The Treatment of Cognitive Impairment Associated with Parkinson's Disease

Abstract

Cognitive impairment and dementia associated with Parkinson's disease (PD) are common and often have devastating effects upon the patient and their family. Early cognitive impairment in PD is frequent, and the functional impact may be underestimated. Optimal management will rely upon better identification of the predominant symptoms and greater knowledge of their pathophysiological basis. The management of dementia in PD (PD‐D) also has to consider the significant neuropsychiatric burden that frequently accompanies the cognitive decline, as well as fluctuations in attention. Atypical anti‐psychotics have a limited role at present in treating PD‐D, although new drugs are under development. The mainstay of drug management for people with PD‐D is cholinesterase inhibitors, although recent trials have suggested that the N‐methyl‐D aspartate antagonist memantine may also have some benefit. Disease modification remains the ultimate goal for preventing the inexorable decline in PD‐D, although effective interventions are still some way off. Limited benefit may, however, be possible through exercise programmes and so‐called “medical foods”, although randomised trials are required to confirm largely anecdotal observations.

INTRODUCTION

Cognitive impairment and dementia associated with Parkinson's disease (PD) are common and frequently devastating problems that represent major management challenges. Although figures vary depending upon populations assessed, test batteries applied and definitions used, between one‐fifth and one‐third of newly diagnosed people with PD will have some form of cognitive impairment at the time of presentation 5, 21. The pattern of dysfunction is heterogeneous and the prognostic implications for one type of cognitive impairment over another are yet to be established. Additionally, clinically significant neuropsychiatric symptoms occur in 27% of incident PD cases, compared with only 3% in age‐matched controls. Depression, apathy, sleep disturbance and anxiety are the most frequently reported symptoms in these patients (4).

Longitudinal studies suggest that 10% of patients with PD develop dementia annually, a rate four to six times greater to that found in age‐matched controls (2). Ultimately, up to 80% of people with PD develop dementia 1, 30. Dementia associated with PD (PD‐D) is associated with increased mortality, earlier institutionalization and reduced quality of life for both patient and carer.

Dementia that occurs earlier in the course of PD may share greater pathological characteristics with dementia with Lewy bodies (DLB), with greater cortical Lewy body loads and amyloid plaque burden (29). Patients who survive for longer before the onset of dementia tend to have greater subcortical pathology, with Lewy body distribution conforming to the Braak staging of PD. They have less neocortical pathology and greater evidence of cortical cholinergic deficits 9, 29.

Taken together, these clinical and pathological observations have major implications for the effective management of PD‐D, and emphasize the need for early intervention, better characterization of the clinico‐pathological phenotypes and consideration of a variety of treatment approaches, including both amyloid and synuclein protein “busting” approaches. This brief review will consider current treatment approaches to cognitive impairment in PD. It will also briefly consider future strategies.

EARLY COGNITIVE IMPAIRMENT IN PD: TO TREAT OR NOT TO TREAT?

As described above, “mild cognitive impairment” is a common finding in newly diagnosed PD. This has been defined on the basis of formal neuropsychometric testing. What is not known is how these deficits contribute toward disability or indeed whether they progress in a predictable way toward PD‐D. The latter almost certainly depends upon the underlying pathophysiological substrate, with certain forms of cognitive impairment having more sinister portent, and higher likelihood of conversion to PD‐D (67). Were this to be confirmed, then such cases would clearly be the target for aggressive disease modifying approaches, in an effort to delay the inevitable cognitive decline and conversion to PD‐D.

