Alzheimer's disease (AD) is a degenerative neurological disorder and the most common cause of dementia and disability in the older patient 1 . With no known cure and the currently available treatments only able to temporarily ease symptoms, additional therapeutic options are required. New therapeutic approaches include minimising accumulation of amyloid-beta (Aβ) peptides in the brain 2 3 or targeting cells and signalling pathways implicated in neuronal destruction associated with neuroinflammation 4 5 6 .

Mast cells, which are found on both sides of the blood-brain barrier (BBB) 7 8 9 , release large amounts of proinflammatory mediators and therefore play a prominent role in sustaining the inflammatory network of the central nervous system 10 . Moreover, their ability to regulate the BBB's permeability may also be of therapeutic significance; a defective BBB being a common finding in neuroinflammatory and neurodegenerative diseases, including AD 8 11 12 13 14 . Masitinib mesilate, the investigatory drug of the present study, is a selective tyrosine kinase inhibitor that targets c-Kit, platelet-derived growth factor receptors (PDGFR), and, to a lesser extent, Lyn, Fyn, and the FAK pathway, without inhibiting kinases of known toxicities 15 . By combined targeting of c-Kit and Lyn, masitinib is particularly efficient in controlling the survival, differentiation, and degranulation of mast cells, and thus indirectly controlling the array of proinflammatory and vasoactive mediators the cells can release. Indeed, promising results have been reported from human clinical trials of masitinib in inflammatory diseases such as rheumatoid arthritis and asthma 16 17 .

To investigate the hypothesis that masitinib's targeted inhibitory action on mast cells may reduce the symptoms of AD, its efficacy and safety was assessed as compared with a placebo. Masitinib was administered orally as an adjunct therapy to standard care in patients with mild-to-moderate AD.

Within the limitations inherent to such relatively small phase 2 studies, these results suggest that oral masitinib may have benefits in patients with mild-to-moderate AD. The mechanisms underlying this response remain to be elucidated; as orally administered masitinib is unlikely to have effectively penetrated the BBB, however, we may assume its mechanism of action must be indirect, originating outside the BBB 18 . A growing body of evidence implicates Aβ peptides (predominantly Aβ42) as being the main mediator of neurotoxicity in AD 2 3 . Additionally, neuroinflammation is thought to be a major contributor in the pathogenesis of AD 4 5 6 . Therapies are therefore being sought that reduce Aβ-peptide accumulation and inflammatory response in the brain 19 . Moreover, it has been proposed that blood-borne Aβ peptides could represent a substantial and chronic source of soluble, exogenous Aβ peptides 11 . Although plasma levels of Aβ peptides are about 20-fold lower than cerebrospinal fluid levels, altered BBB function could provide a route for blood-borne Aβ peptides to contribute to AD. The brain is usually protected from this reservoir of Aβ peptides by the BBB; however, there is evidence suggesting that the BBB is defective in AD patients 11 12 13 14 , conceivably allowing an influx of exogenous Aβ peptides and other blood-borne compounds. Therapies to maintain or reinforce the integrity of the BBB could thus be beneficial in AD.

The possible contribution of mast cells in the physiopathology of AD remains a relatively unknown factor 7 20 . Mast cells reside within the brain, where they are constitutively active or can be activated by a wide range of stimuli, including Aβ peptides 20 . It has been shown that mast cells are able to cross the BBB and their numbers may rapidly increase in response to physiological manipulations 7 8 9 . Because mast cells release large amounts of proinflammatory mediators, they play a prominent role in sustaining the inflammatory network 10 . Additionally, perivascular localised mast cells secrete numerous vasoactive molecules that regulate BBB permeability 21 22 . Masitinib is an effective targeted therapy against mast cells, exerting a direct proapoptotic, anti-migratory, and anti-activation action 15 . We therefore propose that the positive response observed from orally administered masitinib is due in part to its inhibitory action of mast cells. In one possible scenario, inhibition of mast cell mediators and apoptosis of mast cells localised at the BBB would effectively reduce BBB permeability, thereby reinforcing its integrity and stemming the accumulation of exogenous Aβ peptides in the brain with a subsequent decrease in plaque formation, inflammatory response and possibly tau hyperphosphorylation (according to the amyloid hypothesis). Additionally, the influx of proinflammatory molecules released from peripheral mast cells would be reduced, as well as Aβ-induced activation of brain mast cells, further decreasing neuroinflammation and migration of mast cells to the brain. Inhibition of mast cells peripheral to the BBB could therefore impact on the main pathological features of AD.

