Article Text

Download PDFPDF

Cognitive impairment in Parkinson's disease
  1. Jeremy Cosgrove1,2,
  2. Jane Elizabeth Alty1,2,
  3. Stuart Jamieson1
  1. 1Department of Neurology, Leeds General Infirmary, Leeds Teaching Hospitals NHS Trust, Leeds, UK
  2. 2Hull York Medical School, University of York, York, UK
  1. Correspondence to Dr Jeremy Cosgrove, Department of Neurology, Leeds General Infirmary, Great George Street, Leeds LS1 3EX, UK; jeremycosgrove{at}


Cognitive impairment is a significant non-motor symptom of Parkinson's disease (PD). Longitudinal cohort studies have demonstrated that approximately 50% of those with PD develop dementia after 10 years, increasing to over 80% after 20 years. Deficits in cognition can be identified at the time of PD diagnosis in some patients and this mild cognitive impairment (PD-MCI) has been studied extensively over the last decade. Although PD-MCI is a risk factor for developing Parkinson's disease dementia there is evidence to suggest that PD-MCI might consist of distinct subtypes with different pathophysiologies and prognoses. The major pathological correlate of Parkinson's disease dementia is Lewy body deposition in the limbic system and neocortex although Alzheimer's related pathology is also an important contributor. Pathological damage causes alteration to neurotransmitter systems within the brain, producing behavioural change. Management of cognitive impairment in PD requires a multidisciplinary approach and accurate communication with patients and relatives is essential.

View Full Text

Statistics from


Parkinson's disease (PD) was traditionally considered solely a disorder of movement but there is now recognition of the significance of non-motor features which include depression, psychosis, anxiety, apathy and sleep disorders.1 Cognitive impairment is another important non-motor feature and longitudinal cohort studies have revealed that dementia is common in PD, affecting over 80% of those who survive for 20 years.2 Parkinson's disease dementia (PDD) reduces quality of life, increases mortality3 and intensifies caregiver burden.4 A systematic review estimated that the prevalence of PDD in those over 65 years was 0.2–0.5%, accounting for 3–4% of all cases of dementia.5 To put in to context, Alzheimer's disease (AD) accounts for 50–70% of dementia cases and vascular dementia for 15–25%.6 PDD is associated with falls,7 ,8 which along with pneumonia are the most common reasons for hospital admission in those with PD.9 A range of clinicians including general practitioners, acute medical physicians, geriatricians, neurologists, psychiatrists and orthopaedic surgeons are likely to be involved in the management of cases of PDD.

Research studies of mild cognitive impairment in PD (PD-MCI) have increased dramatically over the last decade and the definition was standardised in 2012 by the International Parkinson and Movement Disorder Society (MDS).10 PD-MCI is common and can be identified at the time of PD diagnosis in some cases.11 ,12 It is a risk factor for developing PDD11 ,13 but some studies suggest that there are two distinct types of MCI in PD, each with a different pathophysiology and prognosis.14–16 This finding and others have contributed to the development of the ‘dual syndrome hypothesis’,17 ,18 proposing that executive dysfunction in PD is caused primarily by catecholaminergic deficits with a smaller contribution from cholinergic deficit, whereas the cognitive profile of PDD is caused by early and prominent cholinergic deficit.

One of the objectives behind the standardisation of PD-MCI is to identify those with PD most at risk of developing PDD.10 ,19 This will allow those subjects access to early treatments and enrolment in relevant clinical trials, as well as increase understanding of the pathological drivers of dementia.

The aim of this review is to provide a summary of cognitive dysfunction in PD, which is a common clinical problem that many clinicians are likely to encounter in their practice and a fascinating, fast-moving area of research.


Formal diagnostic guidelines for PDD were proposed by the MDS in 2007.20 Hitherto clinicians had often used generic criteria such as the Diagnostic and Statistical Manual of Mental Disorders to diagnose PDD.21 A diagnosis of PDD requires a confirmed diagnosis of PD as outlined by the UK Parkinson's Disease Brain Bank Criteria22 and the development of a dementia syndrome in the context of established PD.20 A dementia syndrome is defined as “impairment in more than one cognitive domain, representing a decline from premorbid level, with deficits severe enough to impair daily life, independent of the impairment ascribable to motor or autonomic symptoms” (Emre et al).20

