Amyloid β, in its various incarnations, remains a favorite target of big pharma. This year’s AAIC meeting, held July 22-28 in Toronto, saw the arrival of both new BACE inhibitors and the first therapeutic antibody against pyroglutamate forms of Aβ found in plaques.
The latter seems to clear aggregated Aβ from the brain, but researchers expressed concern that this passive immunotherapy turns into active antigen, eliciting a peripheral immune response from nearly every volunteer who receives it. Researchers were unsure what this means for the future of the therapy; some pronounced it dead on arrival but others said this sort of problem can be overcome.
In the increasingly crowded BACE inhibitor arena, researchers from both Novartis and Eli Lilly and Company reported first-in-human studies of their newest compounds, while Janssen reported the first trials of its hopeful in Alzheimer’s patients. These three BACE inhibitors join several others that are well on their way through the trials pipeline. “It is a good thing to have so many BACE inhibitors in clinical trials,” Robert Vassar, Northwestern University, Chicago, told Alzforum. “It is an indication of the high level of confidence that the pharmaceutical industry has in the target. The field is moving forward, and over the course of the next several years, we will know one way or the other whether BACE inhibitors will be safe and effective for Alzheimer’s disease.”
Specificity, Potency, Early Intervention Key for BACE Inhibitors
Johannes Streffer from Janssen Research and Development in Beerse, Belgium, described the first trial of the company’s BACE inhibitor JNJ-54861911 in people with early AD. Streffer previously had introduced the compound at the AD/PD meeting in Nice, where he described the first safety and tolerability study in healthy elderly. Janssen researchers had used state-of-the-art diagnostics to monitor the effects of the inhibitor on APP processing (see Apr 2015 conference news). At AAIC, Streffer reported first results in people who either have prodromal AD or are at risk for the disease.
For the trial, Streffer and colleagues selected 37 people with prodromal AD based on their having a clinical dementia rating (CDR) score of 0.5, and 19 cognitively healthy people who were at risk for AD based on their biomarker profile. The volunteers were randomized (1:1:1) to receive placebo, 10, or 50 mg of JNJ-54861911 once daily for four weeks. Plasma Aβ fell by 83 and 92 percent, respectively, for the low and high dose, and by 67 and 90 percent in the CSF. Similar reductions occurred across the board for all Aβ peptides tested, including Aβ1-37, Aβ1-38, Aβ1-40, and Aβ1-42. Streffer said levels of sAPPβ, the fragment of APP left over by BACE cleavage, fell by similar amounts while there was a dose-dependent uptick in sAPPα, which is shed by α-secretase. The volunteers on the drug performed no better than those on placebo in cognitive testing, but Streffer said he would not expect to see a difference in such a short trial.
All the volunteers completed the trial. Streffer said there were eight adverse events in the high dose, three in the low, and four in the placebo group. He said the pattern did not suggest any safety concern and that the drug was well tolerated.
Streffer said the data matches the lowering seen in Janssen’s previous study on healthy elderly. In a poster, Alberto Russo described how pharmacokinetic/pharmacodynamic modeling of data from that trial accurately predicted lowering of CSF Aβ in the prodromal AD and at-risk population. Streffer told Alzforum that the modeling allows researchers to predict the range of reductions of CSF Aβ in treated populations to ensure that an adequate dose is given to achieve a desired effect. This helps the researchers set dosing levels in upcoming trials. DIAN is pursuing a similar approach of building a disease-progression model from observational data, and using it to predict responses in the DIAN-TU trials (see Part 7 of this series).
Janssen has already begun recruiting for the Phase 2/3 EARLY trial, a safety and efficacy study of people who are asymptomatic but at risk for AD. They plan to recruit more than 2,000 volunteers to receive placebo, 5, or 25 mg of JNJ-54861911 over 4.5 years. Reisa Sperling, Brigham and Women’s Hospital, Boston, told Alzforum she is very excited about this trial, dubbed A5 because it broadens the horizon of the current A4 trial by treating younger, more cognitively intact people. EARLY will recruit volunteers as young as 60. There is no exclusion of “supernormals,” i.e., people who score particularly high on cognitive tests, as there was in A4. Because the trial focuses on younger people and runs longer than A4, those restrictions could be relaxed, noted Sperling. Recruits will have to test positive for amyloid, either by PET scan or through CSF analysis for low Aβ. “CSF is more important in countries that place less emphasis on PET,” said Sperling. “Evidence also suggests Aβ in CSF drops before PET scans test positive, potentially moving us even earlier in disease course,” she added.
