Figure 1. Patient disposition. Patients were balanced across groups for concurrent acetylcholinesterase enzyme inhibitor and Memantine treatment (177 total, 34-37 per group) and lumbar puncture (56 total, 10-13 per group). ECG indicates electrocardiography; CT, computed tomography; MMSE, Mini-Mental State Examination; and MRI, magnetic resonance imaging.
Figure 2. Cerebrospinal fluid (CSF) concentration of exploratory CSF biomarkers over time (as percentage of baseline). A, Aβ1-14; B, Aβ1-15; C, Aβ1-16; and D, Aβ1-34 fragments. Aβ indicates beta-amyloid peptide.
Coric V, van Dyck CH, Salloway S, et al. Safety and Tolerability of the ?-Secretase Inhibitor Avagacestat in a Phase 2 Study in Mild to Moderate Alzheimer Disease [published online August 13, 2012]. Arch Neurol.. Arch Neurol. doi:10.1001/archneurol.2012.2194.
eFigure 1. Mean trough concentrations (Cmin) of avagacestat over time
eFigure 2. Cumulative percent discontinuation of patients due to adverse events
eFigure 3. Percentage of patients experiencing 5- and 10-point changes from baseline ADAS-cog during the 24-week, double-blind treatment period
eTable 1. Mean Changes from Baseline to Week 24 in Trefoil Factor 3 (TFF3) and Hairy Enhancer of Split 1 (HES-1)
eTable 2. Mean Change from Baseline to Week 24 in Renal-Related Electrolytes
eTable 3. Mean Change from Baseline to Week 24 in Thyroid Function Test Results
eTable. 4 Mean Change from Baseline to Week 24 in Immunologic Parameters
eTable. 5 Change in ADAS-cog Scores During a 12-Week Washout Period in a Subset of Patients Who Had Experienced Any Worsening in ADAS-cog Score (>0) During the Double-blind Treatment Phase
Coric V, van Dyck CH, Salloway S, Andreasen N, Brody M, Richter RW, Soininen H, Thein S, Shiovitz T, Pilcher G, Colby S, Rollin L, Dockens R, Pachai C, Portelius E, Andreasson U, Blennow K, Soares H, Albright C, Feldman HH, Berman RM. Safety and Tolerability of the γ-Secretase Inhibitor Avagacestat in a Phase 2 Study of Mild to Moderate Alzheimer Disease. Arch Neurol. 2012;69(11):1430-1440. doi:10.1001/archneurol.2012.2194
Author Affiliations: Bristol-Myers Squibb, Wallingford, Connecticut (Drs Coric, Pilcher, Rollin, Dockens, Soares, Albright, Feldman, and Berman and Ms Colby); Yale University School of Medicine, New Haven, Connecticut (Dr van Dyck); Alpert Medical School, Brown University, Providence, Rhode Island (Dr Salloway); Karolinska Institutet, Solna, Sweden (Dr Andreasen); Brain Matters Research, Delray Beach, Florida (Dr Brody); University of Oklahoma College of Medicine, Tulsa (Dr Richter); University of Eastern Finland, Kuopio, Finland (Dr Soininen); Pacific Research Network, Inc, San Diego, California (Dr Thein); California Neuroscience Research Medical Group Inc, Sherman Oaks (Dr Shiovitz); BioClinica Inc, Newtown, Pennsylvania (Dr Pachai); and Clinical Neurochemistry Laboratory, Department of Neuroscience and Physiology, University of Gothenburg, Mölndal, Sweden (Drs Portelius, Andreasson, and Blennow).
Objective To assess the safety, tolerability, and pharmacokinetic and pharmacodynamic effects of the γ-secretase inhibitor avagacestat in patients with mild to moderate Alzheimer disease (AD).
Design Randomized, double-blind, placebo-controlled, 24-week phase 2 study.
Setting Global, multicenter trial.
Patients A total of 209 outpatients with mild to moderate AD were randomized into the double-blind treatment phase. The median age of the patients was 75 years, 58.9% were APOE ϵ4 carriers, and baseline measures of disease severity were similar among groups.
Intervention Avagacestat, 25, 50, 100, or 125 mg daily, or placebo administered orally daily.
Main Outcome Measures Safety and tolerability of avagacestat.
