AZD3293: Pharmacokinetic and Pharmacodynamic Effects in Healthy Subjects and Patients with Alzheimer’s Disease

Abstract

AZD3293 (LY3314814) is a promising new potentially disease-modifying BACE1 (β-secretase) inhibitor in Phase III clinical development for the treatment of Alzheimer’s disease. Reported here are the first two Phase I studies: (1) a single ascending dose study evaluating doses of 1–750 mg with a food-effect component (n = 72), and (2) a 2-week multiple ascending dose study evaluating doses of 15 or 50 mg once daily (QD) or 70 mg once weekly (QW) in elderly subjects (Part 1, n = 31), and 15, 50, or 150 mg QD in patients with mild to moderate Alzheimer’s disease (Part 2, n = 16). AZD3293 was generally well tolerated up to the highest doses given. No notable food effects were observed. PK following multiple doses (Part 2) were t_max of 1 to 3 h and mean t_1/2 of 16 to 21 h across the 15 to 150 mg dose range. For single doses of ≥5 mg, a ≥70% reduction was observed in mean plasma Aβ₄₀ and Aβ₄₂ concentrations, with prolonged suppression for up to 3 weeks at the highest dose level studied. Following multiple doses, robust reductions in plasma (≥64% at 15 mg and ≥78% at ≥50 mg) and cerebrospinal fluid (≥51% at 15 mg and ≥76% at ≥50 mg) Aβ peptides were seen, including prolonged suppression even with a QW dosing regimen. AZD3293 is the only BACE1 inhibitor for which prolonged suppression of plasma Aβ with a QW dosing schedule has been reported. Two Phase III studies of AZD3293 (AMARANTH, NCT02245737; and DAYBREAK-ALZ, NCT02783573) are now ongoing.

Keywords

Amyloid-beta peptides AZD3293 BACE1 protein-human cerebrospinal fluid proteins early onset Alzheimer’s disease pharmacodynamics pharmacokinetics Phase I clinical trials

INTRODUCTION

Alzheimer’s disease and other dementias affect more than 47 million people and their families worldwide [1], a number which is projected to reach 115 million by 2050 [2]. Current treatments are palliative, offering only modest improvements in some symptoms, and have no effect on disease progression [3, 4]. Therapeutic approaches that act on the underlying pathophysiology of Alzheimer’s disease have the potential to slow disease progression.

Pathological, biomarker, genetic, and mechanistic data suggest that amyloid accumulation, as a result of changes in production, processing, and/or clearance of brain amyloid-β (Aβ) peptides, plays a key role in the pathogenesis of Alzheimer’s disease [5, 6]. Consequently, investigating potential disease modification through modulation of Aβ concentrations has become one of the highest priority therapeutic targets in Alzheimer’s disease [4, 7].

Alzheimer’s disease pathology is characterized by cerebral Aβ accumulation and the formation of amyloid plaques and neurofibrillary tangles, which are associated with inflammatory changes and neurotransmitter loss [5, 8]. The genetics of Alzheimer’s disease also provides compelling evidence that cerebral Aβ accumulation is crucially involved in the pathogenesis of Alzheimer’s disease, with diverse changes in at least five different genes leading to increased cerebral Aβ accumulation associated with familial Alzheimer’s disease [6, 9]. Genetic mutations in the amyloid-β protein precursor (AβPP) have been linked causally to familial early onset Alzheimer’s disease [10 –12], and many of these mutations are clustered around the β- and γ-secretase sites, leading to increased cleavage and production of the toxic Aβ peptide [13 –15].

The formation of Aβ begins with the cleavage of AβPP by the β-site AβPP cleaving enzyme 1 (BACE1, or β-secretase), with the soluble N terminal fragment of AβPP (sAβPPβ) as a direct product [6]. Inhibition of BACE1 at the first step in the processing of AβPP to Aβ peptides is therefore an attractive target for therapeutic intervention to stop Aβ production and theoretically slow disease progression [5, 9]. This hypothesis is supported by studies in transgenic mouse models of Alzheimer’s disease [9 , 16–19] and most compellingly by human genetic associations. Specifically, two mutations in AβPP have been associated with changes in AβPP cleavage by BACE1. Of these, the Swedish mutation, K670N/M671L, increases AβPP susceptibility to BACE cleavage and confers early onset Alzheimer’s disease [12], while the A673T AβPP variant reduces AβPP susceptibility to BACE cleavage and is associated with a reduced risk for Alzheimer’s disease in elderly individuals [20]. Based on this amyloid cascade hypothesis, reducing Aβ accumulation in the brain by inhibiting the rate-limiting BACE cleavage enzyme early enough in the disease process could be efficacious in Alzheimer’s disease [6].

While early studies with the first generation BACE1 inhibitors were hampered by factors including poor blood-brain barrier penetration and hERG (human ether-a-go-go related gene) activity [21 –23], several second generation BACE1 inhibitors with improved drug-like properties have now progressed into clinical trials. Eli Lilly’s LY2811376 was the first to report Phase I clinical trial results but was terminated due to the observation of cytoplasmic inclusions in the retinal epithelium in a chronic toxicology study in rats [24]. LY2886721, also by Eli Lilly, was terminated in Phase II due to abnormal liver biochemistry [25]. Other BACE1 inhibitors currently in clinical development include MK-8931 ([26]; Merck, Phase III), E2609 ([27], Eisai/Biogen, Phase II), JNJ-54861911 ([28]; Johnson & Johnson, Phase II), and CNP520 ([29]; Novartis/Amgen, Phase II) [6 , 29].

The novel, potent BACE1 inhibitor AZD3293 (LY3314814) is a blood-brain barrier permeable, orally active compound with a slow off-rate from its target enzyme BACE1, which robustly reduced plasma, cerebrospinal fluid (CSF), and brain Aβ₄₀, Aβ₄₂, and sAβPPβ concentrations in vitro and in vivo in mouse, guinea pig, and dog (Table 1) [30]. The in vitro 50% inhibitory concentrations (IC₅₀s) of AZD3293 for BACE1 and BACE2 are both sub-nanomolar (0.6 and 0.9 nM, respectively), and the mean Caco-2 apical to basolateral P_app permeability is 34.8 10^–6 cm/sec [30].

