Rationale for combining glutamatergic and cholinergic approaches in the symptomatic treatment of Alzheimer’s disease

Abstract

A total of 40 years of biochemical, clinical and neuropathological research have revolutionized our understanding of the pathophysiology of Alzheimer’s disease, yet at the present moment the only drugs licensed for treatment are targeted essentially at symptoms. Some disease-modifying drugs remain in clinical trials, but many that have used similar approaches have failed. It is therefore of considerable interest to examine the optimal way of using existing medications for the benefit of patients. This article looks at the rationale behind the combined use of acetylcholinesterase inhibitors and the N-methyl-d-aspartate-receptor antagonist, memantine, from both preclinical and clinical perspectives.

Keywords::

• acetylcholine
• acetylcholinesterase inhibitor
• Alzheimer’s disease
• donepezil
• galantamine
• glutamate
• memantine
• N-methyl-d-aspartate-receptor
• rivastigmine

Figure 1. Normal cortical cholinergic and glutamatergic neurotransmission, changes that occur in Alzheimer’s disease, and the proposed mechanism of action of acetylcholinesterase inhibitors and memantine with respect to these changes.

(A) Normal cortical cholinergic and glutamergic neurotransmission. Cholinergic innervation of glutamatergic cortical pyramidal neurons is provided by cells in the nbM in the basal forebrain. Acetylcholine (ACh) is released following depolarization to act on nicotinic and muscarinic receptors located on glutamatergic neurons, with neurotransmitter action being terminated by AChE. Glutamatergic neurons also receive innervation from other glutamatergic cortical pyramidal neurons – in this case, released glutamate acts on α-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid and NMDA receptors and action is terminated mostly by the reuptake of glutamate into associated glial cells. (B) Changes that occur in Alzheimer’s disease (AD). In AD atrophy and dysfunction of cholinergic and glutamatergic pyramidal neurons occurs such that remaining cortical neurons receive less innervation and consequently are less likely to be depolarized by synaptic signals. Signal-to-noise ratio at glutamatergic synapses would be reduced by a combination of a weaker ‘signal’ (from ACh and glutamate release in response to depolarization) and greater ‘noise’ (caused by an increased concentration of glutamate in the synaptic cleft between synaptic activation, due to partial failure of reuptake mechanisms). (C) The proposed mechanism of action of AChEIs and memantine with respect to the changes in AD. The use of AChEIs in AD is proposed to return the concentration of ACh in the synapse towards the normal level by reducing its breakdown, and thereby increasing the chance of the interaction of ACh with cholinergic receptors on glutamatergic pyramidal neurons. Memantine is proposed to reduce ‘noise’ at glutamatergic synapses by preventing the NMDA receptors from responding to the increased concentrations of glutamate present in the synaptic cleft between synaptic activation. In combination, it is proposed that the AChEIs and memantine would have their individual effects, but the potential for synergistic action occurs because an improved signal-to-noise ratio at glutamatergic synapses in the nbM would be expected to increase the firing rate of cholinergic neurons, and AChE inhibition would enhance this effect.

^†Glutamate receptor; α-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid receptor has been excluded for clarity.

ACheE: Acetylcholinesterase; AChEI: Acetylcholinesterase inhibitor; ACh-r: Acetylcholine receptor; nbM: Nucleus basalis of Meynert; NMDA-r: NMDA receptor.

Left part of each panel adapted with permission from [Danysz W, Unpublished Data].

Biochemical rationale for cholinergic & glutamatergic approaches in the treatment of cognitive & noncognitive symptoms

Cholinergic anatomy & pharmacology

Anatomical studies have identified the presence of cholinergic markers (choline acetyltransferase [ChAT], and vesicular acetylcholine transporter) associated with the cell bodies and dendrites of cortical neurons and, in particular, glutamatergic cortical pyramidal neurons [1,2]. Most cortical neurons, and many glia, appear to carry muscarinic and/or nicotinic receptors [3–5] making them sensitive to the activity of cholinergic basal forebrain neurons [3,6,7]. Cholinergic neurons that innervate the neocortex and hippocampus are located in the basal forebrain, with cell groups in the medial septum, diagonal band of Broca, and the nucleus basalis of Meynert (nbM) [8]. Furthermore, the activity of these cholinergic neurons is likely to be regulated by glutamatergic inputs from regions such as the amygdala, the brainstem reticular formation, the hippocampus and the cerebral cortex [9–12]. Consistent with these observations, the nbM expresses large numbers of N-methyl-d-aspartate (NMDA) glutamate receptors [13]. Various studies using microdialysis have investigated this regulation of cholinergic function by glutamatergic systems – the pharmacology appears complex and depends on the site of administration of the compounds. For example, administration of a glutamatergic antagonist intracerebroventricularly stimulated cortical acetylcholine (ACh) release [14,15], whereas the same compound given through the dialysis probe in the nbM inhibited ACh release [14]. It is therefore proposed that there is both a direct and an indirect (via a γ-aminobutyric acid inhibitory interneuron) regulation of ACh release by glutamate [14,16]. The importance of the direct pathway is supported by in vivo studies which showed that both NMDA and α-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid (AMPA), applied through a dialysis probe in the nbM, increased cortical ACh release in a concentration-dependent manner [17]; and by in vitro investigations in which NMDA antagonists inhibited ACh release from cortical slices [18,19], and NMDA-induced tonic firing of nbM neurons in basal forebrain slices [20].