There has been considerable controversy whether dopaminergic drugs used to treat the motor symptoms of PD may have beneficial or indeed deleterious effects upon cognition in early PD. A 2‐year longitudinal, parallel, randomized open study of l‐dopa or pergolide treatment in newly diagnosed PD suggested that both drugs were associated with incomplete but long‐lasting improvements in several cognitive tasks, which declined thereafter (33). Moreover, no significant changes were noted on attentional, short‐term memory and Stroop tests during the study in either group. More recently, the concept of contrasting effects of l‐dopa reflecting the spatio‐temporal progression of dopamine depletion within the striatum has been proposed (14). This persuasive hypothesis suggests that l‐dopa may improve cognitive functions mediated by areas that are initially hypo‐dopaminergic, such as the dorsal striatum, whereas l‐dopa replacement in areas that are initially relatively preserved, notably the ventral striatum, may impair cognitive functions via a relative “over‐dosing”. Overall, therefore, dopaminergic drugs may have some benefits upon mild cognitive dysfunction in early PD, but these effects are partial, and unlikely to be of major clinical relevance.

A recent open‐label study examined the use of atomoxetine, a norepinephrine reuptake inhibitor, in the treatment of disabling executive dysfunction associated with nondemented PD (42). Atomoxetine was associated with improvements in all three primary outcome measures, which included the clinician's global impression of change and behavioral measures of executive dysfunction. Adverse effects included sleep, gastrointestinal disturbance and, in one subject, hypomania. This preliminary study offers a novel neurochemical approach to the management of cognitive dysfunction in PD, although further randomized, placebo‐controlled studies are necessary to better establish both the efficacy and tolerability of atomoxetine.

Glutamatergic agents have also been considered as potential disease‐modifying agents for delaying progression to PD‐D, as well as for symptomatic benefits for early cognitive dysfunction. Thus, Inzelberg and colleagues reported that use of amantadine, a drug with N‐methyl‐D‐aspartate (NMDA) receptor blocking properties, was associated with slower mental decline, delaying the onset of dementia and attenuating its severity (32). Although an intriguing result, this study was nonrandomized and potentially subject to several inherent biases. Safinamide, a unique molecule with both dopaminergic and glutamatergic properties, is currently undergoing evaluation for its effects on motor function in early and late PD (49). Preliminary data also suggest that safinamide may have some benefits upon executive dysfunction in early PD, and this is currently being pursued as part of the drug development program.

In the face of profound cholinergic deficits in both early and advanced PD, with and without dementia (59), one randomized study recently examined the use of the cholinesterase inhibitor galantamine to improve cognition in nondemented PD subjects (28). Disappointingly, the drug did not lead to any improvement in attention/executive function, memory or visuospatial performance. Furthermore, the drug was poorly tolerated, mainly caused by the gastrointestinal side effects and self‐reported worsening of parkinsonian symptoms.

THE MANAGEMENT OF ESTABLISHED DEMENTIA

General principles

The onset of functionally disabling cognitive impairment may be both slow and insidious in a patient who the physician has been following up for several years for the management of motor problems. Significant cognitive dysfunction may be denied by the patient, or covered over by the carer (eg, “we all get a bit forgetful when we get older—I'm no different”). If in any doubt, a thorough clinic re‐assessment, including recording cognitive and neuropsychiatric function, should be undertaken, with reference to recently described Movement Disorder Society Task Force Criteria for the diagnosis of PD‐D 16, 19. The insensitivity of the Mini‐Mental Status Examination (MMSE) to detect impairments in PD‐D has been well highlighted and this scale should not be relied upon to confirm or refute the diagnosis. Some measure of fluctuating attention should also be attempted, as this can have a major impact upon activities of daily living (12).

In the course of assessment, treatable problems should be excluded. These include significant dementia, and ruling out the unlikely possibility of an underlying metabolic disorder, such as hypothyroidism. There should be a high index of suspicion for chronic subdural hematoma, particularly if the clinical course is atypical, as this may be “silent,” and often no clear history is obtained for antecedent head injury. Urinary tract, and other infections, should be considered and actively treated.

The rationalization of anti‐parkinsonian and other medication may have a beneficial impact upon psychotic symptoms in PD, and can preclude the need for antipsychotic drugs in up to 60%, at least in the short term (64). Typical offending agents include anticholinergics (now more likely to be prescribed for urinary symptoms), amantadine, selegiline and dopamine agonists. If necessary, there should be a graded withdrawal of these drugs, with the patient reverting to l‐dopa monotherapy, in the lowest possible dose.

