PARIS—The long frustrating search for a disease modifying therapy for Alzheimer disease (AD) hit one of its lowest points last August when two large pivotal trials of the gamma secretase inhibitor semagacestat were halted by Eli Lilly and Co. after an interim analysis showed that subjects not only did not improve on cognitive or functional measures, they also got worse.
Now, Lilly is back with preliminary follow-up data that show that even after treatment was withdrawn, patients who received semagacestat continued to perform worse than those on placebo, although their decline slowed to a rate similar to those on placebo.
Presenting the data here at the Alzheimer's Association International Conference, Eric Siemers, MD, senior medical director for Alzheimer disease at Lilly, pointed to a positive aspect of the trial, claiming that biomarker studies showed that semagacestat did indeed hit its target, inhibiting gamma secretase in the brain. Moreover, he said: “This was the first compound that has gone into phase 3 testing that actually has had a cognitive effect. Unfortunately, it was not the effect we wanted, but it suggests that we are going along the right track.”
Not everyone agreed with Dr. Siemers' rosy assessment, however. Rachelle Doody, MD, PhD, director of the Alzheimer's Disease and Memory Disorders Center at Baylor College of Medicine chaired the Alzheimer's Disease Cooperative Study (ADCS) committee set up to independently review the data. Emphasizing that the ADCS has yet to complete its analysis of the data, she said the trial so far has only shown that semagacestat does not work and that the development of the drug should stop.
Until the failure of the semagacestat trial, gamma secretase inhibitors had been viewed as one of the most promising strategies for reducing amyloid in the brain and thus preventing disease.
Gamma secretase is an enzyme that cleaves amyloid precursor protein (APP) to produce the fragments that accumulate in the brains of AD patients as amyloid plaque. Inhibition of the enzyme would presumably interfere with the development of plaque. But what was not known when semagacestat was first developed is that APP is just one of at least 50 transmembrane proteins that are substrates for gamma secretase. So one possibility is that an off-target effect of semagacestat led to the cognitive worsening.
One protein that even early in the development process was known to be affected by gamma secretase inhibition is notch. Notch receptors enable cellular communication and are involved in many cellular processes including those related to neuronal function. Previous studies in mice had shown that semagacestat inhibits notch cleavage with potentially toxic effects.
“We have always been a little worried about notch in terms of tolerability, but these effects seemed manageable,” said Dr. Siemers. However, the worsening of cognition seen in the trial raises the possibility that inhibition of notch could be involved, possibly through its known involvement with neuronal stem cells and neurogenesis.
Dr. Siemers noted there are numerous other possible explanations. Indeed, the one aspect of the trial that garners the most agreement is that it raises more questions than it answers.
The strategy that Lilly used in developing semagacestat was somewhat controversial. Rather than relying on a signal of cognitive benefit from phase 1 and 2 trials, they decided to move forward to phase 3 on the basis of tolerability and biomarker evidence — particularly the measurement of plasma and CSF amyloid beta (Abeta) — reasoning that phase 2 studies are rarely large enough to provide the power needed to quantify changes in cognition. Two studies were launched — Identity and Identity II, which together would enroll more than 2600 mild-to-moderate AD patients.
Now that the results are in, patients did not improve, and the trial has been stopped, Dr. Siemers still says he believes the biomarker strategy was the right one. “We had a good package of biomarker data showing that these compounds entered the brain in sufficient quantities to have an effect on the intended target.”
Lilly is now using that same approach in developing another drug that targets amyloid, the monoclonal antibody solanezumab, which binds soluble Abeta with the goal of reducing deposition into amyloid plaques.
Given that the drug development program for semagacestat was unsuccessful, “it's hard to say that the biomarker approach was validated,” said Paul Aisen, MD, director of the ADCS and professor of neuroscience at the University of California, San Diego. However, he continued, “I happen to think that it was a very reasonable approach to select a drug on the basis of the biomarker evidence that supported the semagacestat program.”
But biomarkers can only tell you so much, he continued. “They don't estimate the clinical relevance of the pharmacodynamics effect and do not direct the investigator to the proper population for treatment.”
The proper population is critical, he said, because it may be that people who already show symptoms of AD have such significant brain dysfunction that they are highly susceptible to perturbation that could have resulted in worsening cognition. “Perhaps the same dose in a different population might be effective.”
Dr. Doody agreed that there are limits to what biomarkers can tell you. “The goals of a phase 2 trial are to make sure you figure out the dose to take forward into phase 3, understand the safety, and get a clear signal of efficacy,” she said. “I don't think we can do that from biomarkers alone.”
THE NEXT STEP
The ADCS committee has barely begun its own analysis of the data from the follow up study. Further analysis of the pharmacodynamic and pharmacokinetic data, in combination with measurements of Abeta levels in the blood and CSF and other biomarkers, could help elucidate the meaning of the adverse events, including decline in cognition, and provide some guidance for gamma secretase and anti-amyloid programs in the future, said Dr. Aisen.
Dr. Doody added: “There will likely be new questions asked. We don't have the final conclusions yet.”