Critical Appraisal of Evidence for Improving Gait Speed in People with Multiple Sclerosis

Mobility limitations affect more than 90% of people with multiple sclerosis (MS).¹ Impairments in the ability to walk can lead to reduced community participation and low levels of physical activity, depression, and poor quality of life.^2–4 Considering that the onset of MS typically occurs in people 20 to 40 years old,⁵ which is much earlier in life than the typical onset of other common neurologic disorders (eg, stroke and Parkinson's disease), the potential impact of walking limitations on quality of life in people with MS is substantial. Thus, it is not surprising that people with MS report walking as one of their priorities.⁶ Dalfampridine extended-release (dalfampridine-ER), although a relatively new treatment option, is rapidly growing as a pharmacologic approach for the management of gait-related impairment in MS.⁷

The efficacy of treatment with dalfampridine-ER (10-mg tablets taken twice daily) for improving gait speed in people with MS has been established in two phase 3 clinical trials.^{8 9} The pooled analysis of these two trials¹⁰ showed that individuals treated with dalfampridine-ER had a greater improvement in gait speed from baseline (13.8%) than those treated with the placebo (6.5%, P < .001). However, there has been no research examining how the efficacy of this new treatment approach compares to that of physical therapy, the traditional clinical management approach for treating mobility limitations in MS. The purpose of this review was to critically appraise the evidence for dalfampridine-ER and physical therapy approaches to increase gait speed in people with MS, with an emphasis on evaluating and comparing the magnitude of treatment effects. This review focuses primarily on the effects of dalfampridine-ER and physical therapy for increasing gait speed because gait speed is the primary outcome by which the efficacy of dalfampridine-ER has been evaluated and is the indicated use of the drug as approved by the US Food and Drug Administration.

To locate sources for this review, a systematic search of the research literature was conducted using PubMed, the Cochrane Database of Systematic Reviews, and the Physiotherapy Evidence Database. A health sciences librarian was consulted and gave input regarding the search strategy. The search strategy used in PubMed is shown in Supplementary Table 1 (published in the online version of this article at ijmsc.org), and was last conducted on November 25, 2014. Consideration was given to randomized controlled trials (RCTs), systematic reviews, and meta-analyses only. For physical therapy/rehabilitation studies, only those that involved task-specific gait training and measured treatment effects on gait speed were selected. Task-specific training is defined as “the repetitive practice of a task that is specific to the intended outcome.”^11(p1581) Studies that trained gait through repetitive practice of stepping using any modality (eg, treadmill, overground, and robotic assistance) were considered. This review was not intended to be a formal systematic review but rather an appraisal of the current results and effect sizes from the published clinical trials of dalfampridine compared with gait training.

Effects of Dalfampridine on Gait Speed in MS

The pharmacokinetics of dalfampridine-ER have been examined in several clinical studies^12–16 and summarized in recent review articles^{7 17} and will not be detailed herein. To briefly summarize the mechanism of action, dalfampridine (chemical name: 4-aminopyridine) is a potassium-channel blocker that is believed to improve motor function in people with MS by increasing nerve conduction through demyelinated axons.¹⁸

One phase 2 trial,¹⁹ two phase 3 trials,^{8 9} and a pooled analysis of the phase 3 data¹⁰ have examined the efficacy of dalfampridine-ER in MS. The results are summarized in Table 1. The phase 2 trial¹⁹ was a multicenter, randomized, double-blind, placebo-controlled, parallel-group design involving 206 participants. Participants were individuals with any form of MS who had not had a recent relapse or change in medication, were aged 18 to 70 years, and could complete the Timed 25-Foot Walk test (T25FW) in 8 to 60 seconds, which equates to gait speed ranging from 0.13 to 0.95 m/s. Participants were randomly assigned to one of four treatment groups: twice-daily dalfampridine-ER in 10-, 15-, or 20-mg doses or placebo. After baseline assessment, there was a 2-week single-blind dose-escalation phase, followed by a 12-week double-blind stable-dose phase and a 1-week dose-reduction phase. Although the dalfampridine-ER groups showed larger percentage increases in gait speed from baseline (approximately 7%–11% increases [estimated from a graphical presentation of data; actual values were not reported]) than the placebo group (an approximately 2.5% increase [estimated from a graphical presentation of data; actual values were not reported]), there were no significant differences between any of the groups.

