Cholinesterase inhibitors inhibit the activity of the enzyme acetylcholinesterase (AChE), which hydrolyzes the neurotransmitter acetylcholine left over in the synaptic space, producing choline and acetic acid. This inhibition decreases acetylcholine elimination, hence leading to an increased availability of the neurotransmitter in the synaptic space. Although this has been shown to be effective in improving the cognitive function of patients with mild to moderate AD [50], no evidence indicates that these drugs slow progression of mild cognitive impairment (MCI) in non-demented patients, and it is clear that they do not prevent the development of dementia.

Currently, three cholinesterase inhibitors (rivistagmine, donepezil and galantamine) are being used for AD treatment. Rivastigmine selectively inhibits both cortical AChE in the central nervous system and butyrylcholinesterase (BuChE), which is documented as the predominant cholinesterase in many key regions affected in AD, including the hippocampus, thalamic nuclei and the amygdala [51]. Donepezil is a reversible and highly selective inhibitor of AChE which has also been found to be effective in treating cognitive impairment in patients with mild to moderate AD [52]. Indeed, Birks and Harvey [53] reported that a 10 mg/day treatment of donepezil during 52 weeks improved cognitive function, daily living activities performance and behavior in AD patients with mild, moderate or severe dementia. Finally, galantamine is characterized by two pharmacological mechanisms that involve the inhibition of AChE and the binding to nicotinic acetylcholine receptors in order to modulate allosterically the actions of ligands [54]. It also has very little activity in inhibiting BuChE. The clinical benefits of galantamine in AD patients have been recently demonstrated by Xu and colleagues [55], who reported a significant beneficial effect on cognitive improvement evaluated with the Mini-Mental State Examination. This improvement on cognitive decline, although modest, was persistent. Since the authors confirmed the usefulness of galantamine in the treatment of AD, studies assessing a combinatory therapy of galantamine with other drugs would be interesting.

Learning and memory processes involve long-term potentiation (LTP), a persistent and rapid increase in synaptic transmission mediated by the neurotransmitter glutamate through the NMDA receptor [56]. Indeed, NMDA receptors are abundant in the pyramidal cells of the hippocampus and cortex (areas involved in cognition, learning, and memory). However, high glutamate levels are associated with neurotoxicity through the activation of the NMDA GluN2B extrasynaptic receptors, leading to long-term depression (LTD), spine shrinkage and synaptic loss through a mechanism known as excitotoxicity [57]. In addition, these receptors also induce an increase in intracellular calcium and mitochondrial alterations, as well as an increase in reactive oxygen species (ROS) and nitric oxide (NO) production, which also contribute to neuronal cell death [58].

Memantine is a non-competitive, moderate-affinity NMDA receptor antagonist thought to decrease glutamate-induced excitotoxicity while allowing the physiological actions of glutamate on learning and memory [59]. Nevertheless, a study conducted by Peters and colleagues [60] reported that a combined treatment of galantamine plus memantine did not offer additional cognitive or functional advantages in patients with mild-to-moderate AD patients as compared with MCI patients who received galantamine alone. However, both drugs should not be discarded in a future disease-modifying strategy for AD.

The “amyloid cascade hypothesis” is, probably, the most accepted pathophysiological hypothesis in AD. As already explained, it proposes that an altered cleavage of APP by $\beta$-secretases (BACE1) and $\gamma$-secretases generate A$\beta$ soluble neurotoxic oligomers (A$\beta$Os) capable of producing fibrils that are deposited in the brain as A$\beta$ plaques. In the past, plaque formation was thought to be the cause of neuronal death, but currently is well known that A$\beta$Os are responsible of neurotoxicity through different mechanisms, including excitotoxicity, oxidative stress, mitochondrial dysfunction, and alteration of synapses [63]. Hence, it has been shown that A$\beta$Os bind to NMDA receptors, generating excitotoxicity both in preclinical models and AD patients [64, 65]. Likewise, A$\beta$Os interact, directly or indirectly, with a wide range of presynaptic and postsynaptic neuronal receptors, including ionotropic (iGluR) and metabotropic (mGluR) glutamate receptors, the cellular prion protein (PrPc), neuroligines, neurexins and insulin receptors, among others [63]. For this reason, the amyloid cascade hypothesis is more or less related to most of the other hypotheses trying to explain the pathophysiology of AD. In this way, the tau hypothesis states that the formation of A$\beta$Os leads to the activation of kinases leading to tau hyperphosphorylation and its polymerization into insoluble NFTs [66]. Moreover, the inflammatory hypothesis proposes that the formation of A$\beta$Os is followed by the activation of microglia and the neuroinflammatory response, contributing to neurotoxicity [67].

