Updated 03 August 2011.

With the buzz surrounding amyloid-targeting therapies, it is easy to lose sight of the fact that other treatments are being tested as well. Presentations at the 11th annual Alzheimer’s Association International Conference (AAIC, formerly ICAD), held 16-21 July 2011 in Paris, France, provided snippets of new information on ongoing trials.

One that is being closely watched is the Alzheimer’s Disease Cooperative Study (ADCS) Phase 3 trial of a plasma-derived human antibody product called intravenous immunoglobulin (IVIG), or Gammagard, developed by Baxter Healthcare. In small studies, this treatment has resulted in both improved cognition and increases in brain volume in AD patients (see ARF related conference story). “As far as I know, the only experimental treatment that has shown a positive correlation between a biomarker and a clinical measure is IVIG,” said Pierre Tariot at the Banner Alzheimer’s Institute in Phoenix, Arizona, who serves as a sub-investigator in the ADCS/Baxter IGIV trial.

It is not yet clear whether IVIG decreases amyloid-β in the brains of AD patients, but the antibody mixture does contain some antibodies against Aβ (Dodel et al., 2011 and Szabo et al., 2010). The drug could have other mechanisms of action, as well, such as immune system-modulating properties, which could explain its beneficial effects. This suggestion came from Norman Relkin at Weill Cornell Medical College in New York City in his AAIC presentation. Relkin measured plasma cytokine levels in 24 AD patients who had enrolled in a prior Phase 2 placebo-controlled IVIG clinical trial. Results from that trial indicated that the cognitive symptoms of patients receiving the treatment improved (see ARF related conference story and ARF conference story). In these patients, nine out of 31 blood cytokines tested increased significantly, and in a dose-dependent manner, following IVIG treatment. The nine cytokines were IL-1A, IL-4, IL-5, IL-6, IL-8, IL-13, GCSF, EGF, and VEGF; increases in IL-5 and IL-8 best correlated with six-month cognitive outcomes. Relkin is currently leading the ADCS Phase 3 multicenter trial of IVIG in 390 AD patients. That study is now fully enrolled and expected to read out during the first half of 2013.

AAIC featured a handful of presentations on IVIG. A poster by William Shankle, at the Hoag Neurosciences Institute in Newport Beach, California, reported a slowing of cognitive and functional decline in a subset of 17 patients with AD or Lewy body disease when these patients were treated with IVIG for up to five years. This slowing persisted for up to three years after stopping IVIG in patients who discontinued the treatment. These are not data from a blinded trial; through his private practice in Newport Beach, California, Shankle has for some years administered IVIG to a number of patients, who pay for this treatment out of pocket. “While this is not a controlled, clinical trial, it has the advantage of adjusting each outcome to the patient's own rate of decline before IVIG treatment,” wrote Shankle in an e-mail to ARF.

At AAIC, Anton Porsteinsson of the University of Rochester Medical Center presented some of the data from the Phase 2 trial of ELND005 (aka scyllo-inositol), a potential oral treatment for AD, being developed by Elan Corporation and partner Transition Therapeutics. The drug has been shown to inhibit aggregation of Aβ in transgenic TgCRND8 mice (McLaurin et al., 2006), and based on Phase 2 trial data, the companies announced they would take the compound into Phase 3 (see ARF related news story).

Originally, 353 patients with mild to moderate AD were randomized to one of three doses of ELND005 (250, 1,000, or 2,000 mg), but Elan later dropped the two higher doses because of safety concerns. At AAIC, Elan and academic scientists reported on a poster that there were no deaths in the placebo group versus one death in the 250 mg group and five and four deaths in the two higher dose groups, respectively. Infections also were more common in the high-dose groups. All subsequent functional and biomarker analyses pertain to the lower 250 mg dose.

With regard to function, Stephen Salloway of Brown University, Providence, Rhode Island, showed data indicating that the treatment group and the placebo group did not differ significantly on the neuropsychological test battery (NTB) or AD Cooperative Study-Activities of Daily Living Scale (ADCS-ADL). This means the study missed its co-primary endpoints. The scientists then moved on to subgroup analyses pre-specified in the protocol, and saw that patients with mild AD receiving 250 mg of ELND005 had significantly higher NTB scores at week 78 compared to placebo.