Antemortem diagnosis of PDD can be ‘probable’ or ‘possible’ according to the MDS guidelines.20 Probable PDD requires impairment in two or more of the four cognitive domains that are most characteristic of PDD. These are attention, executive function, visuospatial function and memory20 (figure 1). As in dementia with Lewy bodies (DLB), prominent visuospatial dysfunction and marked fluctuations in attention are usually seen in established PDD.23 Neuropsychiatric symptoms such as apathy, anxiety, hallucinations and depression are not required in order to make a diagnosis of PDD20 but the association between PDD and neuropsychiatric symptoms is well recognised.24 For example, in a study of 537 PDD subjects more than 60% had at least one clinically significant neuropsychiatric symptom.25 Exclusion criteria for a diagnosis of PDD include development of dementia before or within 1 year of typical motor symptoms of PD (in which case a diagnosis of DLB should be considered26), delirium or features compatible with vascular dementia.20

Figure 1

Visuospatial deficits are common in PD-MCI and PDD. This figure demonstrates varying degrees of impairment of visuoconstructional skills when copying a cube in five subjects with PDD (cubes B–F) compared with a PD subject with normal cognition (cube A). PD, Parkinson's disease; PDD, Parkinson's disease dementia; PD-MCI, mild cognitive impairment in PD.

Recommendations about how to diagnose PDD were produced at the same time as the diagnostic criteria.27 Level 1 testing was designed as a screening tool to be performed quickly in the outpatient clinic. Level 2 testing requires detailed neuropsychological assessment and is designed for use in research settings, to monitor pattern and severity of cognitive change over time and for when level 1 testing produces ambiguous results.27 The MDS diagnostic criteria are more sensitive than Diagnostic and Statistical Manual of Mental Disorders criteria.21 Level 1 testing is specific for a diagnosis of PD but lacks sensitivity when compared with level 2 testing.28 ,29

Before MDS guidelines in 2012,10 PD-MCI was most often diagnosed using AD related MCI criteria.30 Along with a diagnosis of PD as per the UK Parkinson's Disease Brain Bank Criteria,22 the MDS guidelines require a gradual decline in cognition although deficits must not ‘interfere significantly with functional independence, although subtle difficulties on complex functional tasks may be present’ (Litvan et al).10 A level 1 category diagnosis of PD-MCI allows the use of a scale of global cognitive function, for example the Montreal Cognitive Assessment (MoCA), to be used to demonstrate cognitive deficits whereas a level 2 category diagnosis requires neuropsychological testing of attention and working memory, executive function, language, memory and visuospatial function.10 Two neuropsychological tests specific to each of the five cognitive domains must be performed and a level 2 category diagnosis can be made if deficits are identified on at least two tests. The abnormal tests can be within the same cognitive domain or within different cognitive domains, therefore allowing subtyping into single or multiple cognitive domains. The definition of impairment in cognitive tests is not precisely defined.10

Large-scale validation of the 2012 PD-MCI criteria is currently being undertaken by an MDS Study Group.31 A number of issues exist regarding the current PD-MCI diagnostic criteria. One is that the formal definition of PD-MCI is a new construct and therefore there is a lack of a reference standard for the validation process.23 Another is the uncertainty regarding the optimum SD cut-off for impairment when compared with age-adjusted normative means.32 Finally, the recommendation that a level 1 category diagnosis can be made using a global cognitive screening tool has been questioned after three of the recommended tests including the MoCA were shown to lack diagnostic accuracy when compared with level 2 category diagnosis,33 although the MoCA performed better in another recent study.34

Key points

  • The MDS guidelines allow a level 1 diagnosis of PD-MCI and PDD to be made using a global cognitive screening tool to demonstrate cognitive dysfunction. This avoids the need for neuropsychological expertise.

  • The core cognitive domains affected in PD-MCI and PDD are executive function, attention, language, visuospatial function and memory.

  • PDD cannot be diagnosed if dementia occurs before or within 1 year of developing motor symptoms. In such cases, DLB should be considered.


A review of the literature which informed the 2012 MDS guidelines10 found that 26.7% (range 18.9–38.2%) of nearly 1000 PD subjects from six cross-sectional and two prospective studies had PD-MCI,19 using cut-offs of between 115 and 2 SD35 below normative mean scores to define MCI. A multicentre pooled analysis study of over 1300 PD subjects found that 25.8% had MCI,36 which was diagnosed if a participant scored 1.5 SD or more below a normative mean in any one of three core cognitive domains (visuospatial, memory or attention/executive function). In both studies PD-MCI was more commonly a dysfunction of a single cognitive domain than multiple cognitive domains. In contrast to AD where memory dysfunction dominates,37 in single domain PD-MCI a cognitive domain other than memory is more likely to be affected.10 ,36

More recently two studies using the MDS criteria10 have suggested that PD-MCI is common even at the time of PD diagnosis, affecting 35%38 to 42.5%.12 The variable reported prevalence of PD-MCI can be attributed to its definition (SD cut-off scores used to define abnormal results on neuropsychological tests, pre-MDS and post-MDS diagnostic criteria, level 1 category or level 2 category diagnoses if using MDS criteria) and the population studied (incident vs prevalent studies, clinic cohorts vs community based studies).32

Since dementia becomes more prevalent in the later stages of PD, studies that prospectively follow cohorts to see what percentage develop PDD are more relevant that those looking at point prevalence. These longitudinal studies have had a profound impact because they have highlighted that a large proportion of PD subjects will develop dementia.