EARLY has begun recruiting in Europe and Australia and starts up in Japan today under the leadership of Takeshi Iwatsubo at the University of Tokyo; it will start in the United States and Canada this fall. Its leaders chose the Preclinical Alzheimer Cognitive Composite, or PACC, as the primary outcome. This test battery was designed to detect subtle changes in cognitively normal people and is being used in the A4 trial as well (see Jun 2014 news). EARLY will measure a wide range of secondary outcomes, including safety parameters, PET and MRI imaging for amyloid and neurodegeneration, and a number of different cognitive and functional tests.
Ulf Neumann from Novartis, Basel, Switzerland, presented the first-in-human data for the company’s BACE inhibitor CNP520. Novartis lags behind some of the other pharmaceutical companies that have BACE inhibitors already in Phase 2 and 3 trials, including Merck, Lilly, and AstraZeneca. “Because we are a bit late to the field, we can learn from what others have discovered along the way to help us better tailor our compounds,” Neumann told Alzforum. He believes since all the BACE inhibitors in development are very potent, safety may be the distinguishing characteristic. “We want to test the compound in the early phase of AD when the chances of slowing the disease are best, but where we need long, large trials. Therefore the drug has to be very safe,” Neumann added.
CNP520 was chosen for the Generation prevention trial being conducted by Novartis, the Banner Alzheimer’s Institute in Phoenix, and Amgen. The trial, which began enrolling this month, will test the BACE inhibitor and CAD106, Novartis’ anti-Aβ active immunotherapy in asymptomatic people who carry two copies of the ApoE4 gene (see July 2014 conference news). “CNP520 was selected because of its relatively preferential selectivity for BACE-1, preliminary indications of its safety, tolerability, and potent Aβ-reducing effects,” wrote Eric Reiman from Banner, who is the principal investigator of the study. Generation is incorporating genetic screening and counseling, and will run for five years. Primary outcome measures are time to MCI or AD and change in the Alzheimer's Prevention Initiative Composite Cognitive test score.
In his AAIC talk, Neumann reviewed preclinical safety data for CNP520. He claimed that CNP520 has about threefold selectivity for BACE1 over BACE2. In contrast to NB-360, an earlier Novartis compound, CNP520 caused no coat discoloration in mice. Hypopigmentation can occur if BACE2 is blocked because it helps with processing the melanocyte protein PMEL that is essential for pigmentation (see Dec 2013 conference news). Vassar said that as far as he knows, all other BACE inhibitors cause hypopigmentation to a degree, but whether this will be a significant problem in people remains to be seen. Neumann thinks CNP520 spares melanocytes partly because of its selectivity for BACE1 and partly because it distributes poorly to the skin. Other scientists said the compound’s high brain penetrance might be as important as its selectivity in reducing the risk of peripheral side effects.
CNP520 had even better selectivity over other proteases, including pepsin, renin, and cathepsin D, a close relative of BACE1, Neumann reported. Both are aspartic proteases and have similar pH optima. CNP520 bound to BACE1 with a 1,000-fold specificity over the cathepsin. Vassar noted that most BACE inhibitors now in development show similar selectivity over this protease. Poor selectivity over cathepsin D likely scuppered some early BACE inhibitors, because the protease is essential for the rapid turnover of photoreceptors in the eye and blocking it can cause lipofuscinosis of the retina (see Mar 2011 conference news).