Effects of dalfampridine extended-release on gait speed

To evaluate the clinical significance of the phase 2 efficacy results, Goodman et al.¹⁹ compared the proportion of participants in each group who had an average improvement in gait speed greater than 20% from baseline, a value they specified a priori to be clinically meaningful. Although the dalfampridine-ER groups had larger percentages of participants with average increases in gait speed exceeding 20% (23.5% taking 10 mg; 26.0%, 15 mg; and 15.8%, 20 mg) than the placebo group (12.8%), none of the differences were statistically significant. The authors conducted an additional post hoc responder analysis, which revealed a significantly higher proportion of responders in the dalfampridine-ER 10-, 15-, and 20-mg groups (35.3%, 36.0%, and 38.6%, respectively) than in the placebo group (8.5%). For this analysis, a responder was defined as a participant who experienced a consistent improvement during the treatment phase, specifically, a person whose gait speed for at least three assessments during the stable-dose treatment phase was faster than his or her maximum speed in the five nontreatment visits (ie, before or after the intervention period). The post hoc responder analysis prompted criticism by Kryscio²⁰ that Goodman et al. had excessively “massaged” the data after the planned analyses yielded nonsignificant results and that by doing so had increased the type I error rate. Subsequently, the phase 3 efficacy studies were powered for this responder analysis approach.

The two phase 3 clinical trials^{8 9} used a similar multisite, randomized, double-blind, placebo-controlled, parallel-group design. The minimum gait speed for inclusion was increased from 0.13 m/s in the phase 2 study to 0.17 m/s (maximum of 45 seconds to complete the T25FW). In the first phase 3 trial,⁹ participants were randomized at a ratio of 3:1 to receive 10 mg twice daily of dalfampridine-ER or placebo. There was a 2-week placebo run-in phase, followed by a 14-week treatment period and then a 4-week no-treatment follow-up. The primary efficacy variable was responder status, defined as consistency of gait speed improvement (ie, gait speed for at least three of the four assessments during the treatment period greater than the maximum speed in any of the off-drug assessments). The results showed that the proportion of participants meeting the responder criterion was significantly greater in the dalfampridine-ER group (35%) than in the placebo group (8%, P < .0001). The average increase in gait speed from baseline during the treatment period was 25.2% for dalfampridine-ER responders (0.16 m/s, 95% confidence interval [CI], 0.12–0.19 m/s), 7.5% for dalfampridine-ER nonresponders (0.05 m/s, 95% CI, 0.03–0.06 m/s), and 4.7% for the placebo group (0.03 m/s, 95% CI, 0.01–0.05 m/s) (Table 1). An analysis of all the dalfampridine-ER participants (ie, responders and nonresponders combined) compared with the placebo group was not reported by the authors because the primary efficacy variable nominated for the phase 3 trial was responder status. Thus, for this review, to provide a more accurate estimate of the effect size of dalfampridine-ER 10 mg twice daily on gait speed, an analysis of the reported data combining responders and nonresponders was performed using Comprehensive Meta-Analysis software, version 3 (Biostat, Englewood, NJ). This analysis showed that in the first phase 3 clinical trial,⁹ the mean change in gait speed from baseline for the dalfampridine-ER group was 0.07 m/s (95% CI, 0.05–0.08 m/s), compared with the reported 0.03 m/s (95% CI, 0.01–0.05 m/s) for the placebo group.

In the second phase 3 trial,⁸ the participants were randomly assigned in a 1:1 ratio to receive 10 mg of dalfampridine-ER twice daily or placebo. The study design was the same as in the previous phase 3 trial⁹ except that the treatment period was 9 weeks instead of 14 weeks and the nontreatment follow-up period was 2 weeks instead of 4 weeks. Again, the proportion of participants who were responders was significantly higher in the dalfampridine-ER group (42.9%) than in the placebo group (9.3%, P < .0001). The average increase in gait speed from baseline during the treatment period was 24.7% for dalfampridine-ER responders (0.16 m/s; 95% CI, 0.13–0.18 m/s), 6.0% for dalfampridine-ER nonresponders (0.04 m/s; 95% CI, 0.02–0.06 m/s), and 7.7% for the placebo group (0.05 m/s; 95% CI, 0.03–0.07 m/s) (Table 1). The overall effect of dalfampridine-ER combining responders and nonresponders (calculated using Comprehensive Meta-Analysis software, version 3) was 0.09 m/s (95% CI, 0.07–0.10 m/s). Importantly, the phase 3 trials showed that although the treatment effects were retained throughout the treatment period, the effects were reversed when treatment was discontinued.

