27 July 2004. Brain imaging has emerged as perhaps the most promising source to date of an antecedent marker for Alzheimer's disease. Such a sought-after, tell-tale giveaway that the underlying process is marching along inexorably in people who do not yet have thinking problems could spot people who will most likely develop overt disease in the next few years. This could prompt them to adopt the modest preventive measures available to date, and perhaps expedite their access to clinical trials, as well as preventive and mechanism-based treatments as they—hopefully—come on line. A biomarker could sharpen the diagnosis of AD, and could monitor whether experimental therapies work. Finally, it could help test the major hypotheses of what causes AD more conclusively than is possible today.
Promising brain imaging approaches have sprouted up in many centers and labs. Attempts are underway across the field to achieve a consensus on which ones will prove to be the most useful in clinical trials and daily practice. The 9th International Conference on Alzheimer's and Related Diseases, held last week in Philadelphia, featured 306 presentations on various aspects of brain imaging. Rather than survey this still-scattered field, this conference story will summarize how a leading candidate is coming along.
Developed by Bill Klunk, Chet Mathis, and colleagues at the University of Pittsburgh, Pittsburgh Compound B (PIB) is a benzothiazole derivative that binds to β-amyloid in brain and is currently used as a PET imaging agent (see ARF related news story). Currently, six academic medical centers worldwide are examining the use of PIB imaging in AD diagnosis in small groups of patients and are conducting natural history studies of incipient AD. They are preparing to conduct larger studies. The Pittsburgh scientists are also planning to use PIB to monitor amyloid pathology in long-term HIV survivors who are developing AD-like pathology in what appears to represent a natural MCI model of sorts (see upcoming news story). However, PIB is not yet available to a wider public, and likely won’t be for at least another three years, Klunk emphasized at a press briefing at the conference. The compound has to clear technical hurdles and must be further validated before it can accompany large multicenter trials. For example, PIB imaging will not accompany the ongoing phase 3 trial of the anti-amyloid agent AlzhemedTM, said Paul Aisen of Georgetown University, who works on the trial.
What’s New with PIB?
In Philadelphia, the Pittsburgh team reported advances on several fronts in PIB development. For one, Mathis investigated a red flag that a separate group had raised when they found a metabolite of PIB in rat brain. This is worrisome because a metabolite lingering in brain could yield false-positive signals. Mathis analyzed blood and brain of mouse, rat, baboon, and human, and reported that, indeed, the troublesome metabolite did occur in rat brain but was absent from mouse or human brain.
For another, Klunk reported on another puzzle with PIB in animals. A problem hobbling microPET studies of PIB in transgenic mouse models commonly used in AD research—Tg2576, PDAPP, PS1/APP—is that PIB yields weak readouts even though these mice’s brains are loaded with Aβ. How could this be? Klunk and colleagues prepared brain homogenates of these models and tested PIB binding at the nanomolar levels used in imaging studies. The researchers found that about two moles of Aβ from human brain bind one mole of PIB, but that it takes 1,000 moles of human Aβ from mouse brain to bind one mole of PIB. “This may be important for nothing but PIB, or it might tell us something about the nature of Aβ aggregation in mouse brain,” said Klunk. Unlike in sporadic AD, mouse amyloid is driven by massive overproduction of Aβ. “In our hurry to aggregate Aβ in mice in a few months, something is not working in the same ways as in the real disease,” Klunk said. The difference could lie in Aβ modifications, chaperone functions, differences in kinetics. Whatever the explanation will prove to be, clearly β-sheet forms of mouse amyloid have a lower percentage of PIB binding sites than the real McCoy.
Pushing Back Diagnosis (Or “If it's a Duck…”)
In another presentation, Klunk showed data of an ongoing human PIB imaging study at the University of Pittsburgh. The researchers inject PIB peripherally and, minutes later, image brain amyloid in people with AD, MCI, and controls. Klunk reported on 15 initial cases. The five AD and control cases had values and imaging dynamics similar to the previous cases and controls reported in the older pilot study at Uppsala University in Sweden (see ARF related news story). More surprising were the data on the five MCI cases imaged to date. When lumped together into a group, these cases at first appeared to have PIB values intermediate between control and AD. But when plotted individually, the MCI cases separated completely, and each person fell either smack into the control group or the AD group.
Data on such small numbers are clearly preliminary. Even so, they indicate that PIB can image MCI, said Klunk. Agneta Nordberg, of the Karolinska Institute in Stockholm, Sweden, presented similar results of MCI cases in the Swedish study.
