Alzheimer’s Imaging Agents Should Be Validated With Autopsies, Panel Says

Autopsy is the gold standard for confirming a diagnosis of Alzheimer's disease and should be used as the standard of truth in studies of imaging agents hoping to gain approval for amyloid plaque detection, an FDA panel unanimously concluded.

The agency, however, expressed concern about the feasibility of conducting such a study in a short time period at the Peripheral and Central Nervous System Advisory Committee's Oct. 23 meeting. FDA convened the panel to discuss validation and clinical utility of three competing amyloid imaging agents in Alzheimer's disease patients.

Avid Pharmaceuticals, Bayer Health Care Pharmaceuticals and GE Healthcare have agents in clinical trials and requested guidance on their proposed pivotal trial designs (see sidebar: " ¹ The Tangled Web Of Imaging Agents ").

Companies are moving away from current treatment paradigms that treat Alzheimer's symptoms, like acetylcholinesterase inhibitors, and toward disease-modifying drugs that target the amyloid beta protein fragments (² (Also see "Anti-Amyloid Still All The RAGE At Alzheimer’s Disease Conference" - Pink Sheet, 4 Aug, 2008.), p. 15).

The protein fragments accumulate and form amyloid plaques, a required component for diagnosis of Alzheimer's disease. The true value of detecting the presence of amyloid plaques with imaging agents and antibodies will likely be correlation with clinical progression, so that the presence of plaque could guide diagnosis and treatment of the disease (including tracking reduction of plaque formations by potential disease-modifying therapies, eventually).

But first, before considering all the possible valuable applications, the panel had to answer the fundamental question of how to define the standard of truth to demonstrate whether the tracers can reliably identify amyloid plaque in the brain, said acting chair Larry Goldstein, Duke University Medical Center.

Three Proposals, Three Standards Of Truth

The dilemma was a divergence on the standard of truth (i.e. a comparator) for the imaging agents to use in their pivotal studies. The three companies' proposed indications were similar, but the suggestions for study designs varied.

Avid is seeking an indication for fluorine-radiolabeled AV-45, which will bind to amyloid plaque in the brain and highlight plaque deposits in a positron emission tomography scan, "for PET imaging of amyloid plaque pathology in the brain to aid in evaluation of patients with signs or symptoms of cognitive impairment."

Avid proposed comparing imaging results to post-mortem plaque histopathology (the standard of truth) in its first prospective, open-label, single-arm Phase III trial. "The patient sample size of this pivotal autopsy correlation study will be modest (due to practical considerations we anticipate 30 to 50 autopsies)," the briefing material states.

In an interview, Avid CEO Daniel Skovronsky noted that the feedback from the advisory panel was in line with Avid's development plan and proposed protocol. "Certainly we'll continue to refine and revise the protocol and work with FDA but we don't anticipate any major shifts in the paradigm," he said.

At the meeting, FDA Medical Officer Qi Feng said the proposal was generally reasonable, but questioned whether the study could be completed within the proposed one to two year timeframe. "The concern is if they can get enough subjects to do the autopsy in the reasonable time. Usually it is quite difficult," he said.

Avid said it will circumvent the usually lengthy time period necessary for autopsy studies by recruiting terminally ill patients with a life expectancy of less than six months. About one third of the subjects should have AD, Avid said.

Another source of patients, Skovronsky said, is the NIH's National Institute of Aging, which is following healthy subjects as well as those with mild cognitive impairment and Alzheimer's. "This huge investment by the NIH is a resource we can plug into by cooperating with them to image patients they're already following, and then work with them to make sure autopsy data is available."

The trial's primary analysis will focus on patients who come to autopsy within 12 months, Avid's briefing documents state, because the link between the imaging data and autopsy result may be less certain as neurodegeneration continues over time.

Both Bayer and GE acknowledge in briefing documents that the current standard of truth for confirming brain amyloid deposition in patients is through autopsy. But such a development plan would be "long and difficult," Bayer says, and "impractical," according to GE.