Based on the amyloidogenic hypothesis of AD, the elimination of oligomers (soluble) and plaques (insoluble) with monoclonal antibodies could be able to decrease the progression of the disease [18]. Monoclonal antibodies develop an immune response against these A$\beta$ peptides causing an increase in their clearance. Indeed, A$\beta$ peptides are a common target of phase II and III drug development programs, and some monoclonal antibodies against oligomers, plaques or protofibrils have been widely studied [19]. However, until 2020–2021, most drugs aiming to promote amyloid clearance failed to improve cognitive performance out rightly, irrespective of whether they were monoclonal antibodies or vaccines. In this sense, aducanumab, a monoclonal antibody targeting A$\beta$ protofibrils, was a turning point. Based on the results of two-phase III clinical trials (EMERGE and ENGAGE), the FDA approved its use for AD treatment in 2021. Notwithstanding, these two phase III clinical trials were stopped prematurely in 2019, due to the low efficacy of the drug. While one of the trials showed a positive trend, the other showed no clinical benefit [68, 69]. Subsequently, it was observed that participants who received sufficiently high doses of aducanumab showed a therapeutic improvement in both trials. Accordingly, the researchers suggested that aducanumab was effective in decreasing dementia at a clinical level (Fig. 1). Nevertheless, and despite its FDA approval, much controversy exists about the efficacy of aducanumab in the treatment of AD, as well as its pricing. Indeed, it has been criticized that the approval of aducanumab is based more on speculation than on contrasted clinical data demonstrating a real efficacy. In this sense, the European Medicines Agency (EMA) [70] recommends the denegation of a marketing authorization, due to many doubts about its efficacy and safety. Moreover, it is unclear how aducanumab affects the different types of amyloid oligomers that could be responsible for cognitive loss. In addition, after years of research, all drugs targeting the amyloid cascade have failed in clinical studies, suggesting that amyloid reduction does not lead to a significant clinical benefit in AD. For instance, solanezumab and bapineuzumab showed no clinical benefit in phase III trials, resulting in the end of their development programs.

Fig. 1.

Mechanisms proposed to explain the efficacy of different drugs in the treatment of Alzheimer’s disease. Pathophysiological mechanisms are depicted in red, whereas the specific therapeutic agents are depicted in green.

In turn, donanemab, which targets A$\beta$ plaques, reported both a decrease in A$\beta$ aggregates and a reduction in the rate of cognitive and functional decline in a phase II study [71]; while a larger phase III trial is underway (NCT # 04437511). Another monoclonal antibody, lecanemab (which targets A$\beta$ oligomer protofibrils) is also in late-stage assays (NCT # 03887455). Swanson and colleagues reported that lecanemab treatment resulted in an effective dose-dependent reduction in clinical deterioration when compared to placebo [20]. In addition, after 18 months of treatment, they found a dose-dependent reduction in brain amyloid (assessed by PET). Taken together, the findings of this double-blind trial across multiple cognitive endpoints and biomarkers support a therapeutic effect for targeting specific oligomeric species (protofibrils) in the process of pathophysiologic amyloid generation in AD.

A$\beta$ peptides are able to bind to serum albumin and cross the blood-brain barrier (BBB) [72]. In this sense, it has been proposed that replacing serum albumin in plasma could be a suitable therapy for AD, since it would allow to remove A$\beta$ from the central nervous system (CNS). Moreover, the antioxidant properties of albumin could provide additional therapeutic benefits. Following this line of thought, combining plasma exchange with intravenous immunoglobulin infusion (which also binds A$\beta$) could further increase the amyloid clearance from the brain. In this regard, Gamunex® is a human intravenous immunoglobulin (IVIG) that is administered concomitantly with human albumin. A phase III multicenter CT (AMBAR) examined the effects of human albumin infusion plasmapheresis combined with IVIG in patients with mild-moderate AD [21]. 347 patients were randomized into three treatment arms and a control group. This study showed that plasma exchange could slow down cognitive and functional decline in AD. However, more studies are needed to clarify the positive effects of this therapy.

