BACE1 Inhibitors for Alzheimer’s Disease: Current Challenges and Future Perspectives

ALZHEIMER’S DISEASE-MODIFYING THERAPIES

Alzheimer’s disease (AD) is the leading cause of dementia. It is characterized by the accumulation of extracellular neuritic plaques, mainly composed by amyloid-β peptide (Aβ), and by intracellular neurofibrillary tangles comprising hyperphosphorylated tau. As the disease advances, a progressive neuronal loss, particularly in the cerebral cortex and in the hippocampus, leads to brain atrophy and cognitive impairment [1]. The amyloid cascade hypothesis has been the leading model of AD pathogenesis since it was proposed by Hardy and Higgins in 1992. The original hypothesis postulates that large insoluble Aβ fibrils deposits in the brain are the initiating event in AD pathogenesis, resulting in subsequent neurofibrillary tangle formation, neuronal loss, and cognitive decline [2]. This hypothesis has been revised over time [3, 4] since several in vitro and in vivo studies have demonstrated the neurotoxic effect of Aβ oligomers by themselves [5–7].

The Aβ peptide is produced by the sequential cleavage of amyloid-β protein precursor (AβPP) by β-site AβPP cleaving enzyme (BACE1) and γ-secretase. AβPP cleavage by BACE1 releases a soluble extracellular fragment called sAβPPβ whereas a 99-residue C-terminal fragment designated CTFβ or C99 remains bound to the membrane. C99 is further cleaved by γ-secretase, leading to the release of the Aβ peptide, which may have different lengths [4, 8]. Although Aβ₄₀ is the most abundant species, Aβ₄₂ and longer Aβ species have a greater propensity for aggregation. Extracellular accumulation of Aβ triggers its aggregation, progressing from monomers, oligomers, and protofibrils and eventually forming insoluble plaques [1, 8]. The causative role of Aβ accumulation in AD pathogenesis has been supported by genetic and non-genetic studies that implicate Aβ aggregates in leading disease progression in both familial and sporadic forms of the disease [9, 10].

Disease-modifying therapies (DMTs) for AD are based on interventions that act on the main drivers of the pathogenesis thus allowing to alter the course of the disease. Therefore, understanding AD pathophysiological mechanisms that underlie neurodegeneration is extremely important to identify therapeutic targets and for the rational design and development of DMTs able to prevent disease progression and leading to improvements in cognitive functions. Indeed, the therapeutic hypothesis of anti-amyloid strategies postulates that reducing either soluble (oligomeric) Aβ, or amyloid plaques, or both, to non-pathological levels will prevent the tau pathology spread that correlates with neuronal loss and cognitive decline, providing a therapeutic benefit [11–13]. Therefore, the most favorable approach would be to initiate a DMT before the onset of dementia, assuming the neurodegenerative process could be prevented or delayed if the therapeutic intervention occurs before clinical manifestations. Paramount to a widespread therapeutic intervention in the prodromal phase of AD or even in the preclinical stage of the disease will be the establishment of easy access and inexpensive biomarkers allowing a biological diagnosis. Recent advances in this field raised expectations this goal will be achieved in a near future [14–16].

Currently, the majority of DMTs under development for AD include the search for drugs that are aimed at blocking the main molecular mechanisms responsible for the deposition of insoluble Aβ fibrils detected in senile plaques and for the formation of neurofibrillary tangles. Over the last years, researchers have focused on anti-amyloid approaches to decrease the levels of Aβ or to neutralize its toxic effects, whether it is in the form of oligomers, fibrils, or plaques, which are based on four strategies: (i) minimization of Aβ production; (ii) prevention of Aβ aggregation; (iii) enhancement of the degradation and the clearance of Aβ and its aggregates; and (iv) neutralization of the toxic effects of Aβ aggregates [4, 17]. In fact, following the discovery of APP mutations, it soon became clear that the secretase enzymes involved in AβPP proteolytic cleavage were potential therapeutic targets for drug development directed towards AD treatment [18]. Thus, the therapies to minimize Aβ production mainly target the secretase enzymes involved in AβPP proteolytic cleavage, which include BACE1 and γ-secretase inhibition, or α-secretase potentiation [18]. Nevertheless, an approach targeting the inhibition of APP expression is also under development [8]. The therapies for inhibiting Aβ aggregation comprise anti-aggregation agents as well as drugs that inhibit enzymes responsible for Aβ post-translational modifications that promote Aβ aggregation, such as the glutaminyl cyclase enzyme [8, 20]. The approaches for promoting Aβ species removal enclose drugs that induce microglia-mediated Aβ phagocytosis as well as active or passive immunotherapies [4, 13]. On the other hand, small-molecules aiming to neutralize Aβ species toxicity comprise a modulator of the metabotropic glutamate receptor type 5 (mGluR5) and an antagonist of the sigma-2 receptor (compound CT1812) which act by blocking the binding of Aβ oligomers to the cellular prion protein that interacts with both mGluR5 and sigma-2 receptor at the synapses [21–24].

