12 June 2009. To determine if a medicine works, scientists need reliable patient data, but the right numbers are not always easy to come by. The problem is particularly sticky for a disease like amyotrophic lateral sclerosis that lacks solid biological measures of drug efficacy. ALS takes life swiftly, usually within three years of diagnosis. Because researchers have nothing better, they are forced to rely on survival rate, a measure fraught with variability, or on functional rating scales that attempt to sum up a person’s abilities with a few questions. Because new ALS medications may make only a small difference to outcomes, trial scientists are rather like referees trying to time a 20-second race with a watch that only has a minute hand. A study in this month’s Archives of Neurology highlights another challenge. It suggests that time to permanent ventilation, which frequently is used as a surrogate for survival on the assumption that people would have suffocated had they not opted for the ventilator, is too variable among centers to be of much use. Many scientists are searching for molecular biomarkers, perhaps in the blood or cerebrospinal fluid, but their use in clinical trials is still a way off (see ARF related news story).
Part of the dilemma is that no ALS treatment so far has made a large difference in disease progression. Riluzole, the only medication approved for ALS, adds a few months to a person’s survival. Scientists must look closely to find such small improvements, and the variability in outcome measures may be as broad as the effects they hope to see. “If we really had effective therapy, we would not be worrying about how many angels are dancing on the head of a pin,” said Stanley Appel of the Methodist Hospital System in Houston, Texas.
Trial scientists have commonly equated the use of permanent ventilation with death, assuming that without ventilation, death would result from respiratory failure. Survival without tracheostomy was the primary outcome for the riluzole studies (Bensimon et al., 1994; Lacomblez et al., 1996). However, it has become apparent that people with ALS may die from other causes such as pneumonia, heart failure, and pulmonary embolism (Corcia et al., 2008). The authors of the Archives of Neurology paper, led by Paul Gordon of the Hopital de la Pitie-Salpetriere in Paris, France, argue that because the use of ventilators varies among patients and among clinics, the best outcome measure is death rate, regardless of cause.
Gordon and colleagues combined data from three Phase 3 trials for ALS drugs, covering 2,077 subjects in all. Of 745 recorded deaths, they found that only 82 percent could be attributed to respiratory failure. The use of ventilators varied widely among the nine countries in which centers participated in the trials. In the United States, for example, 6.6 percent of subjects underwent tracheostomy, the highest rate. In the United Kingdom, with the lowest rate, fewer than 1 percent of patients elected a tracheostomy.
Using both ventilation and death as endpoints, the mean survival time for people in the three trials was 458 days; for death alone, it was 467. The endpoint used should make a difference to study design, the authors wrote. They calculated that to detect a 10 percent difference in survival, in an 18-month trial, 490 subjects would be required using the combination endpoint, whereas 410 would be sufficient using death alone.
“Death and respiratory support cannot be assumed to be equivalent,” Gordon wrote in an e-mail to ARF. “We think that the sensitivity and reproducibility of future trials could be improved if death rate alone is used for the survival endpoint.”
Yet death rate is also an imperfect measure. Other differences among trial clinics—such as the availability of physical therapy—can skew the results, said Robert Bowser of the University of Pittsburgh. “Small things like that can make a large difference in a rapidly progressing disease.” He suggested that death should be a primary endpoint for trials, but time to ventilation should be included as a secondary measure. “I think that is probably the safest route we can take at the moment.”
Another issue with survival as an outcome is that it takes time; trials must last for years in order to collect data. In that time, subjects may move away or drop out of the study. “I think it definitely hinders studies,” said Richard Bedlack of the Duke ALS Clinic in Durham, North Carolina. While it is unlikely that researchers would miss out on a truly effective treatment, they may waste too much time on drugs that are ineffective.
Though survival rate makes sense as the primary outcome for a fatal condition, death versus survival is an inherently unsatisfactory outcome for both scientists and research participants. “From my perspective, I am more concerned with getting measurements while the patient is alive,” Appel said. “If we are telling patients that we have to wait until they die to know whether a drug works, they are not going to be very happy about that.”
During treatment, scientists can use measures such as the ALS Functional Rating Scale, or a scoring system designed by Appel (Appel et al., 1987), to chart progress. These systems rate a person’s ability to perform basic functions such as swallowing and speaking. The scores have several advantages, Bedlack said: they are simple to administer, appear to decline linearly, and correlate with other measures such as muscle strength. “I am a fan of the ALS Functional Rating Scale until we get something better,” Bedlack said.
That “something better” is likely to be a biomarker. Such a marker for ALS would not only improve trial measurements, but also has potential use as a diagnostic. Right now, doctors rely on symptoms and histories, not biochemistry, to diagnose motor neuron disease. Scientists would also like to use a biological marker to determine if a drug is having the desired effect, and to select the proper dosage. For example, researchers testing a cholesterol drug can easily determine if people’s cholesterol levels change. The ALS researcher can only administer the therapy and hope. “We just do not have good surrogate measures,” Bedlack said. “That is what is holding back the whole field, in my opinion.”
A good biomarker, validated as a surrogate endpoint, is “years away,” Bowser said. But with recent advances in ALS research, Appel believes there is reason to be optimistic: “We all have to hang in there, because there have to be some solutions that we can get to.”—Amber Dance.
Gordon PH, Corcia P, Lacomblez L, Pochigaeva K, Abitbol JL, Cudkowicz M, Leigh PN, Meininger V. Defining survival as an outcome measure in amyotrophic lateral sclerosis. Arch Neurol. 2009 Jun;66(6):758-61. Abstract