FormalPara Key Points

Cholinesterase inhibitors produce small improvements in cognitive function in mild to moderate dementia, with marginal effects seen in severe disease, long-term treatment, and advanced age.

Cholinesterase inhibitors are associated with significant cholinergic side effects, including weight loss, debility, and syncope, which can be detrimental in the frail elderly population.

The available evidence indicates that memantine monotherapy produces small benefits in cognitive function in moderate to severe dementia, without significant adverse effects.

1 Introduction

Dementia is a clinical syndrome characterized by acquired global cognitive deficits that interfere with social and occupational functioning [1]. It typically presents in elderly individuals and involves a progressive decline in memory, reasoning, language, and executive function [2, 3]. The estimates for dementia prevalence vary based on geographic location and diagnostic criteria used, but it is thought that at least 5 % of people over the age of 60 years live with dementia, and the total number of these cases is expected to double every 20 years [4]. Incidence of dementia increases exponentially with advancing age, with one population study observing it doubling every 5 years [5]. Due to the rapid expansion of the elderly population, dementia is a serious public health concern, with an estimated worldwide cost of US$600 billion in 2010 [6].

Currently, there is no cure for dementia, and no treatments exist to halt or reverse its progression, so therapeutic interventions are targeted to treat symptoms or to improve cognitive function [7–9]. The two treatments with marketing approval for dementia are the cholinesterase inhibitors (ChEIs) donepezil, galantamine, and rivastigmine, and the low-affinity N-methyl-d-aspartate (NMDA) antagonist memantine [1, 10]. The vast majority of published data available on dementia treatment are in short-term industry-sponsored trials of ChEIs [11]. The trials have mostly been limited to patient populations with a mean age of approximately 75 years, with relatively mild Alzheimer’s dementia and few to no comorbidities [11, 12]. This significantly decreases the clinical applicability of the research to real-world cases of dementia, as the vast majority of patients, such as those with severe disease, advanced age, or frailty, would have been ineligible for these trials, and the benefit of long-term treatment is unclear [13–18]. It is also important to distinguish between statistical and clinical significance, as a small change in cognitive status may not translate into a meaningful change for the patient [1, 2, 19]. In addition, a risk–benefit assessment must include quantification of adverse effects, especially in old frail individuals in whom side effects may have serious consequences for survival and quality of life [2, 11, 20].

2 Methods

We performed a literature search of MEDLINE through November 2014 for English-language studies evaluating the risks and benefits of ChEIs and memantine in the treatment of dementia and mild cognitive impairment (MCI). Additional references were identified by reviewing the bibliographies of relevant papers. Whenever possible, pooled trial data from meta-analyses or systematic reviews were obtained. Search terms for cognitive disorders included dementia, mild cognitive impairment, cognition disorder, Alzheimer, Parkinson’s dementia, Lewy body dementia, and vascular dementia. Search terms for treatment included cholinesterase inhibitor, acetylcholinesterase inhibitor, donepezil, galantamine, rivastigmine, and memantine. Studies were included for review if they were randomized trials or observational studies that evaluated efficacy outcomes (cognitive, neuropsychiatric, or global function, and disease progression) or adverse outcomes associated with treatment. Studies were excluded if they evaluated non-FDA approved treatments for dementia, such as ginkgo biloba, vitamin E, and hormonal treatments [2, 10]. In addition, studies were excluded if they evaluated ChEIs or memantine in disorders other than dementia and MCI.

Two independent reviewers (JSB, SRS) completed data abstraction and evaluated the quality and risk of bias of included studies. The results were then presented in a systematic review, following PRISMA (preferred reporting items for systematic reviews and meta-analyses) guidelines. The primary scales used to measure cognition deficits were the mini-mental state examination (MMSE), the Alzheimer’s Disease Assessment Scale (ADAS) cognitive subscale (ADAS-cog), and the severe impairment battery (SIB). Other domains assessed included the activities of daily living (ADL), instrumental activities of daily living (IADL), and the neuropsychiatric inventory (NPI). Global function and disease progression were also assessed. Adverse events assessed included abdominal pain, nausea, vomiting, diarrhea, anorexia, headache, dizziness, insomnia, abnormal dreams, vertigo, agitation, muscle cramps, tremor, bradycardia, syncope, pacemaker implantation, fatigue, asthenia, weight loss, falls, and drug–drug interactions.

The search identified approximately 6800 potentially relevant articles, of which 540 were retrieved and evaluated. Of those, 257 studies were included in the systematic review. Study inclusion is represented in a flow diagram (Fig. 1).