Beyond anti-amyloid therapies, strategies to prevent tau hyperphosphorylation and phosphorylated tau (p-tau) aggregation, as well as active/passive anti-tau immunotherapies are other examples of DMTs under development [25]. Furthermore, multiple pathogenesis pathways involving deficits on neurotransmission and neurogenesis, chronic neuroinflammation and oxidative stress, which might be triggered by direct and/or indirect effects of Aβ and p-tau, provide additional AD-related therapeutic targets for disease-modifying interventions [26]. Thus, a multi-target strategy is also reported as a promising approach for AD treatment [27]. One approach to achieve a multi-target therapy comprehend the use of multi-target-directed ligands (MTDLs) that consist of single hybrid chemical entities able to hit two or more AD-relevant therapeutic targets. So far, the results obtained have been encouraging and indeed this strategy might present one of the best pharmacological options for tackling the multifactorial nature of AD. Several promising MTDLs candidate drugs with disease-modifying potential are now in the pipeline and have reached clinical trials [28]. A multi-target approach might be achieved as well with combined therapies which have recently been proposed as a strategy to stop the progression of AD [29].

The establishment of an efficacious DMT might provide a crucial pharmacological tool to fight the disease. Currently, there are 187 trials assessing 141 drugs in the AD drug development pipeline, with DMTs comprising 79% of the candidate treatments among which the anti-amyloid approaches are prominent [30].

DEVELOPMENT OF BACE1 INHIBITORS

BACE1 has been extensively studied in the context of brain amyloidogenesis and linked to AD pathogenesis mechanisms [31]. Indeed, high BACE1 enzymatic activity and elevated BACE1 levels were found in the brains of AD patients [32–35]. Particularly, a relatively large accumulation of BACE1 enzyme was verified in neuritic dystrophies in the vicinity of Aβ plaques both in AD amyloidogenic mouse models and in the brain of AD patients [36, 37]. Considering that BACE1 is the initiating enzyme in Aβ generation, and putatively rate-limiting, it is a main drug target for AD treatment.

The therapeutic potential of BACE1 inhibition has been investigated for more than two decades. BACE1 inhibitors can be grouped into peptidomimetics and nonpeptidic or small-molecules, and further subclassification can be made based on the chemical structure of the drug molecule (Fig. 1) [38]. The first generation of BACE1 inhibitors was projected as peptidomimetics, which are designed to emulate the natural substrate of the enzyme, and classified according to their transition state isostere scaffolds [39]. While peptidomimetic inhibitors exhibit high potency in vitro, their limited drug-like properties account for their low in vivo efficacy. The transition from peptidomimetic drugs to effective brain-penetrant small-molecules was demanding [40] but successful due to the application of several medicinal chemistry approaches leading to the discovery of posterior generations of nonpeptidic BACE1 inhibitors [38]. A wide number of structurally diverse classes of small drug-like BACE1 inhibitors have been developed and classified based on the core functional groups that interact with the catalytic dyad (aspartate-binding groups) of the enzyme (Fig. 1).

OPEN IN VIEWER

Fig. 1

Schematic representation of different scaffolds of BACE1 inhibitors.

Moreover, many naturally occurring compounds with BACE1 inhibitory properties have been isolated from medicinal herbs, fungi, and marine flora, namely flavonoids, terpenes, triterpenes, alkaloids, chalcones, phenolic acids and cyanobacterial compounds, among many others (Fig. 1) [41, 42]. Additionally, several MTDLs with the ability to inhibit BACE1 and to modulate synergistically diverse other targets and/or pathogenesis mechanisms leading to neurodegeneration have been reported in recent years [43, 44].

A milestone in the search for BACE1 inhibitors was the obtention of crystal structures of an active form of the enzyme, which facilitated various structure-based drug design projects aimed at discovering new BACE1 inhibitors. To date, more than 431 crystallographic structures of human BACE1 are deposited in the Protein Data Bank (PDB) repository [45]. These data allow researchers to consider important structural features of the enzyme during the drug discovery process, namely the catalytic aspartic dyad, the structural flexibility, and the large binding pocket [46, 47]. Ensuring the selectivity of BACE1 inhibitors over other structurally related aspartic proteases, including BACE2, Cathepsin D/E (CatD/E), pepsin, and renin, is particularly crucial to prevent adverse effects due to cross-inhibition [38]. Noteworthy for drug design, the subsite specificity of each enzyme’s active site may be slightly different hence ligand interactions involving residues in the enzyme specific subsites may improve target specificity. In fact, it was shown that inhibitors targeting the flap region and the S3 subpocket of BACE1 displayed reduced effects on skin/hair pigmentation related to BACE2 cross-inhibition since the BACE1 and BACE2 isoforms have distinct conformational features proximal to these regions [48–50]. In addition, off-target CatD inhibition and its associated ocular toxicity have been avoided by targeting the S3 subpocket of BACE1 [51]. On the other hand, the cardiac toxicity associated to amidine-based BACE1 inhibitors has been minimized by carefully controlling the pKa of the amidine moiety and the compound overall lipophilicity, while liver toxicity has been avoided by shifting to novel chemotypes that are less prone to generate reactive metabolites [40, 52].

