Long-term use of non-steroidal anti-inflammatory drugs (NSAIDs) may delay or prevent the onset of Alzheimer's disease (AD). A subset of NSAIDs, including flurbiprofen, has been shown to selectively inhibit the production of beta-amyloid(1-42) (Abeta42), independently from their cyclooxygenase (COX) inhibiting activity. We evaluated the in vitro and in vivo profiles of CHF5022 and CHF5074, two flurbiprofen analogues. The in vitro Abeta inhibiting activity was evaluated in a human neuroglioma cell line (H4) carrying the double Swedish mutation (K595N/M596L) of the human amyloid precursor protein (APPsw). The in vitro anti-COX activity was evaluated using human recombinant enzymes isolated from transfected Sf-9 cells. The in vivo pharmacokinetic and pharmacodynamic profiles of the two compounds were evaluated in young APPsw transgenic mice (Tg2576) after oral gavage (100 or 300mgkg(-1) day(-1) for 4-5 days) and after medicated diet (375ppm for 4 weeks). R-Flurbiprofen was used as comparator. In vitro, CHF5022 and CHF5074 were found to be 3- and 7-fold more potent than R-flurbiprofen in inhibiting Abeta42 secretion (IC(50)s of 92, 40 and 268microM, respectively). Differently from R-flurbiprofen, CHF5022 and CHF5074 did not affect COX-1 (at 100microM) and COX-2 (at 300microM) activity. Similarly to R-flurbiprofen, no significant alteration in the expression profile of a subset of Notch intracellular domain-responsive genes was observed with either CHF5022 or CHF5074. In Tg2576 mice, CHF5022 was well tolerated when administered by oral gavage (100mgkg(-1) day(-1) for 5 days) or by medicated diet (56mg kg(-1) day(-1) for 4 weeks). R-Flurbiprofen was poorly tolerated in the diet (32mgkg(-1) day(-1)) with 55% of the animals dying during the first week of treatment. After 4-5 days of oral gavage, CHF5022 and CHF5074 plasma and brain levels at 3h were found to increase with the dose, leading to brain concentrations of about 10% and 5% of the corresponding plasma concentrations, respectively. In animals fed for 4 weeks with compound-supplemented diet, mean plasma (580microM) and brain (20microM) Cyrillic) concentrations of CHF5022 were 8 and 15 times higher than those of R-flurbiprofen. Plasma Abeta42 concentration was dose-dependently decreased by CHF5022 and CHF5074. Brain Abeta levels (formic acid-extractable) were not significantly affected by either compound, although Abeta42 levels tended to inversely correlate (P=0.105) with CHF5022 concentration in the brain. CHF5022 and CHF5074 thus appear to have a promising in vitro and in vivo profile. This warrants further evaluation of their long-term effects on Abeta brain pathology.