A range of compounds that have been investigated as cognition enhancers have been shown to increase glutamate release in vivo, including an acetylcholinesterase inhibitor (AChEI) [6], nicotine and subtype-specific nicotinic agonists [21,22], a muscarinic agonist [6], serotonin (5-HT)_1A antagonists [23,24], and a 5-HT₆ antagonist [25]. In this context, it is a tenable hypothesis that a key feature of compounds that act as cognitive enhancers is likely to be their ability to act on receptors located on pyramidal neurons, influencing pyramidal neuron activity and glutamate release [26]. In the case of cholinomimetics, such an effect would be direct for muscarinic and nicotinic agonists, or indirect (by increasing the concentration of ACh at cholinergic–glutamatergic synapses) for AChEIs.

Cholinergic mechanisms in Alzheimer’s disease

The activation of cholinergic receptors underlies improvements in attention and other aspects of cognition [27], and cholinergic dysfunction is associated with the early cognitive impairments observed in Alzheimer’s disease (AD). Studies of patients in the early stages of AD have found a consistent reduction in presynaptic markers of the cholinergic system. One investigation found that the levels of ChAT activity, ACh synthesis and high-affinity choline uptake were reduced to approximately half of the control level in patients with a mean of 3.5 years since symptom onset [26,28]. A similar finding was reported in patients who had displayed symptoms for less than 1 year [28,29]. These changes in cholinergic marker enzymes also appear to correlate with the extent of cognitive decline in AD [28,30]. However, there is also evidence that ChAT activity may show stabilization, or even upregulation, in patients with mild cognitive impairment and early AD [30,31] – although ChAT is not the rate-limiting step of ACh synthesis, which may have an influence on the interpretation of these data.

Further to enzymatic findings, studies have shown that AD brains have a reduced number of some ACh receptor types – for example, fewer nicotinic [32] and M₂ muscarinic [33] receptors, located on presynaptic cholinergic terminals. By contrast, postsynaptic M₁ and M₃ muscarinic receptors appear to be relatively preserved, although there may be disruptions in signaling from M₁ receptors to their secondary messenger systems (via G proteins) [34–36]. There is also evidence that the extent of uncoupling between M₁ receptors and G proteins correlates with dementia severity [37]. These reports of postsynaptic disruption indicate that cholinergic abnormalities in AD may not be restricted to presynaptic locations, as previously thought. With regard to nicotinic receptors, the presynaptic reduction observed in AD reflects a decrease in both the α7 and α4β2-receptor subtypes [38–40]. The importance of the α4β2 receptor in AD pathology is further suggested by observations of a positive correlation between cognitive performance (Mini-Mental State Examination score) and α4β2-receptor binding [Francis PT, Unpublished Data]. Indeed, preliminary studies have shown that agents acting at these nicotinic sites may have therapeutic effects in AD [41], and several nicotinic agonists are currently in clinical development.

The AChEIs were developed to target presynaptic cholinergic dysfunction. They help to address cognitive deficits by preventing the breakdown of the already diminished ACh supply at the synaptic cleft. In this way, cholinergic neurotransmission is facilitated as the neurotransmitter binds and activates cholinergic receptors on postsynaptic cells, including glutamatergic pyramidal neurons [6,26]. The three main AChEIs marketed for the treatment of AD – donepezil, rivastigmine, and galantamine – are licensed for use in patients with mild to moderately severe AD (donepezil is now also licensed in some countries for severe AD). Tacrine, the first licensed AChEI, is also still marketed but is no longer recommended due to safety issues, including hepatotoxicity. A summary of the symptomatic mechanisms (as well as approval dates and indications) of marketed cholinergic drugs is given in Table 1. Details of other cholinergic compounds in development, and those that have been discontinued due to safety concerns or lack of efficacy, are also shown.

Donepezil, rivastigmine and galantamine have similar mechanisms of action, and similar efficacy profiles. However, they can be distinguished by some mechanistic differences. As reviewed by Wilkinson et al. [42], while all three drugs inhibit acetylcholinesterase (AChE), rivastigmine shows equally potent inhibition of butyrylcholinesterase, and in certain studies galantamine has demonstrated allosteric activation of α4 and α7-nicotinic receptors half-life. In terms of administration, donepezil’s long half-life (70 h), means that it can be dosed once daily [43], while oral rivastigmine (half-life = 1 h, with 9 h of AChE inhibition) and galantamine (half-life = 7–8 h) require twice-daily dosing [42,44,45], although a once-daily extended release formulation of galantamine [46] and a once-daily rivastigmine skin-patch [47] are both now widely available.