STRUCTURAL TYPES OF SMALL-MOLECULE BACE1 INHIBITORS AND INSIGHTS FROM CLINICAL TRIALS WITH THE MOST PROMISING COMPOUNDS

Small-molecule BACE1 inhibitors comprising many structural classes have been extensively reported in the literature and their structural evolution has been reviewed somewhere else [53–56]. These inhibitors can be grouped according to the chemical moiety that has the capacity to interact with the catalytic aspartates as summarized in Fig. 1. Several strategies have been implemented to disclose novel derivatives with increased potency and selectivity for BACE1, among other suitable pharmacokinetic properties such as oral bioavailability, brain penetration, reduced susceptibility to P-glycoprotein (P-gp) efflux, decreased human Ether-A-Go-Go ion channel (hERG) activity and risk of drug-drug interactions [57]. This topic will focus on the inhibitors’ classes reported in recent years (at least 5 years), highlighting those that had been in clinical development.

Aminooxazoline and aminooxazines

Small-molecule derivatives with aminooxazoline and aminooxazines scaffolds have been reported as BACE1 inhibitors, including compounds with an aminooxazoline headgroup connected to a central di-spiro scaffold. Compound 1 (listed in Table 1) was identified as the most attractive candidate within this series with greater selectivity over BACE2, thus avoiding potential effects on pigmentation, and exhibiting a good overall profile (including no hERG interaction neither CatD, renin, or pepsin inhibition) [58].

OPEN IN VIEWER

Table 1

Representative structures of BACE1 inhibitors scaffolds and respective biological data

Among the derivatives with aminooxazine scaffold that have been reported as BACE1 inhibitors, compound 2 (NB-360) developed by Novartis Pharma AG (listed in Table 1) that contains a 5-amino-1,4-oxazine headgroup showed an excellent pharmacological profile with a pKa of 7.1 and a very low P-gp efflux ratio enabling high central nervous system (CNS) penetration and exposure, reducing significantly Aβ levels in mice, rats, and dogs both in acute and chronic treatment regimens [59]. Subsequent optimization led to the discovery of a potent inhibitor with superior BACE1/BACE2 selectivity and pharmacokinetics named Umibecestat (CNP520) (compound 13 in Fig. 2) that advanced into clinical studies while the development of the clinical candidate NB-360 was discontinued (due to the side effect of animal hair depigmentation attributed to BACE2 inhibition) [60, 61]. Clinical investigation of Umibecestat was discontinued after phase 2/3 since the potential benefit for participants in the clinical trials did not outweigh the risk since it was observed cognitive worsening and brain atrophy in the treatment group, although these deficits were shown to be reversible after discontinuation of the treatment (conference news reported on Alzforum.com [62]). Furthermore, another inhibitor with a 5-amino1,4-oxazine scaffold, RG7129 (compound 14 in Fig. 2, [63]), developed by Hoffmann-La Roche Ltd., was also clinically evaluated and terminated because of hepatotoxicity.

OPEN IN VIEWER

Fig. 2

Structures of BACE1 inhibitors that were in clinical development.

Multi-target compounds

Given the numerous AD-related targets in the pathogenesis of the disease that have shown promising effects on modifying the course of the disease in preclinical studies but that have failed in clinical trials, suggesting that modulation of a single target might not be sufficient to improve brain function, the current AD medicinal chemistry trend focuses on the design of multi-target drugs. Data in the literature have reported on the success of creating several MTDLs with simultaneously inhibitory effects on BACE1 and multiple targets with important roles in AD-related processes, including acetylcholinesterase (AChE), butyrylcholinesterase (BChE), glycogen synthase kinase 3β (GSK3β), monoamine oxidase (MAO), ciclo-oxigenase-2 (COX-2), lipoxygenase (LOX), and the GABA transporter (GAT), among others [44, 93]. In particular, due to the ability of the natural phytoconstituents to interact with multiple therapeutic targets rather than a single target, multi-target-based strategies using derivatives of naturally occurring compounds with BACE1 inhibitory activity have been explored for AD treatment [44]. Compounds 24–33 from Table 2 are examples of MTDLs recently reported. Furthermore, BACE1 inhibitors acting simultaneously as anti-Aβ aggregation agents, antioxidant, anti-inflammatory, metal chelators, and neuroprotective agents, have also been reported (compounds 24–26, 31, 32, 34, and 35) [43, 44]. The concomitant modulation of BACE1 and other AD-related targets has demonstrated positive impacts on amyloid and tau pathology, neurogenesis, synaptic function, neuroinflammation, and oxidative stress. These effects have led to prevent animal cognitive decline and memory impairment, in numerous studies [93]. These results substantiate the idea that simultaneously targeting several enzymes contributing to different pathogenesis mechanisms could be an ideal therapeutic approach for a DMT. Nevertheless, additional research is needed to ascertain the reliability of the clinical application of MTDL drugs.