The data from the two phase 3 trials were recently pooled for further analysis and examination of participant subsets.¹⁰ The efficacy results were consistent with those of the individual trials, demonstrating a larger number of responders in the dalfampridine-ER group (37.6%) than in the placebo group (8.9%, P < .0001). Further results are presented in Table 1. The pooled analysis also found that the dalfampridine-ER responder rate was independent of demographic characteristics, disease duration, level of disability, baseline walking speed, type of MS, and use of immunomodulatory therapies.

Effects of Gait Training on Gait Speed in MS

The physical rehabilitation RCTs identified by the current search comprised interventions that can be classified as gait training interventions, including conventional gait training and treadmill training with or without body-weight support or robot assistance,^21–26 aerobic exercise,^{25 27} progressive resistance exercise,^27–34 functional electrical stimulation with or without exercise,^{28 29 35} conventional physical therapy,^{31–34 36–38} and other interventions, including whole-body vibration,^39–41 torso weighting,⁴² Wii activities,^{43 44} massage,⁴⁵ and rhythmic auditory stimulation.⁴⁶ Although many of these studies measured treatment effects on walking ability,^{21–28 30 31 34 36–38 42 46 47} this review focuses on only the task-specific gait training RCTs.^21–26

The individual study findings from the gait training RCTs (k = 6) are summarized in Table 2. The results of five of these six RCTs,^{21 22 24–26} as well as one uncontrolled trial⁴⁸ and two case series reports,^{49 50} have been previously synthesized in a systematic review on the effectiveness of treadmill training, body-weight–supported treadmill training, and robot-assisted treadmill training (using the Lokomat [Hocoma Inc USA, Norwell, MA]) in people with MS.⁵¹ The purpose of the previous systematic review was to determine whether one gait training approach is superior to the others. Overall, Swinnen et al.⁵¹ found positive effects of treadmill training on gait speed in people with MS, with effect sizes ranging from small to large. There was no clear evidence from the review that one form of gait training was more effective than another.

Effects of gait training on gait speed

Most of the six gait training RCTs located for this review used robot-assisted or treadmill modalities for gait training, with or without body-weight support (Table 2), including one crossover design of both robot-assisted and treadmill training.²² Ruiz et al.²³ included both robot-assisted and treadmill training independently in each session. Conventional (overground) gait training was examined in three of the studies as a comparison group,^{21 24 26} and two studies had a no-treatment (delayed) comparison.^{23 25} Treatment durations ranged from 9 to 16 sessions (30–40 minutes per session). Frequency varied from two to three times per week to five times per week in the two inpatient studies.^{21 26} Conventional gait training was not well described but included the assistance of a physical therapist to train walking “in the gym room or sometimes outside on uneven surfaces with walking aids of (the patient's) choice,”²⁶ “walking overground (with or without walking aids),”²¹ or “gait and dynamic balance exercises, standing from sitting training, and walking with or without walking aids.”²⁴

To summarize the findings: Beer et al.²¹ found significant within-group differences for conventional (overground) gait training (median increase, 0.11 m/s) and robot-assisted treadmill training (median increase, 0.07 m/s), with a large effect size that was not statistically significant for the between-group difference, in favor of the robot-assisted group. In a randomized crossover design of body-weight–supported treadmill training and robot-assisted treadmill training, Lo and Triche²² found that both groups significantly improved their speed on the T25FW after the crossed-over intervention (mean increases, 0.39 and 0.29 m/s, respectively), but there was wide variation. At the end of the first phase (six sessions over 3 weeks), which provided a direct comparison of the treadmill and robot-assisted interventions, there were no statistically significant differences in changes in the T25FW, although the absolute change scores favored the treadmill group.²² Schwartz et al.,²⁴ who compared robot-assisted gait training on a treadmill with conventional gait training provided by a physical therapist, found that at the end of the 4 weeks of treatment, only conventional gait training showed significant improvements in gait speed (0.10 m/s). Similarly, Vaney et al.²⁶ also found a between-group difference in favor of the overground walking group compared with robot-assisted gait training (adjusted mean difference, 0.05 m/s, 95% CI, −0.03 to 0.13). Van den Berg et al.²⁵ found that aerobic treadmill training for 30 minutes three times per week for 4 weeks significantly improved gait speed (0.12 m/s increase) compared with a nonintervention control group (0.03 m/s increase), although the intervention group walked significantly more slowly (mean ± SD, 0.56 ± 0.17 m/s) than the control group (mean ± SD, 0.71 ± 0.28 m/s) at baseline.