Results Discontinuation rates for the 25-mg and 50-mg doses of avagacestat were comparable with placebo but were higher in the 100-mg and 125-mg dose groups. Trends for worsening cognition, as measured by change from baseline Alzheimer Disease Assessment Scale cognitive subscale score, were observed in the 100-mg and 125-mg dose groups. Treatment-emergent serious adverse events were similar across placebo and treatment groups. The most common reason for discontinuation was adverse events, predominantly gastrointestinal and dermatologic. Other adverse events occurring more frequently in patients undergoing treatment included reversible glycosuria (without associated serum glucose changes), nonmelanoma skin cancer, and asymptomatic magnetic resonance imaging findings. Exploratory cerebrospinal fluid amyloid isoforms and tau biomarker analysis demonstrated dose-dependent but not statistically significant reductions in a small subset of patients.
Conclusions Avagacestat dosed at 25 and 50 mg daily was relatively well tolerated and had low discontinuation rates. The 100-mg and 125-mg dose arms were poorly tolerated with trends for cognitive worsening. Exploratory cerebrospinal fluid biomarker substudies provide preliminary support for γ-secretase target engagement, but additional studies are warranted to better characterize pharmacodynamic effects at the 25- and 50-mg doses. This study establishes an acceptable safety and tolerability dose range for future avagacestat studies in AD.
Trial Registration clinicaltrials.gov Identifier: NCT00810147
Inhibition of the γ-secretase enzyme complex has been pursued as a drug target for the treatment of Alzheimer disease (AD) for more than 20 years.1 Evidence suggests that the sequential cleavage of the amyloid precursor protein first by the β-amyloid cleavage enzyme and then by γ-secretase results in the production of several brain amyloid (Aβ) peptides, including the highly amyloidgenic isoform Aβ42.2 Inhibiting γ-secretase may decrease production of amyloid and potentially modify disease progression in AD.
There are more than 50 known substrates of γ-secretase.1 A major challenge in developing γ-secretase inhibitors for AD is identifying compounds that selectively inhibit amyloid production while minimizing effects on other substrates. In particular, γ-secretase is known to cleave Notch proteins involved in cell fate decisions.3- 5 Consistent with an important role in Notch signaling, long-term administration of γ-secretase inhibitors could produce adverse events (AEs) in the gastrointestinal (GI) tract, thymus, spleen, and skin.6,7
Avagacestat is an oral γ-secretase inhibitor designed for the selective inhibition of Aβ synthesis relative to Notch substrates. In vitro pharmacologic studies8 indicate that avagacestat was approximately 193-fold more selective toward Aβ production than Notch. In vivo pharmacologic studies8- 10 found significant reductions in brain Aβ without Notch-dependent toxic effects. On the basis of these observations, avagacestat was advanced into phase 1 testing, in which single doses of up to 800 mg and multiple doses up to 150 mg/d were well tolerated.11,12 The primary objective of this phase 2 study was to determine the safety and tolerability of avagacestat across 4 dose arms (25, 50, 100, and 125 mg daily) for a 24-week treatment period.
This multicenter, global, randomized, double-blind, placebo-controlled, 5-arm, fixed-dose, parallel-group study was performed in a 24-week treatment period. Written informed consent was obtained from male and female outpatients aged 50 to 90 years with mild to moderate AD. The study was approved by an institutional review board designated by each site and was conducted in accordance with ethical principles and applicable regulatory requirements (clinicaltrials.gov identifier NCT00810147).13,14
Patients met the clinical diagnosis of probable AD based on the National Institute of Neurological and Communicative Disorders and Stroke and the Alzheimer's Disease and Related Disorders Association and Diagnostic and Statistical Manual of Mental Disorders (Fourth Edition, Text Revision) criteria with a Mini-Mental State Examination (MMSE) score of 16 to 26. Patients had a documented cognitive decline for 6 months or longer, brain computed tomography or magnetic resonance imaging (MRI) within 12 months before baseline that had normal results or demonstrated atrophy consistent with AD, no more than mild to moderate white matter disease, 2 or fewer lacunar infarcts, and a Hachinski Ischemia Scale score of 4 or less. Patients were medically stable and had access to a reliable study partner for 10 or more hours per week. Exclusion criteria included the following: a medical condition other than AD that could explain the patient's dementia, previous stroke history, immunocompromised, current diagnosis of peptic ulceration or GI bleeding within the last year, positive fecal immunochemical test result for occult blood in the stool at screening, chronic inflammatory bowel disease, chronic or frequent diarrhea or loose stools, history of GI surgery that resulted in malabsorption, vitamin B12 or folate deficiency, hematologic or solid malignant tumor diagnoses within 5 years, Geriatric Depression Scale score of 6 or higher at screening, or exposure to an investigational agent affecting Aβ levels or function within 12 months before screening. Patients with elevated liver enzyme levels, diabetes mellitus, or a glomerular filtration rate less than 30 mL/min/1.73 m2 were also excluded. Patients treated with approved AD medications had stable treatment for 2 months or longer before screening or were to remain free of such medications.
