This is Part 2 of a two-part meeting report from the 8th International Conference AD/PD, held 14-18 March in Salzburg Austria. See Part 1.
19 April 2007. Differential diagnosis
clinicians face a daily challenge in properly distinguishing AD from related neurodegenerative diseases that can present a similar set of symptoms. Could amyloid imaging help? Presentations in Salzburg converged on the take-home message that PIB separates AD from Parkinson’s with later dementia and from frontotemporal dementia (FTD), but does not separate AD from dementia with Lewy bodies (DLB), and maybe not from cerebral amyloid angiopathy (CAA), either.
David Brooks’s London team reported two small series of 13 cases each. One showed that most people who develop Parkinson disease first and dementia later (i.e., PDD) did not show labeling with PIB even though their glucose utilization dropped. This suggests that they have no Aβ amyloid and that their dementia may be a consequence of Lewy body pathology. The other series scanned people with DLB, who develop dementia alongside their PD symptoms. Most of them did show significant PIB labeling, if not as much as did people with AD, suggesting that they have two different pathologies in their brains. The anterior cingulate was one of the first brain areas to show amyloid in people with DLB. Christopher Rowe’s Melbourne group did not study PDD but reported, like Brooks, that patients with DLB have amyloid deposition similar to AD patients. Rowe showed that all his cases with FTD looked like normal controls in PIB PET scans. Similarly, Timo Grimmer in Alexander Drzezga’s group at Technical University, Munich, Germany, described a study of semantic dementia (SD). This form of temporal lobe degeneration tends to strike at earlier ages than AD and is marked by deficits in word memory and comprehension. The clinical symptoms and the pattern of glucose metabolism with FDG PET of semantic dementia somewhat resemble those of AD. But at postmortem, the brains of people with semantic dementia show a characteristic pattern of neuronal loss without amyloid pathology, Grimmer noted. A comparison of eight SD and eight AD cases showed that the trained eye can distinguish them on FDG PET scans because the SD cases have reduced metabolism more in frontal and temporal cortex while AD cases have it prominently in temporal and parietal areas. But PIB PET distinguished much more clearly in that the SD cases showed no binding at all. (They did show a small amount of white matter binding, as do most PIB studies. That, however, reflects merely the tracer’s slower kinetics through white matter, not a specific binding to amyloid deposits, as presented on a poster by Michelle Teena Fodero-Tavoletti and other Australian researchers.) The German group suggested that semantic dementia arises from a pathology other than amyloid, and that amyloid imaging could help tell the two apart in cases where the diagnosis is in question.
Not Just Plaques: PIB Biology
The rush to explore PIB binding in people has left fundamental biologic research in its dust, said Andrew Lockhart of GlaxoSmithKline in Cambridge, U.K. The excitement about potential antecedent markers and drug effect monitoring has obscured the fact that the underlying biology and binding properties of PIB in tissue remain hazy (see Lockhart, 2006). PIB’s developers, Chet Mathis and William Klunk of the University of Pittsburgh, have conducted biochemical binding studies but have emphasized repeatedly that they do not know exactly which forms of Aβ amyloid PIB binds in people’s brains. This amyloid comes in many guises. Oligomers, diffuse deposits, neuritic plaque, cored plaque, parenchymal versus vascular amyloid—what exactly is behind those colored pictures?
Lockhart and colleagues set out to study the specificity of PIB binding to the major known types of amyloid lesion, and to other forms of non-amyloid deposits. To that end, he prepared a tritium-tagged version of PIB and treated fresh cryosections of human postmortem brain with it for quantitative autoradiography. The scientists used immunocytochemistry on adjacent sections for comparison. The study included sections from various forms of pathologically confirmed disease, including plaque-only AD, plaque-and-tangle AD, CAA, mixed parenchymal and vascular pathology, tangle-only or Lewy body-only cases of dementia. Lockhart found that PIB labeled all forms of amyloid plaque including diffuse plaques throughout all brain regions examined. PIB delineated the same structures that lit up with thioflavin S and the 6E10 antibody.
PIB did not bind to α-synuclein deposits, that is, Lewy bodies. This is consistent with Rowe’s and Brooks’s human imaging studies, in which PIB appeared to label only Aβ amyloid pathology in DLB patients who had a PIB-positive scan, and in which patients with PDD had PIB-negative scans. This implies that PIB will be useful for stratifying people with mixed-amyloid DLB from others with pure α-synuclein disease, Lockhart noted. Lockhart’s group found that PIB clearly delineated neurofibrillary tangles, but the labeling was much less intense than that of Aβ lesions, suggesting that in vivo PIB is essentially a specific reagent that measures total Aβ amyloid load.