OPEN IN VIEWER

Table 2

Structures of multi-target-directed ligands (MTDLs) inhibiting simultaneously BACE1 and other AD-related targets and respective biological activity

OBSTACLES IN THE ESTABLISHMENT OF A THERAPY WITH BACE1 INHIBITORS

In the last decades intense research has been done to find potent, selective, safe, orally bioavailable and brain penetrant BACE1 inhibitors. Some of the small-molecule BACE1 inhibitors showing promising results in animal models, namely reduction of cerebral Aβ levels and cognitive/behavioral improvement, have reached human clinical development as summarized in Table 3. Among these, various BACE1 inhibitors have managed to enter advanced phases of clinical trials, such as verubecestat (MK8931), elenbecestat (E2609), lanabecestat (AZD3293), atabecestat (JNJ54861911), and umibecestat (CNP520). Key findings from these clinical trials were Aβ decrease in a dose-response manner and a stable continuous level of Aβ reduction in participants with AD. Also, PET studies showed that verubecestat and lanabecestat lowered amyloid plaques by an average 5–10% per year (reviewed in [94]). Unfortunately, inefficacy in providing cognitive benefits or toxicity issues led to the discontinuation/termination of all trials. Actually, individuals treated with BACE1 inhibitors in trials targeting both symptomatic and presymptomatic stages of AD showed mild non-progressive cognitive worsening and, in some instances, it was observed a non-progressive reduction in brain volume [94]. Still, the follow up of participants allowed to demonstrate the adverse cognitive effects and MRI volumetric changes in presymptomatic individuals are almost completely reversible within 3 months of treatment cessation uplifting the idea that they are not due to an acceleration of neurodegeneration [95, 96]. Other side effects, although observed to a mild extent and not consistently across distinct BACE1 inhibitors, include anxiety, weight loss, falls, suicidal ideation and sleep disturbances [94].

Noteworthy, the BACE1 inhibitor doses used in the Phase 2 or Phase 3 clinical trials allowed for a reduction in Aβ levels in human CSF of about 90%, which might be excessive considering the relevant BACE1 and Aβ physiological roles. It is possible that an almost total inhibition of the proteolysis of BACE1 substrates may underlie the occurrence of the cognitive and psychiatric effects observed in the clinical trials. Clearly, it would be interesting to know whether patients would experience a cognitive benefit with a lower level of BACE1 inhibition. Unfortunately, trials with lower doses of BACE1 inhibitors were not performed yet. Nonetheless, preclinical studies with AD animal models, including our own work, point that a partial reduction of BACE1 enzyme activity throughout the time leads to a dramatic decrease in insoluble Aβ levels [97], amyloid plaque deposition and synaptic pathology [98]. Also, the sequential and increased deletion of BACE1 in an adult AD mouse model (5xFAD) led to a reversal in amyloid deposition and an improvement in gliosis, neuritic dystrophy, long-term potentiation and behavioral deficits, supporting the view that sustained BACE1 inhibition can rescue AD pathology [99]. Furthermore, the Icelandic APP gene mutation (A673T), that leads to a lifelong reduction in Aβ levels of about 30%, protects against AD and age-related cognitive decline [100, 101]. Taken together, these studies suggest that a chronic partial inhibition of BACE1 might decrease the risk or delay the onset of AD clinical symptoms, particularly if initiated before advanced amyloid pathology, while preventing the occurrence of the mechanism-based side effects observed in clinical trials.

The lack of efficacy of the BACE1 inhibitors on the prespecified primary outcomes, such as those measuring cognition and function in clinical trials may involve many reasons, including the clinical trial design where inhibitors are being administered too late in the time course of the disease following irreversible neurodegenerative damage. Indeed, another issue to be addressed is the criteria for patient population inclusion in these clinical trials namely regarding the disease stage. PET imaging studies showed that Aβ deposits may accumulate in an Alzheimer’s brain about 19 years before expected onset whereas structural decline is observed 5 years before expected onset of clinical dementia in individuals from families with autosomal dominant AD [102]. A change in CSF Aβ₄₂ levels could be detected as early as 25 years before symptoms onset [103]. Thus, evidence suggests that the optimal timing for treatment with BACE1 inhibitors should be as early as possible. Unfortunately, the current lack of reliable AD early diagnosis is restricting the potential of a preventive approach.

