This is Part 2 of a four-part series. See also Part 1, Part 3, Part 4.
Download a PDF of the entire series.
27 October 2011. Stephen Stokes has had amyotrophic lateral sclerosis (ALS) for two years. For the 47-year-old Boston resident, joining a clinical trial with even a sliver of a possibility that the drug would slow his disease was a no brainer. However, choosing the right trial was anything but. Trials vary not only in the drug they test, but also in the study design and what impact it might have on a participant’s activities. Stokes told ARF that he agonized over his options. Part of the dilemma was that one trial he considered would have required a jugular intravenous line, rendering swimming and showering difficult. Part of it was that any large trial came with the risk of being assigned to the placebo group. In the end, Stokes decided to enroll in the Phase 3 trial of dexpramipexole (formerly known as KNS-760704). The drug, he hopes, will protect the motor neurons under attack in his spinal cord. The dexpramipexole study is one of several run by the Northeast ALS Consortium (NEALS) (see Part 1).
Since 1995, NEALS has united clinicians around the common goal of bringing potential medicines to people like Stokes by mounting large, multicenter clinical trials. However, it is not clear what is the best way to run a clinical trial for ALS. The NEALS team has experimented with many trial designs and a variety of outcome measures. “I do not necessarily think that we know what the best approach to ALS trials is,” said NEALS member Richard Bedlack, who directs the Duke ALS Clinic in Durham, North Carolina. Bedlack commended the consortium for not being “stuck in its ways.” The overall trend is toward leaner, more efficient trials, added Robert Miller, who directs the Forbes Norris MDA/ALS Research Center at the California Pacific Medical Center in San Francisco. It is no small issue: Poor design choices can make it difficult to recruit participants or result in invalid or insufficiently powered data. The worst outcome is not a negative trial, but an uninformative one from which scientists cannot learn.
As the behemoth in the ALS clinical trials field, NEALS has the opportunity to test-run trial designs as it experiments with different drugs. Currently, NEALS is testing several study designs in trials. They include the Phase 3 dexpramipexole study, a multiphase study of ceftriaxone to clear excitotoxic glutamate from synapses, and a Phase 2 trial comparing the mitochondrial booster creatine to the inflammation regulator tamoxifen. Earlier in the clinical pipeline, they include a Phase 1 trial with stem cell therapy (one of the first of its kind; see ARF related news story). Targeting select subpopulations, the designs include a Phase 1 trial of an antisense nucleic acid to reduce levels of the mutant enzyme superoxide dismutase 1 (SOD1), and a Phase 2/3 study examining arimoclomol to prevent protein aggregation (Cudkowicz et al., 2008). In the past, NEALS has also tested lithium, which ended up being ineffective; sodium phenylbutyrate, which proved safe (the study was not designed to examine clinical efficacy; Cudkowicz et al., 2008); celecoxib, which was ineffective (Cudkowicz et al., 2006); coenzyme Q, deemed safe (the study was not designed to examine clinical efficacy; Ferrante et al., 2005); topiramate, which gave no benefit (Cudkowicz et al., 2003); and a compound, CK-2017357, that activates skeletal muscle in an attempt to relieve the weakness that characterizes ALS, which appeared promising in a Phase 2 study.
Filling ALS trials is hard, in part because both doctors and patients can shy away from joining trials, Bedlack said. Doctors are concerned that trials will increase their workload. NEALS addresses this in that its established infrastructure makes participation easier for doctors. Many patients are reluctant because they worry that being in a trial will cost them time and money; some even fear they will become helpless "guinea pigs" at the mercy of scientists, Bedlack said. The last concern is unfounded—a person cannot participate in a trial without giving informed consent—but the general specter of the abusive scientist persists in some people’s minds.
Even though new treatments cannot reach FDA approval without ALS patients participating in trials, only 10 percent of people with the disease sign up (Bedlack et al., 2008). That number seems respectable compared to other fields; a recent study suggested that less than 1 percent of people with cancer join trials (Al-Refaie et al., 2011). However, the overwhelmingly larger number of cancer patients still enables many more treatment trials in cancer than in ALS. What’s more, those people with ALS who do enroll represent only a sliver of the spectrum of the ALS population. Data from a recent study suggest that people in trials are younger and took longer to diagnose than people with ALS overall. People in trials were also more likely to be men and to have spinal onset of the disease (Chiò et al., 2011).
Stokes, an environmental consultant, carefully reviewed his options before joining the dexpramipexole study. He calls the decision process “double Russian roulette”: First, he had to bet on which drug was most likely to succeed; second, he had to hope he would wind up in the active compound, not the placebo, arm of the study. “The psychological consequences of that can never be understated,” he said. ”It might all be for nothing. It is kind of a form of cruelty, I think.”
Trial designers try to heed the concerns of people like Stokes without sacrificing the science. For example, NEALS researchers testing lithium two years ago needed to recruit participants who were willing to risk receiving the placebo, even though the drug was already available for bipolar disorder and people with ALS could easily get it off-label. Those eager to obtain lithium were tempted to do so by a small previous study that had suggested the medicine could prolong survival in ALS (see ARF related news story on Fornai et al., 2008), but the claim had not been rigorously tested. People with ALS have so little time that for them, a stay on placebo is particularly costly.
