Introduction

We are in a stage of paradigm shift in the therapeutic field of Alzheimer’s disease (AD) from an era of symptomatic therapies to disease modification, now that a first monoclonal antibody against beta-amyloid has received Food and Drug Administration approval. There are major advances in our understanding of the interaction between amyloid and tau pathologies thanks to in vivo brain imaging using Positron Emission Tomography (PET) (1), as well as the interaction between tau pathology and microglial activation (2). On the other hand, there is awareness that while progressive beta-42 amyloid deposition and spreading of hyperphosphorylated tau fibrils are required findings in the current biological definition of AD (3), they may not sufficient for dementia to be manifest clinically, particularly in older persons, where other pathologies, such as vascular lesions, neocortical Lewy bodies and TDP-43/hippocampal sclerosis may interact with the primary AD pathologies (4).

Anti-amyloid treatments

Most of the randomized clinical trials (RCT) over the past decade aiming at disease modification in AD have targeted various components of the amyloid cascade. The most successful approach so far has been monoclonal antibodies targeting fibrillar Abeta, administered either intravenously or subcutaneously. There is cumulative evidence that high doses are required, despite the risk of Amyloid Related Imaging Abnormalities (ARIA), in order to normalize amyloid brain levels at least for the species measurable using PET imaging, and there is emerging evidence that the amyloid levels remain low for one to two years after the antibody therapy is stopped (5).

An encouraging observation is that anti-amyloid therapy using aducanumab lowers cerebro-spinal fluid (CSF) tau markers as well as plasma p-tau181 isoform (6), suggesting downstream effects on anti-amyloid therapy on tau pathophysiology. Similarly gantenerumab treatment in autosomal dominant AD led to a reduction in CSF Total Tau, p-tau 181 and attenuated increases in neurofilament light chain (7).

The magnitude of the benefit on clinical outcome measures over the 1.5 years of observation in clinical studies of amyloid-removing antibodies (e.g. aducanumab phase 3, and lecanemab and donanemab phase 2 trials) was relatively small and consisted of a reduction in the rate of clinical worsening (8). It is reasonable to speculate that the reduction of the decline trajectory in amyloid removal-treated patients could deviate further from the untreated state beyond 18 months, but it is possible that the antibody treated versus placebo differences in rate of decline will remain fixed. Thus, even if anti-amyloid therapy results can be replicated in ongoing phase 3 trials, the fact that some disease progression continues despite amyloid lowering justifies aggressive efforts to pursue other targets. In our review, the perspective on monoclonal anti-amyloid antibodies as of December 2021 is uncertain: the outcomes of the pivotal trials of donanemab, lecanemab and gantenerumab that will be available in 2022 and 2023 could paint a more or less favorable picture.

Anti-tau treatments

The conceptual logic of tau-directed approaches for symptomatic AD is considerably stronger than that of amyloid-directed approaches because tau accumulation occurs at the right time and the right place to account for clinical cognitive deficit. (9, 10). A large body of nonclinical studies in cell culture and animal models suggests that reduction of toxic tau gain of function (eg., accumulation, misfolding and/or posttranslational modifications) may be an effective therapeutic strategy for AD as well as other non-AD tauopathies (11). Tau protein exists in numerous aggregated states, some of which might be more amenable to treatment than others, and some might actually be central to the disease’s pathogenesis.

