Finally, LY450139 progressed into phase Ib/IIa human trials despite the red flag associated with it due to the inhibition of Notch cleavage. Several other companies abandoned their compounds that displayed reduction in Notch cleavage.

The dosage was 30-40 mg/d over 6 weeks. One serious event of gastrointestinal bleeding occurred, but this was assigned to Barrett's oesophagus, a condition that develops in some people with chronic gastroesophagal reflux.
The histological changes do not indicate impairment of Notch cleavage. However, the impact of LY450139 on cell growth in the oesophagus and on the acid-resistant cells lining the stomach needs to be addressed further. More importantly for AD: Abeta levels in the CSF were not related to the concentrations of LY450139; plasma Abeta levels were reduced by 38 percent. Unfortunately, there are no data on cognitive function. Obviously the trial size and duration were just too short.

View all comments by Boris Schmidt

The observations by Lanz et al. on time- and concentration-dependent changes of Aβ peptide in plasma, CSF, and brain of guinea pigs treated with the γ-secretase inhibitor LY450139 extend our knowledge of a well-recognized, yet still poorly understood property of this enigmatic enzyme. This property is the elevation of Aβ above baseline following acute (and now also chronic) treatment in-vivo with low doses of inhibitor. This effect is most easily detected in plasma, and somewhat more variably in CSF. Lanz et al. also report elevation of plasma Aβ above baseline at later time points (preceded by reduction below baseline at early time points), following acute treatment with high/inhibitory doses of LY411575. Mass spectrophotometric analysis of secreted Aβ species from H4 cells expressing APPsw revealed differential dose-dependent effects on elevation of Aβ1-40, 1-42, and 11-40 species. Elevation of plasma Aβ has been observed with the structurally related LY411575 in Tg2576 by the same authors (Lanz et al., 2004).

The observations...  Read more

The observations by Lanz et al. on time- and concentration-dependent changes of Aβ peptide in plasma, CSF, and brain of guinea pigs treated with the γ-secretase inhibitor LY450139 extend our knowledge of a well-recognized, yet still poorly understood property of this enigmatic enzyme. This property is the elevation of Aβ above baseline following acute (and now also chronic) treatment in-vivo with low doses of inhibitor. This effect is most easily detected in plasma, and somewhat more variably in CSF. Lanz et al. also report elevation of plasma Aβ above baseline at later time points (preceded by reduction below baseline at early time points), following acute treatment with high/inhibitory doses of LY411575. Mass spectrophotometric analysis of secreted Aβ species from H4 cells expressing APPsw revealed differential dose-dependent effects on elevation of Aβ1-40, 1-42, and 11-40 species. Elevation of plasma Aβ has been observed with the structurally related LY411575 in Tg2576 by the same authors (Lanz et al., 2004).

The observations of Lanz et al. have been noted in humans with LY450139 as well (Siemers et al., 2005), and are relevant to Aβ as a biomarker for γ-secretase inhibitors in the clinic. In addition to the mechanistic basis underlying elevation of Aβ at low concentrations of γ-secretase inhibitor, another question is whether elevation of Aβ is model-specific, or compound-specific. Treatment with low doses of LY411575 did not elevate plasma Aβ in TgCRND8 mice (Wong et al., 2004), suggesting the model may be a factor. A structurally unrelated γ inhibitor, the sulfonamide BMS-299897, did not elevate plasma Aβ at low doses in Tg2567 (Barten et al., 2005), nor in guinea pig (Anderson et al., 2005). The benzodiazepene γ inhibitor, compound E, also did not elevate plasma Aβ in guinea pigs at low doses (Grimwood et al., 2005). Together, the latter data suggest elevation of plasma Aβ may be specific to certain classes of γ inhibitors in specific models.

Effects on CSF Aβ following γ inhibitor treatment have also proven to be both compound- and model-dependent. In the current study, Lanz observed elevation of CSF Aβ under limited conditions in guinea pig CSF with LY450139. The dibenzocaprolactam LY411575 elevated CSF Aβ in Tg2576 (Lanz et al., 2004), but not in rats (Best et al., 2005). The sulfonamide MRK-560 also did not elevate CSF Aβ in rats at low doses (Best et al., 2006). No changes in CSF Aβ in humans treated with LY450139 have been reported (Siemers et al., 2005; Siemers et al., 2006). Elucidating the mechanistic basis for Aβ stimulation at low doses of γ inhibitors will further add to the list of novel catalytic mechanisms discovered from study of this unique enzyme.