The Sydney Multicentre Study of PD followed 136 newly diagnosed PD subjects over a 20-year period, by which time 83% of the 30 survivors had developed dementia. Overall, 75% of the total cohort developed dementia before death.2 The Cambridgeshire Parkinson's Incidence from GP to Neurologist (CamPaIGN) study attempted to follow all incident cases of PD between 2000 and 2002 in the Cambridgeshire region, providing a population based cohort. Ten-year outcome of 142 patients showed that the cumulative proportion developing dementia was 46%.14 A French population based study of over 65-years-olds found that 50% of PD survivors had dementia at 8 years39 and in a Chinese clinic-based study the cumulative proportion with dementia was 49% after 11.3 years of follow-up.40 The relative risk of a person with PD developing dementia compared with someone without PD varies across studies with a range of 1.741 to 5.1.42

Key points

  • The reported prevalence of PD-MCI varies because different definitions are used and different populations are studied.

  • Approximately 25% of those with PD have MCI using a SD of 1.5 below normative means to diagnose cognitive dysfunction.

  • PD-MCI more commonly affects a single cognitive domain than multiple cognitive domains.

  • In single domain PD-MCI a cognitive domain other than memory is more likely to be affected.

  • Approximately 50% of those with PD will develop dementia within 10 years of diagnosis increasing to over 80% after 20 years.

Risk factors

Although longitudinal studies are limited there is clear evidence to suggest that PD-MCI is a risk factor for PDD. A 2006 study using a logistic regression model controlling for age, disease stage, education and gender found an OR of 5.1 (95% CI 1.51 to 16.24).13 More recently, a 3-year follow-up study found that the relative risk of developing PDD was 39.2 (95% CI 5.2 to 296.5) in those with PD-MCI at baseline compared with PD subjects with normal cognition.11 This study also found that 22% of those diagnosed with PD-MCI at baseline had reverted to normal cognition, implying that PD-MCI can be reversible in some cases11 although further research is required to investigate this further.

Age is the biggest risk factor for the development of PDD and it appears that it is age rather than age of onset of PD that conveys an increased risk of developing PDD.43 Findings from the Sydney Multicentre Study suggest a significantly longer dementia-free survival in younger onset PD compared with later onset.44 A clinicopathological study of 129 proven cases of PD suggested that four clinical milestones—cognitive impairment, falls, visual hallucinations and the need for residential care—are common features of advanced PD and mark a terminal phase of decline with death occurring approximately 5 years after onset of visual hallucinations.8 However, those with early onset disease had significantly longer survival before developing these clinical milestones than those who developed PD later in life.8

Two broad clinical motor phenotypes are recognised in PD—a tremor dominant (TD) group and a group characterised by postural instability and gait disorder (PIGD).45 The PIGD phenotype is associated with a more rapid cognitive decline than the TD phenotype,16 as is conversion from the TD phenotype to the PIGD phenotype.46 Those with PIGD phenotype are more likely to have PD-MCI at diagnosis that those with TD phenotype.47

The major genetic risk factors currently associated with PDD are in catechol-O-methyltransferase, microtubule associated protein τ and apolipoprotein E. Mutant glucocerebrosidase has also been associated with increased risk of PDD (table 1).

Table 1

Genetic risk factors associated with cognitive impairment in PD

Other risk factors for developing PDD are a lower level of education,52 severity of motor deficit15 ,16 and male gender.36

Whether PDD is an inevitable consequence if someone with PD survives for long enough or is a nosologically distinct condition—with as of yet unidentified pathobiological and genetic biomarkers—from a PD type without dementia is unknown. In other words, it remains unclear whether the 17% of dementia-free survivors at 20 years in the Sydney Multicentre Study2 had a different disorder from those who developed dementia. Large, international, cohort-based biomarker studies such as the Parkinson Progression Marker Initiative53 have been established to help answer this question.

Key points

  • PD-MCI is a risk factor for developing PDD but PD-MCI does not always develop into PDD.

  • Age, rather than age at diagnosis, increases the risk of developing PDD.

  • Those who have the PIGD phenotype are more likely to develop dementia than those who have the TD phenotype.