Fig.1
figure 1

Flow chart of studies search

Full size image

3 Results

3.1 Epidemiology and Assessment of Dementia

There are several types and causes of dementia, with confusion related to variations in diagnostic criteria and the fact that many forms of dementia exist concurrently [2, 21]. Alzheimer’s disease is the most common cause, accounting for roughly 50–70 % of all dementia cases [22]. Another large subgroup is vascular dementia, which is often observed concurrently with other forms, resulting in prevalence estimates that vary considerably [23–28]. Mixed Alzheimer’s disease and vascular dementia is common, accounting for approximately 25 % of all dementia cases [29]. Vascular dementia becomes increasingly prevalent with advanced age, with at least 70 % of those over the age of 90 years having some degree of vascular pathology [13]. Finally, Parkinson’s disease dementia and dementia with Lewy bodies are two major types of dementia with similar neuropathological presentations, indicating that they are part of the same condition or exist within a spectrum of disorders, often collectively called Lewy body dementia [30]. Lewy body dementia may account for 10–20 % of global dementia cases, with the risk increasing with age and years since Parkinson’s disease onset [31–34]. Many other forms of clinically relevant dementia exist, but this review focusses on these three main subgroups.

MCI is classified as a cognitive decline that exceeds the typical aging process yet does not meet the diagnostic criteria of dementia [35]. This cognitive decline usually manifests as a loss in memory that is not significant enough to alter ADL [35]. The prevalence estimates for MCI range from 5 to 20 % in older individuals, with up to one-third of MCI cases progressing to dementia within 2 years [7]. Although MCI is a distinct entity, it is often considered to be a pre-clinical stage of dementia [7]. Studies have evaluated whether pharmacological interventions are capable of delaying the progression to dementia or improving the quality of life of individuals with MCI.

Various assessment tools are available to measure cognitive function, which have been used to determine the presence and severity of dementia as well as the progression of the disease [36]. One standard measurement is the MMSE, which is easy to administer in the clinical setting and measures cognitive function on a 30-point scale, with higher scores corresponding to higher cognitive function [37, 38]. Although the MMSE has its limitations, including decreased utility for detecting dementia in patients with minimal cognitive deficits or those with low education levels, it has been widely used and validated globally, and is quite accurate for diagnosing and tracking mild to moderate dementia [39–42]. Another common tool is the ADAS-cog, a 70-question scale with higher scores indicating worse cognitive performance, which was designed to detect and follow clinical changes in cognition [43, 44]. The ADAS-cog is most sensitive to changes in mild to moderate dementia [44–48]. Another cognitive assessment tool is the SIB, which is most helpful for those with severe dementia [49].

Other tests are commonly used to measure outcomes other than cognition in patients with dementia. An important measure in the determination of quality of life and degree of disability is the assessment of ADL, such as feeding and dressing, and IADL, such as handling money, preparing food, or using the telephone [36, 50, 51]. In addition, behavioral and psychological symptoms can be assessed with the NPI, which focuses on behavioral disturbances such as delusions, anxiety, and disinhibition [51, 52]. Another research tool is the measurement of global dementia status, which integrates in-depth clinical assessments of patients with caregiver input to derive a basic quantitative measure of change [53].

There are no universally accepted cut-offs for defining dementia severity, but it is reasonable to use MMSE scores to divide dementia severity into mild (20–23), moderate (10–19), severe (1–9), and end-stage (0) [40, 41, 54–59]. MCI corresponds to an MMSE score of 24–27, although this is not sufficient as a sole diagnostic criterion [58, 60–62]. Mild dementia can be detected by basic clinical assessment and is characterized by decreased memory of personal and current events and slightly diminished executive function [41]. In moderate dementia, the individual begins to lose independence and conversational skills and may forget significant personal information [56, 63]. By the stage of severe dementia, there is a complete loss of independence in performing ADLs associated with lack of orientation, but with some maintained ambulatory and verbal functioning [63, 64]. End-stage dementia, in which the individual no longer has the ability to walk or speak, is an important clinical stage that may persist for years before death [63, 64].