Importantly, PIB also bound to vascular amyloid deposits, confirming a recent postmortem study (Bacskai et al., 2007). In a poster, Nordberg’s group also showed binding of PIB to blood vessels in brain tissue of people with CAA. Together, the findings imply that in vivo PIB PET images reflect Aβ-related amyloid but say little about the specific kind of amyloid people predominantly have in their brains, Lockhart said. One footnote: curiously, Lockhart found differences in PIB binding to cerebrovascular amyloid in the brains of ApoE4 carriers, an issue that needs more study, especially if PIB is to be used in monitoring preclinical AD in higher-risk people homozygous for ApoE4.
Beyond PIB: Other Ligands
PIB is the most thoroughly studied, but far from the only amyloid imaging probe. Others have entered the scene and are at various stages of exploration. In Salzburg, Rowe presented the first human data on a new ligand that has the great advantage in that it does not require an 11C radiolabel to make it “hot.” The short half-life of 11C requires a separate cyclotron run for each study, as well as specially trained chemists to couple the isotope to the ligand. By contrast, PET using the radioligand 18F is quite widely used in cancer imaging and would facilitate the spread of PIB past a small number of academic centers. Rowe’s center is testing the 18F AV-1/ZK ligand. It is related to the older stilbene SB13 ligand originally developed by Hank Kung at University of Pennsylvania, Philadelphia, and tested once in humans (Verhoeff et al., 2004). Avid Radiopharmaceuticals, also in Philadelphia, and Bayer Schering Pharma are developing the commercial product, and Rowe has an arrangement with them as a clinical investigator.
In Salzburg, Rowe presented first data on 10 AD patients and eight controls. Injected intravenously, the new probe caused no adverse events, and time-activity curves showed ready metabolism in both groups of people. The PET images with this new probe show the same distribution of label as PIB. Most controls did not retain PIB, and all people with AD had the highest binding in their precuneus, posterior cingulate, and frontal cortex, followed by lateral temporal and parietal cortex, as well as low binding in sensorimotor cortex. The non-specific white matter signal was also there. The caudate nucleus showed a little less binding than with PIB. At 3.03, the effect size of the AV-1/ZK is slightly smaller than PIB’s 3.7 but still represents a robust increase in AD over control, Rowe noted. Expressed in different terms, the relative increase in binding in AD of AV-1/ZK is in the range of 60 percent. This compares to 60-80 percent for PIB, and 10 percent for FDDNP, Rowe said.
Other compounds being developed include Japanese ones such as BF-227 (Kudo et al., 2007; Kudo, 2006). Preclinical MRI agents are promising (Higuchi et al., 2005). Several speakers in Salzburg mentioned the 18F compound FDDNP (Small et al., 2006), but the conference featured no presentations on it. Brooks noted that side-by-side comparisons in the same subjects of the different PET tracers would be interesting, but are complicated, thanks to licensing restrictions and academic rivalries.
Inflammation and Amyloid: See Both Together
The search for microglial tracers represents a new frontier in brain imaging, and the AD/PD conference offered some glimpses on how it might begin to dovetail with amyloid imaging. One such approach comes from Brooks’s attempts to visualize the inflammatory response to amyloid deposition with 11C-PK11195, an old compound that was developed originally at Parke-Davis as an anti-arthritis drug. It binds to microglia only when they are switched on. When an injury occurs in the brain (not just in AD), microglia swell, release cytokines, and express peripheral benzodiazepine binding sites that bind PK11195 (e.g., Tai et al., 2007). Brooks showed that wherever amyloid is deposited in AD cortex, his group sees activated microglia, that is, the distribution of the PIB and PK11195 corresponded closely. In PD, microglia are known to become activated, as well (Ouchi et al., 2005; Brooks, 2007). Brooks reported seeing a microglial signal come up not only in PD midbrain, but also throughout the brain in a distribution and strength that correlated with the Braak staging of Lewy body degeneration.
The question is whether activated microglia are helping to clear amyloid or are releasing damaging proteins, or both. Some drugs are known to switch off microglial signals, such as FK-506 (e.g., Yoshiyama et al., 2007), new PPARγ agonists, or minocycline. In the neurodegenerative disease multiple system atrophy, minocycline treatment tamps down the microglial signal, Brooks said. Both Siemens and GE Healthcare are developing newer microglial tracers than PK11195. Brooks hopes that the new compounds will soon put scientists in a position where they can try to treat brain inflammation pharmacologically, and accompany such a trial with both microglial and amyloid imaging as treatment-based biomarkers.—Gabrielle Strobel.