23 September 2011. In the last 10 years, no new Alzheimer’s disease drugs have emerged from clinical trials. (The last one approved in the U.S. in 2003, Namenda, had been on the market in Europe for years, so is not a new drug.) The list of disappointments includes some high-profile anti-amyloid therapies, such as the γ-secretase inhibitor semagacestat (see ARF related news story) and the γ-secretase modulator tarenflurbil (see ARF related news story). The prevailing view in the field is that anti-amyloid therapy needs to start earlier in the disease course (see ARF related news story). However, every other once-hopeful approach has come up empty in late-stage trials as well. Remember Dimebon, estrogen, non-steroidal anti-inflammatory drugs (NSAIDs), docosahexaenoic acid (DHA), valproate, rosiglitazone, simvastatin? The list goes on. The field at large, and indeed the general public, may well think that nothing works. But there are some hidden lessons in this litany of failure. Here is a summary of what clinicians say they learned.
Several opinion leaders in the field told ARF that epidemiological data produce poor drug candidates. They suggested that the field move away from trials based on epidemiology and focus more on potential disease-modifying mechanisms. Despite the trial setbacks, most scientists this reporter interviewed are encouraged by how much researchers have learned about the progression of AD in the last five years, and in particular by the ability to track the disease with biomarkers. By and large, they believe that ongoing and planned trials have better chances of success than previous efforts, although biomarkers still have to prove themselves in this regard. A common fear, however, is that the succession of null trials will discourage the pharmaceutical industry from continuing to invest in AD therapies. “No one has infinite resources to throw at a problem,” said Cliff Jack at the Mayo Clinic, Rochester, Minnesota.
Choosing Better Candidates
Many compounds backed by promising observational data subsequently flopped in randomized controlled trials. This is true for NSAIDs (see ARF related news story), vitamin E (see ARF related news story), DHA (see ARF related news story), and ginkgo biloba (see ARF related news story). One of the latest such disappointments is the cholesterol-lowering agent simvastatin. As reported in the August 9 Neurology, an 18-month trial of 400 participants with mild to moderate AD showed no benefit of simvastatin over placebo, contradicting a great deal of work from epidemiological studies, cell culture, and animal models, which had suggested the compound would have an effect, said senior author Paul Aisen at the University of California, San Diego.
In most of these cases, “The problem is not with the studies,” said John Breitner at McGill University, Montreal, Canada, a noted epidemiologist and trialist. Many of these clinical trials were well designed and sufficiently powered to detect an effect, Breitner said. The drugs simply did not work. One issue that has plagued the field is that some proposed AD drugs enter the brain poorly or do not stay long enough to engage their target. Drug pharmacology must improve to produce a successful AD treatment, those in the field agree. In general, epidemiological data may not translate well to clinical trials, because in population studies the length of exposure can be much longer, said Laurie Ryan, who directs clinical trials in the Division of Neuroscience at the National Institute on Aging in Bethesda, Maryland.
Many of these candidates were tested in animal studies, but Breitner noted that while animal models are important in AD research, they also have severe limitations. “You can have a spectacularly successful mouse experiment that means nothing for human disease, because the mouse models of AD are not human AD.” So despite the wealth of epidemiology and animal data behind these trials, the drugs so chosen still represented long shots. “No one is surprised by the [negative] results,” Breitner said, voicing the view of many. The consensus among scientists is that epidemiological candidates have fallen out of favor. The trials that are being published now were begun five years ago or longer, and represent an older approach. “Nobody is doing these kinds of trials anymore,” Breitner said. Ryan agreed: “We were doing what we could with the state of the art at that time, but the state of the art has shifted.”
If not from epidemiology, then where should compounds come from? Scientists pointed to the need for more basic research to pinpoint the best disease-modifying targets. “There are many potential targets, and none of them are validated,” said Lon Schneider at the University of Southern California, Los Angeles. The current favorite target for early disease stages is amyloid, although the jury is still out on whether this approach will work. P. Murali Doraiswamy at Duke University, Durham, North Carolina, noted that the field has a tendency to move on quickly from failed trials to the next “big thing,” without stopping to analyze what went wrong. “Given how much the field and industry are now vested in the amyloid theory, it is critical that we encourage contrarian thinking and diversify the portfolio,” Doraiswamy wrote to ARF.
At this point in time, most seem to agree that amyloid is most promising as an early-stage target, meaning that later disease stages may require different strategies. Ron Petersen at the Mayo Clinic in Rochester, Minnesota, suggested that because tau is downstream of amyloid, and tau pathology seems to manifest around the time that clinical symptoms first appear, tau might be a logical target in people who are already cognitively impaired. “I would like to see a tau therapy for the MCI stage,” he said. Tau therapies are less well developed than amyloid approaches, but are gaining in popularity (see, e.g., ARF related news story).
