In 2002, Phase 2 trials of the first AD vaccine to reach human testing—Elan/Wyeth’s AN1792—came to a halt when treated participants developed encephalitis (see ARF related news story). Subsequent investigation (see, e.g., Pride et al., 2008) traced the harmful inflammation to Aβ-specific T cells produced in response to immunization. Attempting to sidestep these adverse events, Elan/Wyeth came out with a new AD vaccine using a seven-amino acid immunogenic sequence from the Aβ N-terminal but lacking the C-terminal epitope believed to trigger Aβ-specific T cell activation (see Agadjanyan et al., 2005). Called ACC-001, clinical testing of this vaccine also stumbled in Phase 2 when a patient developed vasculitis, or inflammation of the blood vessels. This serious adverse event forced Elan and Wyeth to temporarily suspend dosing in their trial earlier this spring (see ARF related news story), though the trial resumed and is again recruiting patients (see ARF news story).

Other drug candidates that look promising in mice have met similar ill fates when they entered human testing. To bridge this gap, some scientists are testing their therapies in primates, though most of these preclinical studies are small and unpublished. Several years ago, Cindy Lemere of Brigham and Women's Hospital and Harvard Medical School and colleagues reported that 10 months of active immunization with full-length Aβ reduced CNS amyloid deposition in four aged vervets with early AD pathology, compared to 10 age-matched controls (Lemere et al., 2004). While this pilot study demonstrated that vervets are a suitable AD model with amyloid levels that could be lowered by active Aβ immunization, the researchers could not at the time address whether the vaccine actually improved cognition.

At the SfN meeting, Lemere presented data from a more recent Elan/Wyeth-funded study involving 22 vervets aged 18-24 years. Lemere noted that at the mean age in this study (20.5 years), vervets show some plaque deposition, though most don't have severe AD pathology. At this age, their amyloid pathology roughly corresponds to the early stages of AD, Lemere told the audience. The vervets were randomized into three groups using age, gender, CSF Aβ levels, and behavioral performance as selection criteria. The control group received QS-21, an adjuvant used in the AN1792 formulation, without Aβ. The treatment groups received QS-21 with either full-length Aβ42 or dendrimeric Aβ1-15 (dAβ1-15). With 16 copies of Aβ1-15 peptide linked on a branched lysine core—a molecular trick for beefing up short peptides to drive a strong immune response—dAβ1-15 lacks the C-terminal epitope thought to induce self-reactive, Aβ-specific T cells. In earlier mouse studies, Lemere and colleagues showed that this vaccine triggers a robust antibody response and lowers plaque burden in APP transgenic mice immunized intranasally (Seabrook et al., 2006). T cells activated by this vaccine did not recognize endogenous full-length Aβ, suggesting that this approach might avoid the harmful inflammation that plagued human trials of earlier AD vaccine candidates.

In the study presented at SfN, each group of vervets received seven intramuscular injections over 9.5 months. All animals immunized with full-length Aβ produced whopping specific antibody titers, detected by ELISA using Aβ40-coated plates. In the dendrimeric vaccine group, most animals made Aβ-specific antibodies at titers that were reasonable though lower than those seen in animals that received full-length Aβ, Lemere said. All antibody responders showed reduced plaque burden.

Importantly, these amyloid reductions appeared to do some good. By the end of the study, vervets immunized with full-length Aβ kept up their performance on two cognitive measures, whereas non-vaccinated animals had deteriorated. Vervets in the dendrimeric Aβ group actually showed cognitive improvement on both tests, and it was significant, Lemere said in her talk. She added that her group was surprised as to how such strong cognitive gains could come with specific antibody titers that were lower than those in the full-length Aβ group. The Aβ40 coated ELISA might have missed certain conformation-specific Aβ epitopes, or the dendrimeric vaccine might have induced a different kind of immune response. The team will next address the apparent discrepancy between the cognitive effect and the strength of specific immune response.

While the work of Lemere’s group suggests that N-terminal Aβ vaccines may be a way to go, data from another preclinical study presented at the SfN meeting show that full-length Aβ works well in APP-overexpressing mice that model human disease better than do most other AD rodent lines. Despite their significant amyloid and sometimes tau pathology, many Alzheimer’s mouse models lack a defining feature of AD—significant neuronal loss. By knocking out the gene for nitric oxide synthase 2 (NOS2), Donna Wilcock, Carol Colton, and Mike Vitek at Duke University Medical Center in Durham, North Carolina, enabled the primarily vascular amyloid deposition in Tg-SwDI mice (Davis et al., 2004) to expand to tau pathology and neuronal loss (see ARF related news story).