Demented PD patients are more likely to fall than nondemented patients, and impaired attention is independently predictive of falls (8). Appropriate measures, including the use of hip protectors, an assessment with bone densitometry and prescription of “bone protection” agents if necessary and an occupational therapy home assessment should all be routinely considered. If still driving, the patient should be asked to inform the relevant authority, and warned that their license may well be revoked.

Finally, it should be remembered that medical decision‐making capacity is impaired in PD‐D, and that significant impairments may also occur in cognitively impaired PD patients without dementia (43). The latter particularly applies to understanding treatment consent, with an appreciation of treatment choices and risks vs. benefits.

Atypical antipsychotic agents

Using an administrative health‐care database in Ontario, Canada, Marras and colleagues determined that the cumulative probability of a person with PD, aged 66 or older, requiring an antipsychotic drug was 35% at 7 years (41). The use of “typical” (ie, drugs with high dopamine D2 receptor antagonism) fell from 56% of prescriptions in 1998 to 9% in 2002. These figures imply that a large number of older people are prescribed antipsychotics relatively early in their disease duration. Intriguingly, an observational prospective study has suggested that the early of use an atypical antipsychotic drug in people with PD and mild hallucinations with retained insight might reduce the risk of deterioration to a state of delusions or hallucinations without insight (27). In the absence of a randomized trial, these results are difficult to explain, although at least one possibility is that the drugs in some way are able to modify the natural history of psychosis in PD.

The majority of trials of antipsychotics in PD have either excluded PD‐D, or at best included a relatively small number of subjects, so the efficacy of these drugs in the demented case with active psychosis is not clearly established. In nondemented PD patients, the use of risperidone, olanzepine and aripiprazole has been associated with unacceptable worsening of extrapyramidal symptoms and signs, so these agents are best avoided 23, 25. Two randomized controlled trials (RCTs) have demonstrated that clozapine is efficacious in nondemented PD patients with psychosis 51, 62, although the need for regular blood monitoring makes this a difficult drug to use. Quetiapine has not been shown to be effective in two randomized placebo‐controlled trials in treating PD‐related psychosis 48, 54, although a hypnotic effect and ease of use continue to make this agent a first choice in most centers. In the study of Rabey and colleagues, 29 of the 58 study participants had dementia (54). There were no significant differences in treatment response between demented and nondemented patients. Data are conflicting for whether the use of quetiapine in PD‐D is associated with worsening of parkinsonism: Two studies have suggested that this is the case 20, 56, while two others did not report any obvious deterioration 34, 54.

Ziprasidone can potentially prolong the Q‐T interval. Studies of this agent in PD have been limited to open‐label trials and case reports and the results are highly variable (25). Ziprasidone may also be administered parenterally.

A phase II double‐blind, placebo‐controlled study of pimavanserin, a 5‐HT_2A inverse agonist/antagonist in 60 people with treatment‐induced psychosis reported that the drug did not worsen motor symptoms (the primary end‐point) and also reduced psychosis, as measured by percent change from baseline on the Scale for Assessment of Positive Symptoms (SAPS) (24). A recent phase III multicenter, double‐blind, placebo‐controlled study with this agent did not, however, meet its primary end‐point of antipsychotic efficacy as measured using the SAPS. A total of 298 patients were enrolled in the trial and randomized to one of three study arms, including two different doses of pimavanserin (10 mg or 40 mg) and placebo. Patients showed marked improvements in the SAPS scores across all study arms. At Day 42, mean reductions in SAPS scores were 5.9 points in the placebo arm, 5.8 points in the 10‐mg pimavanserin arm and 6.7 points in the 40‐mg pimavanserin arm. Statistical significance was not achieved in either pimavanserin arm, primarily because of the larger than expected improvement in placebo‐treated patients. Pimavanserin met the key secondary end‐point of not worsening motor function and was safe and well tolerated, with the frequency of adverse events generally similar in the pimavanserin and placebo arms (http://www.acadia‐pharm.com). The patients enrolled in this study were nondemented.