The most recent RCT,²³ which was not included in the systematic review by Swinnen et al.,⁵¹ compared a combination of robot-assisted gait training and body-weight–supported treadmill training (without robot assistance) for 16 sessions over 2 months with a no-treatment control. The sample size was very small (N = 7), and although the median change in gait speed in the gait training group (0.14 m/s) was greater than that in the no-intervention control group (−0.02 m/s), the difference was not statistically significant.

Discussion: Dalfampridine Versus Gait Training

Figure 1 summarizes the mean change in gait speed and 95% CIs for individual studies and overall effect sizes for the gait training RCTs^{22 24–26} and the dalfampridine-ER clinical trials.^{8 9 19} Mean changes in gait speed are for individual treatment groups and reflect within-group change, not between-group differences. For Goodman et al., 2009⁹ and 2010,⁸ data were combined for responders and nonresponders to more accurately reflect the effect size of dalfampridine-ER on gait speed. The wider CIs for the gait training interventions are due to considerably smaller sample sizes relative to the dalfampridine-ER studies. Two gait training studies^{21 23} were excluded from the quantitative synthesis because data were reported as medians and interquartile ranges; the change in gait speed in these studies is shown in Table 2. The synthesis also excludes the no-treatment control groups because the focus of this review was on comparing dalfampridine-ER with gait training. Figure 1 clearly illustrates that across task-specific gait training RCTs in MS, the overall effect size (0.06 m/s; 95% CI, 0.02–0.10) is similar to that observed for dalfampridine-ER when accounting for both responders and nonresponders (0.07 m/s; 95% CI, 0.04–0.09). Although the effect size is considerably larger if only the responders are considered (0.16 m/s; 95% CI, 0.13–0.18) (Table 1),^{8 9} it is not possible to compare this effect size with that of the gait training studies because the gait training studies do not include responder analyses. It is not presently known what proportion of patients respond favorably to gait training in the context of a clinical trial.

Forest plot showing mean changes in gait speed and 95% confidence intervals (CIs) for individual studies and overall effect sizes for gait training randomized controlled trials22 24–26 and dalfampridine extended-release (D-ER) clinical trials.8 9 19 Mean changes in gait speed are for individual treatment groups and reflect within-group change, not between-group differences. For Goodman et al. 20099 and 2010,8 the data represent changes for all participants (combined responders and nonresponders). Analyses were performed using Comprehensive Meta-Analysis software, version 3. bid, twice daily.

Participant characteristics and major exclusion criteria for each of the studies are summarized in Table 3. Across all of the studies, the participants were predominantly women, with a mean or median age in the late 40s or early 50s. All of the studies excluded participants with recent relapses. The cognitive function of the participants was largely unreported. Mean or median disease duration was relatively consistent (typically, 11–15 years) across studies but was not always specified in the gait training studies. Despite similar demographic characteristics, direct comparisons of treatment effects between the gait training studies and the dalfampridine-ER trials should be made with caution because the studies differed in terms of average baseline gait speed and level of disability (Table 3). The dalfampridine trials had considerably larger sample sizes and longer treatment periods than any of the gait training trials. Indeed, several of the gait training trials were reported as pilot studies, with very small sample sizes and insufficient statistical power. Thus, the effect sizes observed for the gait training studies should be viewed cautiously. Of the six gait training RCTs, only Lo and Triche²² reported an effect size greater than the average 0.16 m/s increase in gait speed observed for dalfampridine responders.^{8 9} However, most of the gait training interventions, regardless of modality, produced larger mean increases in gait speed than did dalfampridine-ER in the dalfampridine nonresponders. Given that 62% of people who receive dalfampridine-ER are nonresponders,¹⁰ there is a tremendous need for development of other interventions for gait rehabilitation for people with MS.

Sample characteristics of the gait training and D-ER RCTs

Sample characteristics of the gait training and D-ER RCTs

The therapeutic effects of dalfampridine-ER last only as long as the medication is being taken, and they are reversed when the treatment is discontinued.^{8 9} Post-intervention follow-up was lacking in most of the gait training studies. However, Beer et al.²¹ and Schwartz et al.²⁴ reported that at 6-month follow-up assessments, gait parameters had returned to baseline values. Therefore, based on current evidence, both dalfampridine-ER and gait training treatment options seem to produce only short-lived improvements.