Here we report the favorable findings of the first Phase I clinical trials of AZD3293 in healthy subjects and patients with mild to moderate Alzheimer’s disease. AZD3293 was given as an oral solution of the camsylate salt with dose and concentration units reported as AZD3293 free base equivalents.

Objectives

To investigate the safety/tolerability, pharmacokinetics (PK), and pharmacodynamics (PD) of AZD3293 in two well matched and designed Phase I placebo-controlled trials: a single ascending dose (SAD)/food-effect study in young and elderly healthy subjects; and a multiple ascending dose (MAD) study in elderly healthy subjects and patients with mild to moderate Alzheimer’s disease.

MATERIALS AND METHODS

Study design

Two Phase I, randomized, double-blind, placebo-controlled studies of AZD3293, SAD/food-effect and MAD, were conducted in healthy subjects and patients with mild to moderate Alzheimer’s disease. An overview of the SAD/food-effect and MAD study designs is provided in Supplementary Figure 1. In both studies the primary endpoint was safety and tolerability. Secondary objectives included plasma PK, plasma biomarkers, and PK/PD relationship in plasma. CSF samples were also collected in the MAD study, for analysis of PK and PD biomarkers. Based on selected findings in nonclinical studies, the MAD study protocol was amended to collect baseline and post-treatment skin biopsies in Part 2 of the study (see Procedures).

The SAD/food-effect study was conducted at PAREXEL, Harbor Hospital, Baltimore, MD, USA, from December 2012 to May 2013. The MAD study was conducted at PAREXEL, Early Phase Clinical Unit, Glendale, CA, USA, from March 2013 to March 2014. Both studies were approved by the independent Institutional Review Board, Aspire IRB, Santee, CA, USA. The studies were performed in accordance with the Declaration of Helsinki of 1975 and International Conference on Harmonisation/Good Clinical Practice guidelines. All subjects provided written informed consent prior to enrolment into the studies.

Subjects

SAD/food-effect study

Young (18–55 years) and elderly (55–80 years) healthy subjects with a body weight of ≥50 to ≤100 kg and a body mass index (BMI) ≥19 to ≤30 kg/m², and clinically normal findings on physical examination in relation to age, as judged by the investigator, were included in the SAD/food-effect study (NCT01739647). Exclusion criteria included a history of previous or on-going psychiatric disease/condition, neurologic disease, or use of antipsychotic, antidepressant or anxiolytic drugs.

MAD study

Elderly healthy subjects and patients with mild to moderate Alzheimer’s disease were included in the MAD study (NCT01795339). In Part 1, elderly healthy subjects aged 55–80 years with a body weight ≥50 to ≤100 kg and BMI ≥19 to ≤30 kg/m² and with clinically normal findings on physical examination in relation to age were included. Exclusion criteria included a history of previous or on-going psychiatric disease/condition or neurologic disease, a history of any malignant disease within the past 5 years, or clinically significant illness, medical/surgical procedure or trauma within 4 weeks of the first administration of investigational product. In Part 2, subjects aged 55–85 years with a clinical diagnosis of probable Alzheimer’s disease according to the National Institute of Neurological and Communicative Disorders and Stroke and the Alzheimer’s Disease and Related Disorders Association (NINCDS-ADRDA [31]) criteria, with manifestation of Alzheimer’s disease symptoms at least 6 months before randomization, and a Mini-Mental State Examination [32] score of 16–26 were included. Permitted treatments for Alzheimer’s disease included acetylcholinesterase inhibitors donepezil and rivastigmine or memantine; galantamine was not allowed. Subjects had otherwise clinically normal findings on physical examination in relation to age, as judged by the investigator. Exclusion criteria included significant disease affecting the CNS other than Alzheimer’s disease, including other dementias, other significant neurological or major psychiatric disease, stroke in the last 3 years, active cardiovascular disease, uncontrolled type 1 or type 2 diabetes mellitus or other uncontrolled risk factors for stroke, and a history of any malignant disease within the past 5 years.

Procedures

SAD study

In the SAD study, single oral doses of 1, 5, 15, 50, 150, 450, and 750 mg AZD3293 or placebo were administered in a fasted state through gradual escalation of dose in 56 young healthy subjects, and one dose of 15 mg or placebo in eight elderly healthy subjects. Safety was assessed before and after dosing and during follow-up (7 to 10 days postdose) by recording adverse events (AEs), physical and neurological examinations, Columbia-suicide severity rating scale (C-SSRS [33], Mini-International Neuropsychiatric Interview (MINI [34]), vital signs, ECG, and results of clinical safety laboratory tests. AEs were also recorded at additional follow up visits at 14–21 days and 21–24 days after dose. For PK analysis of plasma AZD3293 and its main metabolite AZ13569724, blood samples were collected at predose and at regular intervals postdose up to 72 h. Blood samples for determination of unbound AZD3293 in plasma were collected 1.5 h postdose. In urine sampling for PK, there were six urine collection periods up to 72 h postdose. Plasma and urine sampling collection time points are provided in full in Supplementary Table 1 and could be modified or extra visits added depending on emerging preclinical data. For plasma PD biomarker analysis (Aβ₄₀ and Aβ₄₂), blood samples were collected predose, at regular intervals in the 72 h postdose and at follow up visits on 7–10, 14–21, and 21–24 days postdose (Supplementary Table 1).

Food-effect study

The food-effect study was a randomized, open-label, single-dose, cross-over, food-effect design at one dose level (150 mg). Healthy young subjects (n = 8) were randomized to start with a single dose of AZD3293 in fed (n = 4) or fasting (n = 4) state, and then following a washout period of at least 7 days, were crossed to fasting or fed, respectively, to receive another single dose of AZD3293. Safety was assessed before and after dosing and during follow-up (7–10 days after dose) by recording AEs, physical and neurological examinations, C-SSRS, vital signs, ECG, and results of clinical safety laboratory tests. Blood samples for PK analysis of plasma AZD3293 and AZ13569724 were collected predose and in the 72 h postdose, as described in Supplementary Table 1.