What about the human data? Novartis tested CNP520 in a Phase 1 safety and tolerability trial consisting of three parts. In one, 38 healthy adults received a single ascending dose up to 1,125 mg; 14 received placebo. The second part repeated the single ascending dose up to 750 mg in 50 healthy older adults, while 17 got placebo. In part three, multiple ascending doses up to 300 mg were given to 55 older adults (20 received placebo) over two weeks. The volunteers were monitored for safety, tolerability of the drug, and pharmacodynamics based on CSF sampling. In Toronto, Neumann said adverse events were generally mild. In the single-ascending-dose arm, one volunteer who received 30 mg of CNP520 became anxious and delusional, but on review these symptoms showed up in the person’s medical history and should have excluded him or her from the study, claimed Neumann.
As for efficacy, the compound dose-dependently lowered Aβ40 in the CSF by up to 95 percent after multiple doses. The drug appears to be cleared relatively slowly, with half of it still remaining in the blood of adults after 70 hours. “That can be both a positive and a negative,” Neumann told Alzforum. On the plus side, he said, it’s a forgiving drug—once it reaches a steady state, its concentration will not change very much, so even if a person forgets to take his or her medication one day it will not matter that much. The downside is that if there is an adverse reaction and the patient needs to stop taking the medication, it will take days for it to clear the system. Novartis has yet to run a trial to test the effect of the drug on cognition, but has run other safety and tolerability trials, including a three-month Phase 2a trial in healthy elderly and an ethnic sensitivity study in Japan.
Lilly has taken a new tack in BACE-inhibitor development, debuting a highly potent compound at AAIC. As Lilly’s Brian Willis reported, 1 mg/day of LY3202626 reduced CSF Aβ1-40 by half after two weeks. “This should be an advantage, because lower dosing means reduced potential for off-target side effects,” Vassar told Alzforum. Neumann agreed, noting that the potency Lilly achieved in vitro translated into improved potency in people. Others claimed this was the first time they had seen this work out for a BACE compound. Neumann noted it has no selectivity over BACE2.
Nevertheless, the inhibitor seems to have a good safety profile. Willis reported results of a Phase 1 trial conducted on healthy controls and people with AD. The trial comprised four sequential parts. Parts A to C enrolled 92 healthy people, while part D enrolled two people with AD. In part A, subjects received placebo or 0.1 to 45 mg of the drug as a single dose and were monitored for safety. For part B, volunteers were given 1.6 to 26 mg or placebo as a single dose, and serial plasma and CSF samples were taken for pharmacokinetic and pharmacodynamic analysis. Part C entailed multiple dosing of 1 to 26 mg or placebo for two weeks. Lastly, the two AD patients received 6 mg/day for two weeks. Willis and colleagues took CSF at baseline from volunteers in parts C and D, and again 24 hours after their final dose.
Willis reported that volunteers tolerated LY3202626 well at all doses, without adverse events up to 42 days after the last dose. The compound’s half-life in plasma was 21 hours, and it readily crossed the blood-brain barrier. The compound dose-dependently reduced Aβ40 and Aβ42 in plasma and CSF. At 1, 6, and 26 mg/day over two weeks, it reduced Aβ40 in CSF by 50, 75, and 90 percent, respectively.
Passive Immunotherapy Turns Active Antigen
Mike Irizarry from Eli Lilly and Company, Indianapolis, presented results of a Phase 1 safety and tolerability trial of LY3002813, an antibody against pyroglutamate-modified forms of Aβ. LY3002813 is a humanized version of a mouse monoclonal that purportedly recognizes only plaques (see Dec 2012 news). For the Phase 1, Irizarry and colleagues administered LY3002813 or placebo intravenously to 49 patients, average age 74, who had prodromal to moderate Alzheimer’s disease. All had memory deficits and a positive florbetapir scan for brain deposits of Aβ. Thirty-seven volunteers in five cohorts received a single initial dose of between 0.1 and 10 mg/Kg of the antibody, while 12 received placebo. During this single-ascending-dose phase, clinicians monitored them for three months for any adverse signs, then in the next multiple-dose phase gave them up to four additional monthly injections. For this phase, each person was given the same dose they got originally, except for people in the 0.1mg/Kg cohort—they received 0.3mg/Kg in subsequent injections.