On another front, the receptor for advanced glycation end products (RAGE) not only binds and transports A$\beta$ peptides from the blood to the brain, but also induces neuroinflammation [73]. For this reason, RAGE has emerged as a potential target for AD treatment, and drugs blocking the A$\beta$-RAGE interaction could show numerous therapeutic actions, including the reduction of neuroinflammation and A$\beta$ levels in the brain and the decrease of cognitive impairment. Azeliragon is a RAGE antagonist that decreases A$\beta$ transport into the brain, thus preventing the toxic effects of oligomers and reducing the neuroinflammatory effects of glial cells [22]. In a phase III CT, azeliragon did not improve functional outcomes, so it was terminated [22]. Subsequently, a phase II CT comparing azeliragon with placebo was performed in patients with AD and T2DM during six months. In December 2020, vTv Therapeutics Inc. announced that the study did not meet the primary efficacy endpoint. The company will continue to analyze the data to determine if there are potential benefits or future applications of this drug in AD and/or other diseases [44].

As already mentioned, the fact that most drugs aiming to promote amyloid clearance have failed, irrespective of whether they are monoclonal antibodies or vaccines, challenges the amyloid cascade hypothesis. Moreover, some studies show that patients treated with these drugs eliminate amyloid from the brain but do not improve cognitively [74, 75]. For that matter, a more reconciling A$\beta$ hypothesis suggests that AD has at least two stages, a first one in which A$\beta$ accumulates over time and may be a key factor involved in AD onset, and a second stage, with an ongoing AD where amyloid loses importance and where tau and glia activation are key pathophysiological mechanisms. However, other factors should not be ruled out. Indeed, according to synaptic deficiency hypothesis of AD, A$\beta$ and tau would be secondary to cell injury and the role of neurotrophins in synaptic maintenance and neuronal survival, hence attributing to anti-amyloid treatments a very limited therapeutic efficacy. In any case, several other therapies are in various stages of clinical development, including drugs directed against tau accumulation or dissemination.

Tau is a microtubule-associated protein that can be found both in the CNS and at the peripheral level. In a physiological context, tau exists as a soluble protein that, apart from promoting correct assembly for the stabilization of the microtubular structure, also plays a role in the balance of axonal transport and synapses [76]. Tau is regulated both by normal homeostatic responses and by stress responses, through a series of post-translational modifications such as glycosylation, ubiquitination, glycation, nitration, oxidation and phosphorylation, the last one being the most important one [77]. Tau phosphorylation is regulated by several kinases, including glycogen synthase kinase 3 (GSK3$\beta$) and extracellular cyclin 5-dependent kinase (CDK5). As we have explained, tau protein is hyperphosphorylated in AD patients, which favors and leads to fibrillation and aggregation of the protein, forming helical filaments that make up intracellular NFTs that destabilize the neuronal cytoskeleton. In neurodegenerative diseases, tau is a biomarker that migrates to the extracellular compartment and increases its concentration in the cerebrospinal fluid (CSF). The tangles appear initially in the entorhinal cortex of the temporal lobe, spreading to limbic areas such as the hippocampus and finally affecting large areas of the neocortex [78]. The histopathological diagnosis of AD includes the assessment of the quantity and location of NFTs, its number and density being correlated with the intensity of the dementia [23]. Although GSK3$\beta$ and CDK5 are the most important kinases responsible for tau hyperphosphorylation, other kinases such as Protein Kinase C, Protein Kinase A, the serine/threonine kinase ERK2, caspase 3 and caspase 9, which can be activated by A$\beta$, also have a prominent role [79]. Beyond tau hyperphosphorylation and formation of NFTs, a “tau propagation hypothesis” has been proposed, stating that poorly folded fibrillar tau aggregates can propagate through cells in a similar way to prions [80].