It is therefore difficult to recommend an antipsychotic agent for use in PD‐D on evidence currently available. If required, low dose quetiapine (25 mg per day, escalating very slowly to 100–150 mg per day) would be our usual agent of choice. In the person with PD‐D, however, cholinesterase inhibitors may have useful antipsychotic potential and can obviate the need for additional atypical antipsychotics.

Cholinesterase inhibitors

Multiple convergent lines of evidence from clinical, neuroimaging, pathological and neurochemical studies indicate a profound loss of cholinergic neurotransmission in PDD, in excess of that found in Alzheimer's disease (AD) 11, 22, 65. This loss underpins several core features of PD‐D. Cholinesterase inhibitor drugs (ChEIs) act to improve cholinergically mediated cognitive and neuropsychiatric symptoms in PDD by reducing enzymatic breakdown of acetylcholine. There is therefore a sound scientific rationale for the use of these agents in the symptomatic management of PD‐D. Using functional imaging, ChEIs have been associated with increased regional cerebral glucose metabolism and blood flow in patients with PD‐D 31, 35.

Open‐label trials have suggested that galantamine may be efficacious in PD‐D 6, 38, but in the absence of randomized studies it is not possible to draw any firm conclusion. Just two adequately sized RCTs have been performed for the use of ChEIs in PD‐D and only one of these (EXPRESS study of rivastigmine, n = 541) has fully reported (18), with the other (EDON study of donepezil, n = 549), published in abstract form (17). The other three randomized trials have been very small, recruiting a total of 52 patients between them 7, 36, 55. Both the EXPRESS and EDON studies recruited PD patients with mild to moderate dementia, with treatment durations of 24 weeks. EXPRESS study participants were also offered the chance to enter an active treatment extension phase over an additional 24 weeks. Using the cognitive subscale of the Alzheimer's Disease Assessment Scale (ADAS‐cog, range 0–70 points max), modest but statistically significant benefits were noted in both studies at 24 weeks in the ChEI group (2.8 points improvement in active vs. placebo groups in EXPRESS and 3.4 points improvement in active vs. placebo groups in EDON). Significant improvements were also noted in the EXPRESS study in several secondary outcome measures, including activities of daily living and burden of neuropsychiatric symptoms. A total of 73% of those assigned to rivastigmine completed treatment, compared with 82% of those allocated to placebo; the main reason for premature withdrawal was adverse events. A total of 93% of patients, receiving rivastigmine, provided follow‐up data on at least one occasion, contributing to the efficacy analysis compared with 92% in the placebo group. The study could be criticized, however, for the use of “last observation carried forward” as the means of imputation of missing data for those lost to follow‐up prior to 24 weeks. There were fewer deaths in the active treated group [4/362 vs. 7/179, odds ratio (OR) 0.27, 95% (confidence interval) CI 0.08–0.95, P = 0.04]. Patients receiving rivastigmine in the active treatment extension study continued to derive a small improvement from the drug at 48 weeks (53). A subsequent report derived from the EXPRESS study assessments highlighted the benefits of rivastigmine compared with placebo at 24 weeks on computerized measures of attention, including sustained attention (66).

A problem with both the EXPRESS and EDON RCTs is that they used the generic Diagnostic Statistical Manual IV (DSM‐IV) criteria to define dementia in PD, and graded severity according to MMSE score. Neither the DSM‐IV criteria nor the MMSE capture the cognitive profile of PD‐D, potentially biasing the sample to patients with a more “Alzheimer‐type” presentation, and excluding those with greater executive dysfunction who arguably may have benefited more from the ChEI intervention. It is possible for a PD patient to have dementia but a “normal” MMSE score, yet such patients would have been excluded from the EXPRESS and EDON studies. Furthermore, the ADAS‐cog as a primary outcome measure is arguably better suited to the cognitive dysfunction observed in AD patients and is not ideally tailored to measure treatment effects in PD‐D. In particular, it is relatively insensitive to the prominent and early dysexecutive problems that characterize PD‐D. No quality of life, carer strain or health economic measures were reported in these studies.