On the other hand, previous clinical trials of both γ-secretase inhibitors and BACE1 inhibitors have failed, in large part due to insufficient drug selectivity and specificity. The toxicity observed in clinical trials with γ-secretase inhibitors was mainly due to the lack of substrate-specific inhibition. In this regard, the toxicity arising from inhibition of Notch-1 cleavage by γ-secretase, which disrupts essential signaling from this receptor, is particularly relevant. Thus, the current trend is to discover drugs that inhibit the cleavage of C99, the immediate precursor of Aβ, while allowing Notch cleavage to pursue. It seems to be the case of verteporfin whose inhibitory effect is C99-specifc [104]. Also, serine169 in PS1 was recently pointed as a potential target for the development of new substrate-selective γ-secretase modulators [105]. The lack of selectivity is also an impediment to the clinical use of BACE1 inhibitors in AD. As mentioned above, several BACE1 inhibitors exhibit cross-inhibition with other members of the aspartyl protease family (e.g., BACE2, pepsin, renin, CatD/E), which is responsible for some adverse/toxic effects observed in the clinical trials. In this regard, the recent advances on the physiological functions of BACE2 are particularly relevant to understand some off-target effects (reviewed in [4, 107]). Since BACE2 cleaves the AβPP within the Aβ domain, finally hampering the generation of Aβ peptides, this secretase might mitigate AD-associated pathology. Indeed, it was proposed a physiological role of BACE2 as a dose-sensitive AD-suppressor gene [108]. Accordingly, identified polymorphisms within the BACE2 locus have been associated with altered AD risk [109] and with an earlier age of dementia onset in individuals with Down syndrome [110]. In plus, BACE2 can also degrade Aβ by itself [111]. Taken together, this evidence suggests a neuroprotective role for BACE2. Regrettably, all BACE1 inhibitors tested in Phase 2/Phase 3 clinical trials inhibit both BACE1 and BACE2. Thus, we cannot exclude the cognitive worsening observed in participants treated with these drugs might be, at least, partially due to BACE2 inhibition. Whether more selective BACE1 over BACE2 inhibitors would hamper such side effect have yet to be determined. On the other hand, although BACE2 selective drugs for AD treatment are not yet developed, it is interesting to note that BACE2 expression may be upregulated by miRNAs [107].

Furthermore, in addition to AβPP processing, BACE1 enzymatic activity is involved in neuronal and non-neuronal physiological functions [31, 113]. Indeed, beyond AβPP, BACE1 has many other substrates (Table 4) that may be important for axon guidance, synaptic plasticity, and synaptic homeostasis hence its physiological role is known to be required for optimal cognitive function. In fact, BACE1 is required for optimal release of synaptic vesicles and BACE1 deficiency reduces the level of pre- and post-synaptic proteins crucial for synaptic functions [114]. Thus, it is worthy of mention that the BACE1 inhibitors tested in clinical trials were not substrate-selective which may underlie the cognitive decline observed in these studies. In this context, animal studies showed that the administration of a mGluR1 positive allosteric modulator mitigates the synaptic deficits induced by the pharmacological BACE1 inhibitors verubecestat and lanabecestat suggesting that a therapy combining BACE1 inhibitors for reducing Aβ production and an mGluR1 positive allosteric modulator for counteracting BACE1-mediated synaptic deficits might be an effective approach for treating AD patients [114]. Significantly, although the knockout of the BACE1 gene in the germline of mice causes multiple neurological phenotypes, these phenotypes may be due to BACE1 enzymatic activity deficits during the nervous system development rather than due to the absence of BACE1 activity in the adult animal. Ou-Yang and coworkers [115] showed that adult conditional BACE1 knockout mice largely lack phenotypes, pinpointing that most BACE1 functions are not required in the adult animal. Still, the adult conditional BACE1 knockout mice exhibited reduced length and disorganization of the hippocampal mossy fiber infrapyramidal bundle, a phenotype that correlates with reduced proteolysis of the neural cell adhesion protein CHL1, a BACE1 substrate implicated in axonal guidance. Thus, non-substrate-selective BACE1 inhibitors may disrupt the adult circuitry architecture in the hippocampus, a brain structure that support humans to process and retrieve declarative memories and spatial relationships. Although additional research is required to comprehensively grasp the biological functions of BACE1 and the implications of its chronic inhibition, it is advisable to carefully titrate drug dosages [106, 116] and, specially, to develop substrate-selective BACE1 inhibitors that may prevent AβPP cleavage without interfering with the proteolysis of other substrates in order to minimize mechanism-based toxicity associated with the long-term use of BACE1 inhibitors. In this regard, considering that AβPPβ cleavage occurs in endosomal compartments [117, 118], strategies favoring the endosomal localization of BACE1 inhibitors might provide inhibition of AβPPβ cleavage without interfering with the proteolysis of the BACE1 substrates that occurs in an endosomal-independent manner. Accordingly, we and others showed that such an approach may spare the cleavage of BACE1 substrates with relevant synaptic functions while inhibiting AβPP cleavage [97, 117]. Indeed, BACE1 remains a well-validated therapeutic target in AD and although past failures in advanced clinical trials, BACE1 inhibitors potentially continue to be a relevant approach in delaying or preventing the onset of AD.