3.2 Pathophysiological Basis of Dementia with Rationale for Treatment

Cognitive decline in dementia is due in part to deficient cholinergic activity in the basal ganglia and neocortex caused by destruction of cholinergic neurons. ChEIs function by inhibiting acetylcholinesterase, thereby increasing available acetylcholine [65–67]. Another cause of cognitive decline common to various types of dementia is excitatory neurotoxicity caused by overstimulation of NMDA-glutamate receptors, and memantine could serve by blocking NMDA activity at a low enough affinity to prevent excitotoxicity without interfering with necessary functions of glutamate in learning and memory [68–71]. Despite very different pathogenic mechanisms, there are key similarities in the pathophysiology of Alzheimer’s dementia, Lewy body dementia, and vascular dementia that result in cognitive decline, and these may indicate ChEIs and memantine as possible treatment options.

In Alzheimer’s disease, the initiating event is thought to be the accumulation of amyloid beta-protein (Aβ) plaques, which in turn trigger hyperphosphorylation of tau protein and neuronal degeneration [72]. The Aβ plaques cause a decrease in acetylcholine release and synthesis, along with increased activity of acetylcholinesterase, impaired muscarinic acetylcholine signaling pathways, and a decrease in cholinergic signaling and function; these changes in turn indirectly result in disrupted NMDA-receptor activity and glutamate neurotoxicity [73, 74]. Since the plaques are concentrated in areas related to memory and executive function, such as the basal ganglia and parts of the temporal lobe and neocortex, the neural damage caused by cholinergic impairment and glutamate neurotoxicity is related to a loss of these cognitive functions [65].

Vascular dementia results from varied cerebrovascular events such as ischemia, infarct, or hemorrhage, so its pathogenesis is multifactorial and poorly defined [75]. The cognitive deficits involved resemble those found in Alzheimer’s disease and other dementia types, as they involve cerebrovascular damage to similar neural regions, such as the temporal lobe, basal ganglia, and neocortex [75]. These vascular changes in the brain are associated with inflammation, atrophy, and decreased cerebral perfusion, as well as decreased inhibition of NMDA channels, leading to chronic overstimulation of glutamate receptors [71, 75]. There is also some evidence that these changes may result in impaired cholinergic transmission [67, 76].

Lewy body dementia is a multisystem disease with various presentations along a spectrum; if the parkinsonian extrapyramidal symptoms precede dementia, it is considered to be Parkinson’s disease dementia, but if the order of presentation is reversed it is typically diagnosed as dementia with Lewy bodies [77]. Both disorders eventually involve the production of proteinaceous cytoplasmic inclusions throughout the brain, called Lewy bodies, with neurodegeneration of dopaminergic neurons in the substantia nigra and denervation of cholinergic neurons in the basal forebrain and cerebral cortex [77–79]. There is some evidence that these conditions may also induce glutamate hyperactivity by altering the expression of glutamate receptors, and this is often worsened by chronic l-dopa administration, which is commonly used in the treatment of Parkinson’s disease [69, 80, 81].

Ultimately, the use of ChEIs may serve to compensate for decreased cholinergic activity by increasing the availability of acetylcholine released from limited cholinergic neurons, and memantine may offer neuroprotective effects from excitotoxic stimulation of overexpressed NMDA pathways, but neither medication can reverse or slow the underlying pathophysiological progression of the disease. Therefore, the target of these therapies is to attempt to reduce symptoms related to the disease, rather than address the neurodegenerative processes themselves.

3.3 Cholinesterase Inhibitors

3.3.1 Benefits of Cholinesterase Inhibitors for Mild Cognitive Impairment

No medication has marketing approval for the treatment of MCI, but many investigators have evaluated whether ChEIs could delay the progression to dementia or provide clinically significant improvements in cognition for individuals with MCI. A few small studies have shown some minimal cognitive benefits of ChEIs in the treatment of MCI [56, 82–84], but pooled data analyses of the trials show that there is no significant difference in progression to dementia, nor in global, cognitive, or neuropsychiatric outcomes [7, 85, 86]. A significant increase in adverse events was observed for MCI patients treated with ChEIs, mainly gastrointestinal side effects such as nausea, diarrhea, and vomiting, as well as cardiac syncope and headaches [7, 54, 84, 85, 87, 88]. Therefore, ChEIs are not recommended for the treatment of MCI [89, 90].

3.3.2 Benefits of Cholinesterase Inhibitors for Dementia

Numerous studies have investigated the efficacy of different treatments for cognitive function in dementia, and several systematic reviews and meta-analyses have worked to pool the data and summarize the results (Table 1) [1, 2, 11, 20, 30, 56, 91–96]. The three commercially available ChEIs, donepezil, galantamine, and rivastigmine, have marketing approval for the treatment of mild to moderate dementia in most countries [10, 56]. The first-generation ChEI, tacrine, was removed from the US market due to poor efficacy and high incidence of serious adverse events, and is not evaluated here [2, 20]. The available ChEIs differ slightly in their pharmacological properties, but they share the main mechanism of acetylcholinesterase inhibition [2, 97]. Studies have shown similar efficacies with these medications so the data will be pooled for this review [2, 91, 98, 99]. A meta-analysis also concluded that results in efficacy for low and high doses of ChEIs were comparable, so results from studies using different doses will also be pooled [20].