A Cochrane review (39) identified only one RCT, the aforementioned EXPRESS study (18), that fulfilled prespecified inclusion criteria and concluded that rivastigmine appeared to improve cognition and activities of daily living in PD‐D, resulting in clinically meaningful benefit in 15% of cases. This review suggested that future trials in PD‐D should involve other ChEIs, utilize analytic tools that limit any bias and also measure health economic factors. Patient and carer quality of life assessments were also recommended as outcome measures.

In 2006, the American Academy of Neurology Practice Parameter document suggested that donepezil or rivastigmine should be considered for the treatment of PD‐D (46). In the same year, the National Institute for Clinical Excellence in the UK recognized that while ChEIs have been used successfully in individual people with PD‐D, further research was recommended to identify those patients who will benefit from this treatment (63).

A 76‐week prospective, multicenter, open‐label trial is ongoing to evaluate the long‐term effect of rivastigmine capsule and transdermal patch on worsening of motor symptoms in patients with mild to moderately severe PD‐D (http://www.clinicaltrials.gov). The rivastigmine patch, which may be associated with superior tolerability over the capsule preparation, is already licensed for use in AD in several countries.

Although some patients report an increase in tremor on ChEI, this is an uncommon reason for discontinuation of the drug (18) and, objectively, motor kinematic measures are unaffected (44). More recently, there have been safety concerns raised in a large population‐based study of a heterogeneous group of patient‐prescribed ChEIs for a range of dementias (the majority having AD) (26). Compared with control subjects, the 19 803 community‐dwelling older adults with dementia‐prescribed ChEIs had more frequent hospital visits for bradycardia [Hazard ratio (HR), 1.69 with 95% CI, 1.32–2.15], permanent pacemaker insertion (HR 1.49; 95% CI 1.12–2.0) and hip fracture (1.18; 95% CI 1.04–1.34). Although the hazard ratios were significant, the absolute number of events was small. A previously published clinical protocol indicated that pretreatment electrocardiograms and 24‐h cardiac monitoring were not justified (57) and that there were no high‐risk groups for whom to target screening. Nevertheless, the paper conceded that checking the patient's pulse was important, and that detection of bradycardia (<50 beats per minute) before or during treatment should be investigated further and the patient monitored more frequently thereafter.

The clinical response to ChEIs is highly variable and attempts have been made to identify those patients likely to respond to the drugs. A highly sensitive and specific predictor would have both clinical and economic benefits, as well as potentially informing regarding the underlying pathophysiology. From a sub‐analysis of the EXPRESS study, visual hallucinations appear to predict more rapid decline and possibly greater therapeutic benefit from rivastigmine treatment in PDD (13). Neurophysiological studies suggest that latency of the P3 event‐related potential may predict response to donepezil (50), while short‐latency afferent inhibition technique may be useful to monitor the efficacy of ChEI treatment (40). A single nucleotide polymorphism in the cytochrome P450 CYP2D6 gene may influence the efficacy of donepezil in mild to moderate AD (52). Patients with the “G” allele had a significantly higher risk of poor response to the ChEI (OR 3.43; 95% CI 1.49–7.9) in this study. The functional effect of this polymorphism on treatment response in PD‐D has not yet been reported.

While debate continues regarding the relative efficacy for ChEIs in the treatment of PD‐D, it would be prudent to withdraw the agents gradually, should it be decided that they are not beneficial in a given patient. Sudden ChEI withdrawal has been associated with acute cognitive and behavioral decline (45).

Memantine

Memantine is a dimethyl derivative of amantadine and possesses NMDA receptor blocking properties. The drug modulates glutamatergic neuronal transmission and could potentially prevent the toxic effects of raised concentrations of glutamate. Memantine has previously been shown to have clinical efficacy in AD and vascular dementia. A small RCT of this agent in PD‐D showed that the drug was well tolerated, but because of the small numbers enrolled, only indirect evidence was gained for efficacy, in that more patients taking the active drug deteriorated at the end of the study compared with those allocated to placebo (37).

