The 5th HAI conference, held in Miami Beach, Florida, concluded on 15 January with a panel discussion of how far amyloid PET has come in the seven years since its first major publication (Klunk et al., 2004) when viewed from the perspective of therapy development. After two days of talks on an expanding breadth of basic science studies with these tracers, Eric Siemers of Eli Lilly and Company summed up the theme by issuing this call to his colleagues: “Don’t get too caught up in the 'gee whiz' of amyloid imaging—and it is fascinating science!—when the goal is to use this to improve clinical trials and make a difference in the treatment of the disease.” To that end, what do scientists know? Which trial purposes are amyloid imaging well suited to accomplish, what can’t it do? Siemers; Howard Feldman of Bristol-Myers Squibb; Michael Grundman, formerly of Elan and now at Global R&D Partners, San Diego; and Paul Aisen of the Alzheimer's Study Group, also there, weighed these questions before the crowd dispersed for the airport, the beaches, or Miami’s hip Art Deco district, as the case may be. Here are the main points.

In the past few years, scientists and clinicians in AD have coalesced around a consensus that amyloid deposition precedes the traditional diagnosis of AD by around 15 years. As cognitive measures become more sensitive and diagnostic criteria evolve, this number shrinks somewhat, but it still leaves a long time for neurodegeneration to eat away at the brain and make a case for intervening early.

Feldman first said that clinicians had put much effort into reframing the concept for the years before frank dementia, and then he placed amyloid imaging into this concept. In a nutshell, trialists are pushing leftward on the disease’s timeline. They have defined the stage immediately prior as prodromal AD (this would capture the large fraction of people with MCI who progress to AD) and the stage before that as asymptomatic at-risk (those are cognitively normal people with biomarker changes; see Dubois et al., 2010; Dubois et al., 2007). A separate set of new diagnostic guidelines is substantially similar, most scientists agree, with differences largely in semantics (see ARF ICAD news story).

Trials in the prodromal and the asymptomatic at-risk groups are new ground, Feldman said. For them to happen, biomarkers need to accomplish several things. Biomarkers need to identify people with brain amyloid. AIBL, ADNI, and other studies showed that about a third or more MCI patients have no or low brain amyloid. These people should not be in the same trial as, or at least not be analyzed together with, people with abundant amyloid. It seems clear that both amyloid PET and CSF assays can do that. Avid’s Phase 3 histopathology study shows that florbetapir detects aggregated forms of brain amyloid, and the ADNI CSF team showed that it is possible to set a cutoff point for Aβ and tau that distinguishes healthy controls from AD in postmortem series, Feldman said.

Further, biomarkers need to ascertain that trial participants will move along the continuum to AD dementia. For progression from the prodromal stage to AD dementia, converging evidence from multiple studies portends that CSF predicts progression, and a florbetapir Phase 2 trial also shows faster progression in amyloid-positive participants. Moreover, CSF biomarkers and amyloid PET correspond well to each other (Fagan et al., 2006; Jagust, 2009). “Again, both amyloid markers can do this,” Feldman said. Data on the progression from the asymptomatic at-risk to the prodromal stage is a bit further behind but heading in the same direction.

With this relentless focus on early-stage trials, what about the current patients? Will they be forgotten after one trial after another in them has failed for the past decade? Not at all, the panel agreed, and biomarkers will hopefully help give trials in this phase a better shot at success, as well. It is generally accepted that about every fifth person diagnosed as having AD turns out after death not to have had amyloid pathology. If AD is defined by its signature pathology, Feldman said, then these 20 percent of cases should not be called Alzheimer’s, and should not enroll in those trials. For his part, Siemers added that in the course of semagacestat testing, some 15 percent of 600 trial participants who had florbetapir scans turned out not to have amyloid. “That will dilute your trial,” Siemers said.