Given that the severity of tau pathology is more correlated with the progression of cognitive impairment than A$\beta$ [81], multiple studies trying to ascertain tau modulation, (de) phosphorylation and hyperphosphorylation processes in AD have been carried out in order to find suitable treatments. For instance, preclinical studies have reported that sodium selenite, a PP2A activator, restores synaptic plasticity and improves learning and memory in Tg mice with an advanced stage of tauopathy [82]. Focusing on clinical studies, GSK3$\beta$ inhibitors such as tideglusib (ClinicalTrials.gov Identifier: NCT01350362) and lithium have been tested [24] (Fig. 1). Specifically, Forlenza and colleagues reported that a long-term lithium treatment in subtherapeutic doses (lithium carbonate was prescribed to concentrations 0.25–0.5 mEq/L), can be safe and well tolerated by patients, which supports its potential administration in patients with cognitive deficits (trial registration at clinicaltrials.gov: NCT01055392) [25].

In turn, the second-generation tau aggregation inhibitor LMTX (TRx0237) has been extensively investigated and it is the only drug in this group that reached clinical phase III. Hence, two clinical phase III trials showed that a LMTX monotherapy seems to slow cognitive decline in patients with mild to moderate AD (NCT01689246, NCT01689233). Consequently, additional clinical phase III studies are underway (NCT03446001) [26].

In addition to these strategies, relatively new approaches based on tau have also been investigated; however, some of them did not show results in clinical trials and have been discontinued. These include inhibition of acetylation and deglucosylation of tau (MK-8719 and ASN120290, respectively), monoclonal antibodies against tau (Gosuranemab or BIIB092, Tilavonemab or ABBV-8E12) and reduction of tau levels by an antisense oligonucleotide (ASO) (IONIS-MAPTRx) [27].

The notion that inflammation and the immune responses play a fundamental role in the pathophysiology of AD is supported by many studies [83]. Much research has been focused on the involvement of astrocytes, microglia and CNS mast cells, but also on the impact of systemic inflammation caused by intestinal and gut microbiota. These have led to the development of different drugs targeting the inflammatory process that could result in neurodegeneration, including tyrosine kinase inhibitors that modulate mast cells (masitinib), monoclonal antibodies that regulate microglial activity as daratumumab (against CD38), AL002 (against TREM2) or AL003 (against SIGLEC-3), and curcumin, which also has antioxidant properties. The use of montelukast (a leukotriene receptor antagonist) and intestinal microbiota transplants, among others, are also being studied.

Microglia are CNS-resident phagocytes that play a vital role in maintaining neuronal plasticity and synapse remodeling [84, 85]. They have been shown to be involved in the maintenance of neural networks by releasing neurotrophic factors such as BDNF, but also contribute to brain homeostasis by exerting synaptic elimination [86]. Microglia can be activated by the accumulation of proteins that act as a pathological trigger, producing the migration of the microglial cells and the onset of an innate immune response. Indeed, microglia actively participate as supportive cells by engulfing A$\beta$ in the AD brain, which may favor a cognitive improvement [87]. The A$\beta$ peptide can trigger the neuroinflammatory process through different microglial receptors, including toll-like receptors (TLRs), receptors for advanced glycation end products, pentraxins, and the complement cascade, among others [88]. One of the most studied receptors involved in this process are TLRs, the TLR2 and TLR4 subtypes being considered critical in the microglial recognition of A$\beta$ [89, 90]. Downstream TLR signaling through NF-$\kappa$B pathways, activator protein 1, and interferon regulatory factor (IRF) lead to transcription of pro-inflammatory genes and secretion of pro-inflammatory cytokines and chemokines. After binding to receptors, microglia cells also endocyte A$\beta$ oligomers and destroy them by endolysosomal degradation, where microglial proteases such as neprilysin and insulin degrading enzyme (IDE) play an important role [91]. However, in severe AD, microglial clearance of A$\beta$ is inefficient due to the increased concentrations of cytokines downregulating the expression of A$\beta$ phagocytosis receptors, such as Myeloid Cell 2 Activation Receptor (TREM2). TREM2 is surface receptor of the Ig superfamily, widely expressed in microglia and involved in the mediation of phagocytic clearance of neuronal remains [92]. It has been shown that a rare mutation in TREM2 (R47H) leads to the inability of the receptors to remove A$\beta$ from the CNS, hence increasing the risk of developing AD [93]. In light of all these findings, microglia constitute an interesting therapeutic target against AD, either because a microglial overstimulation can increase synaptic elimination and lead to cognitive loss, or because increasing TREM2 activity could prevent A$\beta$ accumulation.

