In this study, we sought to test whether ibuprofen demonstrates any GSM activity in
plasma and CSF in nonhuman primates and humans following acute dosing. Contrary to
previous in vitro studies, a single high dose of ibuprofen did not modulate A? levels
as would be predicted if the compound has ?-secretase modulating activity in vivo
at the dose given acutely. At a high dose, but within the therapeutic dosing range
(i.e., 800 mg IV-ibuprofen), there were no changes in A? levels in human plasma. Even
higher doses in nonhuman primates, 100 mg/kg, failed to demonstrate A?42 reduction
in CSF or plasma. In contrast, a second-generation GSM, GSM-1, demonstrated potent
changes in A?42 and A?38 levels in CSF and plasma in monkeys indicative of ?-secretase
modulation. These latter findings showed that ?-secretase complex can be modulated
in primates in a manner consistent with prototypic GSM activity first defined in vitro
and in rodents.
Epidemiological studies have consistently shown that chronic usage of NSAIDs reduced
the risk of AD and, further, that the protective effects are linked with the duration
of NSAID use 17], 18], 42]. A recent study using a Veterans Administration database reported that ibuprofen,
one of the three most commonly used NSAIDs together with naproxen and indomethacin,
had the strongest protective effect against AD when used for more than 5 years 18]. However, there was no correlation between NSAIDs that had GSM activity versus those
that did not 43]. In a 1-year treatment trial of mild to moderate AD individuals, there was no benefit
of 400 mg ibuprofen given twice daily, but a sub-analysis of apoE4 individuals suggested
slowing in cognitive decline 44]. Other treatment trials with NSAIDs have either been negative or demonstrated equivocal
benefit 45]–48].
Unresolved in all the observational studies are the mechanisms by which NSAID use
may lower the risk for developing AD, if this happens 49]. A number of mechanisms have been proposed but none have been rigorously tested 50]. In this study, we tested whether ibuprofen may act as a GSM and reduce production
of the pathogenic A?42 peptides through this mechanism. In preclinical studies, subacute
treatment with 50 mg/kg ibuprofen for 3 days significantly reduced A?42 but not A?40
levels in the Tg2576 line of APP transgenic mice 16], 51]. Chronic treatment with ibuprofen decreased amyloid load in the brain accompanied
by reduction in all A? species, and rescued memory deficits in the same APP transgenic
mouse line 16], 24], 26], 27], but the mechanisms for these effects were not established. Furthermore, we are not
aware of a study in rodents that examined the changes in A? peptides after acute treatment
with ibuprofen to demonstrate GSM activity, nor has the GSM activity of ibuprofen
been examined directly in primates. Results from this study showed that, contrary
to expectations, a single high dose of ibuprofen did not reduce A?42 levels in monkeys
or humans (Figs. 2 and 4). The maximum average concentrations of total ibuprofen detected from plasma in monkeys
and human subjects were 1060.33 ?M and 441.63 ?M, respectively (Table 2). Although the assay used in this study may be less quantitative than other methods,
such as liquid chromatography-tandem mass spectrometry, the drug levels were within
the range of published reports 37]. Further, we did not measure free versus bound concentration of the drug. However,
because ibuprofen is about 99 % bound to plasma proteins in humans 38], the estimated unbound concentration of ibuprofen was ~4.4 ?M. In mice treated with
50 mg/kg ibuprofen daily for 3 days and reported to demonstrate GSM activity, the
average plasma drug concentration was 43 ?M, 2 h after the last dose 51]. Because the unbound fraction of ibuprofen in rodents has been reported to be ~5.5
% 52], the estimated unbound ibuprofen concentration in that study would be ~2.37 ?M, a
level that is comparable to what we detected in humans. Consequently, the lack of
effect of ibuprofen in humans and monkeys cannot be easily attributable to insufficient
drug exposure, at least if treatment for 3 days in mice is comparable to the single
acute treatment given in the present study. Another potential explanation for the
negative results is that, because of lower efficacy, ibuprofen may require more sustained
treatment, as was done in the APP transgenic mouse line. However, in view of the robust
effects of GSM-1 in monkeys, this reasoning is perhaps less convincing. Therefore,
our results suggested that if subchronic to chronic treatment of ibuprofen were effective
in decreasing brain A? levels in primates, it may not be due to direct ?-secretase
modulation, but possibly acting in combination with other mechanisms, such as lowering
pro-inflammatory cytokines to decrease APP expression and A? accumulation, increasing
anti-inflammatory cytokines to facilitate A? clearance, or anti-aggregation properties
to reduce A? fibril formation 50]. Interestingly, in the 5XFAD mice with aggressive early amyloid pathology, 3 months
of treatment with ibuprofen resulted in an increase rather than decrease in soluble
A? levels and worsening of behavioral performance even in the presence of demonstrated
reduction in inflammatory responses. This study in rodents called into question the
beneficial effects of anti-inflammatory treatments targeting the cyclooxygenase pathway
53]. Taken together, the effects of ibuprofen are complicated. If indeed they are beneficial
in preventing AD in humans, the mechanism of risk reduction may occur through multiple
pathways that have thus far not been adequately tested in preclinical or clinical
settings.
In stark contrast to ibuprofen, the second-generation GSM-1 significantly decreased
A?42 ratios in CSF and plasma in monkeys. GSM-1 is a NSAID-derived carboxylic acid-containing
GSM originally developed by Merck by modulating compound structure to reduce lipophilicity
19]–21], 30]. The drug efficacy in transgenic mice was first reported by Page and colleagues where
a single dose of GSM-1 given at even 3 mg/kg lowered brain A?42 by 25 % and, at 30
mg/kg, brain A?42 was reduced by 70 % in APP transgenic mice. As would be expected,
A?38 levels were concomitantly increased 28]. Consistent with these rodent studies, we observed a 40 % reduction in ratio of A?42/A?40
in plasma 4 h after dosing, whereas CSF ratio of A?42 to total A? was decreased by
35 % 12 h after treatment in monkeys (Fig. 2d,f). At the same time, changes in A?38 levels appeared more rapidly and were more sustained
(Fig. 2e). While the reciprocal A?42 and A?38 changes are one of the “signatures†of NSAID-based
GSMs, it is unclear whether these alterations are mechanistically linked or occur
independently. The uncoupling of the two events had been reported in the setting of
presenilin 1 (PS1) or PS2 mutations 28], 54]. In addition, a recent study reported that a novel GSM decreased both A?42 and A?38
levels while increasing A?37 and A?39 peptides 55]. These results suggested this coupling is not necessarily obligatory.
As expected, the second generation GSM-1 is much more potent than ibuprofen in vitro
with an ED
50
of GSM-1 of approximately 200 nM, compared to 250–500 ?M for ibuprofen (Fig. 1). Although not reported for GSM-1, a GSM with similar structure and comparable in
vitro and in vivo GSM activity, GSM-10 h, demonstrated excellent brain penetration
in an APP transgenic mouse line (TASTPM) 56]. In contrast, as estimated from the CSF to plasma ratio, ibuprofen brain penetration
was about 1 % in monkeys (Table 2), a value consistent with other NSAIDs 39], 40], 51]. Furthermore, NSAIDs such as fenofibrate and flurbiprofen have been reported to bind
to APP, presumably one mechanism of action underlying the NSAID-based GSMs 57]. On the other hand, GSM-1, like other second-generation compounds, binds directly
to the PS1 N-terminal fragment and through this interaction is believed to alter ?-secretase
processing in an allosteric manner 58]–61]. Whether these differences in binding sites confer different efficacy or mode of
action has not been established.