MAD study

In Part 1, healthy elderly subjects were dosed with AZD3293 15 or 50 mg or placebo once on Day 1 followed by once daily (QD) doses starting on Day 3 for 12 days, or AZD3293 70 mg once weekly (QW) on Days 1, 8, and 15. Safety was assessed before and after dosing and during follow-up (7–10 days postdose) by recording AEs, physical and neurological examinations, C-SSRS, vital signs, ECG, and results of clinical safety laboratory tests.

In Part 1, QD and QW dosing, PK blood and urine sampling for determination of AZD3293 and AZ13569724 were each conducted at multiple time points as detailed in Supplementary Table 1. Blood sampling for PD (Aβ₄₀ and Aβ₄₂) and investigative (sAβPPα and sAβPPβ) biomarker analysis was performed predose, postdose during treatment and on follow-up visits with QD and QW dosing (Supplementary Table 1). Serial CSF samples were collected by lumbar puncture for PK and biomarker analysis with QD and QW dosing of AZD3293 at multiple time points (Supplementary Table 1). Dosing and all other procedures for the QD and QW dosing patients were identical, with the subgroups differing only with respect to the timing of lumbar punctures and the final PK/PD samples.

In Part 2, which was conducted after Part 1, patients with mild to moderate Alzheimer’s disease received 15, 50, or 150 mg AZD3293 or placebo once on Day 1 followed by once daily for 12 days. In Part 2, the assessments and study duration were similar to Part 1. Safety was assessed before and after dosing and during follow-up (7–10 days postdose) by recording AEs, physical and neurological examinations, C-SSRS, vital signs, ECG, eye examinations performed by an ophthalmologist, and results of clinical safety laboratory tests. The time points when blood samples were collected for PK analysis of plasma AZD3293 and plasma PD biomarkers (Aβ₄₀ and Aβ₄₂), and CSF sampling was performed for AZD3293 PK and PD biomarkers (Aβ₄₀ and Aβ₄₂) are detailed in Supplementary Table 1. In addition, skin biopsies were taken from patients in Part 2 of the MAD study on Day –1 (predose) and Day 15 (after last day of multiple dosing) to microscopically assess any effects on skin pigmentation.

Sample handling and analysis

Similar methods of sample collection and analysis were used in the SAD/food-effect and MAD studies. These methods are explained in detail in the Supplementary Material.

Statistical analysis

Similar approaches to statistical analysis were used in the SAD/food-effect and MAD studies. The analysis of data was based on different subsets according to the purpose of analysis. All randomized subjects who received at least 1 dose of AZD3293 or placebo, and for whom any postdose data were available were included in the safety analysis set. The PK and biomarker analyses used an as-treated approach, with the analysis sets based on all subjects for whom PK or biomarker data, respectively, were available.

In the SAD study, the two age groups were analyzed separately. Similarly, in the MAD study, healthy subjects and Alzheimer’s disease patients were analyzed separately. However, data from both groups were included in the PK/PD analysis where the disease status (healthy subject/Alzheimer’s disease patient) was considered as a covariate. No formal statistical hypothesis testing was performed. Safety, tolerability, PK, and PD data were summarized descriptively.

Due to the investigative nature of the studies, sample size was not based on formal statistical considerations. The sample size for each study was based on experience from previous similar Phase 1 studies with other compounds to obtain adequate safety, tolerability, and PK data to achieve the objectives of the study whilst exposing as few subjects as possible to study medication and procedures.

RESULTS

Patients

SAD/food-effect study

In total, 72 subjects were randomized into the first Phase I trial at one study site. Of these, 64 subjects were randomized into the SAD part of the study and 61 completed that study; one subject each from the placebo and 50 mg dose groups was lost to follow up, and one subject in the 450 mg dose group withdrew consent. Eight subjects were randomized into the food-effect part of the study, of which six subjects received treatment as planned in two periods; two subjects discontinued from the food-effect study, one discontinued due to an AE and one discontinued due to severe non-compliance.

In the SAD study, the mean age was between 32.5 and 42.3 years in the young subjects, and between 61.0 and 66.2 years in the elderly subjects; the majority of subjects were male (58/64), and Black or African American (41/64) or White (22/64). In the food-effect study, all subjects were males, and White (5/8) or Black or African American (3/8). Overall, the treatment groups were well balanced and comparable with regards to demographic characteristics. Demographic and baseline characteristics are summarized by treatment group in Table 2.

MAD study

In total, 47 subjects were randomized into the MAD study at one study site. Of these, 31 healthy elderly subjects were randomized into Part 1 and 16 patients with mild to moderate Alzheimer’s disease were randomized into Part 2.

In Part 1, there were more male (19/31) than female (12/31) subjects; mean ages across all groups were similar, ranging from 61.2 years in the 15 mg dose group to 66.0 years in the 70 mg dose group. In Part 2, there were more female (11/16) than male (5/16) patients; mean ages across all groups were similar, ranging from 60.8 years in the 150 mg group to 66.0 years in the 15 mg group.

Demographic and baseline characteristics were well matched across the different treatment subgroups of the MAD study and are summarized in Table 3.

Safety

SAD/food-effect study

No safety and tolerability concerns were identified in the SAD/food-effect study up to the highest AZD3293 dose given (750 mg). There were no deaths or other serious adverse events (SAEs), and all AEs were resolved at the end of the study. Across all dose groups, the types and frequencies of AEs were comparable. In the SAD study, the most frequently reported AEs were upper respiratory tract infection (3 subjects [9.7%] treated with placebo) and contact dermatitis (1 subject receiving placebo, 2 receiving AZD3293), both of which were considered unrelated to treatment in all cases (contact dermatitis considered secondary to use of ECG lead in two cases). Dizziness/dizziness postural was reported by three subjects, and was considered unrelated to treatment in two cases. In the food-effect study, there were three AEs reported, each by one subject (14.3%), dizziness,contact dermatitis, and rash. The subject with rash had AZD3293 treatment discontinued owing to the AE; the AE was considered mild in intensity and related to study drug. AEs assessed as related to study medication are shown in Supplementary Table 2.