Table 1 Benefit assessment of cholinesterase inhibitors for dementia by type
Full size table

Essentially all of the published trials on ChEI use in dementia were industry sponsored, with the exception of one long-term trial by the Alzheimer’s Disease Collaborative Group [100, 101]. This raises the possibility of publication bias or selective reporting of the results. One study evaluating the potential bias in reporting adverse events found that most of the randomized trials stated that they examined harm, but many did not report mortality data or provide clear definitions and detailed analyses of harm [102].

A recent clinical practice guideline by the American College of Physicians and the American Academy of Family Physicians presented the available evidence on ChEI treatments based on systematic evidence reviews sponsored by the Agency for Healthcare Research and Quality [1, 11, 93]. They concluded that ChEI treatment for dementia results in a small statistically significant improvement in cognitive function that was not considered to be clinically important [1]. Pooled data from ChEI trials in Alzheimer’s disease, Lewy body dementia, and vascular dementia show average improvements in the MMSE ranging from 0.8 to 1.6 points on a 30-point scale, while 3 points or more was generally considered to be clinically significant [1, 12, 30, 93, 94, 103]. Pooled data also showed an average improvement in ADAS-cog of approximately 1.4–2.7 points on a 70-point scale, while several studies have used a change of 4 points or more to indicate clinical relevance [1, 12, 93, 94, 103].

The majority of available data has been in mild to moderate Alzheimer’s disease, with limited data on Lewy body dementia and vascular dementia [11]. The pooled data from meta-analyses on ChEIs have shown relatively similar results for cognitive outcomes in Alzheimer’s dementia and Lewy body dementia, with improvements on the order of 1.5 points in MMSE and 2.5 points in ADAS-cog compared with placebo [12, 30]. In comparison, trials on vascular dementia show marginal cognitive effects of ChEIs (0.8 points in MMSE and 1.5 points in ADAS-cog), with some studies finding no improvements at all [94].

In addition to cognitive assessment, many studies also assessed global and functional outcomes of ChEI treatment. The results from studies on global outcomes have been mixed, with some studies concluding that there is no significant change and others finding slight improvements [12, 30, 93, 94]. Pooled data show a mean improvement of 0.4–0.5 for treatment in Alzheimer’s and Lewy body dementia in a 7-point interview-based global assessment of change scale [30, 93, 104]. Functional measures of ADL showed similarly mixed results for the treatment of Alzheimer’s and Lewy body dementia, with studies achieving improvements on the order of 0.1 standard deviations compared with placebo, which was not considered to be clinically significant. [1, 28, 87, 89, 93] Studies on ChEIs for vascular dementia have found negligible improvements in global and functional measures [94]. A secondary outcome often studied in dementia trials is the incidence and severity of neuropsychiatric symptoms, including behavioral and psychiatric disturbances, common in patients with dementia [104]. As with global and functional scores, pooled data showed a marginal improvement of about 2 of 144 points in NPI for Alzheimer’s and Lewy body dementia, and no improvement for vascular dementia [91, 94, 103].

Very few studies have evaluated severe dementia (MMSE 1–9), and none have looked at end-stage dementia (MMSE 0). In fact, only one published study with 248 participants examined monotherapy of ChEIs in patients with severe dementia (MMSE 1–10) [105]. If the criteria were expanded to define severe dementia as MMSE 1–12, and then post hoc subgroup analyses of larger studies were included, that would allow for the inclusion of a few more small trials of short duration [106–109]. One other study evaluated severe dementia but did not classify patients according to MMSE [110]. These trials had variable results using the SIB, with some showing minimal cognitive benefit (4–7 points of the 133-point SIB compared with placebo) and others showing a trend to worse scores compared with placebo [105–111]. Treatment of severe dementia resulted in minimal to no benefit in ADLs or in NPI scores [105–109].