View all comments by Guriqbal S Basi

This thorough study from researchers at Pfizer demonstrates that a γ-secretase inhibitor elevates Aβ levels at low doses in a guinea pig model (with endogenous Aβ that is identical to human). This phenomenon has been reported for some other γ-secretase inhibitors. However, the compound in the current investigation is in phase 2 clinical trials by Eli Lilly and is the most advanced γ inhibitor to date, so it is particularly important to know its specific effects. The obvious concern is that low doses in humans may have the opposite of the desired effect and may actually exacerbate AD. Although the elevation of Aβ at low doses of inhibitor are seen most clearly for Aβ40 in the guinea pig plasma, elevation of Aβ42 in the CSF also appears elevated. However, due to problems with variability in measuring Aβ42 in the CSF, it is difficult to say whether these elevations reached statistical significance.

The mechanism of these mysterious and paradoxical increases remains unclear. Nevertheless, this study shows the Aβ elevating effects in cell culture as well, suggesting that the...  Read more

This thorough study from researchers at Pfizer demonstrates that a γ-secretase inhibitor elevates Aβ levels at low doses in a guinea pig model (with endogenous Aβ that is identical to human). This phenomenon has been reported for some other γ-secretase inhibitors. However, the compound in the current investigation is in phase 2 clinical trials by Eli Lilly and is the most advanced γ inhibitor to date, so it is particularly important to know its specific effects. The obvious concern is that low doses in humans may have the opposite of the desired effect and may actually exacerbate AD. Although the elevation of Aβ at low doses of inhibitor are seen most clearly for Aβ40 in the guinea pig plasma, elevation of Aβ42 in the CSF also appears elevated. However, due to problems with variability in measuring Aβ42 in the CSF, it is difficult to say whether these elevations reached statistical significance.

The mechanism of these mysterious and paradoxical increases remains unclear. Nevertheless, this study shows the Aβ elevating effects in cell culture as well, suggesting that the effect is due to direct interaction with the protease complex. Perhaps partial inhibition raises substrate levels so that more Aβ is produced by the remaining (uninhibited) protease complexes. γ-secretase likely has further secrets to reveal before this phenomenon can be truly understood.

View all comments by Michael Wolfe

LY-450139 did not have an easy start. There was always the issue of Notch selectivity. One serious case of gastrointestinal bleeding occured in the phase Ib/IIa trials, but that was assigned to Barrett oesephagus and not to impaired Notch signaling. The phase IIa trials revealed a discrepancy between LY-450139 concentration and Abeta levels. Now, Lanz et al. report a concentration dependent modulatory, not full inhibitory, effect on Abeta production in a guinea pig model, which avoids transgenes and overexpression.

The concentration-dependent mode of action as judged by plasma and CSF levels is reminescent of the Scios inhibitors, the Abbenante tripeptide and other early, unselective peptidomimetics. The critical issues of sampling timepoint and the known Abeta overshoot 6-9 h after administration was overcome by steady-state sampling. The steady-state administration, which will be not be easy to achieve in humans, result in increased plasma Abeta levels at low doses and Abeta reduction at high doses. These data rule out stockpiled substrate under low dosage regime as the...  Read more

LY-450139 did not have an easy start. There was always the issue of Notch selectivity. One serious case of gastrointestinal bleeding occured in the phase Ib/IIa trials, but that was assigned to Barrett oesephagus and not to impaired Notch signaling. The phase IIa trials revealed a discrepancy between LY-450139 concentration and Abeta levels. Now, Lanz et al. report a concentration dependent modulatory, not full inhibitory, effect on Abeta production in a guinea pig model, which avoids transgenes and overexpression.

The concentration-dependent mode of action as judged by plasma and CSF levels is reminescent of the Scios inhibitors, the Abbenante tripeptide and other early, unselective peptidomimetics. The critical issues of sampling timepoint and the known Abeta overshoot 6-9 h after administration was overcome by steady-state sampling. The steady-state administration, which will be not be easy to achieve in humans, result in increased plasma Abeta levels at low doses and Abeta reduction at high doses. These data rule out stockpiled substrate under low dosage regime as the reason for the Abeta overshoot at later time points. The 2 binding site model proposed by Mark Shearman at the Madrid Meeting 2006 may hold the key for this observation: At low dosage, one site is occupied only. This results in modulation; whereas at higher dosage both site are occupied, which results in full inhibition.

View all comments by Boris Schmidt