Other obstacle in the establishment of an anti-Aβ therapy concerns the disturbance of the physiological functions of soluble Aβ monomers that include the modulation of the synaptic function and cognition, the support to recovery from traumatic and ischemic injuries, the inhibition of oxidative stress, the suppression of tumor growth, the anti-microbial activity and a possible stimulation of neurogenesis (reviewed in [119] and [4]). Also, AβPP, a type I transmembrane protein highly expressed in neurons, especially at the synaptic level, has important physiological roles in dendritic spine remodeling, synaptic transmission and synaptic homeostasis, which are mediated, at least in part, through the extracellularly released soluble AβPP fragments, sAβPPα and sAβPPβ (reviewed in [9]).

Additionally, the absence of accurate animal models leads to a translational gap between animal research and clinical studies. In fact, the current animal AD models are unable to mimic the complexity of factors influencing the development of late-sporadic AD, a combination of genetic, lifestyle and environmental factors. Also, the extensive neuronal loss observed in the brain of AD patients has not been observed in mice models. Accordingly, a recent study in which human or mouse neurons were xenografted into the brain of a mouse model of AD revealed a human-specific vulnerability to AD, since only the human neurons displayed tangles and considerable neuronal cell loss, among other features [120].

ANTI-AMYLOID PASSIVE IMMUNOTHERAPY: A MILESTONE IN AD TREATMENT

Among the different AD therapies addressing the amyloid pathology that have been under clinical trials, passive immunotherapy with anti-amyloid monoclonal antibodies has been considered as one of the most promising strategies with ability to modify AD progression since it promotes Aβ clearance from the brain reducing Aβ deposits load, thus holding potential to decrease Aβ neurotoxic effects and to prevent cognitive decline.

The monoclonal antibodies that have advanced into clinical development (Table 5) target different Aβ species (e.g., monomers, oligomers, protofibrils, and insoluble fibrils) and recognize different Aβ antigenic sites (epitopes) and thus differ in their capability to reduce brain Aβ burden and in their clinical efficacy [121]. The data gathered from clinical trials pinpoint that antibodies predominantly targeting amyloid plaque, and inducing amyloid plaque removal, showed some clinical benefit as addressed by primary and secondary trial endpoints [11].

Among the most promising antibodies, aducanumab (Aduhelm^®) and lecanemab (Leqembi^®) were recently approved by the US Food and Drug Administration (FDA) under the accelerated approval pathway for the treatment of MCI or mild AD dementia [122]. These agents, which exhibited clinical efficacy in lowering plaque load in clinical trials, are indeed the first DMT against AD reaching clinical usage. In July 2023, the FDA converted lecanemab to traditional approval after a determination that a confirmatory trial verified clinical benefit. Actually, in the Phase 3 Clarity AD study, lecanemab reduced patients decline as assessed by the Clinical Dementia Rating Scale Sum of Boxes score (CDR-SB) by 27% compared to placebo at 18 months. Also, all secondary endpoints, as determined by a change in amyloid burden on PET, the score on the 14-item cognitive subscale of the Alzheimer’s Disease Assessment Scale (ADAS-cog14), the Alzheimer’s Disease Composite Score (ADCOMS), and the score on the Alzheimer’s Disease Cooperative Study–Activities of Daily Living Scale for Mild Cognitive Impairment (ADCSMCI-ADL) were met. Of note, two-thirds of the treated population became PET-amyloid negative at 18 months [123], which is supported by neuropathological autopsy findings [124]. Moreover, in an open-label extension study, lecanemab administered to patients with early AD led to a significant decrease of tangle accumulation in the medial temporal lobe, evaluated by Tau PET, and to a corresponding change in plasma biomarkers (Aβ₄₂/Aβ₄₀ and p-tau181) [125]. Also, additional efficacy studies based on the Clarity trial showed a benefit on health-related quality-of-life [126]. Concerning safety, adverse events that may occur with lecanemab include ARIA and infusion reactions. Thus, Appropriate Use Recommendations were delineated for assisting the introduction of this new immunotherapy in clinical practice [127]. In the meantime, EMA is reviewing the Marketing Authorization Application for lecanemab to treat early AD and is expected to weigh in early 2024.

Regarding aducanumab, FDA required a post-approval trial to verify the clinical benefit of the drug. Conversely, the European Medicines Agency (EMA) had recommended refusing marketing authorization of Aduhelm^® in December 2021 [128]. Indeed, the rationale for approval and the extent of the clinical benefit from aducanumab have been under intense debate [129]. Meanwhile, in January 2024, Biogen announced it would reprioritize the resources allocated to ADUHELM^® (aducanumab-avwa) to advance LEQEMBI^® (lecanemab-irmb) and to develop new treatment modalities. Therefore, the company will discontinue the development and commercialization of ADUHELM^® and will terminate the ENVISION clinical study.