There were no clinically meaningful findings in the clinical laboratory data, vital signs, or ECG results in the SAD/food-effect study.

MAD study

In Part 1 of the study (healthy elderly subjects), there were no SAEs, or treatment discontinuations due to AEs. Treatment-related AEs were reported for subjects taking placebo and across all dose groups; no trend with dose was apparent in the types and frequencies of AEs. AEs assessed as related to study medication are shown in Table 4.

The majority of treatment-related AEs were mild. Two treatment-related AEs of moderate intensity were reported: headache (1 subject [8.3%] in the 70 mg dose group); and orthostatic hypotension (1 subject [33.3%] in the placebo group). All treatment-related AEs were resolved at the end of the study. There were no clinically meaningful findings in the clinical laboratory data (including measures of liver function), vital signs, or ECG results.

In Part 2 of the study (patients with Alzheimer’s disease), there were no treatment discontinuations due to AEs. One SAE was reported following completion of dosing (ventricular tachycardia), which was judged to be due to a previously unreported sporadic prior condition and unrelated to study drug administration. Three patients experienced treatment-related AEs of orthostatic hypotension (Table 4); all were transient in nature and resolved without treatment. No other treatment-related AEs associated with vital signs data and no trends were identified that could be attributed to treatment. There were no clinically meaningful findings in the clinical laboratory data, including measures of liver function; physical examinations; eye examinations, including visual acuity and retinal assessments; or ECG results. There was also no evidence of any drug-related effects on melanin content or melanocyte appearance in skin biopsies performed on Day 15 in patients participating in Part 2 of the study.

Pharmacokinetics

SAD study

After a single oral dose in young and healthy elderly subjects, plasma AZD3293 mean C_max increased from 1.74 ng/mL (1 mg dose) to 5520 ng/mL (750 mg dose). The median t_max for the 5 mg to 750 mg dose levels ranged from 1.1 to 2.5 h. AZD3293 AUC_(0 - t) values increased from 26.0 h·ng/mL (1 mg dose) to 66500 h·ng/mL (750 mg dose). The mean effective half-life (t_1/2) was 11 to 20 h in the young dose groups and 24 h in the elderly dose group. Mean total plasma AZD3293 concentration-time profiles are shown in Supplementary Figure 2.

Mean renal clearance was similar at approximately 1.40 L/h following a single oral dose of 15 mg AZD3293 in young and elderly subjects (see Supplementary Material for a description of AZD3293 urine PK).

Plasma AZ13569724 concentration-time profiles and urinary AZ13569724 excretion varied similarly with respect to both AZD3293 dose and time postdose but at lower levels to parent AZD3293 (see Supplementary Material for information).

Food-effect study

In the food-effect study, statistical assessment of the effect of food on single doses of 150 mg (fed or fasted) on AZD3293 exposure showed a modest decrease in peak exposure (C_max), but no effect on overall exposure (AUC) (see Supplementary Figure 3).

MAD study

Following multiple dose administration only limited plasma AZD3293 accumulation was observed (1.5 fold for C_max) in healthy elderly subjects (Part 1). A summary of plasma AZD3293 PK parameters on Day 10 and Day 14 of once daily administration is provided in Table 5. Maximum CSF AZD3293 concentrations were reached at 3 to 4 h.

No accumulation was observed following multiple weekly doses of AZD3293 based upon C_max-C_min values, but a modest increase in AUC was seen between the first and third weekly dose. PK parameters on Day 15 of QW administration are shown in Table 5.

Mean total plasma AZD3293 concentration-time profiles on Day 1 and Day 14 are shown in Fig. 1 (A, B). In the 50 mg dose group, maximum plasma AZD3293 concentrations occurred with a median t_max of 1.5 h on Day 1 and 1.0 h on Day 14. Geometric mean AZD3293 AUC_tau (AUC_0 - 24 for Day 1) for the 50 mg dose group was 1720 h·ng/mL (range 1190–2450 h·ng/mL) for Day 1 and 2820 h·ng/mL (range 2020–4350 h·ng/mL) for Day 14. Geometric mean t_1/2 for the 50 mg dose group was 13.1 h (range 10.3–18.3 h) on Day 1 and 14.9 h (11.7–21.6 h) on Day 14.

In patients with Alzheimer’s disease (Part 2), plasma AZD3293 multiple dose PK were comparable with those observed in healthy subjects (Table 5). Mean total plasma AZD3293 concentration-time profiles on Day 1 and Day 14 are shown in Fig. 1 (C, D). Maximal plasma AZD3293 concentrations were achieved at 1 to 3 h and mean t_1/2 was 16 to 21 h across the three doses studied (15, 50, or 150 mg).

Across all dose levels studied, in healthy subjects and patients with Alzheimer’s disease, CSF AZD3293 concentrations were much lower (measured predose on Day 14) compared with plasma AZD3293 concentrations. PK findings for AZ13569724 mirrored those of AZD3293 with increasing exposure with increasing dose (data not shown).

Pharmacodynamics: Effects on Plasma and CSF Aβ Peptides

SAD study

Following single doses of AZD3293, both Aβ₄₀ and Aβ₄₂ showed robust and dose-dependent reductions. For single doses of 5 mg and above, a ≥70% reduction was observed in mean plasma Aβ₄₀ and Aβ₄₂ concentrations over time. Maximum plasma Aβ₄₀ suppression was reached at 6 h postdose for nearly all doses of AZD3293, for both young and elderly subjects. Maximum Aβ₄₂ suppression was reached within 2 h postdose for nearly all doses of AZD3293, for both young and elderly subjects. A floor effect was observed with respect to many plasma biomarker results for the AZD3293 dose groups, with concentrations being below the lower limit of quantitation (LLOQ; threshold values provided in the Supplementary Material). Sustained suppression of plasma Aβ₄₀ and Aβ₄₂ was observed for approximately 6 days (147 h) postdose for the 450 mg dose and for up to 3 weeks (504 h) postdose for the 750 mg dose, for both biomarkers. Plasma Aβ₄₀ and Aβ₄₂ suppression was not observed for placebo subjects.