With regard to biomarkers, at week 78, the 84 patients receiving 250 mg of the drug had an increase in their brain ventricular volume compared to the 82 patients on placebo. Ventricular volume expansion indicates atrophy of brain matter. This difference was seen in the subset of patients with moderate AD, not in those with mild AD. Both mild and moderate patients treated with 250 mg of ELND005 also had a 27 percent decrease in CSF amyloid-β 42 (Aβ42) levels after 78 weeks of treatment. CSF total-tau (t-tau) and phosphorylated tau (p-tau) did not change with ELND005 treatment. In toto, then, patients with mild AD showed a positive trend on the study’s primary cognitive outcome and CSF Aβ42 reduction, but no significant change in ventricular volume or CSF t-tau or p-tau, whereas more advanced patients had no cognitive benefit or brain shrinkage.

One of the new kids on the block is PF-04447943, a selective phosphodiesterase-9A (PDE9A) inhibitor that elevates the amount of cyclic GMP in brain and CSF. Pfizer is testing the drug in AD, but as reported in Paris, results of a multicenter, randomized, double-blind, placebo-controlled Phase 2 trial were not particularly encouraging. Although the drug was generally safe, Elias Schwam of Pfizer in Groton, Connecticut, explained that after 12 weeks of treatment, there was no significant difference in ADAS-cog scores or in the global clinical scores between 91 patients with mild to moderate AD who received 25 mg of the drug and 100 patients who received placebo. “This suggests that, within mild to moderate, a target like cyclic GMC does not show efficacy in a short trial,” said Schwam. “Maybe the treatment is effective in earlier stages of disease.”

Too Little Too Late
A growing number of researchers are seeing recent failures in AD clinical trials as a sign that treatment is being given too late in the disease process. “We have had some spectacular Phase 3 failures in mild to moderate disease. It is precisely for these failures that we need to move earlier,” said Reisa Sperling of Harvard Medical School. She will be lead investigator for a therapeutic trial of preclinical AD planned by ADCS (see ARF related Webinar).

A proposal for another trial in preclinical AD has also been submitted for funding by the Alzheimer’s Prevention Initiative (API) of Banner Alzheimer’s Institute in Phoenix (see ARF conference series), according to a presentation by API’s co-director Pierre Tariot. If approved and funded, that trial should begin by the end of 2012, or, more likely, the beginning of 2013, Tariot told ARF.

In his presentation at AAIC, Tariot reviewed several alternative approaches for designing clinical trials for preclinical AD. He suggested that, in the future, when sufficient data become available to conduct the necessary modeling, Bayesian designs may be well suited to such studies. These permit use of accumulating data analyzed at frequent intervals to decide how to modify aspects of the study as it is being conducted (see ARF related news story). In the meantime, the first trial proposed by API incorporates a so-called adaptive frequentist design with a nested cohort study. The primary outcome of the study would be a change in cognition, but it would also measure different biomarkers. Changes in the biomarkers would be used to make decisions during the trial without prespecifying a hierarchy in response pattern. This information would then serve to design future trials in a different way, as well as to establish the predictive, prognostic, and correlateive value of biomarkers. “Multiple hypotheses can be addressed in this type of trial,” said Tariot.

The trial would be carried out in cognitively normal carriers and non-carriers of mutations causing early onset AD, including a large kindred from Colombia, in which AD is caused by the E280A presenilin-1 (PS1) mutation. Tariot and colleagues have calculated that the trial will probably need to enroll 300 participants without clinical symptoms of AD, including both mutation carriers and non-carriers; carriers would be randomized to treatment. API has proposed a prototype treatment in the grant proposal, although all involved agree that the best available agent will need to be used when such a trial is launched, and the choice could change. Tariot could not, however, disclose the drug the investigators have incorporated in the grant proposal. (see ARF related news story).

“I am really impressed with people wanting to go earlier, as it might be easier to go on to conduct another Phase 3 trial,” said Nick Fox of University College London, U.K., who chaired the session at which Tariot spoke. However, in discussions following the presentation, some conference attendees questioned the rationale of going full-steam with a large trial of preclinical AD before there are officially validated biomarkers or cognitive tests for the condition. Both API and a sister effort for preclinal-stage trials, the Dominantly Inherited Alzheimer's Network (DIAN), are doing extensive work on biomarkers and cognition in preparation for clinical trials in their respective populations.

