Subscribe to eTOC

Negative Findings on Minocycline for ALS Prompts Reassessment of Study Designs


  • ✓ Investigators of several large ALS trials provide insights into why past trials have failed, and offer suggestions for new types of study designs.

Chastened by the third major trial in a row to report negative findings for patients with amyotrophic lateral sclerosis (ALS), researchers say they are now moving more carefully before mounting the next major trial.

The investigators concede that — spurred both by the desperation of patients and the allure of the superoxide dismutase (SOD1) mouse model of ALS — they may have been too hasty in the past to rush through the initial phase 1 and 2 studies; but they insist they have learned their lesson.


Dr. Ying Kuen Cheung said that with so many candidate agents it is impossible to test every single one in a separate phase 3 clinical trial.

“Before we do all these fancy phase 3 trials, the field has to go back to do the traditional pharmacokinetic and pharmacodynamic studies to find the right dosage and regimen,” said Merit E. Cudkowicz, MD, associate professor of neurology at Harvard Medical School and co-director of the ALS Clinic at Massachusetts General Hospital. “In the rush from the lab to the clinic, you have to make sure the drug is getting into the brain and doing what you want it to do.”

The latest disappointment — the findings of a phase 3 trial of minocycline published in the December issue of the Lancet Neurology—follows negative findings for both creatine and topiramate. The minocycline study found significantly faster decline and a trend toward greater mortality in the experimental group, even though a 2004 Neurology mouse study had shown significant benefit and two small human trials had shown no signs of danger.

Questions about the ALS Mouse Model

“The trial had to be done,” said Petra Kaufmann, MD, assistant professor of neurology at Columbia University and co-director of the Spinal Muscular Atrophy Clinical Research Center. “But it certainly raises serious doubts regarding the mouse model in ALS. Having significant benefit in the mouse model alone is not enough.”

Eric J. Sorenson, MD, associate professor of neurology and division chair of neuromuscular diseases at the Mayo Clinic in Rochester, MN, said the SOD1 mouse remains the only animal model for testing potential treatments, and so remains an important, if imperfect, tool.

“It's still the closest thing we have,” he said. “But therein lies the value of the phase 1 and 2 clinical trials that have largely been passed over. I wouldn't want to pick on the minocycline study, because none of the large trials that have relied primarily on mouse data have worked out. But the cornerstone of phase 3 trials, even with the best animal model, has to be dose-escalation and toxicity trials.”

While defending the rationale and design of the minocycline trial, the senior author of the study agreed that better spadework will be necessary if future ALS trials are to bear fruit.

“The biggest problem is we just don't yet know the cause of the disorder,” said Paul H. Gordon, MD, associate medical director of the ALS Research Center at Columbia University. “It may turn out that ALS is numerous disorders that we're lumping into one pot. That may be why the trials are failing. We might find that there are numerous causes that have different treatments.”

Need for Innovative Trial Designs

Going forward, Dr. Gordon said: “As we screen more and more drugs in the laboratory, we really need to be sure we're picking winners for phase 3 trials. There are a number of new phase 2 designs that we're hoping will improve our efficiency and ability to eliminate losers early on and pick winners for further evaluation.”

One such approach, called an adaptive design, moves sequentially from phase 1 to 3 within the structure of a single study, using the results of the early phases to shape the dose and regimen of the later phases. Dr. Cudkowicz, who led the topiramate study, is now leading a study of ceftriaxone using an adaptive design.


Dr. Petra Kaufmann: “The trial had to be done. But it certainly raises serious doubts regarding the mouse model in ALS. Having significant benefit in the mouse model alone is not enough.”

In the standard study design, she explained, an investigator completes a phase 1 trial, writes up the findings, and tries to get them published before deciding whether the results warrant moving on to the phase 2 trial. Then recruitment begins for the phase 2 trial; the process repeats itself — analyzing and publishing the findings — before proceeding with recruiting for a larger phase 3 trial. Each phase is a totally separate enterprise.

With an adaptive design, there is an overarching umbrella structure in which each of the three phases fit — even though the exact parameters of the later stages can be altered based on data accumulated in the earlier phases.

“We're doing all the phases of clinical trial development in one study,” Dr. Cudkowicz explained. Beginning with just 60 participants tested first for one week, then for 20 weeks, the trial can eventually ramp up to 540 people given either placebo, 2 grams per day or 4 grams per day, depending on the results of the initial stages.

“The goal is to either stop the trial early based on pharmacokinetics or safety, or to pick the right dosage as quickly as possible,” Dr. Cudkowicz said. “The advantage is you save time. Usually after you do a phase I or 2 trial, you analyze and reboot. The time between these phases can really add up. We thought it would save us two years by not having the breaks between the phases.”