MAD study

In Part 1 (healthy elderly subjects), plasma and CSF Aβ peptides showed robust reductions following multiple doses of AZD3293 (15, 50, and 70 mg), compared with placebo, which did not recover during the sampling period (Fig. 2A–D, showing 15 and 50 mg QD data). Similarly in Part 2 (patients with Alzheimer’s disease), robust plasma and CSF Aβ peptide reduction was observed at all doses tested compared to placebo following single and multiple-dose administration (Fig. 2 E–G).

Mean percent change from baseline concentrations for plasma and CSF Aβ₄₀ and Aβ₄₂ in Parts 1 (daily and weekly dosing in healthy elderly subjects) and 2 (daily dosing in patients with Alzheimer’s disease) of the study are presented in Table 6. Here again, plasma biomarker results for the AZD3293 dose groups fell below the LLOQ, suggesting that if biomarker concentrations could have been determined by more sensitive methods, the mean percent reductions would have been higher than the values reported.

Pharmacodynamics: Effects on CSF Amyloid-Related Investigative Biomarkers

MAD study

In Part 1, a dose-dependent sustained decrease in CSF sAβPPβ and broadly dose-dependent increase in CSF sAβPPα was observed with AZD3293 QD dosing (see Supplementary Figure 4). Similarly, a sustained decrease in CSF sAβPPβ and increase in CSF sAβPPα were observed with AZD3293 when dosed at 70 mg QW (Supplementary Figure 4). Preliminary PK/PD analysis indicated a correlation between steady state average CSF AZD3293 concentrations (for 15 and 50 mg groups) at Day 14 and suppression of sAβPPβ and Aβ₄₂ (Fig. 3).

Plasma amyloid-related biomarkers are briefly described in the Supplementary Material.

DISCUSSION

Here we report favorable findings of the first-in-human studies of the potent BACE1 inhibitor AZD3293 involving 103 healthy young and elderly subjects and 16 patients with mild to moderate Alzheimer’s disease. No safety and tolerability concerns were identified up to the highest doses given, which were a single dose of up to 750 mg or multiple daily doses of up to 150 mg for 2 weeks. AZD3293 produced prolonged suppression of plasma and CSF Aβ peptides in healthy subjects as well as in patients with Alzheimer’s disease, confirming the central target engagement and expected mode of action of the drug. These observations support the potential of AZD3293 to alter the pathogenesis of Alzheimer’s disease in ways that could slow diseaseprogression.

Importantly, no safety issues of special interest were identified with AZD3293. No clinically meaningful elevation of liver enzymes was noted in healthy volunteers or patients with Alzheimer’s disease. This is in contrast to LY2886721, which was voluntarily withdrawn from clinical development at Phase II owing to liver biochemistry abnormalities in a small number of subjects, through a mechanism that appeared to be unrelated to BACE1 inhibition [25]. Clinical development of other small molecule BACE1 inhibitors was discontinued early on, including AZD3839 (AstraZeneca), BI1181181/VTP-37948 (Boehringer/Vitae), HPP854 (High Point), PF-05297909 (Pfizer), and RG7129 (Roche), because of a sub-optimal PK profiles, skin reactions, and for undisclosed reasons [6 , 35–38].

The plasma half-life of AZD3293 was 11–24 h in the SAD study, suggesting that a single daily dose would provide sufficient drug concentrations in vivo. PK results were similar in young and elderly healthy subjects. Food consumption led to a modest reduction in C_max and a minimal effect on AUC. PK/PD modeling of the Phase I data suggests that any effects of food on AZD3293 exposure would not impact on safety or CSF Aβ reductions, and therefore that AZD3293 can be taken with or without food. In the MAD study, similar PK findings were also observed in healthy elderly subjects and patients with Alzheimer’s disease, with t_max of 1.1 to 2.5 h and mean t_1/2 of 16 to 21 h across the 15 to 150 mg dose range in patients with Alzheimer’s disease. AZD3293 appears to have a uniquely slow off-rate from BACE1 (t_1/2 ∼9 h) as observed in in vitro studies. It was suggested that the prolonged reduction of Aβ peptide concentrations was driven not only by the AZD3293 plasma/brain exposure but also the turnover rate of the BACE1 enzyme [30].

Indeed, an interesting observation of the present study was the prolonged suppression of plasma Aβ peptides by a single dose of AZD3293, for up to 3 weeks at the highest dose level studied. Some Phase I studies involving other BACE inhibitors in healthy subjects have also described sustained reductions in plasma Aβ that extend beyond the plasma presence of these compounds, from moderate reductions at around 72–144 h postdose with E2609 and LY2811376 [24, 27] to up to 95% reduction after 14 days following a single dose of JNJ-54861911 [28]. The sustained effects of AZD3293 on Aβ led to the inclusion of QW 70 mg dose groups in the MAD study. Here, results similar to QD dosing at the 15 mg dose level (for mean change from baseline at Day 14) were seen for the reduction in Aβ peptides compared with placebo. At this time, AZD3293 is the only BACE1 inhibitor to have reported prolonged suppression of plasma Aβ with a QW dosing schedule.

Similar to effects in plasma, reductions in CSF Aβ peptides were seen in subjects treated with AZD3293 in the MAD study, which further confirmed the expected mode of action of the drug. In healthy elderly subjects, maximum CSF reductions were seen at the 50 mg dose level, of –78% and –79% for Aβ₄₀ and Aβ₄₂, respectively, both measured predose on Day 14. Similar CSF reductions in Aβ₄₀ and Aβ₄₂ (–76% for both) were seen at the 50 mg dose level at the same time point in patients with Alzheimer’s disease, confirming that PD findings in healthy elderly subjects translate well into patients with Alzheimer’s disease. Similar results have also been observed with another BACE1 inhibitor, MK-8931, which produced sustained mean reductions from baseline in CSF Aβ₄₀, Aβ₄₂, and sAβPPβ concentrations (up to –84, –81, and –88%, respectively) in patients with mild to moderate Alzheimer’s disease [26]. In the MAD study, reductions in CSF Aβ peptides were similar between AZD3293 15 mg QD and 70 mg QW treatment groups (for mean change from baseline at Day 14 and Day 18, respectively). Suppression of CSF Aβ peptides by approximately 50% up to 3 days after the last 70 mg QW dose of AZD3293 suggests that a QW dosing regimen may warrant further investigation. One could assume that with a daily dosing regimen in cognitively impaired patients, a benefit of the prolonged PD effects of AZD3293 would mean that a single or even several unintentionally missed doses would have little effect on maintaining steady suppression of Aβ concentrations in the brain. The possible reason for this prolonged PD effect is the slow off-rate of AZD3293 from the BACE1 enzyme discussed above.