On another front, mast cells (MCs) have an important physiological function and are involved in immunity and inflammation, especially in allergic inflammation. Upon activation, MCs increase the synthesis and secretion of many inflammatory mediators, such as TNF-$\alpha$, serotonin, histamine and heparin, among others [94]. As key components of the immune system, the pathophysiological changes of these cells affect multiple organs. In the CNS, the contribution of MCs to neuroinflammation is beyond doubt, but less is known about their role in neurodegenerative diseases such as AD. In this sense, it has been suggested that MCs could favor alterations in the BBB and interact with microglia. Regarding the latter, some studies have shown that mast cell-glia interaction can occur in different ways [95, 96, 97]. Hence, MCs degranulation and the release of histamine and TNF-$\alpha$ activate microglia [96]. Moreover, during inflammation, ATP derived from MCs activates purinergic P2 receptors in the microglia leading to IL-33 secretion [98]. In turn, IL-33 causes MCs to release IL-6, IL-13 and the monocyte chemoattractant protein-1 (MCP-1), which also regulates microglial activity. In addition, the secretion of tryptase from MCs stimulates microglial release of TNF-$\alpha$, IL-6 and reactive oxygen species (ROS) through the activation of the protease-activated receptor-2 (PAR-2) [99]. Interestingly, PAR-2 also upregulates the expression of microglial ATP-sensitive ionotropic P2X4 receptors which, upon ATP binding, lead to the secretion of BDNF, a key mediator in synaptic plasticity [100].

Masitinib is currently one of the most promising research drugs targeting inflammatory diseases such as rheumatoid arthritis, intestinal disease, asthma and mastocytosis. However, a potential new application of masitinib in neurodegenerative diseases, such as AD and multiple sclerosis (MS), has emerged. Masitinib inhibits c-kit tyrosine kinase, a surface receptor expressed by MCs and many other cells, which plays a prominent role as a regulator of the migration of neuronal stem and progenitor cells to areas of brain injury [101]. Hence, masitinib is thought to exert a neuroprotective effect through its activity on MCs and other non-neuronal cells of the CNS, with a subsequent modulation of inflammatory and neurodegenerative processes. It is currently in the phase 3 of a clinical trial in patients with mild to moderate AD (ClinicalTrials.gov Identifier: NCT01872598). Masitinib is administered as adjunctive therapy to patients who have been treated for a minimum of 6 months with a stable dose of a cholinesterase inhibitor (donepezil, rivastigmine or galantamine) and/or memantine. So far, the company has reported some positive results from this study. Masitinib is also capable of blocking the Src TK Fyn family in a nanomolar range. Fyn is a cytoplasmic tyrosine kinase (TK) belonging to the Src kinase (SFK) family involved in multiple CNS transduction pathways, including synaptic transmission, myelination, axon guidance, and formation of oligodendrocytes [102]. Shirazi and Wood reported that that Fyn is upregulated in the brain of AD patients and demonstrated the presence of a Fyn phosphorylation site in the tau-matched helical filament, supporting a role of Fyn in the neuropathogenesis of AD [103]. Therefore, treatment with masitinib may provide two benefits in the pathology of AD: (i) a reduction in neuroinflammation by modulating the mast cell-glia axis and (ii) a cognitive enhancement through a Fyn inhibition.

After many failures with monotreatments, the combination of cromolyn and ibuprofen (ALZT-OP1) has been proposed as a suitable treatment for LOAD [28]. Cromolyn acts as a modulator of MCs and microglia, as well as an A$\beta$oligomerization inhibitor [104]. In turn, ibuprofen is an anti-inflammatory drug with interesting effects such as $\alpha$-secretase activation -thus modulating A$\beta$ production-, microglia inhibition, activation of PPAR $\gamma$ and modulation of synapses through the inhibition of Rho family GTPases [105, 106, 107].