In the event that masitinib could pass through the BBB and accumulate to a sufficiently high therapeutic concentration - for example, via inflammation-induced permeability or compromised BBB - then several direct mechanisms of action are possible. Neuroinflammation could be reduced through direct inhibition of brain mast cells and modulation of microglial activity via disruption of the SCF/c-Kit signalling pathway 15 23 . Damage caused by neurofibrillary tangles or Aβ protein could be reduced via masitinib's targeting of Fyn or the FAK pathway, kinases that have been implicated in the phosphorylation pathway of Tau protein and Aβ-induced cognitive impairment 24 25 26 . It has also been shown that activation of PDGFR, Src, and Rac1 could be relevant for the generation of Aβ by neurons, and that new targets for therapeutic interventions could be found in this pathway 27 ; masitinib's inhibition of PDGFR might therefore possibly inhibit Aβ generation through disruption of this pathway. These mechanisms are only applicable, however, if masitinib crosses into the brain in sufficient concentration, which was not assessed in the present study.

The current study has shown that masitinib administered as an adjunct to standard treatments during 24 weeks may possibly slow the rate of cognitive decline of AD compared with placebo, as evident from the sustained and statistically significant response in ADAS-Cog. Significant improvement in cognitive function and functional capacity compared with placebo was also evident through the mean change in ADAS-Cog, MMSE, and ADCS-ADL values relative to baseline - findings additionally supported by favourable response in the CIBIC-plus and CDR analyses. Such broad benefits are desirable in AD, effectively translating into an improved quality of life. One should note, however, during the 24-week study period that those patients treated with placebo in association with cholinesterase inhibitors and/or memantine experienced an unusually high rate of decline in their status when compared with reported studies 28 29 . This observation may reflect a bias related to the higher age and baseline ADAS-Cog score of the placebo group, with the possible implication that this group would experience a faster cognitive decline resulting in an overestimation of treatment effect. For progressive diseases such as AD, the use of LOCF analysis is inclined to underestimate cognitive decline; however, due to the relatively high patient attrition rate, exasperated by the closure of one centre, it was considered appropriate to retain the patients from this closed centre via LOCF analysis for the week 24 analysis. This again may have tended to overestimation of the treatment effect at week 24, although it should be emphasised that significant treatment response was observed at week 12 for which not data imputation was performed.

Because it is likely that these factors will have resulted in some degree of overestimation to the observed response, a number of complementary analyses were performed. In general, results of the presented ADAS-Cog analysis were supported by alternative sensitivity analyses (see Table S1 in Additional file 1). For example, in the observed cases dataset (without data imputation for the closed study centre), the mean treatment effect at week 24 was similar at 7.2 (compared with 7.6), although the change in ADAS-Cog score relative to baseline no longer reached statistical significance (P = 0.182) between treatment groups. A higher decline of cognitive function, as assessed by the ADAS-Cog responder rate, was also recorded in the placebo arm compared with the masitinib treatment arm at week 24 (60% vs. 11%, respectively; P = 0.095). Considering analysis of the intent-to-treat population by last observation carried forward, the recorded mean treatment effect was less pronounced at week 24 (3.3 at week 12 and 4.0 at week 24); however, an increase (that is, decline in function) was observed in the placebo arm's ADAS-Cog mean scores at both time points, whereas the masitinib treatment arm registered mean decreases (that is, improvement in function). Additionally, to investigate the impact of treatment groups not being comparable at baseline, a multivariate logistic model was constructed (with adjustment on sex, age and ADAS-Cog score at baseline) to test the effect of masitinib on worsening ADAS-Cog score. This model showed that the parameters sex (P = 0.754) and ADAS-Cog score (P = 0.974) had no particular effect, while age (P = 0.232) showed a nonsignificant effect. Overall, a positive - albeit nonsignificant - treatment response was still observed (P = 0.247). Taken together, these complementary analyses suggest the positive treatment response observed is unlikely to be entirely due to baseline or patient withdrawal effects.