Clinical outcome measures (11-item Alzheimer Disease Assessment Scale cognitive subscale [ADAS-cog], Clinical Dementia Rating Sum of Boxes [CDR-SB], and Alzheimer Disease Cooperative Study Activities of Daily Living [ADCS-ADL]) were performed at baseline, week 12, and week 24.
All patients had trough blood samples drawn for measurement of serum avagacestat concentrations starting at the week 2 visit and at all subsequent visits during the treatment period.
All study participants were invited to participate in a cerebrospinal fluid (CSF) substudy. Trough lumbar punctures were performed in consenting patients at baseline, week 12, and end of treatment. The CSF levels of total tau (T-tau), phosphorylated tau (P-tau), and Aβ1-42 were analyzed at a central laboratory using the Luminex xMAP technique (INNO-BIA AlzBio3 kit; Innogenetics). Aβ38, Aβ40, and Aβ42 were analyzed using the MS6000 Human (6E10) Aβ 3-Plex kit (MesoScale Discovery).
The change in the Aβ isoform pattern (Aβ1-14, Aβ1-15, Aβ1-16, and Aβ1-34) in CSF was also analyzed by immunoprecipitation and mass spectrometry as described previously.15 This exploratory analysis was included based on reports in the literature that it may constitute a potentially sensitive measure of Aβ target engagement.16,17
Baseline (n = 175) and week 24 MRI (n = 129) results were evaluated centrally (Bioclinica).Volumetric MRI measurements were performed between baseline and week 24 on 3-dimensional T1 sequences. A registration-based brain atrophy quantification algorithm using a tensor-based morphometry Jacobian integration technique was used to calculate brain and ventricular atrophy rates.15 Quantification of total hippocampal volume was performed using a semiautomated contour detection technique.18- 20
Patients were randomly assigned in equal proportions across the 5 masked treatment groups: placebo or avagacestat once daily at dosages of 25, 50, 100, or 125 mg. Patients assigned to the 100-mg and 125-mg dose groups received 50 mg daily for the first 2 weeks. Treatment allocation was balanced by dementia severity (mild [MMSE score of 21-26] vs moderate [MMSE score of 16-20]), AD concomitant medications, and consent for lumbar punctures. Patient safety visits occurred every 2 weeks during the 24-week treatment period, with telephone assessments occurring on alternating weeks during the first 12 weeks after randomization. The AEs were identified for up to 30 days after the study and until resolution for serious adverse events (SAEs). All study medications, including the placebo, were identical in appearance to maintain masking.
An independent data monitoring committee had access to all study data and monitored the safety of participants on a quarterly basis throughout the trial. There were 4 interim analyses reviewed by the data monitoring committee. The committee recommended continuation of the study after each meeting.
The sample size was chosen empirically because the primary objective was to assess safety and tolerability. A sample size of 40 patients per treatment arm was estimated to be associated with an 81% probability of observing a specific AE if the true incidence was 4%. The AEs were classified by primary system organ class and preferred term according to the Medical Dictionary for Regulatory Activities, version 13.0. The incidence of AEs and SAEs was tabulated by treatment group and summarized descriptively. The incidence of potentially clinically relevant changes or events in laboratory values was tabulated by status at baseline (ie, normal vs abnormal). Measures of cognition (ADAS-cog), function (ADCS-ADL), and global ratings (CDR-SB) were incorporated as additional assessments.
For each cognition assessment, the change from baseline to postbaseline scores over time from the observed case data set of the randomized sample was analyzed using a mixed-effects, repeated-measures model with a restricted maximum likelihood estimation. The model included the treatment, time, treatment × time interaction, and baseline × time interaction as main effects and included the corresponding baseline score and key baseline factors (eg, MMSE score, APOE ϵ4 status, and baseline status of concomitant AD medication use) as covariates. Time was treated as a categorical variable in the model. An unstructured covariance matrix was used to represent the correlation of the repeated-measures, within-patient errors. The model-based, adjusted mean change score from baseline and the 95% CI for the treatment difference between active doses and placebo were calculated for weeks 12 and 24. The change from baseline ADCS-ADL and CDR-SB from the observed case data set were analyzed similarly to the ADAS-cog. For CSF biomarkers, the geometric mean over baseline of Aβ38, Aβ40, and Aβ42 was analyzed. The mean change from baseline of T-tau and P-tau was also analyzed. No adjustments were made for multiple comparisons. Nominal P values were provided for descriptive purposes.