Translating these mechanisms into clinically relevant effects, large controlled studies have shown that AChEI monotherapy can produce cognitive benefits in most patients with mild-to-moderate AD [48]. In addition, the AChEIs have also been shown to improve noncognitive behavioral symptoms such as apathy, irritability and anxiety [49]. This latter effect may be explained by the reduction in ChAT and other cholinergic markers (e.g., M₂ receptors) that have been seen to correlate with behavioral changes in AD [33,50]. Therefore, cholinergic dysfunction may underlie more than cognitive deficit in AD. However, the symptomatic effects of AChEIs are frequently modest, suggesting that the cholinergic system cannot be the only contributing factor to cognitive (and behavioral) decline. It is likely that other neurotransmitter systems are also involved in mediating cognitive function, with cholinergic processes influencing particular aspects, such as attention and arousal [28,51]. In addition, the potential involvement of postsynaptic cholinergic dysfunction in AD raises the possibility that the limited function of AChEIs is related to their addressing only presynaptic deficits, while downstream signaling remains impaired.

Glutamatergic mechanisms in AD

Glutamate is the principal excitatory neurotransmitter of the CNS. It accounts for approximately two-thirds of synapses in the neocortex (ACh synapses account for ~5%) [52], and is involved in all aspects of cognition and higher mental function. In normal, healthy individuals, glutamatergic neurotransmission acts via both AMPA and NMDA receptors to produce long-term potentiation (LTP). LTP refers to a strengthening of synapses through repeated use and is central to the processes of learning and memory and, in addition to glutamatergic mechanisms, ACh and its receptors play a significant role in both the induction and maintenance of LTP [53,54]. This latter point again emphasizes the intimate relationship of cholinergic and glutamatergic systems in cognitive aspects of brain function.

In AD, the number of glutamatergic pyramidal neurons is significantly reduced [26,55,56], as is reflected in a decrease in wet weight and total protein content in the cortex [57,58]. In the cerebral cortex and hippocampus, there is concomitant loss of glutamatergic neurons [55,59] and abnormalities in NMDA and AMPA-receptor expression [60–63] in patients with AD. There is also evidence that NMDA glutamate receptors may be internalized on exposure to amyloid-b (Ab) [64], perhaps in part as a protective mechanism [65], which may also lead to an apparent reduction in receptor binding in AD as the study method used isolates of cell membranes. However, in severe pathological cases, or severely affected brain regions, these biochemical data may not follow typical patterns when expressed as wet weight or total protein. For example, markers of extrinsic neurons (e.g., ChAT activity for cholinergic neurons) are reduced in cortical regions in severe AD, as they form only a small proportion of the neuropil. By contrast, markers of pyramidal neurons contribute significantly to the total protein content and may, therefore, not always follow this pattern of reduction in severe disease [66]. Certainly, studies of b-tubulin, synaptophysin and the glial marker, glial fibrillary acidic protein (GFAP) have shown that these protein levels were unaltered relative to total protein content in the temporal cortex of patients with AD [66,67]. However, even at the end stage of the disease, GFAP levels were increased in the less severely affected parietal cortex.

With regard to understanding the disease process in AD, it is of considerable interest to consider the temporal relationship between the loss of cholinergic and glutamatergic neurons. Biochemical studies of biopsy samples indicate that cholinergic neurons of the basal forebrain are the first to be affected, as discussed above. However, as the histopathological definition of AD must include the presence of neurofibrillary tangles that occur in glutamatergic cortical pyramidal neurons and are found in preclinical cases of AD [68], it can be argued that changes in the glutamatergic system must be the first to occur. Such a conclusion could have implications for when a particular treatment approach might be expected to be effective, although it must be remembered that only a small proportion of glutamatergic neurons in selected regions will be involved while the loss of a significant proportion of cholinergic innervation of the entire cortex may be more relevant. It is clearly the case that, by the time of current clinical diagnosis, both cholinergic and glutamatergic systems are affected by the disease process.

As well as the consequences of complete neuronal loss, there may also be dysfunction in remaining glutamatergic neurons in the AD brain. A study of human autopsy samples (AD vs control brains) found that the level of the major glial glutamate transporter (excitatory amino acid transporter [EAAT2]) was unaffected in all the studied brain regions [69], whereas the ability of glial cells to remove glutamate from the synaptic cleft has been shown to be impaired in the temporal cortex, as well as other brain regions [70]. This may be explained by the action of free radicals on the glial glutamate transporter, as seen in the rat system with GLT-1 (the rat equivalent of EAAT2) [71]. In addition, although the level of the vesicular glutamate transporter, VGLUT1 (which packages neurotransmitter glutamate into vesicles for synaptic release), was unaffected in the AD temporal (but not parietal) cortex [66,67], its activity was lowered in this region [72]. The mechanism(s) by which these protein activities are altered are, as yet, unknown. However, regarding their consequences, it is proposed that in AD, inadequate removal of glutamate from the synaptic cleft leads to an abnormal build-up of the neurotransmitter between the pre- and post-synaptic neurons [73]. This raised background concentration (‘noise’) of glutamate causes the postsynaptic membrane to become more frequently depolarized, displacing the magnesium ion (Mg²⁺) that blocks the NMDA ion channel under physiological resting conditions [74]. The reduced efficiency of this voltage-dependent Mg²⁺ block impairs the detection of incoming physiological signals by the NMDA receptor and the capacity for generating LTP is reduced. This mechanism, involving dysfunction in glutamate neurotransmission, may contribute to cognitive impairment in AD [73–75]. Furthermore, excessive pathological glutamate stimulation can lead to neuronal death via excitotoxicity. This is thought to be a major factor in neuronal loss in stroke [59,74], and may also contribute to the cell death observed in AD, although the exact extent of this contribution remains unclear [26]. However, despite the correlations observed between the loss of glutamatergic neurons and cognitive decline in AD, very few reports have, as yet, documented a direct link between presynaptic glutamatergic markers and cognitive deficits at the neurochemical level [66,76].

