Riviere M, Meininger V, Zeisser P, Munsat T. An Analysis of Extended Survival in Patients With Amyotrophic Lateral Sclerosis Treated With Riluzole. Arch Neurol. 1998;55(4):526-528. doi:10.1001/archneur.55.4.526
In an attempt to better understand and define the progression of amyotrophic lateral sclerosis (ALS), we developed a classification of 5 discrete health states that reflect patients' activities of daily living. These health states were used to determine whether patients with ALS who are treated with riluzole differed from those treated with placebo.
Clinics for patients with ALS.
Placebo-controlled trial of riluzole treatment in 959 patients with ALS.
Treatment with riluzole or placebo.
Main Dependent Measures
A Cox model was used to assess whether, from the initial randomization to the end of an 18-month follow-up, there was a difference in the times of transition into subsequent health states between patients treated with riluzole and those treated with placebo.
Our analysis showed a significant difference in the time to transit between the riluzole and the placebo groups in less severely affected cases, ie, state 2 and state A (the milder states) of ALS.
Patients receiving riluzole remained in the milder health states longer (P<.05).
AMYOTROPHIC lateral sclerosis (ALS) is a chronic and debilitating neurodegenerative disease for which no cure is currently available. The clinical syndrome associated with the disease is the result of the degeneration of cortical, bulbar, and medullary motor neurons that enable voluntary motor function. Death is usually the result of respiratory failure. The worldwide incidence of the disease is 1 to 3 cases per 100000 population per year.
Patients with ALS can be evaluated using several different functional scales,1- 3 as well as quantitative strength measures.4,5 These scales and measures are used to assess bulbar, respiratory, and upper and lower extremity function. However, they do not allow classification of patients into discrete health states that would allow a closer assessment of their daily functioning.
Riluzole (2-amino-6-[trifluoromethoxy] benzothiazole, RP54274) is an antagonist of glutamatergic neurotransmission6 that prolongs survival in ALS. It is thought that the ability of this compound to reduce glutamate-related excitotoxic effects may involve the synergistic interaction of 3 separate mechanisms: inhibition of glutamic acid release,6 noncompetitive block of NMDA receptor–mediated responses,7 and direct action on the voltage-dependent sodium channel.8 The efficacy of riluzole in extending the survival of patients with ALS has been demonstrated in 2 randomized, double-blind, placebo-controlled trials.9,10
Our objective was to reanalyze the data from the larger ALS trial10 using a discrete 5–health state classification in an attempt to elucidate the mechanism underlying the beneficial effect of riluzole therapy on survival.
To assess the functional status of patients with ALS over time, we developed a new clinical classification system, the ALS Health State Scale (ALS/HSS). The ALS/HSS was developed in 2 stages. First, we interviewed 17 ALS specialists who identified 4 clinically discrete health states in ALS, which are referred to as mild, moderate, severe, and terminal (state 5 is death). We then combined this information with the existing functional scales and developed an algorithm whereby patients with ALS could be semiobjectively classified into the various health states. Clinical classification was based on the ability of the patient to speak, to ambulate, and to perform upper extremity activities of daily living. The 4 distinct states are as follows:
State 1 (mild). Recently diagnosed; mild deficit in only 1 of 3 regions (ie, speech, arm, and leg); and functionally independent in speech, upper extremity activities of daily living, and ambulation.
State 2 (moderate). Mild deficit in all 3 regions or moderate to severe deficit in 1 region, while the other 2 regions are normal or mildly affected.
State 3 (severe). Needs assistance in 2 or 3 regions; speech is dysarthric and/or patient needs assistance to walk and/or needs assistance with upper extremity activities of daily living.
State 4 (terminal). Nonfunctional use of at least 2 regions and moderate or nonfunctional use of the third region.