The researchers tested how active Aβ immunotherapy would fare in these mice (APPSwDI/NOS2-/-) and in another strain they had generated earlier (APPSw/NOS2-/-), which has less severe, predominantly parenchymal Aβ pathology (see ARF related news story). They immunized year-old mice with fibrillar Aβ1-42 or a keyhole limpet hemocyanin (KLH) carrier control, both containing Freund’s adjuvant. At the start of vaccinations, both strains of mice showed extensive amyloid and tau pathology as well as hippocampal neuron loss and cognitive deficits, representing early- to mid-stage AD. By the end of the study at 16 months, vaccinated APPSwDI/NOS2-/- mice showed a 30 percent drop in brain Aβ and 30-40 percent reduced tau hyperphosphorylation. The Aβ immunizations also slowed (but did not halt) neuron loss and partially rescued spatial memory deficits, as measured by a radial arm water maze. Effects in the APPSw/NOS2-/- group were more dramatic. Vaccinated animals had a 65-85 percent reduction in brain Aβ and 50-60 percent reduced levels of hyperphosphorylated tau. Remarkably, the vaccine prevented any further neuron loss and completely reversed memory deficits in these mice.

The only real downer in this study was the presence of microhemorrhages in many of the vaccinated APPSw/NOS2-/- mice. The APPSwDI/NOS2-/- mouse, curiously, does not bleed, normally or with vaccination, Wilcock told ARF. “We hypothesize that is because the CAA in this mouse is primarily capillary in nature, so it is possible the vessel becomes occluded by amyloid prior to the occurrence of leakage,” she said. “This is purely speculative at this point.” She noted that CAA in AD is mostly in smaller arteries and arterioles, as is CAA in APPSw/NOS2-/- mice, which might explain why those animals showed microhemorrhages in the recent study. Though an obvious safety concern, it is unclear if microhemorrhages carry functional ramifications, and studying them is challenging because of their varying extent and localization. “When you’re dealing with seven or eight mice, and each one has (microhemorrhage) in a different place, you really can’t pull out any consequences of that,” Wilcock told this reporter. As reported recently by Dora Games and colleagues at Elan Pharmaceuticals, South San Francisco, California, research in PDAPP mice seems to indicate that microhemorrhage can be mitigated by lowering the dose of N-terminal Aβ antibodies offered for six months as passive immunotherapy (Schroeter et al., 2008 and ARF related news story).

Whether this translates to human studies remains unclear, but a study published last month by James Nicoll of the University of Southampton, U.K., and colleagues supports the idea that Aβ immunization sucks Aβ out of plaques, leading to a transient increase in cerebral amyloid angiopathy (CAA) severity as the Aβ gets drained out of the brain through the perivascular system (Boche et al., 2008). In that investigation of AD patients immunized against Aβ42, Nicoll’s team found that those who survived until four to five years after the initial vaccination had virtually no remaining plaques or CAA, raising the possibility that Aβ is cleared from the cerebral vasculature with time. Games told ARF that her group has unpublished data hinting that a similar clearance may happen in PDAPP mice receiving passive Aβ immunotherapy. “We've done experiments out to nine months, and it looks to us that the incidence of these small microhemorrhages decreases over time,” she said.

Interestingly, even after intense scrutiny of nine brain areas, Lemere’s team did not see microhemorrhage in any of the vervets in their most recent study. When asked about these lesions, Lemere attributed their routine occurrence in mouse studies to the start of vaccinations comparatively later in the disease process, when animals already have severe pathology. “If your animal model has a lot of CAA or vascular amyloid at the time of immunization, whether active or passive, if all of a sudden you've got a buildup of antibodies in the periphery, it makes it more likely you're going to run the risk of having microhemorrhage,” Lemere said. “That's one reason I'm a proponent of vaccinating early, even pre-symptomatically. Not only are younger people better at generating an antibody response, but most younger people don't have amyloid deposition in their blood vessels.”—Esther Landhuis.