Compounding Challenges: Conducting Clinical Trials in Alzheimer Disease

This piece is part of a NeurologyLive series on aducanumab's journey to the FDA and the challenges of conducting clinical trials in Alzheimer disease. For part 1, click here. For part 3, click here.

Lon Schneider, MD, MS

Clinical trials have come a long way since the first systematic trial conducted in the mid-1740s by Scottish physician James Lind, MD.¹ Modern trials now consist of more rigorous design than their predecessors, in part spearheaded by the 1920s Principals of Experimental Design, which included the ideas of randomization, result replication, group blocking, and factorial experimentation.²

These tenets of the trial have helped prove the efficacy of innumerable therapies across the vast spectrum of disease since, including some which were previously thought to be incurable or untreatable. They have push science forward, and as they have been refined over time, have aided researchers in their ability to see the fruits of their labors blossom. Although, despite these arduous processes, the success of trials is still dependent on the efficacy of the therapies being assessed, the recruitment of the correct patients, and the understanding of the disease in question, among other facets—namely, the challenge of showing clinically meaningful change, which presents itself as somewhat subjective in a practice that requires objectivity.

But ultimately, when it comes to determining which of these—determining clinically meaningful change or simply conducting the trials—is the biggest challenge for physicians in the development of therapies for Alzheimer, the physician-scientist is faced with a chicken-or-the-egg question. As Lon Schneider, MD, MS, told NeurologyLive, simply put: “They're both the same.”

“The concept of clinical importance is, in this kind of discussion, a regulatory concept,” Schneider, who is professor of psychiatry and the behavioral sciences, and the Della Martin Chair in Psychiatry and Neuroscience at University of Southern California Keck School of Medicine, said. “And in terms of properly designed trials, we know what properly designed, unbiased trials should be, and what they should look like. But we don't do that.”

“FDA has a phrase: ‘Adequate and well-controlled trials.’ That comes from at least the 1997 Drug and Cosmetic Act, but probably goes back a little further. It's actually in the act, and then it was brought into regulation as well. It's a term of art. It's not a statistical concept,” he added.

The Tools

For Alzheimer disease, these trials have proven to be as equally complex to conduct as the disease itself is to understand. Although progress made in the scientific understanding of pathology, much remains to be learned. Another major roadblock in the conducting of trials in this space has been driven by the limitations of the tools at hand and the difficulty in constructing them, as well as the variation in the disease on both an interpersonal and individual level.

Steven Arnold, MD

“Fundamentally, the designs of trials historically have been the standard parallel-group, randomized, controlled trials that really are looking for an average effect in an average patient. But I'm still waiting to meet an average patient,” Steven Arnold, MD, told NeurologyLive. “The heterogeneity is a real problem. Many of our outcome measures are just too insensitive and too infrequently given.”

Arnold, a professor of neurology at Harvard Medical School, and managing director and translational neurology head of the Interdisciplinary Brain Center at Massachusetts General Hospital, is also an investigator in the phase 2 PEGASUS trial (NCT03533257) of AMX0035, an investigational agent being assessed in Alzheimer disease. He noted that more frequent incorporation of digital technologies might help bolster the regularity of monitoring of patients in trials.

“People have good days and bad days,” he explained. “You do a test at baseline and the person does well, and then you do a test at the end of the trial. Say they slept poorly the night before. They blow the test. Even though they may have been improved all that time, you missed the ability to detect a benefit.”

To overcome that amount of noise, the trials that need to be conducted require large numbers of patients, which are enormously expensive and utilize very blunt measurements. Schneider echoed this sentiment, noting that one must first consider that the measures being used in trials are continuous in nature, and are measuring, mostly, group means. But with a disease as heterogeneous as Alzheimer, it can make it difficult to determine if there is benefit or worsening on an individual level.

“When you're discussing clean outcomes, like death, or survival versus not survival, the concept of clinically important is very much different. Count that 10 people versus 9 people survive—that is clinically meaningful for the 1 person who didn't,” Schneider said. In fact, he noted, a large number of cardiovascular drugs have this level of effect. Similarly, when looking at studies of cancer therapies in which a new combination, or a new therapeutic, is being added on top of the standard, any discrete and clean effect is considered clinically meaningful. The mitigating factors, in that case, are adverse events and the level of performance for the study.

This challenge has existed for quite some time in Alzheimer. In 1993, the first therapy approved for Alzheimer disease was tacrine (Cognex; Shinogi)—which has since been discontinued—based on phase 2 and phase 3 study. Prior to its approval, the conversations at its 3 different FDA Advisory Committee meetings were dominated by what specific point-difference in means actually determined meaningful change. The scale used in those trials was the Alzheimer's Disease Assessment Scale-Cognitive Subscale (ADAS-Cog), a brief neuropsychological assessment used to assess the severity of cognitive symptoms of dementia that’s still in use today.