XPro1595 is a second-generation inhibitor of tumor necrosis factor (TNF$\alpha$), an inflammatory factor implicated in AD pathology [108]. XPro1595 selectively neutralizes soluble TNF (sTNF) and inactivates it. Previous preclinical studies reported that targeting of sTNF/TNFR1 signaling with XPro1595 decreased brain alterations in immune cell populations associated with the neuroinflammatory process in 5xFAD (Tg) mouse model [109]. In addition, XPro1595 treatment rescued impaired LTP and also decreased the production or accumulation of amyloid. Furthermore, XPro1595 has a potential application in inflammatory states associated to obesogenic diets and T2DM. Hence, XPro1595 prevents the development of insulin resistance and risk for AD [110].

From a different point of view, some studies have proposed an association between intestinal microbiota and AD, in which dysbiosis cause neuroinflammation induced by activation of the microglia [111]. In this regard, one of the latest interesting drugs is sodium oligomannate (GV-971), a derivative of brown marine algae based on a mixture of acidic linear oligosaccharides ranging from dimers to decamers [112]. It was developed by Green Valley for reconditioning the intestinal microbiota and treating mild-moderate AD. A preclinical study showed that oral administration of GV-971 for one month markedly altered the composition of the microbiota and reduced the concentrations of phenylalanine and isoleucine in a murine model of AD [113]. These was paralleled by a reduction in microglial activation, as well as in Th1 responses and cytokines in the brain. Simultaneously, oligomannate treatment reduced A$\beta$ deposition and tau phosphorylation. These findings led to a study of GV-971 in patients with mild-moderate AD in a randomized, multicenter, double blind, placebo-controlled phase II clinical trial in China [29, 114]. The objective of this trial was to establish the optimal dose, safety and efficacy of GV-971 capsules. The authors found a cognitive improvement assessed by changes in the ADAS-Cog-12 (Alzheimer’s Disease Assessment Scale) score from baseline to week 24 in patients treated with the drug. GV-971 also demonstrated significant and sustained efficacy in enhancing cognition in his first phase III trial [29]. The effects were more pronounced at the end point and in those patients with the most severe cognitive impairment. In 2019, GV-971 obtained conditional approval in China to treat AD patients; however, additional trials are needed to investigate the full mechanism of action of sodium oligomannate. In 2020, the FDA approved the request for a global phase III trial (Green Memory) to evaluate the safety and efficacy of GV-971 in additional populations (United States, Europe, and Asia). If the results of this study are in the same line that those seen in phase III in China, GV-971 probably could receive a worldwide regulatory approval by the FDA and the EMA.

Beyond the relation between intestinal microbiota and AD, recent studies also support an association between periodontitis, a chronic inflammatory oral disease, and neurodegeneration [115]. It is known that there is a relationship between host immune responses and pathogenic burden of microbial biofilm [116, 117], which is made up of several microorganisms, such as Porphyromonas gingivalis, and their toxic products, such as fimbrins, gingipain and lipopolysaccharides (LPS). Both periodontopathogenic bacteria and their products can enter the bloodstream, promoting the expression of inflammatory mediators that can damage other organs, including the brain. Interestingly, gingipains are proteases that generate pathogenicity factors such as Arg-gingipain (Rgp) and Lys-gingipain (Kgp), which can damage tau [30, 118]. In this regard, atuzaginstat (COR388) is an inhibitor of gingipains, which is being tested in a phase II/III clinical study for AD treatment (NCT03823404) (Fig. 1).

It is widely known that glucose is the main source of energy of the mammalian brain, and several pathologies of the CNS are a consequence of disturbed central or peripheral glucose energy metabolism [119]. In this sense, impaired glucose metabolism has been shown to be related with AD pathophysiology. Hence, alterations in brain glucose uptake have been described in patients with initial symptoms that precede the development of AD [120]. Moreover, it seems that insulin improves glucose uptake in the brain by increasing the activation of hippocampal insulin receptor, which plays a key role in synaptic plasticity, learning and memory [121]. Therefore, the rescue of cerebral insulin signaling and glucose metabolism constitutes a tempting goal to treat neurodegeneration.