Bayer proposes its fluorine-labeled tracer, BAY 94-9172 (formerly AV1/ZK, licensed from Avid in 2007), be indicated "to detect amyloid-ß plaque deposition in the brain, and thereby assist the physician in the diagnosis (exclusion/detection) of Alzheimer's disease."

FDA is "particularly concerned" about Bayer's preferred standard of truth: clinical diagnosis. Typically, AD is diagnosed by a battery of tests in clinical evaluation, and then confirmed with autopsy.

"A clinical diagnosis might not be a fully reliable marker for amyloid," Feng said at the meeting, "especially among patients who may have amyloid in the absence of diagnosis of AD."

Clinical diagnostic accuracy is generally around 85 percent, FDA consultant William Rebeck, Georgetown University, told the panel. While the absence of amyloid plaques rules out Alzheimer's disease, the presence does not necessarily confirm it, as other dementias and diseases exhibit amyloid plaque formation.

The primary objective of the Phase III trial, Bayer's briefing documents note, would be "to determine the diagnostic performance of the ... PET scan results with the tracer in differentiating between patients with probable AD and healthy volunteers when compared to the clinical diagnosis (established by an independent consensus panel)." Bayer aims to recruit 450 to 600 subjects for the open-label, single-dose, non-randomized trial.

GE is seeking an indication for detection of amyloid-beta deposits in the brain for its GE-067. GE's double-dose, open-label trial will compare PET scan images among healthy subjects, probable AD patients, and patients with other types of mild cognitive impairment.

The proposed standard of truth, PIB (Pittsburgh Compound B), is a carbon-labeled tracer commonly used in PET imaging studies but not validated or FDA-approved. GE will compare PIB scans with scans using its GE-067 tracer (identical to PIB, but with a radiolabeled fluorine). Patients will receive GE-067, GE-067 and PIB, or two doses of GE-067.

"In general this approach might be reasonable," FDA's Feng said, "if sufficient data are available to verify the use of PIB as an indicator for amyloid."

Don Black, GE's head of R&D for medical diagnostics, said in an interview that "we all agree the standard of truth is histopathology. I think FDA recognized PIB could be a bridge to pathology. That's why when they described the FDA view [of Bayer's proposal] they didn't think there was enough rigor behind clinical diagnosis. So they thought PIB might be a good bridge."

"I don't think they really addressed PIB per se, they just felt that imaging wasn't enough," Black added. "We agree with them. They really like the idea of histopathology if it could be done. We don't have any issues with that. We feel that PIB does provide good correlations and we would continue to look for more histopathology. But in general, we will have to wait until we talk to FDA to find out if we have to make any changes to our plans, but there's nothing we heard that indicates we have to change our plans."

Panel Supports Autopsy As Standard Of Truth

The advisory committee focused mostly on whether autopsy was an appropriate standard for approval.

Michael Weiner, an industry consultant and principle investigator for the Alzheimer's Disease Neuroimaging Initiative, said during the public hearing that "a good standard of truth is autopsy." But, he added, "I personally don't think [autopsies] should be required for FDA approval." Instead, autopsy studies "should be done in the post-approval stage" because "they will unnecessarily prolong approval."

The idea of requiring autopsy in future clinical studies was furthered by voting member Britt Anderson, University of Waterloo (Ontario), who suggested autopsy confirmation should be the standard of truth "for a disease indication or diagnostic indication given the ambiguity of clinical diagnosis of AD, but not for a biochemical indication."

Other members favored autopsy as standard of truth over the experimental PIB agent, suggesting PIB correlation to disease diagnosis needed more validation. (See ³ (Also see "A More Detailed Image of Alzheimer's Disease" - Scrip, 1 Feb, 2008.) .)

"Would PIB be approved by the FDA? I think that's a pivotal question," said temporary voting member Harvey Zeissman, director of nuclear medicine imaging at Johns Hopkins. "What these companies are being forced to do is think in a larger scale and say, 'What would the FDA be willing to approve?', and therefore they thought through these protocols."