Patients in the riluzole trial10 were assessed with a modified Norris scale every 2 months. These scores were transformed from the Norris scale1 to the ALS/HSS using an algorithm. The 3 neuroanatomical regions considered were bulbar, arm, and leg. This required 2 procedures. Beginning with the Norris Limb scale, all items pertaining to the arm were extracted; each item was scored from 0 (poor function) to 3 (normal function); and these scores were summed. The maximum arm score, based on 11 distinct items was 33, indicating normal function. This procedure was repeated both for the modified Norris Leg scale (5 items) and the modified Norris Bulbar scale (13 items). The scores from the 3 regions were combined, and on the basis of the aggregate health score, patients were assigned to 1 of the 4 health states. Our analysis tested the hypothesis that riluzole therapy specifically delays early disease progression in patients with ALS.9
The time to failure was defined as the number of days after the initial observation (either inclusion or entry into the health state) that passed before the individual entered the next more severe health state. We initially compared the time to failure in riluzole and placebo groups by individual health state, but because of the low numbers of patients in health states 1 and 4, analysis of the 2 milder health states (forming state A) and the 2 more severe states (forming state B) was also carried out.
Health states (individual and combined) were incorporated into a Cox proportional hazard model to perform survival analysis. Time to failure was compared and assessed as relative risk; a value of less than 1 suggests that treatment with riluzole conferred superior survival benefit to treatment with placebo, whereas a relative risk of greater than 1 favored treatment with placebo over treatment with riluzole. Data were used to reproduce 18 months of survival for both the placebo and riluzole arms of our model.
No data were available across any health state for 5 of 959 patients. Table 1 shows the distribution of patients in each health state, according to treatment group, at baseline. State 2 (moderate) had the greatest number of patients. For the initial baseline period, there was no significant difference in the distribution of health states across treatment groups (P=.45).
Table 2 displays the point estimates for the median time to failure (in days) for the analysis performed on the individual and combined health states. For purely descriptive purposes, no P values are presented. The greatest difference in time to transition over the 18-month period was observed in state 2 (moderate): 265 median days in the riluzole group compared with 210 median days in the placebo group. When the 4 health states were collapsed to states A and B, the greatest difference in transition time was observed in state A (mild and moderate combined): 317 median days in the riluzole group compared with 242 median days in the placebo group.
Results of the Cox model analysis using individual health states are shown in Table 3. A statistically significant difference was observed only in state 2 (moderate), where the relative risk (riluzole compared with placebo) was 0.81 and P=.03. Cox model analysis using the combined health states was consistent with that for the individual states; ie, a significant effect of riluzole was observed only in state A, where the relative risk was 0.82 and P=.03 (Table 3).
Analysis of both individual and combined health states showed a significant increase in time to transit in favor of riluzole treatment in the milder disease states. This analysis included 954 of the 959 patients. Although the distribution of patients in the 2 treatment groups did not show any difference among the 4 health states, there was a larger number of patients in state 2 at the beginning of the trial. This is most likely a consequence of the inclusion criteria.
Riluzole therapy did not demonstrate any benefit in patients with more advanced ALS. However, prior therapies may have influenced transition times in these patients. Also, the number of patients included in states 3 and 4, and hence state B, might be too small to show an effect of riluzole. The finding that an effect of riluzole was not observed in the more severe health states may be attributable to the relatively short 18-month follow-up period. Longer follow-up studies might resolve the question of benefit in advanced disease states. Furthermore, it is unlikely that significant benefit can be seen when the total remaining pool of motor neurones is severely depleted.
Our analysis was post hoc and should be considered as a preliminary study requiring future confirmation. The classification system should be further refined, and additional studies of the clinical and economic impact of the treatment of ALS with more effective agents will help to achieve this.
Despite the limitations, our design represents a new approach to the management of ALS that has important implications for the design of future studies, treatment decisions, and health economics. Furthermore, the significant increase in time spent in the 2 milder health states during riluzole treatment provides a tentative explanation of the increase in survival duration observed.9,10
Accepted for publication August 22, 1997.
This study was funded by Rhone-Poulenc Rorer, Antony, France.
Reprints: Theodore Munsat, MD, Neuromuscular Research Unit, New England Medical Center, NEMC 273, Boston, MA 02111.