Even though the community finally came to an agreement of sorts on that matter, Schneider said, “the FDA has never said that there needs to be a minimal difference on a continuous measure. What came out of that was the need for a global measure—the clinical global impression of change—or what's been used more recently in longer-term studies, the CDR, the Clinical Dementia Rating Scale-Sum of Boxes.”

Marwan Sabbagh, MD

"Historically ADAS was it," Marwan Sabbagh, MD, director, Cleveland Clinic Lou Ruvo Center for Brain Health, told NeurologyLive. "Now we're seeing CDR—I think that's a good movement. It's funny because BAN2401, in its phase 2 studies, used ADCOMS [Alzheimer Disease Composite Score], which is a kind of a hybrid scale of CDR, and ADAS, and other things. These are easier instruments to derive clinical meaningfulness."

The CDR, which was utilized in the EMERGE (NCT02484547) and ENGAGE (NCT02477800) clinical trials of Biogen’s investigational (and somewhat controversial) aducanumab, has, in fact, offered a more ordinal measure. Patients change by strata on the clinical global impression of change, with the least amount of change possible being 1 point. Patients can go from “no change” to “minimal change,” with minimal being defined as detectable by clinician—ergo, clinically meaningful change.

This kind of scale, Schneider said, offers the ability to separate patients by who worsened and who improved, and then leads to 2 possibilities of comparison: Chi-square or categorical statistical analysis, or analysis by means. This, in part, has driven some of the discussion around the analysis for aducanumab. When using means, the meaning of the analysis becomes less clear, Schneider explained. That, in turn, presents a possible problem.

“You don't know whether there have been X number more people who were responders in the other group,” he said. “The FDA has been easy going and liberal on that, and saying, ‘You have the scales that measure the absolute change in individuals, and you take the average of them.’ That means that more people, at whatever level you choose, are benefiting from the drugs and placebo. So, this becomes a lot of the quest for clinical meaning.”

Despite the agency previously refusing to approve a therapy based on a cognitive test measurement due to the inconclusive nature of their meaning and generally small effect sizes, in the mid-2010s, that rule was relaxed slightly. Now, cognitive testing can be provided as a provisional sign of efficacy in those who have disease that is considered mild with few symptoms. This is in part due to the FDA’s recognition of the practical difficulty in conducting long-term studies to record individual benefit, Schneider said. The expectation is that these comparison analyses of mean values allow for the marketing of therapies for preclinical Alzheimer, with the necessary analyses being done in post-marketing studies to count individual benefits.

Most of this, Schneider explained, has not yet been widely realized, as only a handful of studies have been conducted in individuals with preclinical disease and a biomarker of increased risk, but no cognitive impairment. Under that circumstance, a neuropsychological composite, which he described as a “work in progress” has been developed for use, called the Preclinical Alzheimer Cognitive Composite (PACC). It is currently being used in the A4 study of solanezumab (NCT02008357).

“For this kind of an early-stage Alzheimer disease study, you can get away with 1 outcome that is a global composite, that brings together both a memory or cognition test and a functional outcome,” Schneider said. “And the poster child for that is the CDR because the scale has memory orientation, judgment, and problem-solving. There's the cognitive stuff. And then it has social activity, basic, basic activities, and community or in-house activities. You can average all those, you can add all those together into the sum of the box score.”

The FDA has noted that a composite of that nature, consisting of activities and cognitive function, such as memory and orientation, can be accepted as a single outcome without the need for a global or an individual patient outcome because the CDR can be considered an individual patient outcome. “That's what you now see with aducanumab. You could also have seen it with other drugs in the past, if we had chosen to make a big deal about it,” Schneider said.

Ultimately, the tools may not be the problem. For Schneider, the instruments used in Alzheimer clinical trials—which have been the same for about 3 decades—are fairly sensitive measures. "If you can't tell the difference, or if you need a very large number of patients to tell the difference, then the drug is not that effective, and probably not effective enough to market," he said. Perhaps, he admitted, the tools are not advanced, but the genuine feeling in the field seems to be that they are good enough.

"We complain about them," Schneider explained. "But we use them. So if you can't find a difference with these instruments, you can find efficacy with these instruments, probably there isn't much there to deal with."

The Timing

Another leading challenge in conducting Alzheimer trials is the timing related to the disease’s progression. In the discussions surrounding the proper targets for therapeutics and the different pathologies of the disease, many have posited over the years that earlier treatment would offer more benefit to patients. This is also reflected partially by the currently approved therapies, all of which are indicated for mild, mild to moderate, or moderate to severe stages of disease.

Richard S. Isaacson, MD

“What happens with clinical trials is that there's always a delay in terms of people may have some early mild symptoms, but there really isn't a push to get people enrolled in clinical trials until those symptoms are causing more and more of a problem,” Richard S. Isaacson, MD, told NeurologyLive. “As we know, Alzheimer disease starts in the brain decades before the first symptom of memory loss. By the time someone already has dementia, that means the pathophysiology has been going on for a very long time. It's a lot harder to make any sort of tangible process in terms of a therapeutic or in terms of clinically meaningful change. I think it just makes sense, based on how Alzheimer progresses, to get people enrolled in clinical trials as early as possible.”