Positive modulation of the glutamate NMDA and AMPA receptors has been tested as a strategy for normalizing glutamatergic function, with some agents undergoing clinical trials in AD, although no such treatment has received regulatory approval. An alternative approach, using an uncompetitive NMDA-receptor antagonist, memantine, has shown greater success in modulating the glutamatergic system in AD. Rather than simply providing a complete block of the NMDA receptor, memantine has moderate voltage dependency and fast blocking/unblocking kinetics. There is evidence that these characteristics allow it to block the NMDA-receptor channel during the constant background ‘noise’ of pathological glutamatergic activation at rest, while still being displaced upon the arrival of a physiological signal [74,75]. In this way, the neuron is likely to be protected against the neurotoxic effects of pathological glutamate stimulation [77], while also being able to manifest LTP [73,74]. As described earlier, AChEIs are able to stimulate glutamatergic neurons. Therefore, in combined use, it may be hypothesized that while memantine would reduce the background ‘noise’ of glutamate in AD, the AChEIs would serve to increase the level of physiological signals (Figure 1) [73].

As listed in Table 2, memantine is currently the only marketed drug for AD that targets the glutamatergic system, although other agents acting at NMDA and AMPA receptors are in development. Memantine is indicated for the treatment of moderate-to-severe disease stages, with support coming from multiple large-scale, controlled clinical studies, including demonstration of efficacy against overall and individual items of cognitive function [78,79]. In addition to cognitive benefits, a range of post hoc analyses (including pooled data) have demonstrated that memantine improves behavioral scores in general, and in particular, subscales associated with agitation and aggressive behavior [80–83]. The potential mechanisms underlying such improvements have been reviewed elsewhere [75].

Preclinical behavioral studies of individual cholinergic & glutamatergic approaches

The biochemical rationale for the use of individual cholinergic and glutamatergic therapies in AD is complemented by findings from cognitive behavioral studies in several animal model systems.

In a recent study using the novel object recognition test in rats, donepezil (and also memantine) was shown to have improved ‘natural forgetting’ when the animals were exposed to a second test after a 4-h intertrial interval [84]. In similar earlier studies, donepezil and metrifonate (also an AChEI) improved discrimination in the same test, particularly in older animals [85–87]. In addition, lesions of the fimbria-fornix, and local microinfusion of scopolamine (a muscarinic-receptor antagonist) into the hippocampus, both impair short-term object recognition in rats, whereas local microinfusion of physostigmine (a reversible AChEI) enhances object recognition [88]. Donepezil was shown to reverse the effects of scopolamine in a wide range of cognitive behavioral tests to a varying degree [89], and also ameliorated sustained attentional deficits in a five-choice serial reaction time test in the triple transgenic mouse model [90].

The effects of memantine on cognitive performance in animals have been the subject of previous reviews [91], with beneficial actions on aspects of memory seen in several models, including amyloid precursor protein APP23, double (APP/presenilin-1, PS1) and triple (APP/PS1/Tau) transgenic mice models. Memantine has also been shown to reverse the effects of intracerebral Ab injections on both object recognition memory and passive avoidance in rats [92].

In the APP transgenic mouse model of cognitive decline, memantine, galantamine and donepezil have all been shown independently to reverse cognitive deficits observed in the water maze test, both immediately [93] and after a 3-week washout, indicating a potentially disease-modifying action [94,95]. In addition, chronic memantine treatment has been shown to fully reverse, and donepezil to partially reverse, the effects of fimbria-fornix lesions on a delayed nonmatch to sample task in rats [16].

Preclinical pharmacological & behavioral studies of combined cholinergic & glutamatergic approaches

The efficacy of individual treatment approaches with contrasting cholinergic and glutamatergic mechanisms of action prompted investigation of the effects of targeting both systems for the treatment of AD.

To date, there have been few preclinical studies investigating the combined effects of memantine and an AChEI (mostly donepezil) on cognitive behavior in animals (Table 3). One study examined the effect of combination therapy on cognitive outcomes in rats, alongside a microdialysis investigation of ACh levels [16]. A strong limitation of the study was that all experiments were conducted in the presence of the AChEI, neostigmine, without which the levels of ACh would not be detected in this system [16]. In anesthetized rats, memantine or donepezil, given systemically, dose-dependently increased extracellular fluid concentrations of ACh in the neocortex and hippocampus, and when the drugs were combined at submaximal concentrations, greater increases were observed when compared with either drug alone [16]. Increases in ACh concentration were also observed following the administration of memantine, and memantine plus donepezil, in rats with partial fimbria-fornix lesions; however, the combination was no more effective than memantine alone [16]. Taken together, these results indicate that memantine is likely to increase ACh release both by increasing the firing rate of cholinergic neurons of the basal forebrain and at the level of cholinergic terminals. This latter action is not likely to be as a result of a direct effect on AChE inhibition, as therapeutic concentrations of memantine do not alter the degree of inhibition of AChEIs [96,97], and memantine alone does not act as an AChEI [96]. Initial attempts in this study to find a direct correlation between treatment-induced changes in ACh levels and improvements in cognitive outcomes were not successful [16].

