Background

AZD3293 is an inhibitor of BACE1, the β-secretase sheddase that cleaves the APP protein to release APP's C99 fragment. This fragment then becomes a substrate for subsequent γ-secretase cleavage and Aβ peptide generation. The rationale of BACE inhibition is that it represents an upstream interference with the amyloid cascade, regardless of which species or aggregation states of Aβ then exert toxicity in the brain. BACE inhibition is sometimes envisioned as long-term maintenance therapy to limit Aβ production after an initial round of immunotherapy to remove existing amyloid deposits. 

AZD3293/lanabecestat inhibits both the BACE1 and 2 enzymes. It is administered in tablet form. This compound's in vitro and in vivo pharmacologic profile in primary cortical neurons, mice, guinea pigs, and dogs, as well as a broader development review, were formally published (Eketjall et al., 2016; Sims et al., 2017).

Findings

In December 2012, AstraZeneca started a Phase 1 study that evaluated the safety and pharmacological effects of single doses of oral AZD3293 in 72 healthy volunteers. Doses ranged from 1 to 1,000 mg.

In 2013 and 2014, six additional Phase 1 studies in the U.S. and Japan further evaluated safety, tolerability, metabolism, and potential drug interactions of single- and multiple-ascending doses of a tablet and an oral formulation in both elderly volunteers and Alzheimer's patients. Effects on biomarkers in plasma and CSF were also measured. Before results were published in peer-reviewed journals, the company reported at scientific conferences that the inhibitor appeared safe and strongly reduced CSF Aβ levels (see Mar 2014 news). In October 2016, a peer-reviewed paper on the first two of these Phase 1 trials reported the compound was well-tolerated across the dose range and caused prolonged Aβ reductions in plasma and CSF. Notably, the paper reported blood Aβ reduction even with once-a-week dosing (Cebers et al., 2017). Likewise, data from the first trial in Japan was published, reportedly showing robust Aβ reduction in blood and CSF, and no safety concerns (Sakamoto et al., 2017). The tablet and liquid formulations had similar pharmacokinetics and bioavailability (Ye et al., 2018).

In September 2014, AstraZeneca and Eli Lilly announced that they would jointly develop AZD3293, with AstraZeneca handling manufacturing and Eli Lilly leading clinical trials (see company press release). The clinical development program of this compound largely skipped Phase 2. Instead of running a medium-size Phase 2 followed by separate, larger confirmatory Phase 3 trials, the sponsors opted for a large, pivotal Phase 2/3 trial. Called AMARANTH, this trial compared AZD3293 to placebo given for two years in 2,202 patients who met NIA-AA criteria for MCI due to AD or mild AD. Each participant or his or her partner was required to report worsening in the past six months, and the participant's MMSE had to be above 21 at screening. To ascertain that they had brain amyloid accumulation, participants either underwent an amyloid PET scan or a lumbar puncture and continued in respective sub-studies monitoring those markers for treatment response. The trial compared two doses given once daily as a tablet to placebo, and measured success by change from baseline on the clinical dementia rating sum of boxes (CDR-SB). The ADAS-cog and ADCS-ADL were secondary outcome measures, along with other clinical markers as well as change in CSF markers, functional and amyloid PET, and MRI. This international trial began enrolling in December 2014, and was set to run until 2019.

In 2015 and 2016, four additional Phase 1 trials in a total of 175 healthy volunteers were conducted. They evaluated a new tablet formulation, as well as the interaction of this BACE inhibitor with certain drugs commonly prescribed in the elderly, such as the blood thinners warfarin and dabigatran, the sedative midalozam, as well as simvastatin and donepezil. Some results are published (Monk et al., 2019).

In July 2016, a second Phase 3 trial started up. Called DAYBREAK-ALZ and conducted at 251 locations worldwide, it enrolled 1,899 patients with mild AD dementia as defined by an NIA-AA diagnosis of probable AD with a biomarker evidence of brain amyloid and an MMSE of 10 to 26. This four-arm trial compared two once-daily doses given for three years to two groups who start out on placebo for 18 months and then switch to either the low or high dose for the second half of the trial. The primary outcome is change on the ADAS-cog13 scale; 16 listed secondary outcomes include clinical, functional, biomarker, and population pharmacokinetic measures. This trial was set to run until 2021.

On August 22, 2016, the FDA fast-tracked AZD3293 for expedited review (Reuters news).

On June 12, 2018, AMARANTH and DAYBREAK-ALZ were discontinued due to lack of efficacy determined at an interim futility analysis (Jun 2018 news). Data were presented at AD/PD in Lisbon, Portugal (May 2019 conference news), and subsequently published (Wessels et al., 2019). In AMARANTH, the larger trial, neither 20 nor 50 mg per day moved any of the primary or secondary outcome measures; the placebo group declined at the expected rate. The drug produced neither ARIA nor the cognitive worsening observed with the BACE 1/2 inhibitors verubecestat and atabecestat. The high dose had more dropouts. The most common serious side effects were psychiatric, 2 kg weight loss, and hair discoloration. The last is likely due to BACE2 inhibition in pigment cells, and echoes skin and hair depigmentation observed in mice, rats, rabbits, and dogs treated with BACE1/2 nonselective inhibitors (Shimshek et al., 2016; Cebers et al., 2016). Likewise, DAYBREAK revealed no treatment benefit on any endpoint, with similar side effects. Biomarker data demonstrated target engagement. Lanabecestat reduced blood Aβ40 and Aβ42 levels by 70 to 80 percent in both trials. CSF Aβ, measured in AMARANTH, dropped by 50 and 73 percent at the low and high doses, respectively.  Lanabecestat dose-dependently reduced brain amyloid on florbetapir-PET imaging.

Data on additional cognitive endpoints in AMARANTH were subsequently published (Dec 2020 news; Wessels et al., 2020). Lanabecestat groups showed worsening on the RBANS Total Score, Immediate Memory, Visuospatial/Constructional indexes, and Digit Symbol Coding, but improvement in verbal fluency tests. 

According to published results of imaging from both studies, lanabecestat did not change tau accumulation, brain glucose metabolism, or blood flow, but led to a greater reduction in brain volume compared to placebo (Zimmer et al., 2021).

Biomarker data demonstrated target engagement. Lanabecestat reduced blood Aβ40 and Aβ42 levels by 70 to 80 percent in both trials. CSF Aβ, measured in AMARANTH, dropped by 50 and 73 percent at the low and high doses, respectively.  Lanabecestat dose-dependently reduced brain amyloid on florbetapir-PET imaging.

For all clinical trials of this compound, see clinicaltrials.gov.

Clinical Trial Timeline