There is also limited evidence for efficacy of ChEIs in long-term treatment, with most randomized trials lasting only 3–6 months despite the fact that patients often continue therapy indefinitely [12, 112]. Only four trials have lasted 1 year or longer, all in mild to moderate Alzheimer’s disease, and they provide some evidence for a waning of effect over time [100, 113–115]. One 1-year trial of donepezil found an improvement in MMSE of 1.5 points by 1 year [113], while another trial showed improvement at 6 months but no significant effect by 1 year [114]. One 2-year trial of galantamine found an improvement in MMSE of 0.7 points by 2 years [115]. However, another 2-year trial of donepezil, which had a complicated protocol with two randomization periods, found an improvement in MMSE of 0.9 points by 12 weeks that did not significantly change in the second randomization period for the duration of the trial [100]. The two primary outcomes of that trial, time to institutionalization and loss of critical ADLs, did not significantly differ between treatment and placebo by 2 years [100].

Finally, a paucity of data exist for the effectiveness of ChEIs in the treatment of those with advanced age; in pooled trial data, the mean age was approximately 75 years, with few participants included over the age of 85 years [11, 12, 94, 116, 117]. A single trial examined the effectiveness of donepezil in a population of nursing home residents with a mean age of 85 years and found no significant improvement in MMSE or neuropsychiatric outcomes at 24 weeks [118].

3.3.3 Risks of Cholinesterase Inhibitors

ChEIs can cause adverse effects due to increased cholinergic stimulation, both centrally and peripherally [119, 120]. Acetylcholine-containing neurons exist in the brain, mainly in the cerebral cortex, and in the periphery, both in the autonomic parasympathetic nervous system and skeletal muscle [121]. The central effects result in gastrointestinal and neurological disturbances, while peripheral cholinergic activity related to vagal stimulation results in cardiovascular and gastrointestinal events, as well as general effects such as weakness or fatigue (Table 2). The ChEIs donepezil, galantamine, and rivastigmine have slightly different mechanisms of action but they share the common effect of reversible acetylcholinesterase inhibition with relative selectivity for central nervous system effects; they have been found to have similar adverse effects, so the data for these drugs will be reported together [2, 121, 122]. In addition, similar effects have been seen with Alzheimer’s disease, vascular dementia, Lewy body dementia, and MCI [7, 11, 12, 30, 94].

Table 2 Adverse effects of cholinesterase inhibitors compared with placebo in Alzheimer’s disease
Full size table

Pooled data from dementia trials show that ChEI treatment is associated with a twofold increase in drop-outs due to adverse events compared with placebo [12, 103]. The absolute increase in risk is 10 %, indicating that the number needed to treat over 3–6 months to cause one significant drug reaction is 10 [12]. For patients with advanced age, frailty, and comorbidities, and those using concomitant medications, there is heightened concern for clinically relevant adverse effects associated with ChEIs. These patients, in general, have not been included in the trial data, as less than 10 % of patients treated in clinical practice would have been eligible for the published trials [15]. Therefore, the exact magnitude of risk for real-world patients is uncertain. Another major difficulty in evaluating the risks of treatment is that the vast majority of randomized trials are industry sponsored [102]. One study evaluated 27 ChEI trials that stated they reported on the safety and tolerability of the medication; however, 90 % of these trials did not report standard regulatory agency-defined serious adverse events, making interpretation of the data very difficult [102]. We provide best estimates from randomized trial data as well as real-world experiences.

3.3.4 Risks of Cholinesterase Inhibitors: Gastrointestinal Effects

Procholinergic drugs such as ChEIs significantly increase gastrointestinal side effects, which can be attributed to both central and peripheral cholinergic pathways [119, 122]. Abdominal pain, nausea, vomiting, diarrhea, and anorexia are all increased by two- to fivefold compared with placebo in pooled data of Alzheimer’s trials lasting at least 6 months (Table 2) [12]. These effects may be particularly prominent in the initial dose escalation phase but can remain throughout long-term treatment, often resulting in discontinuation of the drug [12, 119]. The degree of adverse gastrointestinal side effects appears to be dose dependent [122].

In a meta-analysis of ChEI treatment of Alzheimer’s dementia, a threefold increase in clinically significant weight loss compared with placebo was observed, related at least in part to these gastrointestinal side effects [12]. Little inter-study heterogeneity was observed in the results, indicating a consistent class effect on weight loss for all ChEIs [12]. Evaluation of trial data indicates a dose dependence related to weight loss, with a loss of ≥7 % of body weight occurring in 10–25 % of patients receiving high-dose ChEIs [122–125]. In one randomized trial of ChEI use in nursing home patients with dementia, a twofold increase in significant weight loss was observed for treatment compared with placebo, with those aged over 85 years having almost twice the risk of those aged under 85 years [118]. Weight loss can be quite detrimental in the frail elderly population and has been associated with decreased quality of life, functional decline, institutionalization, and increased mortality [126–130]. Observational studies have indicated that subsequent weight gain can occur when ChEIs are discontinued [131, 132].