Another factor complicating an understanding of the role of tau protein in tauopathies is the diversity of repeats (3R vs 4R) across diseases, and the fact that most of the current animal models of tauopathies utilize FTDP-17 or to MAPT mutations, which are more closely related to fronto-temporal dementia than to AD. AD is often considered a secondary tauopathy, since amyloid pathology precedes the accumulation of tau throughout the brain and spreading from the mesial temporal regions to the neocortex (12). Similarly, tau possesses numerous post-translational modifications (13), and currently it is not clear whether unmodified tau, C-terminal epitopes or N-terminal ones would be the best therapeutic targets. Recent studies using cryo-electron microscopy showed that the core of insoluble tau aggregates in AD and other tauopathies is primarily composed of peptides in the microtubule binding and C-terminal domains of tau, and that there are specific three dimensional protein conformations for each different tauopathy (14). These core regions of tau pathology are distinct from the extracellular soluble tau species that have been used as CSF and plasma biomarkers and that first generation anti-tau antibodies have targeted (15, 16). To date, four N-terminally targeted anti-tau monoclonal antibodies have failed to demonstrate signs of efficacy in Phase 2 clinical trials in prodromal-mild AD and/or progressive supranuclear palsy (PSP). These include gosuranemab, (17) tilavonemab, (18) semorinemab (19) and zagotenemab. All four antibodies target N-terminal regions of tau that are not central to the disease associated protein aggregates, suggesting that despite evidence that the antibodies bind N-terminal tau fragments in human patients’ CSF, (20, 21) one potential reason for lack of efficacy may be that N-terminal tau fragments are not central to tau toxicity. Even if N-terminal tau fragments are not central to AD related tau toxicity, they may have important physiological roles in the brain such as regulating neuronal excitability (22, 23). The recent report that Semorinemab seemed to slow cognitive decline (ADAS cog) at least in in mild-to-moderate AD remains to be understood, especially in light of the failure of treatment to show an effect on the co-primary outcome of functional ability (ADCS- IADL) (24), or earlier stages of disease, but if replicated, might conceivably be explained by effects on N-terminal tau fragments.

There are also tau antibodies now targeting the MTBR and C-terminal domain of tau, which is the region that deposits in tau aggregates in the brain. These antibodies still have the challenge of being extracellular and may or may not reach their intended target, but they have a potential advantage in targeting the tau species that make up tau pathology. Several programs are being developed (Eisai 2814, UCB0107, and others) and are now in clinical studies.

An alternate approach that may better recapitulate nonclinical studies of the role of tau in neurodegeneration is based on suppression of tau gene (MAPT) expression using antisense oligonucleotide (ASO) or other generic drugs in human patients. Recently, on such ASO was shown to be safe and well tolerated with AD showed a reduction in CSF total tau levels similar to that observed from heterozygous inactivation of MAPT in transgenic mice (25). While it remains to be seen whether this suppression will lead to a clinical benefit, such genetically targeted approaches are best situated to recapitulate nonclinical animal and cell culture experiments.

Five general approaches to anti-tau therapies are listed in Table 1.

Table 1 Anti-tau therapeutics

The NIH funded Alzheimer’s Clinical Trial Consortium (ACTC) is planning a platform trial that would allow efficient testing of multiple tau therapies alone or in combination with an anti-amyloid drug. The ACTC Tau Platform (ATP) is currently under review for funding by the NIA and will look at key issues such as when in the course of disease can we measure tau treatment effects, who to enroll in terms of AT(N), and what effect size such as 30% SUVr reduction would correlate with a clinically meaningful response to treatment. Another issue is whether to use an active anti-amyloid comparator arm and a placebo arm in this new generation of randomized clinical trials. Platform trials offer many efficiencies over traditional clinical trial designs (36) and may reduce the amount of time necessary to determine proof of concept as well as the costs of clinical development of new tau therapies for AD.

The NIH funded DIAN-TU trials have launched a combination anti-amyloid and anti-tau program as part of the next generation platform (37). The design (NCT01760005) incorporates combined treatment with lecanemab and E2814, an antibody targeting the MTBR of tau in two populations of pre-symptomatic mutation carriers within 10 years of symptom onset and a symptomatic population up to CDR 1. The trial launched in December 2021 (38). The platform has two funded additional tau drug arms to launch which may or may not include anti-amyloid treatments and will focus on tau drugs with different targets.