A total of 338 patients were enrolled in the study, and 209 were randomized into the double-blind treatment phase (Figure 1). Demographic variables across treatment groups are summarized in Table 1. Median age was 75 years, and baseline disease characteristics were similar between groups.
Steady-state plasma concentrations of avagacestat were achieved by 2 weeks (eFigure 1). The mean concentrations of avagacestat at week 24 were 21, 51, 121, and 175 ng/mL for the 25-, 50-, 100-, and 125-mg dose groups, respectively.
Doses of 25 and 50 mg daily were relatively well tolerated with low discontinuation rates, whereas doses of 100 and 125 mg daily had higher discontinuation rates, primarily attributable to GI and skin AEs (Table 2). Discontinuation rates appeared to plateau by month 3 of the study (eFigure 2). In general, the incidence and intensity of GI-related and dermatologic AEs also appeared to be dose dependent. The overall incidence of AEs in the category of infections and infestations was similar in the placebo, 25-mg, 50-mg, and 100-mg dose groups, whereas in the 125-mg dose group there was an increased incidence of infections without a clear pattern of specific infections. Infection-related AEs were not associated with higher discontinuation rates.
The GI-related AEs (diarrhea, nausea, and vomiting) typically occurred within the first 6 weeks of treatment and plateaued for those individuals who were able to tolerate study medication. Diarrhea was clinically characterized as an increase in frequency of loose stools and was managed clinically with antidiarrheal agents or interruption and downward titration of study medication. The fecal immunochemical test for occult blood was performed in 41 patients with significant GI symptoms while undergoing treatment. Two patients (1 in each of the 25-mg and 125-mg dose groups) had a positive fecal immunochemical test result. Three patients in the 100-mg dose group experienced GI ulcers; 1 additional patient in the 100-mg dose group experienced an ulcer more than 30 days after the last dose of study medication. Two cases of ulcer were characterized as SAEs because of the severity of symptoms and led to discontinuation of use of the study medication. Patients with ulcers tended to have recognized risk factors (chronic reflux, concomitant antiplatelet use, prescribed nonsteroidal anti-inflammatory drugs, cholinesterase use, or Helicobacter -positive culture). The GI-related biomarkers (plasma trefoil factor 3 and messenger RNA hairy enhancer of split 1) showed minimal changes and were not suggestive of a pattern of Notch-related toxic effects (eTable 1).
The overall incidence of SAEs was comparable across groups. The SAEs that occurred in more than 1 patient across dose arms included vasogenic edema (VE) (25 and 50 mg), pneumonia or lobar pneumonia (100 and 125 mg), squamous cell carcinoma (25 and 50 mg), and syncope (25 and 100 mg).
Dermatologic AEs included rash, pruritus, and nonmelanoma skin cancer. Maculopapular rashes occurred on the trunk, extremities, and face, with no reports of desquamation or occurrence on the palms or soles of the feet. Rashes generally resolved within 3 to 4 weeks of discontinuing use of the study medication. No discontinuations occurred because of rash in the placebo or 25-mg or 50-mg dose groups. All cases of skin cancer were clinically manageable by local excision without recurrences during the study period.
No discernable clinical pattern of potentially clinically significant vital sign abnormalities was found. The 100-mg and 125-mg dose groups (13.0% and 20.0%, respectively) had higher rates of weight loss (≥7% decrease in body weight from baseline) than the other groups (≤3.0%) at week 24.
Treatment-emergent glycosuria, defined by any one-time positive urine glucose test result, was observed in a dose-dependent manner but not associated with treatment discontinuations, serum glucose changes, or evidence of glomerular injury (Table 3). More specifically, no decreases in glomerular filtration rate or cystatin C level and no clinically meaningful changes in albumin to creatinine or protein to creatinine ratios were found. Observed glycosuria resolved either with treatment (20%-40% of patients) or within 1 month of discontinuation of treatment.
Few clinically significant changes occurred in renal-related electrolytes, thyroid function, or immunologic parameters (eTables 2-4). Slight decreases over time in mean urinary calcium, uric acid, and potassium plateaued by week 12 and reversed within 1 month of discontinuing use of the study medication. Mild mean decreases in B-cell subtypes (total, naive, or nonswitched memory) were observed in the 100-mg and 125-mg dose groups at week 24 but not at lower doses. Avagacestat was not associated with other clinically meaningful laboratory changes in serum electrolytes or liver function test results.
