outcome measure that would have been.28 For grip strength and arm MVIC, topiramate-treated patients began to decline faster than placebo-treated patients within 3 months of study onset. Although the reason for this effect is unknown, it suggests that the difference in rate of disease progression may be related to an adverse symptomatic effect of topiramate rather than direct motor neuron toxicity. Depression and thinking abnormalities did not have any apparent effect on MVIC measurements.
Topiramate was associated with several clinical adverse events, particularly anorexia, depression, thinking abnormalities, and renal calculi, all of which had been identified as adverse events in previous controlled trials in patients with epilepsy. The number of kidney stone events was higher than expected from those trials and might reflect a superimposed vulnerability of patients with ALS for developing kidney stones, perhaps because of dehydration, or the relatively high dose in this study compared with previous studies. Topiramate was associated with a risk for thromboembolic events (deep venous thrombosis and pulmonary emboli) greater than that seen in other controlled trials of topiramate. This observation led to the early termination of the open-label phase of the study, at which time patients were advised to discontinue their use of topiramate.
Although the evidence from preclinical studies supported testing topiramate in patients with ALS, this study did not demonstrate any benefit of topiramate on muscle strength. This failure may be because of adverse events, possibly because the dose was too high for this patient population, or unfavorable receptor selectivity. It is not possible to determine in this study whether topiramate, at the dosage given, blocked glutamatergic AMPA receptors or decreased glutamate release in the CNS of patients who received topiramate, although these are the presumed mechanisms of action. Topiramate was recently tested in a mouse model at 50 mg/kg daily by oral gavage and not found to have an effect on survival.18 More recent studies of topiramate in other experimental models suggest that, at concentrations >30 μmol/L, neuroprotective effects seen at lower concentrations may be reversed.29 Based on these observations, studying topiramate at lower doses in the ALS transgenic mouse model may be warranted. Although this model is frequently used to screen potential therapies for development in ALS, it is not yet known how well the model predicts what will occur in humans with the disease. As is the case with topiramate, several interventions that have failed in patients with ALS have also failed to alter disease course in the transgenic mouse model. Riluzole, a presynaptic glutamate release inhibitor, prolongs survival in patients with ALS and in the transgenic mouse model by approximately 10%. At this point, the decision to test a therapy in patients with ALS remains based on a rationale hypothesis and supportive preclinical data, which may or may not include efficacy in the mutant SOD1 transgenic mouse model.
The pattern of early decline in muscle strength in topiramate-treated patients closely resembles the pattern seen in a previous study evaluating the efficacy of ciliary neurotrophic factor (CNTF) in patients with ALS.22 In that study, MVIC was also used as an outcome measure, and a dose-related acceleration of strength loss was noted by 1 month after treatment onset, although the curves became more closely aligned with time. As was the case for topiramate, CNTF administration was associated with significant adverse patient experiences, including weight loss. The authors concluded that CNTF was unlikely to have caused worsening of the underlying disease process but rather the decline in muscle strength was the result of lack of tolerance to the drug. Topiramate use was not associated with increased mortality or changes in pulmonary function, suggesting that the more rapid strength decline in treated patients may similarly be a function of drug tolerability. However, the question of whether topiramate, at the doses used here, causes more rapid progression of ALS cannot be conclusively answered. Lessons to be learned from this study include the importance of frequent interim data analysis by an independent SMC for patient safety. An interim analysis for futility may also be useful to keep trial from going to end if likelihood of success is extremely low.
Maximum voluntary isometric contraction testing was also used in a study evaluating the efficacy of gabapentin in patients with ALS.23 There was no significant change in rate of decline in muscle strength as measured by MVIC in this study. The current study was designed to detect an approximate 35% slowing in functional decline. This is the first controlled trial in patients with ALS to measure a difference between treatment and placebo groups using MVIC as the primary outcome measure, albeit in a negative direction. A difference between the topiramate group and the placebo group was not detected with either FVC or ALSFRS. If the decline in muscle strength was caused by lack of drug tolerability, this suggests that ALSFRS and FVC may be more forgiving of such effects. However, such forgiveness may come at the expense of reduced sensitivity to change in a positive direction. Because the three outcome measures assess different areas of the neuraxis, direct comparisons of sensitivity of each measure may not be possible.
This study provides evidence that topiramate, at the highest dose tolerated, is not beneficial for patients with ALS. Whether a lower dose may be better tolerated and therefore be associated with a positive effect is not known. There was a wide variation in average dose among the treated patients, but the ultimate dose for each subject was limited by tolerability. Thus, the effect of topiramate at a dose better tolerated by most patients is not possible to assess from these data.