3.3.5 Risks of Cholinesterase Inhibitors: Neurological Effects

Central nervous system effects are frequently reported with ChEIs [119]. In pooled data from Alzheimer’s trials, a statistically significant increase, on the order of two- to fivefold, was observed for headache, dizziness, insomnia, abnormal dreams, and vertigo (Table 2) [12]. Observational evidence also exists that agitation may be increased with ChEI use [119, 133]. Muscle cramps with ChEI use, which may be related to direct cholinergic stimulation at the neuromuscular junction, is increased 13-fold for ChEIs compared with placebo, and could result in muscle weakness or falls [12, 119].

Tremor and Parkinsonian symptoms are also significantly increased by ChEIs, in those with and without Parkinson’s disease, related to central cholinergic activation [12, 30, 103, 122]. Tremor is increased sevenfold with ChEI treatment in patients with Alzheimer’s dementia, while tremor and Parkinsonian symptoms increase two- to threefold in patients with Parkinson’s disease being treated with dopamine antagonists compared with placebo [12, 103, 117].

3.3.6 Risks of Cholinesterase Inhibitors: Cardiovascular Effects

Procholinergic medications such as ChEIs can produce bradycardia through vagal stimulation of the heart [119]. The degree of bradycardia is not an outcome generally measured in randomized trials of ChEIs. Only one trial reported this outcome and found a statistically significant 50 % increase in bradycardia in patients with MCI [84]. Several large observational studies have assessed the risk for bradycardia and found a significant association between ChEI use and hospital visits for symptomatic bradycardia, with higher doses of medication associated with higher risk [134–136]. Those with symptomatic bradycardia were significantly more likely to fall, experience syncope, or need a pacemaker implantation [134–136].

Pooled data from randomized trials in dementia have found a statistically significant increase, by approximately twofold, in the risk of syncope for ChEI treatment compared with placebo (Table 2) [12, 54]. A meta-analysis of ChEIs in patients with MCI also found a twofold increase in risk of syncope [7]. Inter-study heterogeneity in these pooled analyses was low, indicating a consistent increased risk for syncope associated with ChEIs [7, 12]. Similar results have been found in a large population-based observational study of community-dwelling subjects with dementia, which, after adjusting for potential confounding variables, found that ChEI use was associated with a significantly increased risk for hospital visits for syncope and for syncope-related events such as pacemaker insertion and falls with hip fracture [136]. The absolute increase in risk for syncope in the study was 13 per 1000 person-years, indicating that the number needed to treat for 1 year to cause one hospitalization for syncope was 76 [136]. The absolute increase in hip fracture was 3 per 1000 person-years, indicating a number needed to treat of 333 patients for 1 year to cause one hip fracture [136].

3.3.7 Risks of Cholinesterase Inhibitors: General Effects

Acetylcholinesterase inhibitors can also cause some general adverse effects through cholinergic stimulation of central and peripheral pathways [122]. These include fatigue and asthenia, each of which have been shown to be increased two- to fourfold compared with placebo. Weight loss is also increased threefold; this is discussed above under gastrointestinal effects but is probably also related to general effects of ChEIs on fatigue and debility [12, 122]. In older patients with frailty, these effects can be quite serious and may precipitate further decline. As mentioned above, the risk for significant weight loss with ChEIs essentially doubles for those aged over 85 years compared with younger patients [118]. Data from pharmacovigilance studies have also suggested the risk for adverse drug reactions related to ChEIs are twofold higher for those aged over 85 years than for younger individuals with dementia [137].

The effect of ChEIs on falls is unclear, as the definition of what constitutes a fall varies considerably among studies and fall-related events are significantly under-reported [54]. As mentioned above, ChEIs are associated with an increased risk for syncopal episodes, with some evidence for an increase in syncope-related injuries such as hip fractures [12, 54, 136]. However, most falls are not due to syncope and many do not result in injury [138–140]. One meta-analysis of randomized ChEI trials found no significant effect on falls, despite a 50 % increase in syncopal episodes compared with placebo [54]. Some observational studies of dementia populations have found ChEIs to be associated with an increased risk for syncope, falls, and fall-related injuries such as hip fractures [136, 141], while other studies have found no effect or reduced risk [142–144].

