Alzheimer’s disease: many failed trials, so where do we go from here?

Abstract

Alzheimer’s disease (AD) is a neurodegenerative brain disorder associated with relentlessly progressive loss of of memory and cognitive impairment. AD pathology proceed for decades before cognitive deficits become clinically apparent, opening a window for preventative therapy. Imbalance of clearance and buildup of amyloid β and phosphorylated-Tau proteins in the central nervous system are believed to contribute to AD pathogenesis. However, multiple clinical trials of treatments aimed at averting accumulation of these proteins have yielded little success and there is still no disease-modifying intervention. Here, we discuss current knowledge of AD pathology and treatment with an emphasis on emerging biomarkers and treatment strategies.

Introduction

Alzheimer's disease (AD) is a progressive, irreversible disabling neurodegenerative disorder characterized by memory loss, cognitive dysfunction and behavioral changes (1-3). It is the leading cause of dementia, affecting approximately 46.8 million people globally (4-6.). At the molecular level, pathological changes in the brain that are hallmarks of AD include intracellular neurofibrillary tangles (NFTs) containing hyperphosphorylated tau protein and insoluble extracellular β-amyloid (Aβ) plaques (7, 8). Pathological changes in the brain precede clinical symptoms and, therefore, diagnosis, by decades (9-11). The debilitating effects of AD impose an enormous social, emotional and economic burden on patients and their families (6, 12). The etiology of AD remains unclear and despite research programs costing billions of dollars and numerous successful approaches in mouse models, at this time, there are no effective treatments for human AD (13-16). Mild symptomatic benefits are all that can be offered. This symposium will explore the underlying reasons for failure in developing effective drugs and suggest possible new directions to improve prognosis for this devastating disease.

APP Processing and Aβ Formation

The amyloid cascade hypothesis is a widely accepted model of AD pathogenesis that postulates an imbalance between production and clearance of the Aβ peptide leading to brain deposition of Aβ as the cause of AD (17, 18). Despite multiple failures, this hypothesis continues to drive the development of potential AD treatments (19, 20).

The proteolytic processing of APP determines whether Aβ will be generated (Figure 1). The three enzymes that control this process are α- secretase, β- secretase (β-site APP-cleaving enzyme, BACE1) and γ-secretase. APP can enter either an amyloidogenic or a non-amyloidogenic pathway, depending on which of these secretases act upon it. The amyloidogenic pathway begins with BACE1 releasing a soluble APP fragment (sAPP-β) and leaving in the membrane 99 amino acid C-terminal fragment (CTF-β) (21). This fragment is then cleaved by γ-secretase at slightly different positions, producing Aβ_1-40 and Aβ_1-42 as well as a cytoplasmic peptide, the APP intracellular domain (AICD). The non-amyloidogenic route starts with APP cleavage by α- secretase in the middle of the Aβ domain, freeing a soluble ectodomain (sAPP-α) and a membrane-bound C-terminal APP fragment of 83 amino acids (CTF-α). Further cleavage within the transmembrane domain by the γ-secretase complex yields the p3 peptide fragment and the AICD.

Figure 1. APP processing via non-amyloidogenic and amyloidogenic pathways.

The left side of the figure shows non-amyloid-forming reactions in which APP is cleaved by α-secretase to form sAPP-α, then by γ-secretase to yield P3 and AICD (amyloid intracellular domain). The right side of the figure shows amyloid-forming reactions in which APP is cleaved by BACE1 to form sAPP-β, then by γ-secretase to yield Aβ and AICD

Rodent Models

Transgenic mice overexpressing amyloid precursor protein (APP) have been used in myriad studies and have proven to be a valuable in vivo model, contributing tremendously to our understanding of AD pathophysiology (22-27). However, there is a disconnect between efficacy of treatment in rodent models and failure when attempts are made to translate to humans. In theory, Aβ removal or interference with Aβ aggregation will improve the signs and symptoms of AD. Unfortunately, this holds true in mice, but for reasons that we have to work out, it does not yield the same improvement in people (26, 28-30) (Table 1).

Table 1.

Problems with Current Strategies to Develop AD Treatment

Mouse studies succeed, but human trials fail

Multiple attempts to use anti-amyloid antibodies by drug companies have been unsuccessful, and some have given up, rather than trying to go in another direction.

No good predictor of who will develop AD. Accurate prediction allows intervention before too much neuronal loss

No way to screen potential treatments for human efficacy other than 5 or 10 years of administering to humans or trying it on animals.

Lack of direct access to the neurons in the brain of the affected individual

No personalized approach or precision medicine as we see for many cancers

Alzheimer Pathology and Imaging

The AD brain is notable for the presence of senile plaques and neurofibrillary tangles (31, 32). The plaques are composed primarily of fibrillar Aβ deposited extracellularly while the tangles are intracellular and consist of paired-helical filaments of hyperphosphorylated tau protein (33, 34). These pathological changes occur decades before initial clinical symptoms manifest (35, 36).

The American Colleges of Radiology and The American Colleges of Neurology each recommend structural imaging, such as noncontrast MRI in the evaluation of patients suspected of having AD (37, 38). Atrophy of the brain is a frequent finding on MRI in AD and correlates to the degree of cognitive impairment. Hippocampal atrophy is a validated AD biomarker (39). In general, the pattern of progression begins with early atrophy in the superior temporal region and the hippocampus, then subsequently in the amygdale and the remaining temporal regions and even later in the frontal association cortices (40, 41). Ventricular enlargement accompanies atrophy of the gyri and expanded sulci, indicating loss of both grey and white matter. As expected, a decrease in brain weight has been observed.