Other studies in healthy, uncompromised animals, showed that while either approach alone could improve performance [98,99], no additional benefit was obtained when the drugs were combined [100,101]. Similarly, while individually memantine and donepezil ameliorated the effects of Ab injections in a delayed nonmatch to position task in rats, the combination was no more effective [102]. It is important to consider, however, that with drug combinations ceiling effects may come into play in situations where one drug is being used at an effective dose.

In more recent studies, memantine treatment alone and the combination of memantine and donepezil, significantly improved spatial memory (acquisition and retention in the Morris water maze) in 6-month and 15-month triple-transgenic mice with cognitive impairment and severe AD pathology (Ab plaques and neurofibrillary tangles) [103]. By contrast, donepezil treatment alone significantly improved the retention (but not the acquisition) of spatial memory in younger mice, but no significant effects were seen in older mice [103]. In both age groups, the memantine–donepezil combination was the only treatment that significantly improved the latency aspect of retention [103]. A further study in the APP23 transgenic mouse model has suggested that memantine and donepezil may have a synergistic effect on spatial memory, improving moving time (reflecting memory acquisition) and resting time (reflecting retrieval time), respectively, in a dry land maze test [104].

Clinical experience of combination treatment

Due, in part, to the logical nature of the approach, combination therapy is already widely used in clinical practice. Although compounds with a combined cholinergic–glutamatergic action have been trialed, none has yet been successful in the later stages of development (Table 4). Consequently, combination therapy currently involves the administration of one cholinergic drug (donepezil, galantamine or rivastigmine) and one glutamatergic drug (memantine).

There have been no clinical trials involving simultaneous initiation of therapy with memantine and an AChEI, rather all studies have involved memantine being added on in patients receiving stable AChEI therapy. There are also no studies looking at AChEI addition in patients already on stable therapy with memantine. This may partly be explained by their usual licensed indications being slightly different – mild to moderately severe AD for AChEIs, and moderate-to-severe AD for memantine.

In line with in vitro, in vivo and ex vivo preclinical findings [96,97,105], no pharmacokinetic interactions have been observed in clinical studies of memantine combined with donepezil, galantamine or rivastigmine [106–108]. These pharmacokinetic studies were performed in healthy volunteers, and in patients with mild to moderate AD, and their findings are consistent with the favorable safety profile observed in Phase III and postmarketing studies of combination treatment to date [79,109,110].

Efficacy data from the clinical studies have been systematically reviewed elsewhere [111], and are listed in Table 5. In summary, two large-scale randomized, double-blind, placebo-controlled Phase III studies have assessed the efficacy of memantine in patients already receiving stable doses of AChEIs. One of these studies included patients with moderate-to-severe AD [79], and the second included patients with mild-to-moderate disease [109]. In addition, pooled and meta-analyses and several post hoc assessments have been based on these study data (Table 5). Overall, the available studies indicate that combination treatment is superior to treatment with AChEI monotherapy in both general and specific measures of cognition, function, behavior and global status in patients with moderate-to-severe AD [79,80,82,83,112–114]. Furthermore, there is evidence that combination therapy may reduce the occurrence of clinical worsening (concurrent worsening across symptomatic domains) [115], and prevent the emergence of behavioral symptoms including agitation/aggression [83]. By contrast, combination therapy has shown no significant benefits over AChEI monotherapy in patients with mild disease [109,116–118]. Another prospective, placebo-controlled study of the memantine–donepezil combination (vs monotherapy with either agent) in patients with moderate-to-severe AD has recently been published [119]. Although the power of this study was limited by significant difficulties with patient recruitment, the efficacy of memantine–donepezil combination treatment was seen on the coprimary end point measures of cognition and function [119]. The superiority of combination therapy over monotherapy did not reach statistical significance, despite appearing differentiated for a large part of the treatment period – a finding which has since been described as ‘clinically relevant’ in expert commentary, which also suggests that there is no actual evidence of lack of effect of adjunctive memantine in this study due to issues such as small sample sizes at the end of the study and differential dropout rates [201].

Outside the clinical setting, observational studies have also provided interesting preliminary findings regarding the long-term and ‘real world’ application of combination therapy for AD. Although these studies are, by their nature, nonrandomized and open-label and there are potential confounds since the cohorts receiving the different treatments are potentially dissimilar and largely from different time periods, results indicate that combination therapy may retain its benefits over AChEI monotherapy for a period of years [120] and have an impact on meaningful outcomes, such as time to nursing home admission [121].

Expert commentary

The use of combination therapy with a cholinomimetic (AChEI) and a glutamatergic (memantine) agent for the treatment of AD is justified by findings from biochemical studies, and preclinical investigations in established animal models of AD. This is borne out by most clinical experience and, indeed, combination therapy is already considered standard treatment practice in many countries throughout the world. Despite this extensive use, the clinical evidence support for combination therapy is not complete, as it is based solely on the addition of memantine to existing AChEI treatment, with no data on the opposite order of initiation or the simultaneous treatment de novo with memantine and an AChEI. These appear to be gaps, but the available data actually reflect the real-world situation in which AChEIs (licensed for mild to moderately severe AD) are given early in the disease course, while memantine (licensed for moderate-to-severe AD) is initiated later. Therefore, the natural order of events is that patients who receive combination therapy will have started their treatment with AChEI monotherapy.