A priori, many drugs used to treat T2DM could be recycled to also treat AD, including 116 (glucagon-like peptide 1 receptor agonist), dapaglifozine (SGLT2 inhibitor) and metformin, among others. Regarding the latter, metformin is currently the first-line treatment for T2DM, widely prescribed due to its safeness and virtual absence of side effects. Preclinical results in murine models of AD showed that metformin delayed the progression of cognitive impairment [122]. In turn, Luchsinger and colleagues reported that metformin treatment significantly improves the total recall in the selective reminding test, but not the Alzheimer Disease Assessment Scale-Cognition (ADAS-Cog), in a study with patients with amnestic MCI (ClinicalTrials.gov Identifier: NCT00620191). The authors concluded that additional clinical studies with a larger number of patients are necessary to demonstrate a greater efficacy of the drug. Other studies evaluating the efficacy of metformin on AD biomarkers and cognitive ability (ClinicalTrials.gov Identifier: NCT01965756) and for dementia prevention (NCT04098666) have been conducted. The results of the study NCT01965756 showed that metformin improved executive function and there was a trend in improving the learning process, memory and attention [31]. Regarding the study NCT04098666, no results have yet been published.

NE3107 is another drug which also reached the phase III (Clinicaltrials.gov: NCT04669028). Previous studies have shown that this drug has a dual therapeutic effect, presenting both neuroinflammatory and antidiabetic properties [32]. Hence, NE3107 has been reported to inhibit the inflammatory ERK pathway and to the improve insulin signaling. This dual effect makes this drug interesting in the treatment of AD (Fig. 1).

On a different note, intranasal administration of insulin glulisine is also being evaluated with the aim to increase insulin signaling in the brain [123]. Promising results have been published regarding intranasal insulin administration in patients with amnestic MCI and AD (ClinicalTrials.gov Identifier: NCT01767909). However, more studies are needed to better characterize the neuroprotective effect of insulin and brain insulin receptor stimulation in neurodegeneration.

An association between cognitive impairment in non-demented individuals and cardiovascular risk factors has been reported [124]. Hence, the inflammatory process associated with aging, atherosclerotic cardiovascular diseases and dementia could share common molecular mechanisms. In this sense, the control of cardiovascular risk factors could be an appropriate strategy to reduce or prevent the incidence of dementia [125].

Following this line of thought, drugs acting on cardiovascular and cerebrovascular dysfunctions, BBB and neurovascular unit, hypertension, atherosclerosis, amyloid cerebral angiopathy and lymphatic/glyphic system dysfunction could be useful for AD treatment. For instance, losartan, an antihypertensive drug, reduced plasma and brain A$\beta$1-42 levels in a murine model of AD [126, 127]. In this sense, different antihypertensive drugs that could be used against neurodegeneration are currently under study, including ARAII drugs (candesartan, telmisartan) and combination treatments such as telmisartan + perindopril (ARAII + ACE inhibitors), and losartan + amlodipine + atorvastatin + exercise (ARAII + Ca${}^{2+}$ blocker + an anti-cholesterol agent) (NCT02913664) [33, 128, 129, 130]. These drugs can modulate APP/A$\beta$ metabolism, hence preserving cognitive function in addition to improve cerebrovascular function.

Previous studies have reported a dysregulation of synaptic functions in AD [131, 132]. In fact, synaptic loss and dysfunction is strongly correlated with the cognitive decline observed in AD patients [133]. Indeed, surviving neurons in AD’s neurodegenerative process have been shown to lose synapses, and synaptic dysfunction has been shown to precede amyloid plaque deposition, as LTP impairment is present in early stages in the hippocampus of AD mice [134, 135, 136]. Hence, the loss of synaptic homeostasis or the integrity of the neural network would precede neuronal death and be key to AD development, an idea that fits with the proposed theories and clinical manifestations. Consequently, memory deficits in AD may even begin two decades before the first symptoms appear. Synapse degeneration is thought to begin in dendritic spines and, specifically, with a decrease in the number of molecules that regulate spinal signaling.

It has been shown that glutamate is involved in the development of dendritic spines [137]. On this basis, some drugs targeting neurotransmitters and mechanisms of neurogenesis are being studied (Fig. 1). An example is troriluzole, also known as BHV4157 (proriluzole), which is a prodrug of riluzole. Riluzole inhibits voltage-gated sodium channels and reduces synaptic glutamate by increasing its uptake and inhibiting its release [138]. Indeed, riluzole reduces synaptic glutamate levels by increasing the expression and function of glial glutamate transporters responsible for synaptic glutamate clearance [34]. Its efficacy in the treatment of AD is currently being evaluated in phase 2–3 clinical trials (Clinicaltrials.gov NCT03605667), although so far, the results do not seem to be encouraging.