The panel agreed easily, then, that the time has come for using amyloid markers to select patients for trials in all stages of AD. The question of whether these markers make good outcome measures is more difficult. In theory, amyloid imaging could serve as an endpoint, said Grundman. It could do any of this: show that the drug reaches and affects the target, tell how quickly the drug does this, where in the brain this happens, help with dose selection, help decide whether to continue or abandon development of a given treatment, and tie the target engagement to a subsequent clinical outcome, i.e., the result that ultimately matters to the patient. How did amyloid imaging fare on these challenges when it was put to the test in a small PIB-PET Phase 2 Bapineuzumab trial (Rinne et al., 2010)? The treatment did engage the target (amyloid went down a bit in people on antibody, up a bit in people on placebo), this happened slowly (over 78 weeks), and amyloid went down in all regions evaluated. On the remaining points, this one study alone provided no answers.

Despite these early hints, scientists don’t have their ducks in a row on outcomes nearly as well as on enrichment. For example, in Rinne et al., PIB uptake went down, but CSF Aβ42 did not. Similarly, in AN1792’s Phase 2, amyloid pathology went down as per postmortem pathology, but CSF Aβ42 did not. In AN1792, CSF total tau went down; in the Bapineuzumab Phase 2, it did not, though phospho-tau showed a trend. With semagacestat, there was no effect on CSF Aβ or tau in the trial, but a separate study done with the Silk method of continuous CSF monitoring did show that effect for therapeutic doses used in the trial (Bateman et al., 2009). In a Phase 2 trial, PBT-2 lowered CSF Aβ42 but not tau.

Clearly, the field has not validated an amyloid endpoint as an outcome measure. “If you take a landscape view, it is that we struggle to understand the nuances of what these various results are telling us. We will have to reconcile these differences between CSF and PET. We are certainly not out of the woods,” Feldman said. On the million dollar question of how amyloid correlates with clinical benefit, Aisen urged the field to keep an open mind. “If, indeed, Aβ oligomers are the most toxic species, then in theory, a drug that increases PIB might be beneficial. The answer is likely a long way off,” Aisen said.

All agreed that it may take many trials to sort these questions out. Where to start? Aisen, who, as head of The Alzheimer's Disease Cooperative Study (ADCS), can speak more freely than his industry colleagues about trial designs he has up his sleeve, suggested a design that presented itself out of a recent analysis of ADNI data. It turns out that the normal controls fell into an amyloid-negative and an amyloid-positive group. What’s more, as observation went on, these groups clearly diverged on brain atrophy and ventricular enlargement, and showed a difference on cognition even on a measure as crude as the MMSE. Together, those differences among the "normals" are big enough to support a trial, and power estimates based on them call for group sizes of 300. “That is entirely doable,” Aisen said. Consequently, the ADCS is proposing a trial to the National Institute on Aging. It aims to screen cognitively normal people in their seventies, either with a lumbar puncture or amyloid PET, and would treat the amyloid-positives among them with an anti-amyloid treatment for two years. The trial would measure a fleet of outcome measures including MRI, FDG-PET, sensitive cognitive tests using free or cued selective reminding, and amyloid PET.

This would be secondary prevention. The planning happening at this point in time would constitute groundwork for more such trials. The price tag for amyloid PET will influence how much of this groundwork can be realized, scientists agreed. Screening with this technology may seem prohibitive to a society strapped by deficits and budget cuts in towns, states, and at the federal level; however, Feldman pointed out that primary prevention trials are much larger and more expensive. “In comparison, this seems manageable,” he said.

In the day’s discussions, these points came up:

Question: How will you select what you test in these secondary prevention trials? Nothing has worked yet!

Aisen: There are two basic ideas about that: the safety and the efficacy rationale. Some suggest we should pick an extremely safe drug because it is an asymptomatic population. I am in the other camp, which emphasizes the scientific rationale behind the drug and clear data that the drug hits its target. If we launch this with NIA funding, we will ask an external panel to help us select. It will be an imperfect process.

Reisa Sperling, BWH, Boston: In selecting a drug, we do need evidence of target engagement, but we should not require clinical signals, i.e., an effect in mild to moderate AD. In cancer and heart disease, some drugs help early but not late.

Question: will the FDA accept this?

Aisen: The FDA and EMA support trials with biomarkers anchored by some subtle cognitive marker for familial AD. The bar may be higher for the population I envision, but I see regulatory acceptance for this even for LOAD on the horizon.

Question: What about cognitive reserve?