Significant decreases were found in week 24 trough measurements of Aβ38 in the 100-mg dose group and Aβ38, Aβ40, and Aβ42 in the 125-mg dose group (Table 4). With regard to exploratory analysis of other CSF Aβ fragments, there was a dose-dependent change for the Aβ isoforms Aβ1-14, Aβ1-15, Aβ1-16, and Aβ1-34 (Figure 2).
Repeated-measures analysis of week 24 CSF samples showed declines in T-tau across all doses, but the changes did not reach statistical significance (Table 4). For P-tau, the estimated changes from baseline were less pronounced than T-tau and also did not reach statistical significance.
Mean percent changes from baseline hippocampal, whole-brain, and ventricular volumes were comparable among the treatment groups (Table 5). Retrospective review of baseline and routine week 24 MRI safety scans revealed 3 cases of apparent VE (1 in the 25-mg dose group and 2 in the 50-mg dose group), including 1 patient with apparent VE and numerous cerebral microbleeds (CMBs) at baseline. All cases of VE were clinically asymptomatic and resolved during follow-up. Cases of new CMBs were reported in all study groups, including placebo. The CMBs were defined radiologically as areas of decreased signal intensity on gradient-echo MRI sequences that measured between 2 and 10 mm in diameter. The overall incidence of new CMBs in the placebo and 25-, 50-, 100-, and 125-mg dose groups was 10%, 13%, 16%, 19%, and 13%, respectively. New CMBs observed during routine surveillance MRI were not reported to be associated with clinical symptoms.
Clinical outcome measures are summarized in Table 6. Changes in cognition as measured by the ADAS-cog were similar to placebo at the 25-mg and 50-mg doses but were unfavorable in the 100-mg and 125-mg dose groups. At week 24, the nominal P values for the comparison to placebo were .02 and .09 for 100 and 125 mg, respectively; however, there were no adjustments for multiplicity of tests. The trends for a decrease in clinical performance on the ADAS-cog in the 100- and 125-mg dose groups were not associated with an increase in brain atrophy (as measured by volumetric MRI) or increase in CSF tau levels (Tables 4 and 5). The number of individual patients experiencing potentially clinically meaningful changes in ADAS-cog score (as defined by 5- or 10-point changes from baseline) during the double-blind, 24-week treatment period is shown in eFigure 3. Approximately 58 patients who experienced any increase (>0) from baseline ADAS-cog score during the double-blind treatment period were followed up during a 12-week washout at the completion of the study. No clear pattern of improvement or worsening in the ADAS-cog was seen across treatment groups during washout from study medication (eTables 5). Using a repeated-measures model, we found no statistically significant differences from placebo among treatment groups with regard to the ADCS-ADL except for a trend for a favorable effect in the 25-mg group at week 12 (nominal P = .03). The 50-mg dose group had a mean positive change (improvement) in the ADCS-ADL at week 24 (0.5), whereas the other groups had negative changes (25 mg: −0.7; 100 mg: −1.4; 125 mg: −3.1). Compared with placebo, no statistically significant mean changes in CDR-SB scores were found except for a favorable effect in the 25-mg dose group at week 12 (nominal P = .03).
This study met its aims of identifying safe and tolerable doses of avagacestat for further clinical investigation. Doses of 25 and 50 mg daily were well tolerated with low discontinuation rates, whereas doses of 100 and 125 mg daily were associated with unacceptably high discontinuation rates (primarily attributable to GI and skin AEs) and potentially negative effects on cognition.
The GI-related AEs were dose limiting and accounted for the most common AE category. Although GI-related AEs resulted in significant tolerability issues for patients, overall SAE rates were similar across groups, and circulating GI-related biomarkers did not suggest significant Notch-related GI toxic effects. Even though Notch inhibition can contribute to an increased incidence of nonmelanoma skin cancers,21- 23 elderly individuals are known to be at increased risk for skin cancer, and many of the observed cases of nonmelanoma skin cancer in this study occurred in individuals with risk factors (eg, fair skin or hair, history of skin cancer, and frequent sun exposure). The annual incidence of nonmelanoma skin cancer in similarly aged elderly patients and AD cohorts is estimated to be approximately 4% to 8%.24,25 Clinical experience in this phase 2 study suggests that the occurrence of nonmelanoma skin cancers, if drug related, is manageable with comprehensive baseline skin examinations and ongoing assessment for suspicious skin lesions.