Isaacson, who is director of the Alzheimer’s Prevention Clinic at Weill Cornell Medicine, noted that the need to look into patients earlier in the disease is paramount. No small task, as this requires a clinical picture made clear by the presence of a biomarker. But for Isaacson, it offers a much more robust time point to show potential clinical benefit.

“What that means is when someone is starting to have cognitive changes, but they're still able to take care of themselves—and this is proven to be due to or Alzheimer disease, meaning there's an amyloid biomarker, or tau, whether it's through a brain imaging study or a spinal fluid test, maybe one day in the future—pretty soon actually—a blood test,” he said. The other side of that coin, he admitted though, is that perhaps even at that point, it might be too late. “I believe possibly, but hopefully not.”

He noted that in this sense, treating mild cognitive impairment could operate essentially as tertiary prevention of dementia due to Alzheimer. Current estimates suggest that 47 million people in the United States have preclinical Alzheimer's disease, which would offer quite a large group of potential individuals for trials.³ “That is a striking number. Maybe that's when we need to try to study disease-modifying therapies, when the pathology has started but the symptoms haven't begun yet,” Isaacson said.

Although, attempting to assess therapies even earlier can produce additional issues, even if the measures of the disease are the most sensitive and reliable that are available. On top of that, developing these scales can become even more difficult when done earlier in the disease process, according to Stephen Salloway, MD, MS, director, Neurology and the Memory and Aging Program, Butler Hospital.

Stephen Salloway, MD, MS

“Let's say someone is showing pathological changes of Alzheimer with amyloid or tau buildup and no symptoms. In that case, detecting change on a cognitive scale is very challenging,” Salloway told NeurologyLive. “The more impaired the patient, the easier it is to detect the change—and patients change more during the trials. A very important factor is the rate of change. If you get a lot of people in a trial who are stable, then it's very hard to show a drug’s benefit. You need at least a substantial group of people that are progressing in a detectable way during the trial.”

Salloway noted that a number of advances have been made that might help address this difficulty, particularly the ability to visualized amyloid plaques safely, and earlier in the disease process. He explained that the progress made with brain scans, blood tests, and plasma tests have been surprisingly quick—more so even than what much of the field anticipated. The plasma tests, though, might offer the biggest advantage.

“That's going to be a real game-changer. Once we have a plasma test that's reliable for detecting the disease state and also helping to measure the progression of change, that's going to be really helpful,” he said. “Now we can define a population that is amyloid positive and then you can target whatever stage of the disease, either preclinical, early Alzheimer, or later Alzheimer, and verify that they have the pathology.”

As Salloway pointed out, much of the challenge of timing goes hand-in-hand with the challenge of identification of the disease. As many others have said, the earlier that the disease can be detected, the earlier we can intervene. But this only brings up more questions. When exactly is the optimal time frame to intervene? What is the correct dose? Is there a single correct mechanism of action? Can the intervention be used alone, or does it require accompanying lifestyle approaches? And unfortunately, the questions don’t simply end there.

“The other thing is, we have the amyloid hypothesis, and tau, and glucose, and hypometabolism, inflammation, and infection. Then, insulin resistance and vascular disease—the list just goes on and on and on,” Isaacson said. “The number of things that can promote Alzheimer progression and fast-forward decline is high. We have to be cautious about how much we invest our efforts into 1 specific mechanism.”

Of course, progress has been made in certain areas that warrant continued research, Isaacson explained. But many areas of research in Alzheimer—in fact, almost every area—have experienced failures. But fortunately, all of it has led to an increased understanding that simplicity is not the answer. A complex disease will require complex trials and complex therapeutic approaches.

“We need to kind of have an all-hands-on-deck approach, and really keep hitting this just like any chronic condition that's associated with aging,” Isaacson said. “High blood pressure, diabetes—you don't just treat these with 1 magic drug or pill or infusion. It's multiple, with blood pressure control, ACE inhibitors, calcium channel blockers, and other drugs. And then we need to do some dietary changes, like a low salt diet, and exercise and losing weight. There is just no perfect answer for these complex diseases, especially when it involves the brain. Alzheimer is as much a medical condition as it is a neurological condition, in my opinion, anyway, and we need to treat it accordingly.”

REFERENCES
1. Carlisle, Rodney (2004). Scientific American Inventions and Discoveries, John Wiley & Songs, Inc., New Jersey. p. 393. ISBN 0-471-24410-4.
2. Creswell, J.W. (2008). Educational research: Planning, conducting, and evaluating quantitative and qualitative research (3rd). Upper Saddle River, NJ: Prentice Hall. 2008, p. 300. ISBN 0-13-613550-1
3. New Study Estimates 47 Million Americans With Preclinical Alzheimer’s Disease. News release. UCI Mind. December 8, 2017. Accessed December 15, 2020. https://www.mind.uci.edu/new-study-estimates-47-million-americans-with-preclinical-alzheimers-disease