3.3.8 Risks of Cholinesterase Inhibitors: Drug–Drug Interactions

Cholinesterase inhibitors could be involved in several drug–drug interactions, both because of their complex pharmacodynamic and pharmacokinetic properties, and also because elderly patients with dementia are typically receiving multiple concomitant medications [2, 119, 145–147]. The use of medications with anticholinergic properties, such as some antidepressants, antipsychotics, antihistamines, and bronchodilators, is quite common in the elderly, and the action of either the ChEI or the anticholinergic agent could be inhibited [145, 146]. In addition, medications with anticholinergic effects could cause cognitive worsening or delirium in patients with dementia [147]. Observational studies indicate that over one-third of patients on ChEIs also receive at least one anticholinergic drug [148, 149]. Drugs with additional cholinesterase inhibitory activity, such as ranitidine, could theoretically enhance cholinergic activity and cause greater adverse toxicity [145, 146]. The use of ChEIs together with antiparkinsonian medications could limit the efficacy of either drug when treating Parkinson’s disease and dementia [147]. In addition, as mentioned above, ChEIs have been shown to increase tremor in patients with Parkinson’s disease two- to threefold [12, 103, 119]. Medications with bradycardic effects, such as beta-blockers, digoxin, amiodarone, and calcium-channel blockers, cause vagal stimulation or sympathetic blockade, and may increase the risk for syncope or heart block [145–147, 150].

3.4 Memantine

3.4.1 Benefits of Memantine

Memantine has marketing approval for the treatment of moderate to severe Alzheimer’s dementia in most countries [10, 56, 151, 152]. Studies that have assessed memantine, either as monotherapy or combination therapy with ChEIs, are limited and all are short-term (3–6 months) industry-sponsored trials [56, 95]. A recent meta-analysis from the UK National Institute for Health and Care Excellence pooled data on trials of memantine monotherapy versus placebo in moderate to severe Alzheimer’s dementia and found a small benefit at 3 months of treatment (4 points on the 133-point SIB scale), but no significant change was seen by 6 months of treatment [56]. In addition, memantine monotherapy was found to produce a small improvement of 0.3 points in a 7-point global assessment of change scale at 24–28 weeks of treatment [56]. However, pooled data from the meta-analysis showed no significant benefit in functional outcomes as measured by ADL, or by the Functional Assessment Staging instrument [56]. In addition, no statistically significant gain from memantine monotherapy on behavioral outcomes in moderate to severe dementia was observed as measured by the NPI at 24–28 weeks [56]. When the meta-analysis pooled data on combination therapy with memantine and ChEIs, no statistically significant benefit on cognitive, functional, behavioral, or global outcomes was observed when memantine was added to ChEI treatment [56].

Data on the effect of memantine in the treatment of mild to moderate Alzheimer’s dementia are minimal. Two meta-analyses found that memantine treatment for up to 6 months in moderate Alzheimer’s disease produced a marginal benefit in cognitive function (1 point on the 70-point ADAS-cog scale) and in global outcomes (0.1 points on a 7-point global assessment of change scale) but had no effect on mood and behavior [95, 153]. However, when evaluating trials of mild Alzheimer’s disease, no significant effect of memantine on any outcome was observed, either in any single trial or when the trial data were pooled together [153].

Data on the effect of memantine on vascular and Lewy body dementia are even more scarce [1, 30, 93–95, 154–156]. Memantine treatment in vascular dementia for 6 months showed minimal improvement in cognitive status (approximately 2 points on the ADAS-cog scale) without a significant effect on global outcomes or neuropsychiatric scores [94, 154, 155]. Pooled trial data of memantine treatment in Lewy body dementia found no statistically significant effect on cognition, global outcome, or behavioral symptoms [30, 156–158].

3.4.2 Risks of Memantine

In pooled trial data of memantine, the most common side effects seen were constipation, dizziness, headache, hypertension, and somnolence, but there was no statistically significant difference in the number of withdrawals or patients with at least one adverse event between memantine and placebo [93, 95, 159, 160]. A small reduction was observed in the incidence of agitation recorded as an adverse event, with a number needed to treat of 17 patients for 6 months to prevent one occurrence of agitation [95]. Agitation was not a primary outcome for efficacy of memantine in these trials, and there was no evidence that agitation or other behavioral problems could be treated with memantine [95]. Memantine had no significant effect on falls, weight loss, or confusion [95, 159, 160]. One trial found a threefold increase in hypertension for memantine compared with placebo, but other studies did not evaluate this effect [95].

