Effects of a γ-secretase inhibitor in a randomized study of patients with Alzheimer disease

Although the cause of Alzheimer’s disease (AD) is unknown, many lines of evidence point to a central role of the peptide amyloid β (Aβ).¹ Aβ, particularly Aβ_1-42, is a major constituent of amyloid plaques, and mutations of the amyloid precursor protein have been associated with the early onset of AD in several families. Mutations in either presenilin-1 or presenilin-2, thought to code for a component of the γ-secretase complex, also lead to early-onset familial AD. Additionally, the ε4 allele of APOE is a major risk factor for the development of AD, and APOE is thought to be important in the trafficking of Aβ.²

LY450139 is a functional γ-secretase inhibitor that reduces the rate of formation of Aβ in whole cell assays³ and in transgenic mice⁴ carrying the amyloid precursor protein V717F mutation (“PDAPP” mice).⁵ After administration of LY450139 dihydrate to PDAPP mice, reductions in Aβ in brain, CSF, and plasma are observed.⁴ In a previous clinical study in our laboratories, after 14 days of treatment with 5 to 50 mg LY450139 dihydrate, a dose-dependent decrease in plasma Aβ was observed in volunteers.⁶ In the current study, we determined the safety, tolerability, and change in plasma and CSF Aβ in 70 patients with mild to moderate AD treated for 6 weeks with LY450139 dihydrate.

Methods.

A detailed description of the trial design and methodology is provided in Appendix E-1 (available on the Neurology Web site at www.neurology.org). Subjects with mild to moderate AD were randomly assigned to receive LY450139 or placebo. The dose of LY450139 was 30 mg QD for 1 week, followed by 40 mg QD for an additional 5 weeks. Plasma and CSF concentrations of Aβ were determined along with adverse events and cognitive changes.

Results.

Change in CSF Aβ.

The changes in CSF Aβ concentrations are shown in figure 1. After 6 weeks of administration of LY450139 dihydrate, mean CSF concentrations of Aβ_total, Aβ_1-40, and Aβ_1-42 were numerically less for patients taking drug compared with those taking placebo; however, none of the changes were statistically significant.

• Download figure
• Open in new tab
• Download powerpoint

Figure 1. Change in CSF Aβ concentrations. Each value is the mean percentage of baseline concentration at end point with 95% CIs. N = 33 for LY450139 dihydrate–treated patients, and n = 33 for placebo-treated subjects (a second lumbar puncture could not be obtained for one subject in the placebo-treated group).

The changes in CSF Aβ concentrations were not clearly related to plasma or CSF concentrations of LY450139 dihydrate. Figure 2 shows the percent change in Aβ_1-40 in plasma compared with that in CSF.

• Download figure
• Open in new tab
• Download powerpoint

Figure 2. Percent change in plasma and CSF Aβ_1-40. The horizontal axis represents the time after administering a dose of LY450139 dihydrate. The solid line shown for plasma values represents the average percent change in plasma Aβ_1-40 as a function of time and was obtained using population pharmacokinetic/pharmacodynamic analysis techniques as described in Appendix E-1. CSF values are based on lumbar punctures performed at baseline and after the final dose of LY450139 dihydrate.

Discussion.

In this study, administration of a γ-secretase inhibitor to patients with AD for 6 weeks resulted in a reduction in plasma Aβ very similar to that seen in our previous clinical⁶ and preclinical⁴ studies. A reduction of 38% in plasma Aβ was present, whereas a change in Aβ concentration in CSF could not be demonstrated. The tolerability of LY450139 dihydrate when given at a dose of 40 mg QD seemed to be good.

Several reasons may account for the lack of change in CSF Aβ concentration in this study. First, the magnitude of the change in plasma Aβ was only 38% for Aβ_1-40. Given the variability in CSF Aβ change seen in this study (see figure 2), demonstration of a change in CSF concentration may not have been possible. In a previous study,⁶ the concentration of LY450139 dihydrate in CSF determined 6 hours after a 40-mg dose was 33.1 ng/mL (83.3 nM), which is above the whole cell IC₅₀ of the compound (approximately 15 nM, data on file). Nevertheless, a higher dose and greater reduction in CSF and plasma Aβ may allow a decrease in CSF Aβ concentration to be demonstrated.

The magnitude of acute Aβ effect in plasma or CSF necessary to modify the rate of AD progression is unknown. In transgenic PDAPP mice treated for 5 months with 30 mg/kg/day LY450139 dihydrate, a reduction in brain Aβ compared with vehicle-treated animals has been demonstrated.⁷ A 60% maximum reduction in plasma Aβ was present using a 30-mg/kg/day dose.

A potential cause of toxicity due to γ-secretase inhibitors is their inhibition of NOTCH cleavage,⁸ which has been associated with goblet cell hyperplasia in intestinal epithelium.⁹ One subject treated with LY450139 dihydrate in this study was withdrawn because of gastrointestinal bleeding associated with a Barrett esophagus. The histologic changes reported in the esophagus in the affected patient were not consistent with goblet cell hyperplasia; nevertheless, gastrointestinal adverse events should continue to be monitored closely in trials of γ-secretase inhibitors.

In preclinical studies, changes in the immune system including atrophy of the thymus and a decrease in lymphocytes in the spleen and thymus have been associated with some doses of LY450139 dihydrate (data on file) and other γ-secretase inhibitors.¹⁰ In this clinical trial, T-lymphocyte and eosinophil counts were slightly but statistically increased in the LY450139 dihydrate–treated patients (see table). A decrease in a mitogen stimulation (CD69) assay was found after 6 weeks of treatment (see table); however, this change was no longer present 90 days after treatment discontinuation (data not shown). Although these assessments suggest that any effects of LY405139 dihydrate on lymphocyte function in this study were relatively mild and without clinical consequences, studies of γ-secretase inhibitors should include assays monitoring immune function.

Acknowledgment

The comments of Daniel Ness, PhD, regarding preclinical toxicology studies are greatly appreciated. The expertise of Thomas Brown in the determination of LY450139 concentrations and of Mark Willey, Daphne Farrington, and Betsy Cunningham in the determination of Aβ concentrations are gratefully acknowledged.