Whatever the fallout of the AN1792 debacle, it has not stopped commercial or academic interest in immunotherapy. Elan Pharmaceuticals and its partner Wyeth have moved a passive vaccine based on a version of the plaque-binding 3D6 antibody into a multicenter phase 2 trial. Other companies, too, are beginning to report first clinical experiences with their own vaccines. In Madrid, two such vaccines that have entered human studies were made public in some detail, one by Novartis and its partner Cytos Biotechnology in Zurich, Switzerland, and another by Eli Lilly and Co. in Indianapolis.

Matthias Staufenbiel, of Novartis Institutes for BioMedical Research in Basel, disclosed preclinical data for a new active vaccine, dubbed CAD106, that the company is currently testing in humans in Sweden. Staufenbiel described the vaccine as having grown out of efforts to avoid the T cell activation widely blamed for the inflammatory side effect that stymied Elan/Wyeth’s AN1792 trial. The Swiss scientists fashioned a therapeutic vaccine out of the first six N-terminal amino acids of Aβ—a snippet able to stimulate human B cells but devoid of epitopes that arouse human T cells. The trick lies in hitching this stub of Aβ to a virus-like particle that generates the sort of T cell reaction needed to mount a full-fledged antibody response and to break the body’s self-tolerance against Aβ (see also Li et al., 2004). The virus-like particle comes with the added benefit that it is sufficiently potent at marshaling the immune system’s troops as to render additional adjuvants unnecessary.

All mice injected with this vaccine, young and old, mounted an antibody response with high titers, Staufenbiel said. Rabbits did, too. The researchers used two different strains of APP-transgenic mice—one depositing mostly diffuse amyloid but none around blood vessels, and one depositing plaques on blood vessel walls as well as between neurons (see part 3 for more on vessel amyloid). Both mice strains had reductions in Aβ levels and in their predicted amyloid pathology. Vaccination of young versus old mice indicated that the vaccine is more potent at reducing the accumulation of new plaques than removing existing plaques, Staufenbiel said. CAD106 removes parenchymal and vascular amyloid but, importantly, it does not cause the microhemorrhages that some other vaccines are reported to have caused in mice, Staufenbiel noted.

Tests with spleen cells isolated from immunized mice showed that CAD106 does not appear to stimulate T cells. Studies of CAD106 in rhesus monkeys confirmed the mouse data in that the primate antibodies stain plaques in APP-transgenic mouse brain and AD postmortem brain, do not cross-react with APP, and block Aβ-induced toxicity in cell-based assays. Results from the phase 1 trial are expected in 2007, according to a press release issued by Cytos Biotechnology.

Lilly’s Eric Siemers described an initial single-dose trial of a so-called capture antibody. The rationale of this trial is based on the peripheral sink hypothesis. The approach grew out of a widely noted study showing that a single injection of the m266 antibody, which sticks with high affinity to soluble Aβ, improved cognition in mice overnight (Dodart et al., 2002). Together with a paper describing how peripheral injection of the m266 antibody bound plasma Aβ (DeMattos et al., 2001), this line of investigation raised hope that certain antibodies might be able to “draw” Aβ out of the brain by way of shifting a series of presumably connected transport equilibria across the brain, CSF, and blood toward the side of the blood. The vision of a peripheral therapy for AD, or perhaps a diagnostic test similar to an insulin challenge shot for diabetes, took shape.

After reviewing preclinical research in PDAPP mice and rats, Siemers described the first human study in Lilly’s clinical program using LY206430. This humanized version of the m266 monoclonal antibody binds to Aβ16-23, the peptide’s midsection. Investigators infused one of four doses into the veins of four AD patients per dose, plus three placebo controls. Their mean age was 69, mean MMSE 20. The investigators took CSF samples and ran MRI scans at baseline and 21 days later. They tested the participants’ performance in the ADAS-Cog battery at baseline, three days later to check for immediate effects as seen in the animal studies, and again at 21 days. The patients were then followed for one year.

This is what the Lilly scientists found: The antibody appeared safe in terms of standard laboratory values such as liver enzymes, and it also produced no evidence of the side effects that make AD vaccinologists jittery these days, that is, inflammation or microhemorrhage. The antibody produced a typical infusion reaction in five patients, which soon resolved on its own, Siemers said.

Plasma Aβ40 levels shot up between roughly 150- to 600-fold, depending on how much antibody was infused. CSF Aβ40 levels nudged up roughly 1.2- to 1.8-fold; Siemers estimated that 0.1 percent of this antibody enters the CSF and binds Aβ there. He added that the ADAS-Cog values signaled a hint of improvement but that this single-dose study was not designed to support a statement on efficacy.

Other scientists noted that more work lies ahead for this approach. One question they debate concerns the origin and precise nature of the Aβ that binds to the antibody in plasma. How much of it comes from the brain as compared to coming from other large organs known to secrete Aβ, such as muscle? The body-wide economy of Aβ will become clearer as this approach progresses. In that regard, a poster by Yona Levites, in Todd Golde’s group at the Mayo Clinic in Jacksonville, showed that peripheral injection of an anti-Aβ1-16 monoclonal antibody caused a steep rise in plasma Aβ but did not change brain Aβ in wild-type and young APP-transgenic mice. This raises questions about stabilization of peripheral Aβ by the antibody, and appears to complicate the notion of a diagnostic challenge test based on injected antibody. Peter Seubert, of Elan Pharmaceuticals in South San Francisco, reported that in his group’s hands, even long-term treatment of PDAPP mice with the m266 capture antibody did not lower the mice’s cerebral amyloid burden or improve neuritic dystrophy, but it did reduce soluble brain Aβ levels significantly and produced some improvement in tests of acute function and of synaptic health. Seubert's and Golde’s findings would suggest that a capture antibody might prolong the peripheral clearance rate of Aβ by stabilizing Aβ in the blood.

On a more general note, the plasma half-life of all antibodies that are being developed as passive AD vaccines is roughly around a month. This raises questions about how often it would have to be infused and what price for such a therapy health care systems would be able, or willing, to sustain.—Gabrielle Strobel.