While the safety profile in the present study population showed a higher rate of toxicity (approximately 1.7-fold increase) with masitinib as an adjunct therapy compared with standard (placebo) therapy, the majority of AEs reported were mild to moderate and transient, with few severe side effects. The most frequent masitinib-associated AEs were consistent with the known safety profile of tyrosine kinase inhibitors - notably oedema, rash, nausea, vomiting, and diarrhoea, which are generally considered manageable with symptomatic treatments. A comparison of masitinib's safety profile in this study with that of other masitinib phase 2 nononcology studies shows comparable results, indicating that treatment of this older population with masitinib (median age 75.5 years vs. 49 years for all other studies) remains manageable with no indication of additive toxicity when used in combination with cholinesterase inhibitor and/or memantine (Table 4). This comparison of masitinib-related AEs according to population age also revealed that although the current study population experienced lower rates of overall and severe AEs compared with the pooled masitinib population, it had an equivalent rate of AE-related patient premature withdrawal. This discrepancy may reflect an understandably cautious approach to AEs given that this was an older population. The common misperception that tyrosine kinase inhibitors are primarily chemotherapeutic agents now being applied outside their designated field of use may also have been a contributing factor. On this latter point, it is a common misnomer to describe masitinib, and similar tyrosine kinase inhibitors, as a chemotherapeutic agent because - unlike cytotoxic chemotherapies that kill all dividing cells, including healthy cells - masitinib is a targeted therapy. Moreover, depending on which kinases are targeted, tyrosine kinase inhibitors are equally well suited for the treatment of nononcology diseases, as has been previously demonstrated for masitinib in inflammatory and autoimmune diseases with mast cell involvement, such as rheumatoid arthritis 16 , asthma 17 , mastocytosis 30 , and atopic dermatitis 31 , as well as experimental allergic encephalomyelitis, an animal model of brain inflammation.

Masitinib, an oral tyrosine kinase inhibitor with high activity against mast cells, administered as add-on therapy to standard care during 24 weeks showed promising signs of retarding the rate of cognitive decline of AD with an acceptable tolerance profile. Masitinib may therefore represent an innovative avenue of treatment in AD. Confirmatory phase 3 trials are justified to further investigate the long-term efficacy and safety of masitinib as an adjunct therapy with cholinesterase inhibitors and/or memantine for treatment of mild-to-moderate AD.

Aβ: amyloid beta; AD: Alzheimer's disease; ADAS-Cog: Alzheimer's Disease Assessment Scale - cognitive subscale; ADCS-ADL: Alzheimer's Disease Cooperative Study Activities of Daily Living Inventory; AE: adverse event; BBB: blood-brain barrier; CDR: Clinical Dementia Rating; CIBIC-plus: Clinician's Interview-Based Impression of Change-plus caregiver input; LOCF: last observation carried forward; MMSE: Mini-Mental State Examination; PDGFR: platelet-derived growth factor receptors.

The present study was financially supported by AB Science. Masitinib is under clinical development by the study sponsor, AB Science (Paris, France). The sponsor was involved in the study design; data collection, analysis, and interpretation; and manuscript preparation and submission. AM is an employee and shareholder of the study sponsor. OH and PD are consultants and shareholders of the study sponsor. JB and LH-D have received honorarium from AB Science. No other conflicts of interest have been declared.

FP, JB, HV, NS, SP, MV, CM, and JM participated in the study design, enrolled patients, collected data, and were involved in editing the manuscript. LH-D participated in the patient evaluation forms and editing of the manuscript. J-PK, PD, AM, and OH contributed to the study design and editing of the manuscript. All authors critically reviewed and approved the final manuscript.