"I think these protocols have a lot to say for themselves. I think all of them ought to be done, with modifications," Zeissman added. "The problem is, with the FDA approval process, what the companies tend to do is to do the least they have to do in order to get FDA approval and then nothing else gets done afterward."

"I'm all in favor of rapid approval but I'm also in favor of good science. And I think that's one thing the FDA process does force to a certain extent is good science. I think PIB has some good science but the presentation I saw today didn't convince me - my perspective on the FDA on whether they would approve something like that is not very optimistic. To me, autopsy studies are mandatory. As difficult as they are to do, we need to have that pathological confirmation."

Voting member Peter Herscovitch, NIH Clinical Center, expressed concern that neither Bayer's nor GE's proposals would adequately confirm correlation with AD. "I'm very concerned about only a clinical diagnosis being used." Since the accuracy of a clinical diagnosis varies, "there will always be uncertainty."

Herscovitch suggested the inclusion criteria should be broadened to include other dementias if clinical diagnosis were used to validate the imaging agents, especially if the agents would be used to rule out AD.

"Also, I'm not too sure what studying normal patients would show if there won't ever be a confirmation of presence of amyloid. It's been said that 20 percent to 30 percent of subjects in their 70s and 80s have amyloid, and I think that will be a false specificity figure because there really is no answer."

The proposal he had most difficulty with was GE's, because it would use both clinical diagnosis and PIB.

"Initially I thought that seemed quite reasonable, but then I began to wonder if there's a logical conundrum in using PIB in terms of what we are faced in terms of regulation - given the data on the table today about PIB, would it be registered, would it be approved, to exclude the presence of amyloid? We assume it's safe, we assume manufacturing will be fine, but just on basis of studies to date, are they sufficient?"

GE's Black defended their proposal, saying the company was trying to cover every possibility. "So if the agency at the end of the day was more interested in the clinical diagnosis, we would include that, rather than excluding that. We would certainly look for autopsy cases; we just think it's going to take a long time to find positive cases."

"The idea of doing negative cases didn't occur to us, which is the Avid proposal, because you'd be approving a drug by not seeing something. You could inject saline and not see something too, but what does that really tell you about the drug? We'd have the positive aspect then also by relating back to PIB. We're willing to do all of these things and try to make correlations. Why wouldn't this be useful?"

Although all members said histopathology was the preferred standard of truth, two of the 16 panelists said they were willing to accept PIB as a potential alternative. Three said they did not support PIB, especially as the agent is not FDA-approved.

The Era Of Comparison

The sense of urgency in completing studies to maximize the imaging agents' potential was clear, but whether autopsy studies could be done in an expeditious manner was up for debate (See ⁴ (Also see "The Alzheimer's Divide" - In Vivo, 1 Sep, 2008.) .)

FDA "would like to see something in a year," GE's Black said. Though not simple, Avid does think that is feasible. Black pointed out that "if we find in a year there aren't any autopsy studies, maybe FDA would change their mind."

At the meeting, FDA's Office of Drug Evaluation I Director Bob Temple instead suggested the three companies collaborate on a comparative study of their agents in the same population. "This is the era of comparison," he added.

Zeissman supported Temple's suggestion, saying the companies should get together and support a comparative trial. "All these different approaches could be looked at as a whole and we would be much further ahead in our knowledge of the value of these agents," he added. "I know that's difficult in the business world, but I think in the end it would help them all." Avid would support collaborative efforts, Skovronsky said.

Black offered a potential solution: "One way of working to together is if everyone relates back to PIB." Access to PIB data is readily available through Weiner's AD Neuroimaging Initiative, he noted. The ADNI is a large five year research project studying biomarkers and disease progression. Clinical and imaging data are publicly available on a Web site, allowing external investigators to pool data (⁵ (Also see "Alzheimer’s Biomarker Data Ready For Release; Analyses Pose Next Challenge" - Pink Sheet, 7 Jan, 2008.), p. 30).

- Becky Jungbauer ([email protected])