Dose dependent reduction in CSF sAβPPβ with AZD3293, an exploratory finding of the present study, provides direct evidence of BACE1 inhibition in the human CNS. In preclinical modeling, the maximum effect of AZD3293 on sAβPPβ was observed after that on Aβ₄₀/Aβ₄₂ [30]. Preliminary PK/PD evaluation in the present study indicated a correlation to C_avg,ss with sAβPPβ and Aβ₄₂, which suggests that measures of amyloid metabolism other than Aβ₄₂ and Aβ₄₀ may have significant utility in determining PK/PD relationships in BACE1 inhibition. These observations are consistent with the hypothesized mechanism of action of AZD3293 and indicate that sustained inhibition of BACE1 leads to reduced sAβPPβ and Aβ₄₂ concentrations and increased flux through the non-pathogenic α-secretase-mediated pathway (sAβPPα). Increased brain [39] and CSF [25] sAβPPα, suggestive of enhanced non-amyloidogenic processing of AβPP, have been observed as a result of BACE1 inhibition in a preclinical and clinical study, respectively.

In addition to BACE1 inhibitors, studies of Aβ-specific therapeutic antibodies have provided evidence to support the therapeutic potential of Aβ-directed treatment in Alzheimer’s disease. Most notable is the recent delayed-start analysis of pooled data from the solanezumab EXPEDITION program [40]. The original, Phase III studies (EXPEDITION 1 & 2) of solanezumab, in patients with mild to moderate Alzheimer’s disease, did not reach their primary endpoint [41]. However, delayed-start analysis of the pooled data from the entire EXPEDITION program (including data from the open-label extension study EXPEDITION-EXT), which focused on patients who were defined as ‘mild’ before the start of the placebo-controlled period, provided evidence to support the Aβ hypothesis. The pooled analyses showed that treatment differences at 108 weeks after randomization between early-start and delayed-start groups in both cognition and function were similar to the between-group differences at the end of the placebo-controlled period, i.e., the delayed-start group did not “catch up”. This difference remained through 132 weeks and suggests a disease-modifying effect of Aβ reduction [40].

Recent evidence in mouse models treated with BACE1 inhibitors with distinct differences in disposition in periphery and brain adds support to the hypothesis that reduction of Aβ via BACE1 inhibition needs to be carried out in the brain, arguing against the peripheral sink hypothesis [4] where reduced Aβ load in the brain might be achieved through lowering of Aβ peptides in peripheral organs [42]. Animal studies of AZD3293 confirm that dose- and time-dependent reductions in plasma and CSF Aβ₄₀, Aβ₄₂ concentrations are also seen in brain tissue [30]. Similarly, Eli Lilly’s BACE1 inhibitors LY2811376 and LY2886721 also significantly reduced concentrations of Aβ and sAβPPβ in the PDABPP mouse brain [24, 25].

The two successive first-in-man Phase I studies described here were rationally designed to allow rapid investigation of the novel BACE1 inhibitor AZD3293 in healthy young and elderly subjects and patients with mild to moderate Alzheimer’s disease, and the identification of appropriate doses to be taken forward into future trials. The efficiency of the design is highlighted by the speed of the studies, which took <16 months from the first subject enrolled into the SAD study to the last subject visit of the MAD study. Consequently, AZD3293 has advanced in clinical development and is now being evaluated in two Phase III studies (AMARANTH, NCT02245737; and DAYBREAK-ALZ, NCT02783573). These multicenter, randomized, double-blind, placebo-controlled studies are testing the disease-modifying potential of AZD3293 at daily doses 20 or 50 mg for 18 to 24 months, in over 4,000 patients with mild cognitive impairment due to Alzheimer’s disease and mild Alzheimer’s disease.

CONCLUSION

The BACE1 inhibitor AZD3293 is a promising new potentially disease-modifying treatment for Alzheimer’s disease with a uniquely slow off-rate on its enzyme target. AZD3293 provided potent and sustained inhibition of BACE1 and produced prolonged suppression of plasma and CSF Aβ peptides in healthy subjects and patients with mild to moderate Alzheimer’s disease. AZD3293 is the only BACE1 inhibitor to have demonstrated prolonged suppression of plasma Aβ with a QW dosing schedule. No safety and tolerability concerns were identified up to the highest single or multiple doses studied. Two Phase III studies of AZD3293 (AMARANTH, NCT02245737; and DAYBREAK-ALZ, NCT02783573) are now ongoing.

Footnotes

ACKNOWLEDGMENTS

Medical writing services were provided by Catherine Spanswick, PhD, and were funded by AstraZeneca.

These studies were sponsored by AstraZeneca.

Authors’ disclosures available online ().

Appendix

References

Dementia. Fact sheet number 362 Geneva: World Health Organization, March 2015,WHO, http://www.who.int/mediacentre/factsheets/fs362/en/ Accessed September 6, 2016.

Wortmann

(2012) Dementia: A global health priority – highlights from an ADI and World Health Organization report. Alzheimers Res Ther 4, 40.

Cole

, Vassar

(2007) The Alzheimer’s disease beta-secretase enzyme, BACE1. Mol Neurodegener 2, 22.

Karran

, Mercken

, De Strooper

(2011) The amyloid cascade hypothesis for Alzheimer’s disease: An appraisal for the development of therapeutics. Nat Rev Drug Discov 10, 698–712.