A larger randomized trial of memantine in the Lewy body dementias (ie, patients with either PD‐D or DLB) enrolled 72 patients, 40 of whom had PD‐D (3). Concomitant stable prescription of a ChEI was permitted. The clinical global impression of change was the primary outcome measure, and at 24 weeks there was a significant difference between those taking memantine vs. those taking placebo (mean difference 0.7; 95% CI 0.04–1.39; P = 0.03) in favor of the NMDA antagonist. With the exception of the quick test of cognition, there were no significant differences detected in any of the secondary outcome measures. A total of 78% of the 72 participants completed the study, with a similar proportion of withdrawals occurring in the two groups. A further Lundbeck A/S sponsored randomized 6‐month study of memantine in the Lewy body dementias has completed (http://www.clinicaltrials.gov). In this study, participants were not permitted to take ChEI. The results are eagerly awaited, as there is still uncertainty regarding memantine's efficacy in PD‐D, and whether it should be prescribed with, or instead of, a ChEI. Intriguingly, one study has reported that donepezil can markedly potentiate memantine neurotoxicity in the adult rat brain, possibly via disinhibition‐mediated increased excitatory tone (15). The implications for this finding in humans treated with both memantine and a ChEI are uncertain at present, and further work is clearly required to clarify the issue.

FUTURE DIRECTIONS

A range of nonpharmacological approaches should be explored in the management of cognitive impairment associated with PD. For example, there is increasing interest in the potential for exercise to improve cognitive function. A recent small study has reported that a multimodal exercise program can improve executive function in PD (61). “Medical foods” are also attracting interest, with preclinical animal studies suggesting that a combination of nutrients may increase synapse formation and reduce beta‐amyloid production. RCTs are underway in AD, after encouraging preliminary results (58).

Multifunctional drugs may offer hope in the dual management of motor and cognitive aspects of PD. Safinamide, previously mentioned, would be one example of such a drug. Other potential agents in development include ladostigil (68) and adenosine receptor antagonists (60).

If synuclein‐related pathology underpins the onset of dementia in people with PD, then the development of anti‐synuclein protein aggregating (so‐called “synuclein busting”) agents could represent a significant development in delaying or averting the onset of cognitive decline, particularly because they could be introduced at a relatively early stage in nondemented PD patients. Such agents may end up coming from a variety of backgrounds and therapeutic mechanisms. The cholesterol‐lowering compound methyl‐β‐cyclodextrin, for example, has been shown to reduce neuronal α‐synuclein accumulation in mice, via putative lipid‐synuclein interaction (10).

From in vivo imaging and pathological studies, however, it seems probable that the pathological basis for PD‐D is heterogeneous, with Alzheimer‐type lesions (particularly amyloid plaques) playing an important role. Further work is therefore required to better characterize the clinical variability of early cognitive impairment associated with PD as well as PD‐D and how these phenotypes relate to existing biomarkers such as ¹¹C‐PIB positron emission tomography and cerebrospinal protein analysis (ie, synuclein, amyloid and tau concentrations). In doing so, this could provide an opportunity to therapeutically target the predominant protein aggregate(s) in an earlier window of opportunity (47).

CONCLUSION

This review has considered the management of the spectrum of cognitive dysfunction associated with PD. Such dysfunction is now regarded as an integral and, as our patients survive for longer, an almost inevitable component of the disease. The variability in onset and rate of cognitive decline, together with the heterogeneous clinical phenotype argues strongly for a multifactorial pathological basis for PD‐D. Symptomatic treatments can lead to gratifying benefits in established dementia, but more work is required to better evaluate the clinical relevance of such interventions, as well as their cost‐effectiveness. Not all PD‐D patients improve with these treatments, and better predictors of response would be welcome. Beyond neurotransmitter modulation, there is an urgent need for the development of disease‐modifying therapies that target the responsible pathological processes. To maximize response, such therapies may well need to be individualized, and based upon results from imaging studies, cerebrospinal fluid analysis and other biomarkers.

DISCLOSURES

The author has no relevant disclosures to make regarding this manuscript.