There is a great hope that disease-modifying treatments will succeed in the near future although the risk is that those based on Ab are being investigated through clinical trials too late in the disease course. New initiatives are being considered where people with autosomal dominant inherited AD and those with genetic risk factors such as apolipoprotein E4 (APOE4) are given anti-Ab compounds. This would arguably be the strongest possible test of the amyloid cascade hypothesis.

The direction taken by AD management over the coming years will very much depend on the effectiveness of these treatments; if effectiveness is less than clear cut, then the adoption of such novel therapies is likely to be limited. However, regardless of the outcome with disease-modifying medications, current symptomatic treatments will continue to be the foundation of therapy for many patients with AD, including those diagnosed too late to benefit from any disease-modifying option – with particular relevance in developing countries. Therefore, combination therapy with memantine and an AChEI is likely to remain a widely used approach. In addition, it is recommended that new treatments for cognitive symptoms, and especially noncognitive symptoms, ought to be a major priority for development. These treatments should be identified by a combination of pragmatism – assessing what is available and what might work – as well as an effort to pinpoint mechanism-based approaches. In this context, it is relevant to recognize an important initiative by the American Alzheimer’s Association in bringing together scientists, clinicians and pharmaceutical companies in a ‘round table’ on behavior in dementia. This has resulted in the formation of a Professional Interest Area (PIA) on neuropsychiatric symptoms and the establishment of several working groups to address individual symptoms [122].

Five-year view

The future of the AD treatment landscape in 5 years’ time will depend almost entirely on two factors – whether disease-modifying treatments are proven effective and whether they represent value for money. The expectation is that the current symptomatic drugs will still be used extensively, with clinical trials of other drugs to target behavioral disturbance near to completion. One of the current disease-modifying approaches may show promise at this time, but may not be widely available due to the limitations of cost, complexity of administration or side effects.

Diagnosis would be expected to present an enormous challenge worldwide, with the need to facilitate diagnosis in primary care balanced against the development of more sophisticated and expensive techniques (e.g., positron-emission tomography scanning) to improve diagnostic accuracy. Furthermore, in line with the push towards disease modification, the characterization of biomarkers to aid in early diagnosis is likely to have advanced.

Together, the increased range of options for diagnosis and treatment will undoubtedly present hard choices, creating (one would hope) the potential for enhanced treatment effects, but also producing greater disparity in care availability within and between countries.

Table 1. Cholinergic drugs in clinical development/marketplace.

Drug	Receptor activity	Phase	Key results of clinical trials	Approval date and indication
Donepezil	Reversible, noncompetitive, selective AChEI [43]	IV (marketed) [123]	Positive effect on global status, cognition, function and behavior; gastrointestinal side effects [124]	1997; all AD stages (mild to moderately severe AD in Europe) [43,125]
Rivastigmine	Pseudo-irreversible AChEI and BChEI [44]	IV (marketed) [123]	Positive effect on global status, cognition and function in mild-to-moderate AD; gastrointestinal side effects [126]	1998; mild to moderately severe AD [44]
Galantamine	Reversible, competitive, selective AChEI; nicotinic-receptor allosteric modulator [45]	IV (marketed) [123]	Positive effect on cognition in mild-to-moderate AD; possible positive effect on function and behavior; gastrointestinal side effects [127]	2000; mild to moderately severe AD [45]
Tacrine	Reversible AChEI and BChEI [128]	(Marketed; no longer recommended) [128]	Considerable side effects, including hepatotoxicity [128]	1993; mild to moderately severe AD
Phenserine	AChEI [123]	III (discontinued) [123]	Possible positive effect on cognition; nonsignificant [123]
Huperzine A	Reversible, competitive, selective AChEI [129,130]	II/III [123,129,130,202]	Possible positive effect on global status, cognition (memory) and function; insufficient evidence [128–131]	1994; AD (China only) [130]
Ispronicline (AZD-3480)	Selective nicotinic α4β2-receptor agonist [123,132]	II/III [123,132,202]	Possible positive effect on cognition (memory); nonsignificant [123]
AZD-1446	Nicotinic α4β2-receptor activator [203]	II [202]	No cognitive or global benefits in AD [204]
EVP-6124	Selective nicotinic α7-receptor agonist [123,132]	II [123,132,202]	Possible positive effect on cognition (nonverbal learning, memory, and executive function) [132,133]
Ladostigil (TV-3326)	AChEI; BChEI [134]	II [202]	Reverses memory deficits in animals [134,202]
ST101 (ZSET1446)	Potentiates nicotine-induced acetylcholine release; not an AChEI [135]	II [202]	Improves learning and memory and reverses cognitive deficits in animals [135,136,202]
Varenicline	Selective nicotinic α4β2-receptor partial agonist [137]	II [202]	No cognitive benefits in mild-to-moderate AD [202]
Pozanicline (ABT-089)	Nicotinic α4β2 and α6β2-receptor partial agonist [123]	II (discontinued) [123,202]	Reverses hyoscine-induced cognitive deficits in healthy volunteers; no cognitive benefits in mild-to-moderate AD [123]
RG 3487	Selective nicotinic α7-receptor partial agonist [132]	II (discontinued) [132]	Improves cognitive function in mild-to-moderate AD [132]
AF-102B	M1 muscarinic-receptor agonist [123]	I (discontinued) [123]	Cognitive benefits in animals [138]; undesirable side effects in humans [123]
Talsaclidine (WAL 2014)	M1 muscarinic-receptor agonist [123]	I (discontinued) [123]	Modest cognitive benefits in animals [139]; undesirable side effects in humans [140]
Lecithin	Dietary source of choline [141]	N/A	Insufficient evidence [141]
Nicotine	Nicotinic α4β2 and α7-receptor agonist [142]	N/A	Insufficient evidence [142]
Physostigmine	Reversible AChEI and BChEI [128]	(Discontinued) [143]	Possible positive effect on cognition (memory); gastrointestinal side effects [128,143]
Metrifonate	Irreversible, nonselective AChEI and BChEI [144]	(Discontinued) [144]	Positive effect on global status, cognition and function in mild to moderate AD; respiratory failure and death [144]
Velnacrine	Reversible AChEI [145]	(Discontinued)	Considerable side effects, including hepatotoxicity [145]