Unanticipated laboratory and diagnostic findings included glycosuria and asymptomatic VE. The observed cases of glycosuria were generally not associated with electrolyte or serum glucose abnormalities, and treatment with avagacestat was not associated with decreases in glomerular filtration rate. Although the mechanism of these renal findings remains unknown, dose-dependent decreases in serum uric acid, calcium, and potassium may suggest alterations in proximal tubule function. Potential mechanisms of the glycosuria include a mild proximal renal tubulopathy, a direct effect on a glucose transporter, or increased fibroblast growth factor 23 signaling due to γ-secretase inhibition of Klotho cleavage.26- 29 Preserved glomerular filtration rate, the reversibility of glycosuria, and the plateau in electrolyte changes that readily reversed on washout suggest that these renal effects can be safely monitored and managed in future studies. Nonclinical toxicologic and phase 1 studies (based on symptoms) did not suggest avagacestat was associated with central nervous system disease. Treatment-emergent cases of VE and new CMBs were reported, but all were asymptomatic (eg, discovered on routine, regularly scheduled MRI scans without any clinical correlate to the neuroimaging findings). Whether the development of CMBs is related to a background incidence of cerebral amyloid angiopathy or represents a treatment-associated change remains unknown. Additional data from future studies will help clarify any potential relationship with study medication.
The CSF biomarkers demonstrated significant reductions in CSF Aβ1-38, Aβ1-40, and Aβ1-42 trough β-amyloid concentrations at the 125-mg dose. At the well-tolerated doses of 25 and 50 mg, no mean reductions in CSF Aβ1-40 or Aβ1-42 were observed 24 hours after dosing. The absence of detectable changes in CSF Aβ1-40 and Aβ1-42 is consistent with phase 1, 50-mg, single-dose studies in healthy volunteers, which showed peak CSF amyloid reductions of 11% to 25% that returned to baseline by 24 hours after dosing.30 Given the dynamic variability in CSF Aβ levels and assay limitations, it may not have been possible to demonstrate less than a 15% to 20% reduction in CSF Aβ1-40 and Aβ1-42 in the approximately 8 patients who were enrolled in the CSF substudy in each of the low-dose arms. Well aware of the current methodologic challenges in measuring pharmacodynamic changes in CSF amyloid and the limitations of such small sample sizes, we included an exploratory analysis of CSF isoforms as an alternate method of demonstrating target engagement. The observed dose-dependent decrease in CSF Aβ1-34 provided additional evidence of γ-secretase target engagement. Increases in CSF Aβ1-14, Aβ1-15, and Aβ1-16 were also observed. The increase in these shorter α-secretase–related isoforms is likely due to shunting from the γ-secretase to the alternative α-secretase pathway due to increased C99 substrate availability.31 Levels of these short Aβ isoforms have been shown to change dose dependently in response to γ-secretase inhibition, even when CSF Aβ1-42 is unchanged, thereby providing indirect evidence of target engagement.16,17 The changes observed in levels of short Aβ isoforms with avagacestat thus also provide preliminary evidence of target engagement. These changes in short Aβ isoforms are consistent with data from phase 1 studies that demonstrated dose-dependent changes in mean plasma Aβ1-40 when avagacestat was administered at dosages ranging from 5 to 800 mg in healthy individuals in a single ascending-dose study.11 Finally, numerical reductions in T-tau and P-tau of approximately 10% to 20% were observed. Tau-related reductions are consistent with potential downstream effects on reducing neurodegeneration; however, larger samples will be required to reach any conclusions because the observed changes were not statistically significant. Overall, the CSF biomarker data from this study provide evidence of avagacestat target engagement in patients with AD.
The observed trends for worsening cognition at the 100-mg and 125-mg doses was an unexpected finding that is not yet fully understood. There was no clear pattern of change in cognitive performance in individual patients who were followed up on discontinuation of study medication use. The evaluation of this data is limited by the sample size and intrinsic variability of the clinical scales. The cognitive results observed in this study may also have been confounded by the lack of the expected decline in the placebo group. More important, the decrease in clinical performance on the ADAS-cog in the 100-mg and 125-mg dose groups was not associated with an increase in brain atrophy (as measured by volumetric MRI) or increase in CSF tau levels. The volumetric MRI and CSF tau results support the notion that the decline in performance on cognition at the 100-mg and 125-mg doses was not due to a worsening of the underlying neurodegenerative process.