Adam Fleisher, Banner Alzheimer Institute, Phoenix, Arizona: The consequence of amyloid varies strongly by cognitive reserve. We do not understand the relationship between the amount of amyloid and whether it predisposes to clinical symptoms very well, so we can’t define the boundary between normal aging and AD in this way. But we know in general that if you have elevated amyloid, you are at higher risk of developing AD.

Sperling: Within the group of amyloid-positive normals, there might be a mix of people who are on their way to AD and other people who are good at withstanding the effects of amyloid. If that is so, then we need a way to split these to make sure we are not enrolling the people with greatest reserve into the trials.

The HAI conference then ended with a dual message: Stephen Salloway from Butler Hospital in Providence, Rhode Island, coupled kudos to Bill Klunk and Chet Mathis for opening up this new field with a plea to all investigators to also keep track of the patients who don’t fit the concept. Those would be people who have amyloid but primarily suffer from a different disease, such as dementia with Lewy bodies, or people who have amyloid but also vascular disease, and people who have dementia but no amyloid. It may also include the oldest old, in whom dementia is caused by amyloid and a number of comorbid conditions, challenging clinicians to determine the relative contribution of each of these factors to the patient’s cognitive decline.—Gabrielle Strobel.

This is Part 6 of a seven-part series. See also Part 1, Part 2, Part 3, Part 4, Part 5, and Part 7. View a PDF of the entire series.

Comments

  1. Re: "It is generally accepted that about every fifth person diagnosed as having AD turns out after death not to have had amyloid pathology. If AD is defined by its signature pathology, Feldman said, then these 20 percent of cases should not be called Alzheimer’s."

    As someone with a strong interest but not a professional in this field, I have to ask, How can what is merely a hypothetical signature pathology be used to exclude cases that do not exhibit that pathology?

    All that would prove is A = A!

  2. If we accept that the pathological accumulation of amyloid-β in the brain precedes dementia for over a decade, prevention trials in Alzheimer’s disease with disease-modifying drugs, including anti-amyloid therapy, should be conducted in subjects in their early sixties and not in their early seventies, when the disease is probably too advanced, unless aggressive brain repair/regenerative treatments are administered, also.

Make a Comment

To make a comment you must login or register.

References

News Citations

  1. Noisy Response Greets Revised Diagnostic Criteria for AD
  2. Miami: Women Rock at Human Amyloid Imaging Meeting
  3. Miami: Amyloid in the Aging Brain—What Does It Mean?
  4. Miami: Multimodal Imaging, New Way to Test Amyloid Hypothesis
  5. Miami: Astrocytes, Antidepressants, Microbleeds, and More
  6. Miami: Updates on J-ADNI, 18F Tracers, Biopsies
  7. Miami: HAI Amyloid Imaging Conference Abstracts

Paper Citations

  1. . Imaging brain amyloid in Alzheimer's disease with Pittsburgh Compound-B. Ann Neurol. 2004 Mar;55(3):306-19. PubMed.
  2. . Revising the definition of Alzheimer's disease: a new lexicon. Lancet Neurol. 2010 Nov;9(11):1118-27. PubMed.
  3. . Research criteria for the diagnosis of Alzheimer's disease: revising the NINCDS-ADRDA criteria. Lancet Neurol. 2007 Aug;6(8):734-46. PubMed.
  4. . Inverse relation between in vivo amyloid imaging load and cerebrospinal fluid Abeta42 in humans. Ann Neurol. 2006 Mar;59(3):512-9. PubMed.
  5. . Relationships between biomarkers in aging and dementia. Neurology. 2009 Oct 13;73(15):1193-9. PubMed.
  6. . 11C-PiB PET assessment of change in fibrillar amyloid-beta load in patients with Alzheimer's disease treated with bapineuzumab: a phase 2, double-blind, placebo-controlled, ascending-dose study. Lancet Neurol. 2010 Apr;9(4):363-72. Epub 2010 Feb 26 PubMed.
  7. . A gamma-secretase inhibitor decreases amyloid-beta production in the central nervous system. Ann Neurol. 2009 Jul;66(1):48-54. PubMed.

Other Citations

  1. View a PDF of the entire series.

Further Reading

Papers

  1. . Anti-aβ therapeutics in Alzheimer's disease: the need for a paradigm shift. Neuron. 2011 Jan 27;69(2):203-13. PubMed.