An important limitation in the interpretation of our results is the large proportion of patients for whom outcome data were not available. Approximately 55% of participants completed 12 months of treatment. Some of the patients died before 12 months, and others withdrew participation voluntarily. Previous clinical trials in patients with ALS have also had a large proportion of subjects not complete the study for reasons similar to those reported in the topiramate study. The amount of missing data raises important issues on the choice of outcome measures for a progressive lethal neurologic disorder and methods of analyses to accurately account for the missing data. Our secondary analysis imputed values for missing data and supported the primary conclusion that there was no benefit of topiramate on muscle strength for patients with ALS. It should be noted that any method of imputation requires making assumptions about the true values of missing data and thus may misrepresent the true effect in the study. The primary outcome measure chosen for this trial depended on the ability of patients to return for study visits and to survive a 12-month study. Difficulty returning for study visits was a common reason for study withdrawal. It may be that outcome measures, such as the ALSFRS and survival, which can be obtained over the telephone, will be better tools for clinical trials in patients with ALS. Previous clinical trials in patients with ALS that have used MVIC as the primary outcome measure have largely ignored missing data.18,19⇓ Developing analytical methods to handle missing data for a progressive disorder such as ALS would be beneficial for future clinical trials.
There were 13 protocol inclusion violations. Site investigators were not required to contact the coordination center to review a research participant’s relevant screening and baseline data before randomization of subjects. Although education of site investigators was done to minimize these occurrences, it is preferable in the future to require that inclusion and exclusion criteria are monitored, either by telephone or electronically, before issuance of a randomization number.
The current findings demonstrated that topiramate treatment at a target dose of 800 mg/day in this population did not benefit patients and may have accelerated the loss of arm muscle strength. It was also associated with increased risk for several side effects. In our view, these results argue against further study of topiramate at these doses for patients with ALS. However, it may be premature to conclude that topiramate does not have a neuroprotective effect for patients with ALS. Whether topiramate may be neuroprotective and safer at lower doses is unknown. Future studies of this and other potential neuroprotective agents are needed to clarify their role in treating patients with ALS.
Appendix.
The NEALS consortium consists of (affiliation at time of study):
Participating coordinators and clinical evaluators: L. Clawson, MSN, CRNP, Johns Hopkins University; R. Smart, BA, S. Hilgenberg, BA, Massachusetts General Hospital; J. Wilson, PA, Baylor College of Medicine; L. Haas, RN, A. Micheels, PT, Indiana University; J. Taft, PA, C. Bagley, MA, SUNY Upstate Medical University; T. Paylor, RN, C. Cook, RN, K. Lind, MCP Hahnemann; C. Ashburn, RN, G. Hyde, RN, Wake Forest University School of Medicine; S. Conn, RN, J. Levin, RN, Hennepin County Medical Center; J. Falcone, Lahey Clinic Medical Center; P. Krusinski, University of Vermont College of Medicine; D. Andrews-Hinders, RN, T. Wheeler, Cleveland Clinic Foundation; C. Kiely, RN, University of Connecticut Health Center; J. Smith, RN, R. King, The Carolinas Medical System; J. Choi, George Washington University Hospital; S. Motta, R. Dickinson, RN, Rhode Island Hospital; J. Steele, LPN, University of Miami School of Medicine; Beth Israel Deaconess Medical Center; Bay State Medical Center; A. Sherr, RN, Robert Wood Johnson Medical School; L. Rescorl, L. Marshall, Yale University School of Medicine.
Steering Committee: M. Cudkowicz (principal investigator), R.H. Brown, D. Schoenfeld (biostatistician), Massachusetts General Hospital; Jeremy Shefner (co-principal investigator), SUNY Syracuse Health Center.
Coordination and Data Management Staff: H.L. Johnson, D. Goodman, M. Qureshi, M. Jacobs, H. Zhang, Massachusetts General Hospital.
Safety Monitoring Committee: G. O’Neil (Chair), D. Hoch, Massachusetts General Hospital, T. Munsat, D. Weinberg, New England Medical Center.
DSMB Committee: M. Welch (Chair), H. Mitsumoto, J. Thompson, B. Barton, S. Horowitz
NIH-National Institute of Neurological Disorders and Stroke Sponsor: P. Sheehy, C. Moy
Ortho McNeil Pharmaceuticals: J. Hulihan, L. Kraut, M. Kamin
Acknowledgments
This work was supported primarily by NINDS (5K08NS01896, 1R01NS39988). Support was also provided by the Muscular Dystrophy Association, Ortho-McNeil Pharmaceutical, Inc., and General Clinical Research Centers (grants RR01066, M01RR06192, M01RR00109, M01RR07122, RR01032). Dr. Cudkowicz has received grant support in excess of $10,000 from Ortho-McNeil Pharmaceuticals. Dr. Hulihan has received support in excess of $10,000.
Footnotes
-
See also page 434
-
*See the Appendix for a complete listing of the members of the Northeast ALS Consortium.
-
Additional material related to this article can be found on the Neurology Web site. Go to www.neurology.org and scroll down the Table of Contents for the August 26 issue to find the title link for this article
- Received November 16, 2002.
- Accepted June 5, 2003.
References
Disputes & Debates: Rapid online correspondence
NOTE: All authors' disclosures must be entered and current in our database before comments can be posted. Enter and update disclosures at http://submit.neurology.org. Exception: replies to comments concerning an article you originally authored do not require updated disclosures.
- Stay timely. Submit only on articles published within 6 months of issue date.
- Do not be redundant. Read any comments already posted on the article prior to submission.
- 200 words maximum.
- 5 references maximum. Reference 1 must be the article on which you are commenting.
- 5 authors maximum. Exception: replies can include all original authors of the article.
- Submitted comments are subject to editing and editor review prior to posting.