Further data regarding avagacestat are needed to better understand whether on-target amyloid-lowering properties, Notch-related toxic effects, or other previously unidentified mechanisms account for the observed cognitive worsening in the 100-mg and 125-mg dose arms. Cognitive worsening has been noted with another γ-secretase inhibitor, semagacestat. Published data regarding semagacestat dosed up to 140 mg have not demonstrated pharmacodynamic effects on CSF amyloid levels, and the cause of worsening in cognition remains to be established.32,33Preliminary in vitro selectivity data comparing avagacestat with semagacestat suggest differences on key Notch-sparing properties and greater selectivity of avagacestat for amyloid precursor protein compared with Notch. Avagacestat is estimated to be 15-fold to 32-fold more selective for Aβ synthesis than for Notch processing compared with semagacestat.8,34 The trends for worsening in cognition at the 100-mg and 125-mg doses of avagacestat may reflect loss of selectivity and greater inhibition of γ-secretase cleavage of substrates other than amyloid precursor protein at higher doses. Although contrary to current assumptions regarding the amyloid hypothesis, it is also plausible that the observed worsening in cognition at the higher doses may be due to direct amyloid-lowering properties of avagacestat. More important, avagacestat was not associated with worsening cognition when dosed at 25 and 50 mg daily, but additional biomarker data will be needed to verify the extent of actual target engagement at those lower doses.
Limitations of this study include relatively small sample sizes, short duration of treatment, inherent variability with the psychometric measurements, small number of patients consenting to the CSF substudy, and lack of placebo decline.
The amyloid cascade hypothesis of AD remains a compelling model for targeted therapeutics but has undergone limited testing in the clinic.35 γ-Secretase inhibitors may adversely affect cognition if doses activate Notch or other critical substrates independent of effects on amyloid. This study demonstrated that avagacestat was safe and tolerable when dosed at 25 and 50 mg daily in patients with mild to moderate AD during a 6-month period. Although the exploratory CSF fragment data are suggestive of a pharmacodynamic effect, further studies are warranted to more fully characterize the effect on amyloid processing at the 25- and 50-mg doses. Other ongoing studies with avagacestat will provide additional biomarker data and increased sample size to better characterize the extent of amyloid lowering at these well-tolerated doses. Overall, the safety and tolerability profile of avagacestat dosed at 25 and 50 mg daily supports continued clinical investigation in AD. Close clinical and safety monitoring, even at these lower doses, is warranted because treatment duration is extended and sample size increased in future studies.
Correspondence: Vlad Coric, MD, Neuroscience Global Clinical Research, Bristol-Myers Squibb, 5 Research Parkway, Wallingford, CT 06492 (firstname.lastname@example.org).
Accepted for Publication: June 8, 2012.
Published Online: August 13, 2012. doi:10.1001/archneurol.2012.2194
Author Contributions:Study concept and design: Coric, Salloway, Brody, Colby, Albright, Feldman, and Berman. Acquisition of data: Coric, van Dyck, Salloway, Andreasen, Brody, Richter, Soininen, Shiovitz, Pachai, Portelius, Andreasson, Blennow, Soares, Feldman, and Berman. Analysis and interpretation of data: Coric, van Dyck, Salloway, Thein, Shiovitz, Pilcher, Rollin, Dockens, Pachai, Portelius, Blennow, Soares, Feldman, and Berman. Drafting of the manuscript: Coric, Andreasen, Thein, Rollin, Pachai, Andreasson, Soares, Feldman, and Berman. Critical revision of the manuscript for important intellectual content: Coric, van Dyck, Salloway, Brody, Richter, Soininen, Thein, Shiovitz, Pilcher, Colby, Rollin, Dockens, Pachai, Portelius, Blennow, Soares, Albright, Feldman, and Berman. Statistical analysis: Rollin and Pachai. Administrative, technical, and material support: Coric, van Dyck, Andreasen, Brody, Richter, Thein, Colby, Andreasson, Blennow, Feldman, and Berman. Study supervision: Coric, van Dyck, Soininen, Albright, Feldman, and Berman.