Menting

, Claassen

(2014) beta-secretase inhibitor; a promising novel therapeutic drug in Alzheimer’s disease. Front Aging Neurosci 6, 165.

Vassar

, Kuhn

, Haass

, Kennedy

, Rajendran

, Wong

, Lichtenthaler

(2014) Function, therapeutic potential and cell biology of BACE proteases: Current status and future prospects. J Neurochem 130, 4–28.

Hardy

, Selkoe

(2002) The amyloid hypothesis of Alzheimer’s disease: Progress and problems on the road to therapeutics. Science 297, 353–356.

Selkoe

(1991) The molecular pathology of Alzheimer’s disease. Neuron 6, 487–498.

Yan

, Vassar

(2014) Targeting the beta secretase BACE1 for Alzheimer’s disease therapy. Lancet Neurol 13, 319–329.

10.

Goate

, Chartier-Harlin

, Mullan

, Brown

, Crawford

, Fidani

, Giuffra

, Haynes

, Irving

, James

, Mant

, Newton

, Rooke

, Roques

, Talbot

, Pericak-Vance

, Roses

, Williamson

, Rossor

, Owen

, Hardy

(1991) Segregation of a missense mutation in the amyloid precursor protein gene with familial Alzheimer’s disease. Nature 349, 704–706.

11.

Hendriks

, van

Duijn CM

, Cras

, Cruts

, Van Hul

, van Harskamp

, Warren

, McInnis

, Antonarakis

, Martin

, Hofman

, Van Broeckhoven

(1992) Presenile dementia and cerebral haemorrhage linked to a mutation at codon 692 of the beta-amyloid precursor protein gene. Nat Genet 1, 218–221.

12.

Mullan

, Crawford

, Axelman

, Houlden

, Lilius

, Winblad

, Lannfelt

(1992) A pathogenic mutation for probable Alzheimer’s disease in the APP gene at the N-terminus of beta-amyloid. Nat Genet 1, 345–347.

13.

Citron

, Oltersdorf

, Haass

, McConlogue

, Hung

, Seubert

, Vigo-Pelfrey

, Lieberburg

, Selkoe

(1992) Mutation of the beta-amyloid precursor protein in familial Alzheimer’s disease increases beta-protein production. Nature 360, 672–674.

14.

Cai

, Golde

, Younkin

(1993) Release of excess amyloid beta protein from a mutant amyloid beta protein precursor. Science 259, 514–516.

15.

Suzuki

, Cheung

, Cai

, Odaka

, Otvos

Jr , Eckman

, Golde

, Younkin

(1994) An increased percentage of long amyloid beta protein secreted by familial amyloid beta protein precursor (beta APP717) mutants. Science 264, 1336–1340.

16.

Kawarabayashi

, Younkin

, Saido

, Shoji

, Ashe

, Younkin

(2001) Age-dependent changes in brain, CSF, and plasma amyloid (beta) protein in the Tg2576 transgenic mouse model of Alzheimer’s disease. J Neurosci 21, 372–381.

17.

Roberds

, Anderson

, Basi

, Bienkowski

, Branstetter

, Chen

, Freedman

, Frigon

, Games

, Hu

, Johnson-Wood

, Kappenman

, Kawabe

, Kola

, Kuehn

, Lee

, Liu

, Motter

, Nichols

, Power

, Robertson

, Schenk

, Schoor

, Shopp

, Shuck

, Sinha

, Svensson

, Tatsuno

, Tintrup

, Wijsman

, Wright

, McConlogue

(2001) BACE knockout mice are healthy despite lacking the primary beta-secretase activity in brain: Imlications for Alzheimer’s disease therapeutics. Hum Mol Genet 10, 1317–1324.

18.

Laird

, Cai

, Savonenko

, Farah

, He

, Melnikova

, Wen

, Chiang

, Xu

, Koliatsos

, Borchelt

, Price

, Lee

, Wong

(2005) BACE1, a major determinant of selective vulnerability of the brain to amyloid-beta amyloidogenesis, is essential for cognitive, emotional, and synaptic functions. J Neurosci 25, 11693–11709.

19.

McConlogue

, Buttini

, Anderson

, Brigham

, Chen

, Freedman

, Games

, Johnson-Wood

, Lee

, Zeller

, Liu

, Motter

, Sinha

(2007) Partial reduction of BACE1 has dramatic effects on Alzheimer plaque and synaptic pathology in APP Transgenic Mice. J Biol Chem 282, 26326–26334.

20.

Jonsson

, Atwal

, Steinberg

, Snaedal

, Jonsson

, Bjornsson

, Stefansson

, Sulem

, Gudbjartsson

, Maloney

, Hoyte

, Gustafson

, Liu

, Lu

, Bhangale

, Graham

, Huttenlocher

, Bjornsdottir

, Andreassen

, Jonsson

, Palotie

, Behrens

, Magnusson

, Kong

, Thorsteinsdottir

, Watts

, Stefansson

(2012) A mutation in APP protects against Alzheimer’s disease and age-related cognitive decline. Nature 488, 96–99.

21.

Butini

, Brogi

, Novellino

, Campiani

, Ghosh

, Brindisi

, Gemma

(2013) The structural evolution of beta-secretase inhibitors: A focus on the development of small-molecule inhibitors. Curr Top Med Chem 13, 1787–1807.

22.

Oehlrich

, Prokopcova

, Gijsen

(2014) The evolution of amidine-based brain penetrant BACE1 inhibitors. Bioorg Med Chem Lett 24, 2033–2045.

23.

Vassar

(2014) BACE1 inhibitor drugs in clinical trials for Alzheimer’s disease. Alzheimers Res Ther 6, 89.

24.

May

, Dean

, Lowe

, Martenyi

, Sheehan

, Boggs

, Monk

, Mathes

, Mergott

, Watson

, Stout

, Timm

, Smith Labell

, Gonzales

, Nakano

, Jhee

, Yen

, Ereshefsky

, Lindstrom

, Calligaro

, Cocke

, Greg Hall

, Friedrich

, Citron

, Audia

(2011) Robust central reduction of amyloid-beta in humans with an orally available, non-peptidic beta-secretase inhibitor. J Neurosci 31, 16507–16516.