Therapeutic perspectives beyond reduction of amyloid and tau pathologies

Working within the framework of a pathogenic sequence of AD that includes the over- or aberrant-production of the APP and the Aβ peptide, and then proceeds eventually to 3R/4R tau deposition in the entorhinal cortex with eventual expansion of tauopathy into the temporal and parietal neocortices, there are numerous unanswered questions about the controls at each step (39). If it turns out that either Aβ accumulation or tau accumulation are not the only direct causes of neurodegeneration that produces cognitive impairment, targets along this mechanistic chain other than Aβ or tau reductions might prove to be more effective points of intervention. Or, even if Aβ and tau accumulation are indeed directly in the key mechanistic chain, there could be treatable rate-limiting steps with non-Aβ, non-tau targets.

Several general domains suggest themselves from what is known about the biology of AD. One approach is to understand synaptic homeostasis and how to recover from dyshomeostasis. Treatment with low-dose levetiracetam is one such example (40). Second, understanding the early steps of protein misfolding and how to prevent misfolded proteins from remaining in the cytoplasm is another strategy. Epichaperome inhibitors are being tested in pursuit of this approach (41). Third, dysregulation of neuronal autophagy and the complex machinery of proteostasis may have druggable targets (42). Blocking GTP-Rab5 overactivation with neflamapimod is one example (43). Fourth, understanding dysfunctional inflammatory or cytotoxic responses that are generated by upstream process such as the generation of toxic Aβ or tau species could lead to targeting kinase inhibitors that are specific to activated microglia (44) or to stimulating microglial activity with a proinflammatory cytokine granulocyte macrophage colony-stimulating factor to promote clearance of misfolded proteins (45). Attempting to intervene in the interaction of cellular senescence and neurodegeneration is yet a different approach (46, 47). These examples are but a few of the promising targets that should be pursued.

One aspect of the etiology of the disease that has been largely neglected until recent years is the role of aging (48). The National Institute of Aging, highlighted geroscience for AD drug discovery. Importantly, the paper also outlines specific areas where attention to the pillars of aging might be fruitful in our efforts against Alzheimer’s. A better understanding of the biology of aging paves the way to pharmacological compounds linked to the hallmarks of aging, including senolytics, inflammasome inhibitors, metformin, rapamycin, resveratrol mesenchymal stem cells, mitochondria agents (49)

The general lack of specific pharmacodynamic markers for non-Aβ, non-tau targets is perhaps the greatest conceptual and practical barrier to the discovery of therapeutics for this class of putative therapies. A pharmacodynamic biomarker is of critical importance for determining dosing and for obtaining early-stage evidence to justify moving from phase 2 to phase 3. To be sure the amyloid and tau therapeutic approaches were also stymied for some time by this same limitation. A sponsor seeking to move a therapeutic agent into the clinic should be simultaneously developing a fluid or imaging pharmacodynamic biomarker which would support the mechanism of action of the therapy. This is hardly a trivial matter considering the long gestation and multiple laboratories involved in the development of amyloid or tau biomarkers, and the ongoing frustration of lacking a biomarker for TDP43, for example. Investments in new pharmacodynamic biomarkers must be a major focus of funding and research in AD therapeutics. Geroscience will also probably help to determine biological age in the future (50)

From a RCT perspective limitations for testing these approaches are the inability of smaller and shorter trials to identify relevant disease-modification and impact on clinical symptoms emergence or progression. Fortunately, biological characterisation of potential participants using biomarkers will help recruit homogeneous populations for proof-of-concept studies.

Conclusions

There is strong epidemiologic, genetic, and experimental support for both amyloid and tau as therapeutic targets in the prevention and treatment of AD. There may be a ceiling limiting how much anti-amyloid and anti-tau monotherapies independently slow down clinical progression, despite clear target engagement, especially if initiated at the symptomatic stage of dis. Combination studies where distinct pathologies could be targeted by precise timing of administration ease are critical and inevitable, looking for synergistic effects. Immune system modulation at a strategic time in the course of disease may further modify disease progression.

Though beyond the scope of this year’s EU/US Task Force meeting, the treatment options for persons with an Alzheimer-like phenotype but that are amyloid negative deserves further discussion beyond exclusion from anti-amyloid RCT.