Conflict of Interest Disclosures: Dr van Dyck reported serving as a consultant to Bristol-Myers Squibb, Janssen Alzheimer Immunotherapy, Pfizer Inc, GlaxoSmithKline, and Elan Pharmaceuticals. He reported receiving research support from Bristol-Myers Squibb, Elan Pharmaceuticals, Janssen Alzheimer Immunotherapy, Pfizer Inc, Eli Lilly, Baxter Pharmaceuticals, GlaxoSmithKline, Medivation, Inc, and the National Institute of Aging and Alzheimer's Association. Dr Salloway reported serving as a consultant to Elan Pharmaceuticals, Janssen Alzheimer Immunotherapy, Astra-Zeneca, Avid-Lilly, Baxter, Pfizer Inc, Eisai, Athena, and Bristol-Myers Squibb. He reported receiving honoraria from Eisai, Pfizer Inc, Elan Pharmaceuticals, Bristol-Myers Squibb, and Athena Diagnostics. He reported receiving research support from Elan Pharmaceuticals, Janssen Alzheimer Immunotherapy, Bristol-Myers Squibb, Eisai, Pfizer Inc, Medivation, Inc, Genentech, Bayer, GE, NIA Alzheimer's Disease Neuroimaging Initiative, NIA Dominantly Inherited Alzheimer's Network, the Alzheimer's Association, The Norman and Rosalie Fain Family Foundation, the John and Happy White Foundation, and the Champlin Foundation. Dr Brody's institution reported receiving grant and travel support from Bristol-Myers Squibb. Dr Soininen reported receiving grant support from Bristol-Myers Squibb and the Academy of Finland and serving as a consultant to ACImmune. Dr Thein reported receiving grant support from Abbot, Astellas, Biogen Idec, Chiesi, Elan Pharmaceuticals, Genentech, Janssen Alzheimer Immunotherapy, Merck, Eli Lilly, Medivation, Inc, Pfizer Inc, and Toyama. He reported serving as a consultant to Eli Lilly, Guidepoint, Gerson Lehman Group, and Clinical Advisors LLC, and reported receiving travel support to investigator meetings from Bristol-Myers Squibb. Dr Shiovitz reported receiving research support from Abbott, Astellas, Astra-Zeneca, Bayer, Bristol-Myers Squibb, Chiesi, Elan Pharmaceuticals, Forest, Eli Lilly, Eisai, Epix, GlaxoSmithKline, Janssen Alzheimer Immunotherapy, Johnson & Johnson, Lundbeck, Novartis, Merck, Organon, Otsuka, Pfizer Inc, Praecis, sanofi-aventis, Shire, Solvay, Sunovion, Takeda, and Targacept. Dr Portelius reported receiving travel support from Bristol-Myers Squibb. Dr Blennow reported receiving grant support from Bristol-Myers Squibb, serving as a speaker for Janssen Alzheimer Immunotherapy, and sitting on Advisory Boards for Pfizer Inc, Immunogenetics, and Probiodrug.
Funding/Support: This study was supported by Bristol-Myers Squibb.
Additional Contributions: Our deepest appreciation to the patients who participated in this clinical trial. We are grateful for the hard work and efforts of the following principal investigators and their clinical staff: Niels Andreasen, Jeffrey Apter, Vinod Bhatnagar, John Brockington, Mark Brody, Anna Burke, Craig Curtis, Vithalbhai Dhaduk, Martin Farlow, Mildred Farmer, Stephen Flitman, Gary Gerard, Joshua Grill, Lawrence Honig, Saleem Ismail, Marvin Kalafer, Bruce Kohrman, David Margolin, Lennart Minthon, Mahmoud Okasha, Omid Omidvar, Joseph Pittard, Ralph Richter, Juha Rinne, Joel Ross, Marwan Sabbagh, Stephen Salloway, Douglas Scharre, Thomas Shiovitz, Amanda Smith, Hilkka Soininen, John Stoukides, Leslie Taylor, Stephen Thein, Christopher van Dyck, Anders Wallin, Myron Weiner, Richard Weisler, Kerri Louise Wilks, Lene Wermuth, and Tanya Vapnik. We also thank the γ-Secretase Inhibitor Development Team for their outstanding implementation of this study protocol, including Caroline Clairmont, PhD, James Hazel, BSN, RN, Stephen Kaplita, MS, Olive Watson-Coleman, RN, MPH, Tamara Bratt, MHS, Leah Burns, MPH, Randy Slemmon, PhD, Nina Cerruti, BS, Jian Han, PhD, Sandeep Kumar, MD, Sue Behling, BS, MT, ASCP, Christina Smith, PhD, Kathleen Szymczak, BS, Yong Lu, MS, Maria Klockare, BSc, Katherine Ciarlone, Jenny Bjorkvall, BA, Laura Ruggiero, BS, Deborah Howe, BS, Beth Morris, BA, and Patricia Ambrose, MBA. We are grateful to Jun-Sheng Wang, PhD, Malaz AbuTarif, PhD, and Brian McHugh, PhD, for their pharmacokinetic modeling, simulation, and analyses in supporting our dosing strategy. We thank the biomarkers group for their support including Paul Rhyne, PhD, Flora Berisha, MS, and Holly Soares, PhD. We also thank Thomas Blaetter, MD, for his contributions to the study protocol while employed at BMS. Finally, we are appreciative for the scientific guidance and insight provided by Elliot Sigal, MD, PhD, Brian Daniels, MD, Doug Manion, MD, and Jane Tiller, MD. We also acknowledge the editorial and technical support of Brian Atkinson, PhD, Bristol-Myers Squibb.