Dr. Gordon co-authored a paper in the November 2006 Neurology proposing an even more radical approach to selecting promising ALS therapies in clinical trials: a “selection paradigm,” in which multiple agents are tested against each other to see which one performs best in a large, randomized, controlled trial.

“It's like a horse race, to see which one outperforms the others,” Dr. Gordon said. “It's an idea we've used for a phase 2 study that we'll be presenting soon.”

Another author of the paper proposing the selection paradigm acknowledged that it might be difficult to convince companies and academics working on a variety of agents to pool their resources for a single trial. But, he said, it would be worth the effort.

“There are so many candidate agents out there, and potential combinations of treatments, that it's impossible for researchers to test every single one in a separate phase 3 trial in humans,” said Ying Kuen Cheung, PhD, associate professor in the department of biostatistics at the Columbia University School of Public Health. “The selection paradigm would require fewer patients, and far less time and money, than testing each one individually.”

Other Agents in Development

Even as the design of future trials is debated, a number of ongoing trials will soon be reaching completion. Dr. Sorenson, for instance, is leading a phase 3 trial of insulin-like growth factor 1 that has already completed patient follow-up and awaits only the data-analysis stage. The drug, previously tested in both the SOD1 mouse and humans, produced conflicting results that led patients to advocate for a final, unequivocal trial.

“We've already started biting our nails,” Dr. Sorenson said of the two-year randomized trial involving 330 patients. “The dice have been rolled.”

Also nearing completion is a phase 2 trial led by Dr. Kaufmann, in which coenzyme Q10 or placebo was given to 185 patients. The dietary supplement is a free radical scavenger that has shown benefit in SOD1 mice and is already taken by many patients with ALS. “I'm still blinded,” she said, “but we hope to have the results for the next AAN meeting in April.”

In addition, a previous pilot trial comparing the HIV drug ritonavir against both hydroxyurea and placebo found that while the former caused harm, the latter showed a trend toward benefit. “We're debating now whether to go forward with a larger study of hydroxyurea,” said Catherine Lomen-Hoerth, MD, PhD, assistant professor of neurology at the University of California-San Francisco, and director of the ALS Center at UCSF Medical Center. The drug, approved for use in leukemia and sickle cell anemia, had anecdotally been observed to be effective in a few ALS patients and had a potentially interesting mechanism of action, Dr. Lomen-Hoerth told Neurology Today.

More Questions on Minocycline

Another issue raised by the disappointing minocycline results is whether neurologists were right to have prescribed the drug to patients off-label and outside the confines of the trial.

“Many patients had been taking minocycline off-label,” Dr. Kaufmann said. “This trial shows again how important it is to take new medications only in controlled trials.”

But Dr. Lomen-Hoerth said that many of her patients were desperate to take the drug outside the trial. “I did prescribe it for some patients, and when I let them know they needed to stop because the study found it caused harm, they were not upset at having tried it,” she said. “It gave them hope for something that might be effective. Sometimes there are patients in remote areas who have no chance to participate in a clinical trial who very much want to try something.”

Beyond ALS, according to Dr. Gordon's paper in Lancet Neurology, the trend toward harmfulness “has implications for trials of minocycline in patients with other neurologic disorders.” But researchers conducting studies of the drug for multiple sclerosis and Huntington disease said they are not deterred.

“There are data from animal and humans suggesting a small window between toxicity and efficacy,” said Dr. Cudkowicz, who is running a randomized trial using a dose of 200 mg per day, half that of the ALS trial, in 100 patients with Huntington disease. “We don't know now whether 400 mg was the right dosage for ALS, or whether a smaller dose would have perhaps done better,” she said.

A highly respected British ALS specialist, Nigel Leigh, MD, professor of clinical neurology and head of neurology at Kings Hospital in London, is seeking approval for a small trial of minocycline in ALS that would hope to find a dosage that causes no harm. In an interview in the Nov. 23 Science, he said that it is possible that the high dosage of the antibiotic was neurotoxic for patients. Such a prospect underscores yet again that selecting the right dosage and trial design are at least as important as finding the right drug.


• Gordon PH, Moore DH, Tandan R, et al., for the Western ALS Study Group. Efficacy of minocycline in patients with amyotrophic lateral sclerosis: a phase III randomized trial. Lancet Neurol 2007;6(12):1045–1053.
•Gordon PH, Moore DH, Miller RG, et al. Placebo-controlled phase I/II studies of minocycline in amyotrophic lateral sclerosis. Neurology 2004;62:1845–1847.
•Cheung YK, Gordon PH, Levin B. Selecting promising ALS therapies in clinical trials. Neurology 2006;67:1748–1751.
•Shefner JM, Cudkowicz MD, Burke D, et al., for the NEALS Consortium. A clinical trial of creatine in ALS. Neurology 2004;63:1656–1661.
•Couzin J. Clinical research: ALS trial raises questions about promising drug. Science 2007;318:1227.