Another promising monoclonal antibody under clinical trials is Donanemab, a monoclonal antibody directed against pE-Aβ developed by Eli Lilly that in phase 3 showed to slow cognitive and functional decline in people with early symptomatic AD. Nearly 50% of individuals in the early stage of the disease treated with Donanemab showed no clinical progression after one year and exhibited a 60% reduction in the rate of decline compared to those who received a placebo [130]. Importantly, significant reductions were observed in plasma biomarkers p-tau217 and glial fibrillary acidic protein following Donanemab treatment in patients with early AD [131]. Eli Lilly, that has completed its FDA submission for Donanemab’ full approval to treat patients with early AD, recently shared positive data regarding the first active comparator study (with Aduhelm^®) in early symptomatic AD [132].

Noteworthy, a recent preclinical study regarding a novel and high-affinity antibody against N-terminal pyroglutamate-Aβ used in combination with the BACE1 inhibitor atabecestat (compound 18, Fig. 2) showed increased efficacy and favorable safety profile in plaque-depositing mice [133]. Together, these data confirm the significant potential for passive immunotherapy using anti-Aβ in combination with another approach such as BACE1 inhibition in AD treatment.

Weighing down the widespread clinical use of these immunotherapies is the increased risk of ARIA, a side effect known to occur with the class of antibodies targeting amyloid. Thus, individuals receiving aducanumab and lecanemab should be monitored closely for brain edema [127, 134]. Moreover, anti-Aβ immunotherapies might also contribute to accelerate brain atrophy [135]. Noteworthy, a recent study in an AD mouse model highlighted that antibodies with the Fc fragment reduced Aβ burden but also induced acute microglial synapse removal and rapidly exacerbated cognitive deficits and neuroinflammation suggesting that Aβ-targeting antibodies that lack the Fc fragment, or with reduced Fc effector function, may provide a safer and more efficient therapeutic alternative for passive immunotherapy for AD treatment [136].

Taken together, the anti-Aβ immunotherapy trials showed evidence that the removal of abnormal Aβ from the brain of symptomatic patients might prevent the progression of AD, giving the research community the first clear clinicopathological indication that a DMT for AD is achievable [122]. Moreover, immunotherapy may reduce not only cerebral Aβ content but also decrease the cerebral tauopathy and astrocytic activation [123, 138]. Importantly, the clinical trials with anti-Aβ antibodies showed that to be efficacious the anti-amyloid drugs need to reduce brain amyloid significantly and their trials need to be kept ongoing time enough after amyloid burden removal in order to observe a clinical benefit [11, 139]. These observations should be considered when planning clinical trials with other anti-amyloid drugs.

CONCLUSIONS AND FUTURE PERSPECTIVES

Although the flaws and failures of past anti-amyloid clinical trials, that sparked the debate on the amyloid cascade hypothesis, the knowledge gathered in interdisciplinary studies has contributed to improve our understanding of the biochemical, physiological, and pathological features of the Aβ pathway, including its spatial and temporal relationship along the AD continuum, highlighting its relevance as a driver of disease pathophysiology [4, 9–11]. Importantly, biomarker-based longitudinal studies provide evidence that the amyloid pathology initiates decades before the onset of dementia symptoms and upstream to the development of other pathophysiological hallmarks of AD, namely the spread of tau pathology to limbic regions and the neocortex, the hypometabolism and the structural decline [102, 140], which should be taken into consideration when planning interventions meant to alter the course of disease.

Presently, anti-amyloid therapies are the mainstream line of clinical research in AD therapeutics. The emerging results from passive immunotherapies provide clinical reinforcement to the importance of Aβ aggregates in the pathogenesis of AD. Nevertheless, to develop more refined Aβ-based therapies, namely BACE1 inhibitors, it should be kept in mind that several drug trials failed due to a lack of specificity and thus medicinal chemistry should explore unique features of enzyme specific subsites in order to achieve target specificity. Additionally, the development of substrate-selective inhibitors [97, 117] and the identification of a more accurate dose regimen [94] will surely help to prevent mechanism-based adverse effects [97, 117]. In this regard, a further understanding of BACE1 and BACE2 biological functions and physiological roles of their substrates, as well as of AβPP and Aβ physiological homeostasis, will prove useful. In fact, the results from failed clinical trials with BACE1 inhibitors highlight the need to uncover the relationship between the level of BACE1 inhibition, amyloid load, and cognitive status. Interestingly, in the heterozygous carriers of the AβPP A673T mutation, which have a lower risk to develop AD, it was observed a decrease by 28% in Aβ₄₀ and Aβ₄₂ plasma levels suggesting that a lifelong small reduction in Aβ production might be protective against AD [101]. Thus, in plus to the level of BACE1 inhibition, another issue to be addressed is the eventual need for a prolonged treatment with a BACE1 inhibitor, particularly to observe a possible clinical benefit within the scope of a preventive trial. Indeed, given the complexity of AD pathogenesis, the advantageous cascade of events induced by anti-amyloid therapies ultimately involves several biological steps, making it unlikely that an amelioration of cognitive decline will be temporally near a therapy-mediated reduction in amyloid burden [10, 11].