Minimal drug–drug interactions have been seen with memantine use [160]. However, little is known about the use of memantine with other NMDA antagonists, such as amantadine, ketamine, and dextromethorphan, so caution is recommended [160, 161].

4 Conclusions

This systematic review summarizes the available data on the risks and benefits of dementia treatments in the elderly. Treatment with ChEIs for mild to moderate Alzheimer’s and Lewy body dementia for durations of up to 1 year produces small statistically-significant improvements in cognitive, functional, and global outcomes, but whether these improvements result in clinically relevant effects is unclear. The three available ChEIs have similar efficacy, with minimal differences seen between low and high doses of these agents. Evidence exists that the beneficial effects of ChEIs wane over the course of 1–2 years, and essentially no data is available to evaluate treatment beyond 2 years. No evidence exists for any beneficial effect of ChEIs in the treatment of vascular dementia, and their use is not recommended for this disease [94].

Trial data on ChEI treatment for severe Alzheimer’s dementia are limited and have variable results, and no studies have evaluated severe end-stage dementia. This lack of data may indicate a publication bias on the part of the pharmaceutical industry, with the assumption that patients with more severe dementia are less likely to benefit from treatment [11]. Over half of community-dwelling and roughly 90 % of institutionalized Alzheimer’s patients have dementia rated as moderate to severe, so further research on this population is clearly needed [55, 162]. The present evidence indicates that ChEIs have minimal benefit in patients with severe dementia, and discontinuation of treatment should be considered once the disease becomes severe [163, 164].

The limited trial data indicate that ChEIs do not significantly improve cognitive or neuropsychiatric outcomes in patients aged over 85 years. Of note, the prevalence of vascular dementia increases with advancing age, with 70–90 % of those aged over 90 years showing vascular pathology at autopsy [13]. As ChEIs showed marginal efficacy for the treatment of vascular dementia, this could explain why ChEIs have decreased efficacy in older populations.

ChEIs produce side effects through increased cholinergic stimulation of central and peripheral neural pathways. Pooled trial data show a twofold increase in withdrawals due to adverse effects compared with placebo, with a number needed to harm of ten. Individuals aged over 85 years have almost double the risk for an adverse event compared with younger patients. For this older population, with decreased efficacy of treatment and increased occurrence of adverse effects, we believe the risks of treatment outweigh the benefits.

Gastrointestinal effects are common with ChEI treatment, with a significant increased risk for abdominal pain, nausea, vomiting, diarrhea, and anorexia, and clinically significant weight loss. In addition, the rate of headache, dizziness, tremor, insomnia, abnormal dreams, fatigue, asthenia, vertigo, and muscle cramps are higher than with placebo. Finally, hospitalization for bradycardia and syncope is doubled, and is associated with an increased risk for syncope-related conditions such as hip fracture. There appears to be a dose-dependence related to adverse effects but not efficacy, so treatment with lower doses may be preferable. It is clear that real-world conditions are quite different from the controlled trial environment, involving an older, frailer population with more severe disease.

Memantine monotherapy in moderate to severe Alzheimer’s and vascular dementia has been shown to have small beneficial effects on cognitive and global function in trials of 3–6 months duration, with some waning of effect by 6 months. In limited pooled trial data, memantine has not been shown to be effective in the treatment of mild dementia or Lewy body dementia, or as an add-on treatment to ChEIs. From the available evidence, memantine seems to have a favorable side-effect profile.

In summary, ChEIs may provide some cognitive, functional, and global benefits in mild to moderate Alzheimer’s and Lewy body dementia, but at the risk of significant cholinergic side effects, the most serious being weight loss, debility, and syncope. These effects could be especially detrimental in the frail elderly population, in which the risks of treatment outweigh the benefits. The effectiveness of ChEI treatment may be short lived, with minimal evidence for benefit over 1 year or in those with more advanced disease. No significant efficacy of treatment is seen for those with vascular dementia or for those aged over 85 years. Memantine monotherapy may provide some benefit for patients with moderate to severe Alzheimer’s and vascular dementia, but the benefit is small and may wane over the course of several months. Memantine has a relatively favorable side-effect profile, at least under controlled trial conditions.

This systematic review provides the available evidence on efficacy and risk of dementia treatments, but leaves a lot of uncertainty on how to use these drugs in clinical practice. Good diagnostic criteria for identifying various dementia types are lacking, and no well-defined markers for evaluating therapeutic response across the spectrum of memory disorders exists. How efficacy should be assessed for individual patients in order to understand whether treatment is providing continued benefit or should be stopped is unclear. These decisions will need to be made on a case-by-case basis.