25.

May

, Willis

, Lowe

, Dean

, Monk

, Cocke

, Audia

, Boggs

, Borders

, Brier

, Calligaro

, Day

, Ereshefsky

, Erickson

, Gevorkyan

, Gonzales

, James

, Jhee

, Komjathy

, Li

, Lindstrom

, Mathes

, Martenyi

, Sheehan

, Stout

, Timm

, Vaught

, Watson

, Winneroski

, Yang

, Mergott

(2015) The potent BACE1 inhibitor LY2886721 elicits robust central Abeta pharmacodynamic responses in mice, dogs, and humans. J Neurosci 35, 1199–1210.

26.

Forman

, Kleijn

, Dockendorf

, Palcza

, Tseng

, Canales

(2013) The novel BACE inhibitor MK-8931 dramatically lowers CSF beta-amyloid in patients with mild-to-moderate Alzheimer’s disease. Alzheimers Dement 9, P139.

27.

Lai

, Albala

, Kaplow

, Aluri

, Yen

, Satlin

(2012) First-in-human study of E2609, a novel BACE1 inhibitor, demonstrates prolonged reductions in plasma betaamyloid levels after single dosing. Alzheimers Dement 8, P96.

28.

Timmers

, Van Broeck

, Slemmon

, De Waepenaert

, Russu

, Bogert

(2015) Profiling the dynamics of CSF and plasma Abeta reduction with JNJ-54861911, an oral BACE inhibitor. Neurodegener Dis 15, 310.

29.

Yan

(2016) Stepping closer to treating Alzheimer’s disease patients with BACE1 inhibitor drugs. Transl Neurodegener 5, 13.

30.

Eketjall

, Janson

, Kaspersson

, Bogstedt

, Jeppsson

, Falting

, Haeberlein

, Kugler

, Alexander

, Cebers

(2016) AZD3293: A novel, orally active BACE1 inhibitor with high potency and permeability and markedly slow off-rate kinetics. J Alzheimers Dis 50, 1109–1123.

31.

McKhann

, Knopman

, Chertkow

, Hyman

, Jack

Jr , Kawas

, Klunk

, Koroshetz

, Manly

, Mayeux

, Mohs

, Morris

, Rossor

, Scheltens

, Carrillo

, Thies

, Weintraub

, Phelps

(2011) The diagnosis of dementia due to Alzheimer’s disease: Recommendations from the National Institute on Aging-Alzheimer’s Association workgroups on diagnostic guidelines for Alzheimer’s disease. Alzheimers Dement 7, 263–269.

32.

Folstein

, Folstein

, McHugh

(1975) Mini-mental state. A practical method for grading the cognitive state of patients for the clinician. J Psychiatr Res 12, 189–198.

33.

Columbia-Suicide Severity Rating Scale (Japanese Version), Columbia University Medical Center, http://www.cssrs.columbia.edu, Accessed September 6, 2016.

34.

Sheehan

, Lecrubier

, Sheehan

, Amorim

, Janavs

, Weiller

, Hergueta

, Baker

, Dunbar

(1998) The Mini-International Neuropsychiatric Interview (M.I.N.I.): The development and validation of a structured diagnostic psychiatric interview for DSM-IV and ICD-10. J Clin Psychiatry 59(Suppl 20), 22–33; quiz 34-57.

35.

Bell

, O’Neill

, Brodney

, Hajos-Korcsok

, Lu

, Riddell

(2013) A two-part adaptive design to evaluate safety, pharmacokinetics and pharmacodynamics for modifying beta-amyloid in a first-in-human study. Alzheimers Dement 9, P287.

36.

Quartino

, Huledal

, Sparve

, Luttgen

, Bueters

, Karlsson

, Olsson

, Paraskos

, Maltby

, Claeson-Bohnstedt

, Lee

, Alexander

, Falting

, Paulsson

(2014) Population pharmacokinetic and pharmacodynamic analysis of plasma Abeta40 and Abeta42 following single oral doses of the BACE1 inhibitor AZD3839 to healthy volunteers. Clin Pharmacol Drug Dev 3, 396–405.

37.

BACE Inhibitor BI 1181181 Voluntarily Put on Temporary Clinical Hold for Safety Evaluation, Vitae Pharmaceuticals, http://ir.vitaepharma.com/phoenix.zhtml?c=219654&p=irol-newsArticle&ID=2020814, Accessed September 6, 2016.

38.

MacLeod

, Hillert

, Cameron

, Baillie

(2015) The role and therapeutic targeting of α-, β- and γ-secretase in Alzheimer’s disease. Future Sci 1, FSO11.

39.

Eketjall

, Janson

, Jeppsson

, Svanhagen

, Kolmodin

, Gustavsson

, Radesater

, Eliason

, Briem

, Appelkvist

, Niva

, Berg

, Karlstrom

, Swahn

, Falting

(2013) AZ-4217: A high potency BACE inhibitor displaying acute central efficacy in different in vivo models and reduced amyloid deposition in Tg2576 mice. J Neurosci 33, 10075–10084.

40.

Liu-Seifert

, Siemers

, Holdridge

, Andersen

, Lipkovich

, Carlson

(2015) Delayed-start analysis: Mild Alzheimer’s disease patients in solanezumab trials, 3.5 years. Alzheimers Dement 1, 111–121.

41.

Doody

, Thomas

, Farlow

, Iwatsubo

, Vellas

, Joffe

, Kieburtz

, Raman

, Sun

, Aisen

, Siemers

, Liu-Seifert

, Mohs

(2014) Phase 3 trials of solanezumab for mild-to-moderate Alzheimer’s disease. N Engl J Med 370, 311–321.

42.

Georgievska

, Gustavsson

, Lundkvist

, Neelissen

, Eketjall

, Ramberg

, Bueters

, Agerman

, Jureus

, Svensson

, Berg

, Falting

, Lendahl

(2015) Revisiting the peripheral sink hypothesis: Inhibiting BACE1 activity in the periphery does not alter beta-amyloid levels in the CNS. J Neurochem 132, 477–486.