On the other hand, the discontinued clinical trials with BACE1 inhibitors did not evaluated BACE1 concentration nor enzymatic activity in body fluids, which are biomarkers that could be useful as proof of mechanism, and for dose-finding and efficacy/safety measures [31, 141–143]. It will be important to establish BACE1 biomarker-drug co-development programs that would support the evaluation of drug response, better go/no-go decisions and a reduction of side effects due to excessive BACE1 inhibition [144]. Indeed, epigenetic and posttranslational factors that influence BACE1 expression and BACE1 enzymatic activity may underly interindividual differences in BACE1-mediated pathophysiological mechanisms and in drug response [31]. Thus, BACE1 biomarker-guided trials would contribute to de-risk the study and, importantly, to personalize dose selection which may allow the administration of BACE1 inhibitor lower doses keeping in line with the evolving concept of personalized medicine [144]. Likewise, the comprehension of the pathophysiology of the AD continuum is important to determine the optimal stage of disease at which to treat. Actually, the success of anti-Aβ therapies might be dependent on the stage of AD. Data from monoclonal antibody clinical trials suggest that a threshold amount of brain amyloid is required to drive cognitive impairment [11]. Ideally, the therapeutic intervention should be carried out before attaining that threshold so that the tau pathology, which correlates with cognitive deficits, is less extensive and with a lower risk to a self-sustained propagation. Indeed, it is generally accepted that BACE1 inhibitors should be administered in a primary prevention setting, in persons at greatest risk of developing AD, which includes Aβ-negative individuals with 60–70 years old who are homozygous for APOEɛ4 as well as dominant inherited AD mutation carriers one or two decades before symptoms onset [94]. Also, BACE1 inhibitors could prove useful for secondary prevention, in individuals with low levels of Aβ deposition, before the onset of clinical symptoms, which includes Aβ-positive individuals heterozygous for APOEɛ4 [94]. In this context, the discovery of new biomarkers that can identify patients at very early stages of sporadic AD is crucial for the success of AD prevention. Emerging studies on blood-based biomarkers unveiled plasma p-tau 217 and plasma p-tau 231 as promising low-invasive and low-coast biomarkers to identify a preclinical population for AD clinical trials [15, 16]. Beyond the classical biomarkers (Aβ and tau), the CSF levels of other molecules such as progranulin and neurogranin that are associated with inflammation and synaptic plasticity, respectively, may also be useful to identify patients with preclinical and/or at early clinical stage AD [145, 146]. The recent advances in the discovery of new biomarkers for AD will be essential not only to select individuals with preclinical AD that will benefit from a DMT, but also to demonstrate target engagement and to support biomarker-based outcomes and end points in clinical trials, thus contributing to de-risk AD drug development.

Considering the latest progresses in this theoretical framework, in the years to come the development of AD treatments will possibly see another rise in BACE1-targeting therapies, that continue to represent a rational approach for a disease-modifying preventive strategy for AD with clear advantages over passive immunotherapies. In fact, BACE1 inhibitors can be administered orally while immunotherapies need an intravenous infusion and are expected to have a lower production cost. Nonetheless, it will be advisable the next clinical trials will be designed bearing in mind a sustained low-level BACE1 inhibition with selective BACE1 inhibitors to minimize mechanism-based and off-target side effects. It is also interesting the emerging development of multi-target drugs which allow for the simultaneous modulation of multiple disease pathogenesis pathways. Furthermore, combination therapies addressing different targets within the amyloid pathology (such as BACE1, glutaminyl cyclase, Aβ aggregates or pyroglutamate Aβ) or, alternatively, combination therapies targeting different pathogenesis mechanisms, are expected to be better suited to tackle distinct stages of a multifactorial and complex disease such as AD. Within this context, a BACE1 inhibitor could be used in an enduring manner to keep amyloid plaques under low levels after its removal by passive immunotherapies. Rivetingly, the Dominantly Inherited Alzheimer Network Trial Unit (DIAN-TU) recently initiated a clinical trial combining anti-Aβ and anti-Tau passive immunotherapies in individuals with familial AD mutations [147]. The results of these and other clinical trials regarding anti-Aβ therapies are expected to contribute to establish effective AD-modifying treatments in a near future.

AUTHOR CONTRIBUTIONS

Judite R. M. Coimbra (Writing – original draft; Writing – review & editing); Rosa Resende (Writing – original draft; Writing – review & editing); José B. A. Custódio (Writing – review & editing); Jorge A. R. Salvador (Writing – review & editing; additional corresponding author); Armanda Emanuela Santos (Conceptualization; Writing – original draft; Writing – review & editing).