Aβ pathology can be imaged in human brain tissue in vivo by positron emission tomography (PET) using amyloid-β radiotracers (42, 43). The use of fluorodeoxyglucose positron emission tomography (FDG-PET) has allowed documentation of decreased glucose metabolism in the AD brain, generally occurring earliest in the posteromedial cortex and also in the temporal regions and correlating with neuronal or synaptic loss (44, 45).

Although imaging can show signs of incipient dementia before it becomes clinically overt, by the time imaging is abnormal, substantial neuronal loss has occurred (46). Further, progression from MCI to AD may be predicted with reasonable accuracy, but this does not lead to better outcome (47).

Current and Future Treatment: The Obstacles to Success

There is no disease modifying therapy for AD at this time (15, 48). Only modest symptomatic relief for between 6 and 18 months can be achieved, generally by using cholinesterase inhibitors to prevent degradation of acetylcholine and by using the glutamate receptor antagonist memantine to attenuate excitotoxicity (49). Acetylcholine is a neurotransmitter and neuromodulator that plays a critical role in forming and retrieving memories and maintaining attention. Cholinesterase inhibitors are the first-line medications in the treatment of AD. They slow the degradation of acetylcholine in the synaptic cleft, thus treating AD by increasing cholinergic function in the brain (50). The currently available cholinesterase inhibitors most frequently prescribed are donepezil, rivastigmine, and galantamine (51, 52). Although cholinesterase inhibitors can slow symptoms of AD dementia, they do not alter the inevitable neurodegeneration and cognitive decline (53). The N-methyl-D-aspartate antagonist memantine may delay cognitive decline, although not as effectively as cholinesterase inhibitors, and can be helpful in controlling agitation (54-56). Memantine, either as part of a multidrug regimen or by itself, is most clinically beneficial in people with moderate-to-severe AD (57). Memantine and anticholinesterase therapy are often combined, but it is not clear whether this is superior to the cholinesterase inhibitor alone (58).

In theory, Aβ removal or interference with Aβ aggregation will improve the signs and symptoms of AD (59, 60). Unfortunately, what is effective in mice, for reasons that are numerous and yet to be fully explained, does not translate to humans. Further, some anti-amyloid treatments have caused unacceptable side effects such as meningoencephalitis, brain edema, and brain microhemorrhage. The many failures in clinical trials of active and passive immunotherapies to remove Aβ have led Pfizer to close its neurology division and cease working on AD and Merck to suffer a major failure with verubecestat (61, 62).

Another impediment to designing treatments is the lack of early predictors of who will develop AD (Table 1). Identification of AD patients before they have undergone significant neurodegeneration becomes crucial if we are to preserve cognitive function (61). Current methods have limited accuracy in predicting progression to AD and the search for better biomarkers and imaging techniques is ongoing (63, 64). Currently, CSF biomarkers for AD that are measured in practice for some patients are Aβ42, total tau, and phosphorylated tau (65, 66).

Our lab and others are looking for early biomarkers. One possibility is micro(mi)RNAs, small endogenous non-protein-coding RNAs that influence the post-transcriptional regulation of gene expression and are involved in many neuronal processes (67, 68). A number of miRNAs show differential levels in the circulation and CSF in AD. These include miR-133b and miR-193a-3p which are downregulated in AD serum and miR-206 which is elevated in AD plasma (69-71). The predictive accuracy of any of these miRNAs has yet to be proven or brought into clinical use.

Not only can miRNAs serve as biomarkers, they may be targets for treatment because they can affect signaling pathways crucial to neuronal function (72). Circulating miRNAs are often carried in exosomes, small extracellular vesicles shed from all cells that contain cellular proteins, mRNA transcripts, miRNAs and lipids from their cell of origin (73). They are a fundamental mechanism of communication in the nervous system, allowing bidirectional cell signaling (74). The miRNAs within exosomes can transfer between neurons and microglia and can influence their phenotype. They can affect genes involved in Aβ generation (75). Extraction of exosomes that are shed from the neurons of the brain and CNS of humans with and without AD, may allow us to distinguish miRNAs from brain neurons that are altered by AD. This knowledge could then be used to identify signaling pathways relevant to nerve health and synaptic function that are modulated in AD and potentially to manipulate these in a beneficial direction to mitigate negative effects.

Our group is exploring a human cell culture model of AD that may circumvent or complement the need for rodents. It is not practical to do human brain biopsies and study brain cells in culture. If we could, that would be comparable to our approach to cancer (76). We can, however, approximate neurons from the brain using human induced pluripotent stem cells from patients clinically diagnosed with AD and differentiated to neural stem cells and then further differentiated to human cerebral cortical neurons (26, 77).

Neurons do not exist in isolation. To put a brain model together, microglia are essential because inflammation is undoubtedly part of the process of AD. With injury, the microglia proliferate and transform into active “brain macrophages,” also called reactive microglia. We know that amyloid and tau activate microglia (78, 79). In our model, we are looking at neurons and microglia together. There are differences in the mitochondrial bioenergetics in these AD-derived neurons and our unpublished results show that they respond differently than neurons from non-AD subjects to both direct exposure to high glucose and to conditioned medium from microglia exposed to high glucose. This study is in progress and as we and others continue to look beyond amyloid and tau, breakthroughs may be on the horizon.

Conclusions

AD is a common neurodegenerative disorder that leads inexorably to deterioration of cognitive functions, memory loss, motor impairment, and ultimately death. The need to develop new treatments is urgent and innovative thinking must prevail over repeated attempts to use the same approaches that have failed previously. In the interim there are fundamental steps to take now: encourage AD patients and healthy persons to eat a high quality diet, engage in regular physical activity, increase social connections and intellectual activities, avoid head trauma, and minimize heart disease risk factors: hypertension, obesity, high cholesterol, diabetes (keep A1C in normal range). These seem like very basic lifestyle behaviors that apply almost universally, but so many of us do not put in the effort to adhere to them.