Aβ rebounds in the plasma in both humans and animal AD models when administration of GSIs is stopped (see, e.g., Siemers et al., 2006; Lanz et al., 2006; ARF related news story). In fact, you can end up with more Aβ than if you never gave the inhibitor to begin with, noted Bateman. That would obviously be a concern if it happened in the brain. “The other concern is that a rapid change in concentration, especially a spike, may, on theoretical grounds anyway, increase the likelihood of Aβ to fibrillize and seed,” he said.

Bateman, together with joint first authors Jacquelynn Cook from Merck Research Laboratories, West Point, Pennsylvania, and Kristin Wildsmith from WashU, used a rhesus monkey model to test for rebound in cerebrospinal fluid (CSF). In this model, the fluid is sampled through a port and catheter permanently inserted into the cisterna magna at the base of the brain. This allows the researchers to study a single animal longitudinally with repeat sampling, without the need for multiple lumbar punctures, as would be done in humans.

Following an intravenous pulse of carbon-13-labeled leucine, Cook and colleagues found that brain Aβ synthesis and clearance were not significantly different. Analyzing for the γ-secretase inhibitor MK-0752, they found that after giving it orally (at either 60 or 240 mg/Kg), the drug rapidly appeared in the plasma and CSF and then slowly dissipated, being totally cleared from both fluids by 36 hours. During the initial few hours, plasma Aβ levels fell, but by 27 hours rose to baseline levels in three monkeys given the lowest dose, and continued to rebound slightly higher than baseline over the next two days. The rebound in three other animals given the high dose of MK-0752 took longer but was greater, reaching 150 percent of baseline levels and remaining there for at least seven more days. There was no such rebound in the CSF.

The lowest dose of the GSI substantially blocked Aβ production in the brain, and the highest dose almost completely blocked it, such that within 18 hours, CSF Aβ was halved and almost fully eliminated, respectively. Though the levels eventually crept back up to baseline, they never rose higher.

This acute study suggests that Aβ rebound does not occur in the CSF in rhesus monkeys, and by extrapolation might not in humans, who have similar Aβ synthesis and clearance rates in the brain. What happens after withdrawal of chronically administered drug is not clear, said Bateman. Because of time and cost restraints, he does not plan to investigate this. But he suggested that chronic withdrawal may, in fact, closely mimic reality. “If the concentration of the drug falls below the EC50 before you take the next dose, effectively you are withdrawing every day,” he said. And since GSIs, unless they spare Notch, have known and serious side effects, you don’t want round-the-clock inhibition, said Bateman.

The lack of rebound indicates that APP fragments are not accumulating in the brain when γ-secretase is blocked. That, in turn, suggests the precursor protein is being metabolized by some other pathway. Strong evidence has emerged from in-vitro studies, animal models, and from analysis of human CSF that α-secretase may take up the slack. Researchers led by Erik Portelius and Kaj Blennow at the University of Gothenburg, Sweden, identified small Aβ fragments (Aβ1-14, 1-15, and 1-16) in the CSF of patients enrolled in Ely Lilly’s Phase 2 GSI trial (see ARF related news story). These fragments are consistent with β- followed by α- rather than γ-secretase cleavage of APP. Now Bateman and colleagues report a similar finding, with a two- to threefold jump in synthesis of Aβ1-14, Aβ1-15, and Aβ1-16 in the brain within a few hours of giving MK-0752 to the monkeys. “This is exactly what we saw in humans treated with a GSI,” wrote Blennow’s colleague Henrik Zetterberg, University of Gothenberg, in an e-mail to ARF (see full comment below). In contrast, CSF Aβ1-17, a likely cleavage product of Aβ42, drops in parallel with the latter.

The results suggest that some GSIs, if given at just the right dose, not only block production of Aβ42, but also divert APP processing toward non-amyloidogenic pathways. Slam-dunk for GSI development—eventually?—Tom Fagan.

Reference:
Cook JJ, Wildsmith KR, Gilberto DB, Holahan MA, Kinney GG, Mathers PD, Michener MS, Price EA, Shearman MS, Simon AJ, Wang JX, Wu G, Yaraheski KE, Bateman RJ. Acute gamma-secretase inhibition of nonhuman primate CNS shifts amyloid precursor protein (APP) metabolism from amyloid-beta production to alternative APP fragments without amyloid-beta rebound. J Neurosci. 2010 May 12;30(19):6743-50. Abstract