Other scientists noted that once clinical dementia develops, significant brain degeneration has already occurred. At that stage of the disease, “the correct target is going to have to be neuroprotective,” Jack suggested. One example of such a trial is a Phase 2 gene therapy study that delivers nerve growth factor (NGF), which supports the cholinergic neurons that degenerate in AD. Phase 1 work looked encouraging in terms of being able to improve cholinergic function, Ryan said, and the hope is that this treatment could help restore cognition lost to dementia. A positive result from this trial might energize the twin fields of growth factor therapeutics and direct delivery to the brain, and help them overcome the current perception that this sort of approach is less suitable to a mass therapy than the proverbial once-a-day little white pill.
For his part, Michael Rafii at UCSD said he would like to see more trials testing combination therapies at late stages of the disease. “By the time you are in a dementia stage with behavioral and cognitive issues, multiple circuits in the brain are malfunctioning. To restore function, you may need to target multiple neurotransmitter systems,” Rafii suggested. The Food and Drug Administration, recognizing how hard it has been to translate research into success in the clinic this past decade, now formally encourages the community to work out protocols for combination trials of multiple investigational drugs for serious diseases such as cancer and AD (see FDA guidance issued December 2010).
Will Biomarkers Help?
At McGill, Breitner is taking a different approach to finding new drug candidates. He is setting up the Centre for Studies on Prevention of Alzheimer’s Disease. It will conduct small pilot studies on pre-symptomatic human volunteers, looking for the ability of interventions to change biomarkers that scientists believe reflect underlying disease progression. A change in biomarkers does not prove that the intervention will help prevent AD, Breitner notes, but he believes the converse is even more true: If an intervention does not alter the pre-symptomatic disease process as measured by the biomarkers, then it is very unlikely to help prevent the disease. “Moving biomarkers is the name of the game,” Breitner said. “Within a year or two, we hope to have several of these pilot biomarker endpoint trials in the field.”
He is encouraged by early progress. “There are quite a number of new interventions that have been shown, at least in small experiments, to have potential clinical benefit and ‘move’ biomarkers in people who have mild symptoms,” Breitner said. This kind of preliminary data may identify the best candidates to take into large-scale, risky prevention trials, he suggested. Specifics, please? Too early, Breitner demurred.
Other investigators elsewhere are pursuing a similar strategy as well. Jeffrey Cummings at the Lou Ruvo Center for Brain Health, Las Vegas, Nevada, plans to select patients for trials based on their specific biosignatures, rather than on severity of cognitive decline. “For example, high isoprostane levels might be used to identify patients for antioxidant trials,” Cummings wrote to ARF. “This would allow us to choose patients whose disease is appropriate for a specific mechanism of drug action, and it provides a linked outcome measure.” (See ARF related news story.)
In general, the scientists ARF interviewed agreed that the knowledge of biomarkers provided by the Alzheimer’s Disease Neuroimaging Initiative (ADNI) and similar studies has given researchers powerful new tools. “We have a better understanding of the disease process,” Ryan said. “Almost all clinical trials now try to incorporate some of the ADNI-style methods,” i.e., biomarker endpoints. Researchers noted that biomarkers will allow them to do trials in pre-symptomatic populations, track disease progression, spot subtle treatment-related changes, select the most promising drug candidates to take into larger trials, and make trials more efficient by better participant selection.
Amid the optimism, there are voices of skepticism as well. AD biomarkers still need to be validated as true disease surrogates, Petersen said, meaning they not only move with the disease, but also respond to treatment. Current knowledge on biomarkers is much stronger when it comes to predicting who will become symptomatic and enrolling the right kinds of patients for a given trial than when it comes to measuring drug responses. In other words, at this point biomarkers are quite well substantiated to serve as inclusion criteria for a trial, but less so as outcome measures. The new buzzword in the field is "theranostic marker." Eventually, biomarkers need to predict the clinical outcome, Jack said, noting, “Unless there’s an observable clinical improvement on treatment versus placebo, all the biomarker evidence in the world is not going to convince the FDA to approve a drug, and rightfully so.” For their part, leading FDA scientists have urged the clinical trials community to embed biomarkers in its trials so that this exact information—whether change in biomarker does or does not predict subsequent clinical change—can be gathered as quickly as possible.
Across the board, researchers emphasized the need to keep trying despite the setbacks. “Given the global extent of the problem, I certainly think we should be doing more clinical trials,” Aisen said. Given the number of people suffering from AD—5.4 million at present in the U.S. alone—two dozen or so Phase 3 trials in the past decade arguably represent a trickle of effort compared to what’s needed. For comparison, about 12 million people in the U.S. have cancer, and more than 25 times as many Phase 3 trials for cancer interventions have taken place in the last 10 years. Despite the tremendous need, a range of leading researchers expressed worry that pharmaceutical companies may become disillusioned with AD. “I am concerned that a sufficient number of people are going to become so discouraged if [upcoming] trials are negative that they will turn their attention elsewhere,” Petersen said. The health consequences of giving up would be disastrous, the scientists agreed. “In my mind, AD is the biggest public health problem we face,” Jack said.—Madolyn Bowman Rogers.
Sano M, Bell KL, Galasko D, Galvin JE, Thomas RG, van Dyck CH, Aisen PS. A randomized, double-blind, placebo-controlled trial of simvastatin to treat Alzheimer disease. Neurology. 2011 Aug 9;77(6):556-63. Abstract