AChEI: Acetylcholinesterase inhibitor; AD: Alzheimer’s disease; BChEI: Butyrylcholinesterase inhibitor; N/A: Not applicable.

Table 2. Glutamatergic drugs in clinical development/marketplace.

Drug	Receptor activity	Phase	Key results of clinical trials	Approval date and indication
Memantine	Uncompetitive, voltage-dependent NMDA-receptor antagonist [123]	IV (marketed) [123]	Positive effect on global status, cognition and function in moderate-to-severe AD; reduces agitation development; well tolerated [146]	2002; moderate-to-severe AD [147]
Neramexane	Uncompetitive NMDA-receptor antagonist [98]	III [202]	Positive effect on memory in animals [98]
CX-516	AMPA-receptor positive allosteric modulator [202]	II [202]	Possible positive effect on memory and cognition [148]
D-cycloserine	NMDA-receptor modulator [149]	II (nondemented elderly) [202]	Positive effect on memory in healthy subjects [150]; no benefits in AD [149]
LY451395	AMPA-receptor potentiator [132]	II [132,202]	No cognitive benefits in mild-to-moderate AD; possible behavioral benefit [151]
EVT 101	NMDA-receptor antagonist (at NR2B subunit) [132]	I (healthy subjects) (discontinued) [132,202]	Possible positive effect on brain function in healthy subjects [132]

AD: Alzheimer’s disease; AMPA: α-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid; NMDA: N-methyl-d-aspartate; NR: NMDA receptor.

Table 3. Preclinical investigations of memantine–acetylcholinesterase inhibitor combination treatment.

Study (year)	Title	Ref.
Ihalainen et al. (2011)	Effects of memantine and donepezil on cortical and hippocampal acetylcholine levels and object recognition memory in rats	[16]
Martinez-Coria et al. (2009)	Combination of memantine and donepezil reverses cognitive deficits in transgenic mice with both amyloid-β plaques and neurofibrillary tangles	[103]
Neumeister and Riepe (2012)	Synergistic effects of antidementia drugs on spatial learning and recall in the APP23 transgenic mouse model of AD	[104]
Wise and Lichtman (2007)	The uncompetitive NMDA-receptor antagonist memantine prolongs spatial memory in a rat delayed radial-arm maze memory task	[100]
Woodruff-Pak et al. (2007)	Preclinical investigation of the functional effects of memantine and memantine combined with galantamine or donepezil	[101]
Yamada et al. (2005)	Effects of memantine and donepezil on amyloid β-induced memory impairment in a delayed-matching to position task in rats	[102]

APP: Amyloid precursor protein; NMDA: N-methyl-d-aspartate; NR: NMDA receptor.

Table 4. Mixed cholinergic and glutamatergic drugs in clinical development/market place.

Drug	Receptor activity	Phase	Key results of clinical trials
Latrepirdine (dimebon)	Weak AChEI and BChEI; glutamate-receptor modulator; mitochondrial modulation [152]	III [132,202] (discontinued) [205]	Insufficient evidence from Phase III (despite positive Phase II findings) [152]
Nefiracetam	Nicotinic-receptor potentiator; NMDA-receptor potentiator; modulates the glycine binding site of the NMDA receptor [153]	II [202]	Possible positive effect on cognitive function [202]
Piracetam	α-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid-receptor allosteric modulator [154]; muscarinic cholinergic activity [155]	NA	Insufficient evidence [155]

AD: Alzheimer’s disease; AChEI: Acetylcholinesterase inhibitor; BChEI: Butyrylcholinesterase inhibitor; NA: Not applicable.

Table 5. Studies examining the efficacy of memantine–acetylcholinesterase inhibitor combination treatment in patients with Alzheimer’s disease/dementia.