To attract participants, NEALS used a “time to event” trial design, promising that anyone on placebo whose condition significantly worsened would be automatically switched to the lithium arm. Worsening by six points or more on the ALS Functional Rating Scale, a measure of daily activities such as preparing food, was considered an “event” and reason to switch. To obtain an answer quickly, NEALS preset checkpoints to examine the data, rather than wait until the end of the trial. The team planned the first checkpoint to occur once 84 participants had joined; the second was planned for six months after that, or once 55 events had occurred; and the third was intended to occur after 100 events. As it turned out, the study did not make it past the first checkpoint. At that time, the investigators saw that not only was there no benefit to lithium, but the medication might have done harm. They were able to halt the trial and announce their results within just eight months (see ARF related news story on Aggarwal et al., 2010). This, then, is an example of a negative trial that is nonetheless successful because it was informative.
About the placebo arm that so frequently troubles patients, Miller said it might be possible to eliminate it altogether from Phase 2 studies. Miller heads another trial team, the Western ALS (WALS) consortium, which also tested lithium in a recent Phase 2 trial. WALS compared data obtained from 107 people who took lithium during the trial to published data on 249 placebo control subjects from six previous ALS trials of other drugs (NEALS’ lithium, creatine, and celecoxib trials; WALS’ minocycline trial; Gordon et al., 2007; Kaufmann et al., 2009 on coenzyme Q; and Miller et al., 2007 on TCH346). Like NEALS, WALS found no evidence for a benefit from lithium (Miller et al., 2011). Although placebo groups are required for Phase 3 trials, Miller believes that smaller screens and Phase 2 trials could proceed without them, using these kinds of historical controls instead. For this approach to work, he noted that researchers must make sure the new participants and old controls match on clinical features. In addition, outcomes for placebo-group people must have remained constant since the historical studies. (Over time, outcomes in people with ALS have improved due to better symptom management; see Qureshi et al., 2009.)
For its part, NEALS is pursuing placebo-free results in its Phase 2 trial of tamoxifen and creatine. In this “selection design” trial, participants will receive either creatine or one of two tamoxifen doses. (NEALS has tested creatine before and the drug provided no benefit, but researchers now suspect that the doses used in the previous trial were too low; Shefner et al., 2004). Why combine the two drugs in one trial? It is a fast way to get results when you are not sure if either drug warrants a large trial, said NEALS co-founder Jeremy Shefner of the State University of New York (SUNY) Upstate Medical University in Syracuse. Selection design allows researchers to obtain maximum data on different drugs from a small number of patients, 60 in this study. In the current trial, investigators are looking for one arm to provide 20 percent benefit over the other two on the ALS Functional Rating Scale. The winner is the drug, or dose, worth pursuing further. If no clear winner emerges, then the results will not help the team decide whether to drop or continue to study either drug.
Right now, one of NEALS’ highest-profile projects is the 62-site, 600-person ceftriaxone trial. The “adaptive design” study allows the researchers to modify their plans based on early data coming from the trial, and flows seamlessly from Phase 1, to 2, to 3, eliminating downtime among trials (for more on adaptive designs, see ARF related news story). The researchers have completed the first two phases, examining ceftriaxone uptake into the cerebrospinal fluid and drug safety; now they aim to discover if the drug fights ALS. The original participants can stay on for further phases, and the team added more participants at each stage. The results of the study are expected in the fall of 2012.
For dexpramipexole, NEALS took a different approach to Phase 2 trial design. The study divided 100 participants into four groups, one placebo and three different drug doses. Participants swallowed their assigned pills for 12 weeks, after which they took a break from treatment so the doctors could determine whether any unpleasant side effects or any improvements in their ALS Functional Rating Scale score or lung capacity disappeared once the drug was discontinued. Then, the researchers re-randomized people to one of two drug doses for another 24 weeks. At some point during that period—timing was blinded to the subjects—people received placebo for four weeks.
This novel study design allowed the researchers to examine multiple doses and assess potential efficacy at the same time as they made sure the drug was safe and tolerable. Plus, the multiple doses and two randomizations increased the chance that any one participant would receive active medication at some point in the trial, increasing the appeal to potential volunteers. The research team found that in both the first and second run, the drug slowed down the disease in a dose-dependent manner. If the current Phase 3 trial confirms that the drug works, the earlier trial “might end up being the new model for Phase 2,” Cudkowicz said.
Like Cudkowicz, Stokes and 799 other patients are hanging on the results of the Phase 3 study. Stokes takes his pill morning and night, and is disappointed that he has not noticed any difference in his condition. Even unpleasant side effects would be welcome, he said, if it meant he might be on the active drug that could potentially keep him alive longer. In the end it is that hope, more than anything that trial designers do, that sells the idea of joining a study. Many other participants are further motivated by the desire to help future patients.—Amber Dance.
This is Part 2 of a four-part series. See also Part 1, Part 3, Part 4. Download a PDF of the entire series.