Beyond that, this data separation speaks to an ongoing debate about what exactly is presymptomatic AD. The PIB result supports the view of John Morris, Washington University, St. Louis, that MCI is not a transitional stage between normal aging and AD, but that it already is AD, Klunk said. At the Challenging Views conference, held a day before the main conference, Morris and Jeffrey Cummings had debated Ron Peterson and David Bennett on the question of whether MCI is a prodromal condition from which some people progress to AD, or whether properly diagnosed MCI cases represent nothing other than early AD. Morris summed up his view by saying: “If it’s a duck, call it a duck!” Peterson holds that only a fraction of people with MCI will progress to AD, while others recover, stay stable, or come down with something else. He countered by issuing this advice to prevent an early AD diagnosis: “Don’t move to St. Louis!” While to the casual observer, the details of this ongoing debate have a slightly semantic whiff to it, the underlying issue of when to label incipient AD as indeed being AD has real consequences for clinical trial design and serious implications for the affected person (see also Morris interview).
What If? Would S(he) Have Gotten AD?
In theory, PIB imaging of MCI cases should be able to put to rest the circular argument that comes up when researchers discuss postmortem exams of people who died with florid amyloid pathology but no cognitive decline. Are these brains proof that amyloid is peripheral to the disease, a mere sideshow that has distracted research for too long? Or would these people surely have developed symptoms had they lived a few years longer?
To address this question, Klunk and collaborators are planning to image normal elderly people to find those who have amyloid but no symptoms. “In natural history studies of cognitive performance with age, the older groups perform worse but their curve is quite broad. Many elderly actually perform very well. So the question we are now addressing with PIB is: Do the poor performers have amyloid? Once we identify them, we want to follow them to see if and when they progress to clinical diagnosis.”
In her talk about the ongoing Uppsala study of PIB, Nordberg presented first data on longitudinal PIB imaging, if not yet with presymptomatic people. A woman imaged two years ago at the age of 51, when she had mild-to-moderate AD, has in the meantime progressed to moderate AD clinically, and both FDG PET and PIB were able to track this progression. Maximum levels of PIB signal in human brain are not known, Klunk said.
Nordberg also presented new information on her study’s attempts to correlate PIB binding with CSF values of candidate biochemical biomarkers. The researchers found that PIB and CSF tau values correlated, but Aβ did not. This confirms a broader perception throughout the field that, to date, tau/phospho-tau appears to be a stronger candidate biomarker than Aβ.
PIB and the Vaccine Trial
These evolving data imply that amyloid-clearing treatments should start as early as possible, ideally before tau pathology as fueled by amyloid has taken a considerable toll on the brain, Klunk said. Can PIB improve the analysis of such treatments in trials? When an amyloid-removing drug fails, it is important to distinguish whether it had no clinical effect and no amyloid effect, or whether it had no clinical effect but a large amyloid effect. “These send you into different directions. In the first case, you’d want to make a better amyloid-removing drug; in the second, you’d conclude that amyloid removal does not improve the disease,” Klunk said.
Toward this goal, Klunk reported initial data of using PIB on postmortem tissue of two cases in the aborted AN-1792 Aβ vaccine trial (see ARF related conference story). In one case, studied in Eliezer Masliah’s laboratory, amyloid had vanished from wide swaths of the frontal cortex. PIB binding in homogenates of these areas was down to control levels. The other case, by Isidro Ferrer (see Ferrer et al., 2004) was described as showing more spotty, focal amyloid removal, and PIB binding in homogenates of those areas fell just below that established for AD cases. "We are very encouraged that PIB binding will be able to detect immunotherapy-induced changes in brain amyloid," Klunk said. He added that the spotty amyloid reduction shows why scanning patients will be necessary to obtain precise results.
How early could PIB go? And exactly what forms of amyloid does it bind to in brain? “We don’t know,” Klunk said. “We have not yet found a good way to address what particular oligomeric forms PIB binds to. We know that PIB relies on β-sheet structure, so at some point an oligomer is going to be big enough to bind PIB, but we do not know at what point that is.” On the issue of which forms of amyloid species ought to be removed, Klunk cautioned against focusing narrowly on one species. A growing number of scientists believe that different Aβ species exist in an equilibrium, and that ridding the brain of one will affect the concentration of the others. For example, treatment of mice with antibodies against soluble Aβ tends to lead to a drop of insoluble Aβ, as well. “I see plaques as a landfill where Aβ has been thrown away. When you go in to clean things up, the remaining garbage pile begins to leach back out,” Klunk said. —Gabrielle Strobel.