The good news was that at the highest dose the antibody seemed to reduce plaque load in the brain. Irizarry stressed that at six, the number of people in this group was small. Overall, the mean standard uptake value ratio for florbetapir in the high-dose group was 1.65 at baseline, and this fell by 0.26 over seven months. Irizarry said this translates into a mean 40 percent reduction on the centiloid scale of amyloid positivity (see Nov 2014 news). Researchers at the meeting were impressed by the data, which seemed stark on florbetapir scans. LY3002813 joins aducanumab and gantenerumab as the only immunotherapies that have been shown to reduce brain amyloid to date (see Mar 2015 news). Aducanumab had a considerable rate of ARIA-E and ARIA-E-related discontinuations at the higher doses.
“The significant reduction of PET signal, without showing any ARIA-E side effect, is a very important advance over other antibodies in development,” noted Hans-Ulrich Demuth, from the Fraunhofer Institute for Cell Therapy & Immunology, Leipzig, Germany. Demuth advises Probiodrug, a company developing anti-pyroglutamate antibodies and glutaminyl cyclase inhibitors to prevent formation of pyroglutamate Aβ, which is reportedly more aggregation-prone than the unmodified peptide (see Mar 2013 conference news). Indeed, Irizarry reported no cases of ARIA-E (aka vasogenic edema) in this small trial, but he did report two asymptomatic cases of ARIA-H (microhemorrhages). “It’s difficult to tell if these were related to the antibody since microhemorrhages are common in Alzheimer’s disease, but we will continue to monitor closely,” said Irizarry. Others still saw this as cause for concern.
Researchers were even more concerned that this antibody elicited strong immunogenicity. In the multiple-dose phase, six of the 37 patients developed an infusion reaction, such as chills, flushing, dizziness (postural hypotension), rash, and fever. Anti-drug antibodies turned up in the plasma of almost everyone treated. Dave Morgan, University of South Florida, Tampa, saw this as the end of the road, at least for this version of the antibody. “With seven of eight people mounting an immune response, this is basically dead,” he told Alzforum. “They could try to identify the epitope causing the immune response and modify it,” he suggested. Demuth agreed. He speculated that in designing an antibody with high avidity for pGluAβ, a new epitope was created that was highly immunogenic. Morgan said this was most likely in the variable region of the antibody, the business end that binds pGluAβ. Epitopes in antibody-variable regions, known as idiotypes, can, in some cases, elicit an immune response, he explained. Irizarry told Alzforum that his colleagues are characterizing the anti-drug antibodies, but said he does not know as yet the cause of the immunogenicity.
One problem with having an immune reaction is that it shortens the half-life of the therapeutic antibody, and indeed, Irizarry reported exactly that. At low doses, half the antibody was gone from the participants’ blood within four days, as opposed to the 21 days the researchers expected. Irizarry said he was very surprised at the clearance rates. “None of the preclinical work predicted either the immune response or the rapid clearance,” he said. He suspects the two are related, but noted that even after one low dose the antibody was cleared rapidly. “The clearance may not be fully explained by an immune reaction to the antibody,” he told Alzforum. “To understand the relationships with immunogenicity, immune safety, amyloid clearance, and pharmacokinetics, we have to take the therapy forward to another study,” he said. Irizarry said the company has begun a Phase 1b.
On the other hand, plaque-clearing antibodies might not have to be given long-term. They could help eliminate existing plaque, and its recurrence could then be kept in check with BACE inhibitors. Companies including Lilly are experimenting internally with combination regimens of antibodies and BACE inhibitors.—Tom Fagan
- At AD/PD Meeting, New BACE Inhibitor Struts Its Stuff
- At Age 8, DIAN Is Churning Out Data and Growing into a Movement
- Test Battery Picks Up Cognitive Decline in Normal Populations
- Novartis to Partner with Banner Health on ApoE4 Prevention Trial
- Blocking BACE—Do Adult Mouse Phenotypes Predict Side Effects?
- Barcelona: Out of Left Field—Hit to The Eye Kills BACE Inhibitor
- Could Antibodies Against Pyroglutamate Safely Break Down Plaques?
- Paper Alert: Centiloid Scale Aims to Unify Amyloid PET
- Biogen Antibody Buoyed by Phase 1 Data and Hungry Investors
- Can Dousing PyroGlu-Aβ Treat Alzheimer’s Disease?
No Available Further Reading