In turn, blarcamesine (ANAVEX2-73) is an agonist of the Sigma-1 receptor (S1R) and modulates cholinergic muscarinic receptors in mice [135, 139, 140]. It has been reported that an oral dose (30 or 50 mg) of blarcamesine, followed by an intravenous (IV) dose (3 or 5 mg) in a second period have suitable safety and tolerability in patients with mild-to-moderate AD in a Phase IIa clinical study [35]. Furthermore, phase IIb/III clinical studies consisting in 48 weeks of daily treatment with blarcamesine or placebo, and primary outcomes of ADAS-Cog and ADCS-ADL (Activities of Daily Living) evaluations are being conducted (ClinicalTrials.gov Identifier: NCT03790709). Additionally, in preclinical models of Rett syndrome (RTT), a neurodevelopmental disorder associated with increased risk of cognitive impairment, blarcamesin improved calcium homeostasis, which favors an improvement in mitochondrial and synaptic functions in all brain regions [135]. Besides that, the anticonvulsant levetiracetam has also shown potential as synapse modulator against neurodegeneration. Levetiracetam is an antiepileptic drug that binds to SV glycoprotein 2A (SV2A), a constituent of synaptic vesicle membranes at presynaptic terminals, involved in vesicle trafficking and exocytosis [141]. SV2A is expressed in excitatory and inhibitory synapses in the brain, including the hippocampus, and alterations in this protein have been associated with AD [142, 143]. Interestingly, Rao and Savas [36] reported that levetiracetam lowers A$\beta$42 levels in APP knock-in mice models of amyloid pathology by normalizing the abundance of presynaptic endocytosis machinery, hence favoring a change in the processing of amyloid precursor protein towards the non-amyloidogenic pathway. Other study showed that a treatment with a low dose of levetiracetam (125 mg twice daily), significantly improved memory performance [36].

Currently, a phase II-III study is evaluating the efficacy of AGB101, a proprietary extended-release formulation of levetiracetam, on slowing cognitive and functional impairment in patients with MCI (ClinicalTrials.gov Identifier: NCT03486938). No results have been reported yet.

The fact that diet quality and composition play a role in virtually all health conditions, affecting incidence, complications, management, recovery and quality of life is well supported by mounting evidence [144]. AD is not an exception, and currently several supplements and dietary strategies claim to promote cognitive enhancement. However, to date, no evidence-based product on the market has demonstrated a clear capacity to prevent or slow AD’s progression.

Omega-3 polyunsaturated Fatty Acids ($\omega$3 PUFAs) are among the nutritional and dietary factors that have shown most consistent positive research findings to prevent cognitive decline in older adults [145]. Linoleic and $\alpha$-linolenic acid are examples of $\omega$3 PUFAs which are obtained directly through diet or supplementation. Long-chain $\omega$3 PUFAs (eicosapentaenoic acid and docosahexaenoic acid), have been investigated in different studies as a therapeutic strategy to mitigate cognitive loss, and progression to AD [146, 147]. These compounds can play an important role in improving cognitive impairment through multiple mechanisms such as modulation of the inflammatory process, or acting on genes that regulate the retinoic acid receptor and peroxisome proliferator-activated receptor, as well as the rigidity of the cell membrane [148]. In this context, icosapent ethyl (IPE), a purified form of the omega-3 fatty acid eicosapentaenoic acid (EPA), is under investigation in a clinical study (ClinicalTrials.gov Identifier: NCT02719327). The aim of the research is to evaluate the efficacy of 18 months of IPE therapy in 150 cognitively healthy veterans aged 50 to 75 with increased risk of developing AD due to a parental history of the disease and increased prevalence of the APOE4. The evaluation will be carried out through magnetic resonance imaging (MRI) and the evaluation of cognitive biomarkers for AD in CSF. The study is projected to be finished in early 2023. In the same line, a phase III clinical study (NCT00440050) determined whether chronic supplementation with docosahexaenoic acid (DHA) was capable of slowing the progression of cognitive and functional decline in mild to moderate AD. However, the results showed no differences between DHA supplementation and placebo groups [37].