9 April 2008. Antibody therapies for AD have galvanized researchers, affected families, and observers from academia, industry, and even the investment community. Speakers at the Keystone meeting, held March 24 to 29 in Keystone, Colorado, deferred major news announcement on this topic to the upcoming ICAD conference this July in Chicago, but some morsels did slip out. Eli Lilly and Company’s passive immunotherapy reportedly was safe in an initial Phase 1 trial, though it may not act solely through a peripheral sink effect as previously proposed. Instead, a small proportion of the antibody enters the brain, where it appears to sequester forms of Aβ that are in equilibrium with aggregation and plaque formation. This increases the total CSF Aβ pool; however, a CSF Aβ fraction that is not bound to antibody appears to decrease. Besides this treatment, a range of other immunotherapies have by now entered human trials.
Ron DeMattos from Eli Lilly and Co., Indianapolis, introduced his talk with an overview of immunotherapies that are wending their way through the clinical pipeline. Active vaccines include Elan’s ACC-OO1 in Phase 2, Novartis’s CAD-106
in Phase 1, and Merck’s V950 in Phase 1. All three are based on the N-terminal of the Aβ peptide and are thought to stimulate an antibody-driven immune response that clears Aβ primarily through a mechanism of Fc-mediated endocytosis of antibody-decorated amyloid into microglia. Passive immunotherapies, where patients receive an antibody infusion, include a pharmacogenomic set of two Phase 3 trials for Elan’s bapineuzimap; Lilly’s LY2062430, which concluded a Phase 2 trial last month; Baxter’s small Phase 2 trial of its pooled IVIg preparation (aka Gammaguard ); as well as Phase 1 trials by Pfizer, GlaxoSmithKline, and Hoffman-La Roche/Morphosys.
While the trials play out, a major research question concerns the mechanisms by which these experimental therapies act. The three that have been proposed—phagocytic clearance, inhibition of fibrillogenesis, and peripheral sink—are not mutually exclusive, but it’s still debated which ones are at play in humans or whether there will be a single consensus mechanism. In his talk, DeMattos focused on the peripheral sink hypothesis, an indirect mechanism of amyloid clearance that his and colleagues’ research had suggested might cause a net efflux of Aβ from brain to plasma (DeMattos et al., 2001). In essence, the idea was that a peripheral antibody might be able to draw Aβ from the brain indirectly by shifting transport equilibria between the brain and blood and delivering brain Aβ to clearance in the liver and kidneys. The star in this hypothesis is m266. Called a “capture antibody,” this IgG binds epitopes 16 to 24 in the mid-section of soluble Aβ very tightly, prying it away from chaperones and other endogenous proteins that otherwise stick to Aβ. By contrast, m266 does not bind Aβ deposited in plaques, as do many other candidate immunotherapy antibodies.
In preparing m266 for the clinic, Lilly scientists humanized it. In parallel, they studied in detail how m266 perturbed the Aβ transport equilibrium between the plasma and CSF in PDAPP mice and non-transgenic rats. In the process, they also developed CSF biomarkers tailored to this specific treatment. In describing this research, DeMattos seemed to be gingerly stepping away from the original peripheral sink mechanism, though that appeared not to affect the therapeutic promise of m266 immunotherapy.
DeMattos reported that, as previously seen in PDAPP mice, intravenous m266 in rats caused m266-bound Aβ to shoot up 250-fold in plasma within a day. But the antibody also showed up in the CSF two hours after the injection, and in this compartment it reached an equilibrium of 0.08 to 0.14 percent with plasma antibody within the day. In the CSF, then, Aβ40 and 42 levels also increased. To understand what that meant, the scientists developed an assay that can distinguish between the total CSF Aβ pool (i.e., Aβ bound and unbound to IgG) and the pool of unbound Aβ. A subsequent rat study injecting three different doses of m266 showed a dose-dependent increase in total CSF Aβ but a dose-dependent decrease of the pool that is not bound to m266.
By this mechanism, plasma m266 would not initially draw Aβ out of the CSF but instead enter the CSF and disrupt an equilibrium there. The idea behind it is that soluble and insoluble Aβ are in a pathogenic equilibrium, and that decreasing the former by means of m266 would reduce the supply of Aβ available for aggregation and deposition, gradually shrinking amyloid pathology in that way. DeMattos noted that because both the peripheral and central mechanisms of the antibody occur simultaneously, it was impossible to identify which one was primarily responsible for the decreased unbound CSF Aβ. How local m266 mechanisms in the CSF interact with peripheral mechanisms in plasma is at yet unclear, DeMattos said.
DeMattos then offered a brief summary of the Phase 1 trial, promising full data of the Phase 2 trial for ICAD this July. In brief, study volunteers received placebo or one of three doses of m266. Their plasma Aβ40 increased as expected, though with a slower time course than seen in the animal studies. Their CSF likewise showed an increase in total Aβ, both 40 and 42. This trial did not have an assay for free Aβ, but that critical piece will come with the Phase 2 data, DeMattos said.
It’s unclear at present what happens to the antibody-bound Aβ accumulating in the CSF—whether it gets swiftly degraded or might cause complications. Antibody-Aβ complex that forms in the CNS may traffic to the periphery and get eliminated via normal IgG catabolism. That would be consistent with the original premise of the peripheral sink, but the time course of this traffic and degradation remains unknown. What is known, DeMattos noted, is that m266’s mechanism does not involve inflammatory processes. Nor have the Lilly scientists seen effects on CAA or CAA-related microhemorrhages with this antibody, at least in PDAPP mice.
This conference data comes as the latest word in an ongoing debate about m266. Earlier this year, Peter Seubert and colleagues at Elan Pharmaceuticals reported that, in their hands, m266 failed to shrink amyloidosis in PDAPP mice; it even tended to increase it. These scientists also noted that binding to m266 prolonged the normal degradation of Aβ (Seubert et al., 2008). As is often the case, this scientific discrepancy may find its resolution in the clinic.—Gabrielle Strobel.