Study (year)	Title	Publication	Ref.
Prospective clinical studies
Tariot et al. (2004)	Memantine treatment in patients with moderate to severe AD already receiving donepezil. A randomized controlled trial	JAMA 291(3), 317–324 (2004)	[79]
Porsteinsson et al. (2008)	Memantine treatment in patients with mild to moderate AD already receiving a cholinesterase inhibitor: a randomized, double-blind, placebo-controlled trial	Curr. Alzheimer Res. 5(1), 83–89 (2008)	[109]
Choi et al. (2011)	Tolerability and efficacy of memantine add-on therapy to rivastigmine transdermal patches in mild to moderate AD: a multicenter, randomized, open-label, parallel-group study	Curr. Med. Res. Opin. 27(7), 1375–1383 (2011)	[116]
Dantoine et al. (2006)	Rivastigmine monotherapy and combination therapy with memantine in patients with moderately severe AD who failed to benefit from previous cholinesterase inhibitor treatment	Int. J. Clin. Pract. 60(1), 110–118 (2006)	[156]
Howard et al. (2012)	Donepezil and memantine for moderate-to-severe AD	N. Engl. J. Med. 366, 893–903 (2012)	[119]
Olin et al. (2010)	Safety and tolerability of rivastigmine capsule with memantine in patients with probable AD: a 26-week, open-label, prospective trial (study ENA713B US32)	Int. J. Geriatr. Psychiatry 25(4), 419–426 (2010)	[157]
Riepe et al. (2007)	Domain-specific improvement of cognition on memantine in patients with AD treated with rivastigmine	Dement. Geriatr. Cogn. Disord. 23(5), 301–306 (2007)	[158]
Post hoc analyses
Cummings et al. (2006)	Behavioral effects of memantine in AD patients receiving donepezil treatment	Neurology 67(1), 57–63 (2006)	[83]
Farlow et al. (2010)	A 25-week, open-label trial investigating rivastigmine transdermal patches with concomitant memantine in mild-to-moderate AD: a post hoc analysis	Curr. Med. Res. Opin. 26(2), 263–269 (2010)	[117]
Feldman et al. (2006)	Activities of daily living in moderate-to-severe AD: an analysis of the treatment effects of memantine in patients receiving stable donepezil treatment	Alzheimer Dis. Assoc. Disord. 20(4), 263–268 (2006)	[112]
Schmitt et al. (2006)	Cognitive response to memantine in moderate to severe AD patients already receiving donepezil: an exploratory reanalysis	Alzheimer Dis. Assoc. Disord. 20(4), 255–262 (2006)	[113]
van Dyck et al. (2006)	A responder analysis of memantine treatment in patients with AD maintained on donepezil	Am. J. Geriatr. Psychiatry 14(5), 428–437 (2006)	[114]
Winblad et al. (2006)	Memantine in moderate-to-severe AD: a meta-analysis of randomized clinical trials	Dement. Geriatr. Cogn. Disord. 24, 20–27 (2007)	[80]
Observational & long-term studies
Atri et al. (2008)	Long-term course and effectiveness of combination therapy in AD	Alzheimer Dis. Assoc. Disord. 22, 209–221 (2008)	[120]
Lopez et al. (2009)	Long-term effects of the concomitant use of memantine with cholinesterase inhibition in AD	J. Neurol. Neurosurg. Psychiatry 80(6), 600–607 (2009)	[121]
Schneider et al. (2011)	Treatment with cholinesterase inhibitors and memantine of patients in the AD Neuroimaging Initiative	Arch. Neurol. 68(1), 58–66 (2011)	[118]

AD: Alzheimer’s disease.

Key issues

• • Cholinergic deficit is a feature of Alzheimer’s disease (AD) that correlates with disease severity.

• • Glutamatergic dysfunction is another major contributory factor to AD.

• • Preclinical evidence supports the targeting of cholinomimetic and glutamatergic mechanisms for the treatment of AD, with improved cognition observed in pharmacological and behavioral studies in animal models.

• • In the clinical setting, cholinomimetic therapy (with acetylcholinesterase inhibitors) and N-methyl-d-aspartate-receptor antagonism (with memantine), have shown significant symptomatic efficacy and are approved for the treatment of AD.

• • The acetylcholinesterase inhibitors and memantine have been widely used as monotherapies for AD over the past decade but, increasingly, the application of both approaches in combination treatment is seen as the standard of care, although no studies looking at the simultaneous use of the two types of drug de novo have been conducted.

• • Preclinical and clinical evidence indicates that combination therapy is the way forward for the treatment of AD.

Financial and competing interest disclosure

PT Francis has received speaker bureau honoraria and grant support from H. Lundbeck A/S; speaker bureau honoraria from Novartis and Eisai; and has provided expert testimony to the UK High Court in a patent challenge against Novartis. CG Parsons is an employee of Merz Pharmaceuticals GmbH. RW Jones has received speaker bureau honoraria and/or consultancy fees from H Lundbeck A/S, Merz, Pfizer, Eisai and Novartis, as well as other companies involved in the development of new compounds for the treatment of Alzheimer’s disease and other dementias. The authors have no other relevant affiliations or financial involvement with any organization or entity with a financial interest in or financial conflict with the subject matter or materials discussed in the manuscript apart from those disclosed.

Writing assistance was utilized in the production of this manuscript. The authors gratefully acknowledge the editorial support provided by Cambridge Medical Communication Limited, funded by H. Lundbeck A/S.