This review examined only task-specific gait training rehabilitation interventions, which were considered to provide the optimal comparison with dalfampridine-ER given the large amount of evidence that task-specific training is optimal for improving functional activities in people with neurologic disorders.⁵² Nonetheless, a more comprehensive review of other physical rehabilitation clinical trials on gait performance in people with MS may be warranted to fully appreciate the effect size of physical rehabilitation on gait speed. Indeed, there is strong evidence that exercise therapy improves mobility-related activities, although specific effects on gait speed are not clear.^{53 54} This review focused exclusively on gait speed because this was the primary efficacy measure for the pivotal phase 3 dalfampridine-ER trials. Although gait speed is considered the “functional vital sign,”⁵⁵ treatment effects on other mobility parameters and self-reported measures of perceived disability should also be taken into account to evaluate clinical significance. A review of these outcomes was beyond the scope of the present article.

There have been questions raised about whether the treatment benefit of dalfampridine outweighs the costs⁵⁶; however, the cost-effectiveness of dalfampridine and gait training have not been assessed. The cost of dalfampridine in the United States is high: the wholesale price of a bottle of sixty 10-mg dalfampridine-ER tablets (a 30 days' supply) was $1267.39 in 2010.¹⁷ In the United Kingdom, the cost of dalfampridine is £4700 per year (2014), approximately $7200 USD.⁵⁷ Gait training costs could not be estimated because of variations in the content, duration, and number of treatment sessions.

This review highlights some important priorities for future research. First, there is a need to directly compare dalfampridine-ER with gait training and other physical therapy interventions. Moreover, combining gait training/physical therapy (such as exercise) with dalfampridine-ER may increase the effect size on gait speed and increase the proportion of patients who respond (ie, experience a clinically meaningful treatment effect). To date, gait training studies have not included a responder analysis; much larger gait training trials are needed to afford such an analysis. Cost-effectiveness should also be considered in future research trials of dalfampridine and rehabilitation to enable comparison of the costs and benefits of each treatment approach. As Hayes⁷ argues, another important priority for future research is to examine whether dalfampridine-ER can produce clinically meaningful improvements in instrumental activities of daily living and employment. Indeed, walking in the real world is far more complex than walking in the clinic or laboratory setting, and real-world ambulation typically involves the concurrent performance of cognitive tasks, such as talking (ie, dual tasking). Therefore, future research should also examine the effects of rehabilitation (drug and/or physical) on more realistic locomotor tasks, such as dual-task walking and obstacle negotiation.

Conclusion

There are presently no studies that have directly compared oral dalfampridine-ER 10 mg given twice daily with other treatment options for gait impairment, such as gait training. Thus, comparisons of treatment efficacy for these two options can be made only from relative treatment effects in individual studies. The quantitative summary of existing evidence in this review demonstrates that the overall effect size of dalfampridine-ER is similar to that observed in a pooled analysis of gait training RCTs, although important differences in the sample populations mean that these comparisons should be viewed cautiously. In addition, a wide variation in the response to both forms of therapy is expected. Although 38% of participants in RCTs demonstrated a consistent therapeutic response to dalfampridine-ER, the proportion of responders in gait training studies has not been assessed, and the correlation between study responders and real-life responders has not been established. Both treatment approaches provide only short-term improvements; however, retention was infrequently assessed in the gait training studies. Furthermore, the clinical significance of the changes in gait speed observed with dalfampridine or gait training has not yet been fully determined. In summary, the current evidence is insufficient to conclude whether dalfampridine-ER or gait training is superior for improving gait speed in people with MS. Trials comparing the two approaches directly and examining the effects of a combined intervention are warranted.

PracticePoints

The mean increase in gait speed with dalfampridine in people with MS is 0.07 m/s (95% confidence interval [CI], 0.04–0.09 m/s). A proportion of patients (referred to as “responders,” 37.6%) experience larger treatment effects: 0.16 m/s (95% CI, 0.13–0.18 m/s).

Mean increases in gait speed for individual gait training interventions range from 0.01 m/s (95% CI, −0.05 to 0.07 m/s) to 0.39 m/s (95% CI, −0.20 to 0.98 m/s). The CIs are wide due to very small sample sizes.

The current evidence is insufficient to conclude whether dalfampridine or gait training is superior for improving gait speed in people with MS. Direct comparisons of dalfampridine and gait training are needed, as well as examination of dalfampridine combined with physical therapy.