Despite evidence that amyloid levels were lower, daily lanabecestat 20 mg or 50 mg failed to slow cognitive or functional decline in people with early Alzheimer's disease or mild Alzheimer's dementia in the AMARANTH and DAYBREAK-ALZ trials, John Sims, MD, of Eli Lilly and Company in Indianapolis, and colleagues reported in JAMA Neurology.
Lanabecestat, a brain-permeable inhibitor of human beta-site amyloid precursor protein-cleaving enzyme 1 (BACE1), was developed to modify the clinical course of Alzheimer's by slowing disease progression.
"We designed the molecule to attack [amyloid], and that we could do quite robustly," Sims said in a JAMA Network interview. "We could block, with the doses that we were choosing, somewhere between 50 to 75% of the protein that gets formed. And so the idea was that by blocking that from being released, we would prevent progression of the disease."
Lanabecestat is not the first BACE inhibitor to fail at slowing Alzheimer's disease: a phase III trial of verubecestat in prodromal Alzheimer's patients also had been terminated for futility, while a preliminary report of atabecestat showed the drug was linked to worse cognitive decline in patients with preclinical Alzheimer's disease.
"Unlike some of the other BACE inhibitors, lanabecestat did not worsen cognition," noted David Knopman, MD, of the Mayo Clinic in Rochester, Minnesota, who was not involved with the study.
"Instead, lanabecestat simply failed to produce any cognitive benefits in persons with mild dementia clinically diagnosed as Alzheimer's disease and who had elevated amyloid on a PET scan," he told MedPage Today. "These results definitively demonstrate that amyloid lowering via BACE inhibition is not a therapeutic approach to be pursued in persons with symptomatic Alzheimer disease dementia."
AMARANTH was a phase II/III randomized, placebo-controlled trial lasting 104 weeks from September 2014 to October 2018; DAYBREAK-ALZ was a phase III trial of 78 weeks from July 2016 to September 2018. The studies were designed differently: "AMARANTH was designed to go for about 2 years; that was partly because we were enrolling mild cognitive impairment due to Alzheimer's disease, which is a slightly earlier stage of the dementia process," Sims said. "DAYBREAK-ALZ was only 18 months and focused mostly on mild Alzheimer's disease."
In both studies, men and women who met the National Institute on Aging-Alzheimer's Association criteria for early Alzheimer's or mild Alzheimer's dementia were screened with cognitive assessments and had the presence of amyloid confirmed. The researchers excluded people with unstable medical conditions or medication use, significant cerebrovascular pathologic findings, a history of vitiligo, or current evidence of post-inflammatory hypopigmentation.
AMARANTH randomized 2,218 patients (average age 71; 53.1% women) and 539 patients completed the study. DAYBREAK-ALZ randomized 1,722 patients (average age 72; 59.4% women) and 76 completed the study. Patients were randomized 1:1:1 to once-daily oral doses of lanabecestat 20 mg, 50 mg, or placebo.
The primary outcome measure was change from baseline on the 13-item Alzheimer disease Assessment Scale-cognitive subscale (ADAS Cog-13). Secondary outcomes included efficacy evaluations of lanabecestat in a variety of functional, cognitive, and clinical tests. Efficacy analyses were on the intent-to-treat population.
In June 2018, both lanabecestat studies were terminated early with no consistent, reproducible dose-related findings on primary or secondary measures. The primary efficacy endpoint showed numerical Worsening at certain time points in the lanabecestat arms, but no clear pattern in the two studies.
"Overall, what we saw was that neither of the doses were sufficiently slowing or having an effect on the disease," Sims said. "In the first trial, AMARANTH, it looked like the low dose might have a tiny effect towards the end there, but that certainly was not replicated in DAYBREAK, where actually the low dose looked slightly worse than placebo."
In AMARANTH, lanabecestat produced substantial dose-related reductions in cerebrospinal fluid amyloid markers, greater reduction in beta-amyloid neuritic plaque density than placebo, and greater hippocampal volume loss.
AMARANTH generated a total drug exposure of 1,112 person-years, and DAYBREAK-ALZ 3,027 patient years; over both studies, the drug appeared to be well tolerated. Psychiatric adverse events were numerically greater in lanabecestat groups than placebo and lanabecestat exposure was associated with hair color changes and weight loss. Gray hair was expected, Sims noted, but "the psychiatric symptoms and the weight loss were probably a bit of a surprise for us."
What these findings mean is still being assessed, Sims noted. None of the BACE inhibitors have slowed Alzheimer's "and so we are now in the process of putting all that data together and see if we can understand is there a space here that is still tractable for this mechanism," Sims said.
"In the end, we are still ignorant as to why these molecules did not work as we had anticipated," he continued. "Certainly one hypothesis has been that the amyloid hypothesis isn't a valid one, and that's received a lot of credence lately with the trials that have failed. But I think we're still hopeful that the amyloid hypothesis is true, and we've got a lot of scientific data that still supports that."
Last Updated November 25, 2019
Sims and some co-authors disclosed employment with and/or holding stock in Eli Lilly. Other co-authors disclosed multiple relevant relationships with industry.