Others question the conclusion that the current failure of AD drugs to succeed in trials of mild to moderate AD means researchers should focus on treating preclinical stages of disease. “People make the argument that, by the time you have AD, you have too much neuronal loss and it is too late. I do not think there are enough data to say that,” said Eric Siemers of Eli Lilly and Company. “The other drugs to date never got into the brain in sufficient quantities to really test the hypothesis. But if you can make the rate of cognitive decline faster [as it happened with semagacestat], you might also make it slower.”—Laura Bonetta.

This is Part 2 of a series on clinical trial news. See also Part 1.


  1. I'd like to expand on Bonetta's coverage of our work. The implication of our presentation on IVIG is that IVIG therapy may delay progression of more than just AD. The most parsimonious explanation for a disease-modifying effect of IVIG in both AD and LBD is that they both produce misfolded proteins, and that there are IgG molecules that bind to the misfolded, three-dimensional structure of these proteins. In AD, β amyloid oligomers cause a misfolding of its three-dimensional structure, which renders them neurotoxic. In LBD, α-synuclein causes a misfolding of its three-dimensional structure which renders it neurotoxic. Therefore, the IgG antibodies may not be specific to either α-synuclein or β amyloid, but rather bind to the misfolded three-dimensional structure caused by these proteins. If true, then IVIG should also work for other misfolded protein disorders, including Huntington's disease, the tauopathies of frontal temporal lobe disease, and amyotrophic lateral sclerosis, in which case, therapeutic molecules should be developed to bind against the three-dimensional structure of misfolded proteins, and not necessarily to the specific protein involved in the misfolding, which differs for different diseases.

    In addition, and not covered in this news report, we presented a poster showing that a more sophisticated analysis of the Myriad trial data actually suggests a permanently harmful effect of a γ-secretase modulator, flurizan. In an analysis of a sub-sample of just 14 patients from the Myriad Phase 3 trial of flurizan versus placebo, a newly developed method of measuring the underlying processes involved in memory task performance using the ADAS-cog wordlist memory task, showed that flurizan not only harmed the storage of the learned list of words required to be able to recall them after a delay, but that after the trial was discontinued, the decline accelerated. The novel method of analysis, called Hierarchical Bayesian Cognitive Processing Models, was directly compared to the method used by the FDA. The FDA method showed no treatment group difference in this sub-sample, just like it didn't with the full sample analysis of more than 1,600 patients. The importance of this finding is that Hierarchical Bayesian Cognitive Processing Models can be applied to Phase 1 trials involving small samples of patients to determine more sensitively whether a drug is likely to be beneficial or harmful. In the case of the Myriad trial, and I believe also the Lilly trial, had such analysis been done on a small sample prior to doing the Phase 2 and 3 trials, the larger number of patients whose memory was harmed could have been avoided. From an economic perspective, such testing also seems advantageous in helping companies avoid subsequent losses.

Make a Comment

To make a comment you must login or register.


News Citations

  1. Toronto: In Small Trial, IVIg Slows Brain Shrinkage
  2. Chicago: More Phase 2 News—PBT2 and IVIg
  3. Anti-Aβ Oligomer Headed for Phase 3 Clinical Trial
  4. Can Adaptive Trials Ride to the Rescue?
  5. Colombians Come to Fore in Alzheimer’s Research, Mass Media
  6. Paris: Semagacestat Autopsy and Other News of Trial Tribulations

Therapeutics Citations

  1. ELND005

Webinar Citations

  1. Treating Before Symptoms—ADCS Invites Ideas for Clinical Trials in Very Early AD

Paper Citations

  1. . Naturally occurring autoantibodies against beta-amyloid: investigating their role in transgenic animal and in vitro models of Alzheimer's disease. J Neurosci. 2011 Apr 13;31(15):5847-54. PubMed.
  2. . Measurement of anti-beta amyloid antibodies in human blood. J Neuroimmunol. 2010 Oct 8;227(1-2):167-74. PubMed.
  3. . Cyclohexanehexol inhibitors of Abeta aggregation prevent and reverse Alzheimer phenotype in a mouse model. Nat Med. 2006 Jul;12(7):801-8. PubMed.

Other Citations

  1. Gammagard

Further Reading