The presence of Lewy bodies (located within the cell body) and Lewy neurites (located within the axons) in the limbic system and neocortex appears to be the major cause of PDD but AD pathology in the form of τ and amyloid deposition is also important,54 and many of those with PDD have mixed AD and PD pathology at postmortem.55

α-synuclein (α-syn) is a protein involved in vesicular transport and is the principal component of Lewy bodies and Lewy neurites (now referred to as Lewy based pathology (LBP)). Misfolding of α-syn leads to conformational change from an α-helical to β-pleated sheet structure, which further aggregates into higher-order structures such as amyloid fibrils.55 The insoluble amyloid fibrils are thought to exert a neurotoxic effect via various mechanisms including oxidative stress, synaptic dysfunction and impaired axonal transport.55

A six-stage pathological staging system based on α-syn staining of 168 human brains was published by Braak et al in 2003.56 Based on the assumption that non-symptomatic cases would have the mildest pathology and the most clinically severe cases would have the most dramatic pathology, a caudorostral progression of LBP was proposed, beginning in the medulla oblongata and ascending to the neocortex.56 Stage 1 involves LBP deposition in the medulla oblongata; stage 2 relates to involvement of the pontine tegmentum; stage 3 corresponds to deposition in the midbrain including the substantia nigra pars compacta (SNpc) (figure 2); stage 4 includes involvement of the hippocampus and the transentorhinal cortex (ie, the limbic system) while stages 5 and 6 involve progressive involvement of the neocortex.56 Classification was based on the consistent anatomical pattern of lesion distribution rather than the lesion load at each site, which was found to vary between cases.56

Figure 2

H&E staining of the substantia nigra in an age matched control (A) shows a normal density of large pyramidal neurons and neuromelanin pigment within the neurons (arrow). In a subject with PD (B) the neuronal population is reduced, there is extraneuronal neuromelanin deposition (arrowhead) and two bright red round Lewy bodies within pyramidal neurons (arrows). Magnification × 200. Images provided by Dr Ismail, Consultant Neuropathologist at Leeds Teaching Hospitals NHS Trust.

The caudorostral propagation of LBP is supported by the premotor phase of PD that is identified in some patients and characterised by impaired sense of smell (olfactory pathology), autonomic dysfunction (raphe nuclei) and rapid eye movement (REM) sleep behaviour disorder (subcoeruleus nucleus). However, some pathological studies have identified cases where LBP did not follow a caudorostral propagation, suggesting that neocortical LBP is not always dependent on the presence of subcortical pathology and that simultaneous cortical and subcortical LBP development is possible.57

Braak pathological stage56 has subsequently been linked to cognitive function in subjects with PD by some studies8 ,58 but not all brains with evidence of LBP within the cortex are associated with a history of PDD. Some authors have described this as ‘incidental Lewy body disease’55 and it suggests that the distribution of LBP cannot predict antemortem clinical status. There have also been a few cases reported with PDD that do not have evidence of significant cortical LBP,56 implicating other factors in the development of PDD.

There appears to be an association between the pathological changes associated with AD—namely τ neurofibrillary tangles (NFT) and amyloid β plaques (Aβ)—and cognitive function in some people with PDD. In one postmortem study it was found that a combination of LBP, τ and Aβ pathology was a better neuropathological correlate of PDD than any of the pathologies in isolation.59 In another study LBP was found to have a greater sensitivity for predicting PDD than AD pathology although AD pathology had a greater specificity.60 It is suggested that 40–50% of pathologically proven cases of PDD have enough NFT and Aβ pathology to have an additional diagnosis of AD, that is, PDD and AD.55 Furthermore, work in mice models suggests that LBP, τ and Aβ are synergistic.61

The role of vascular disease, cerebral amyloid angiopathy and hippocampal sclerosis in the development of PDD remains uncertain. The former two pathologies are common in the general population with increasing age and therefore present to variable degrees in autopsy studies of PD subjects, where the average age is most often greater than 75 years.62

Only a limited number of pathological studies have been performed on those with PD-MCI. As with PDD, it seems that a combination of LBP and AD related pathology coexists and the exact role of each is yet to be determined.62

Key points

  • α-syn infiltration of the limbic system and neocortex in the form of LBP is the major pathological correlate of PDD.

  • AD pathology—NFT and Aβ—also variably contributes to PDD development.

Linking neurotransmitter changes and cognitive dysfunction

Establishing links between pathological change and patterns of cognitive dysfunction in PD is difficult because every affected individual has a different degree of involvement of LBP and AD-related pathology at any one time. This means that the rate and degree of damage to the major neurotransmitter systems in the brain of each individual will vary. Radiotracers that bind amyloid (Pittsburgh compound B) are not routinely available, a suitable α-syn binder has not yet been found63 and autopsy studies represent the end point of pathological damage in an individual. The situation is further complicated by age, cognitive reserve (usually measured by proxy, eg, via years in education and occupation; those with increased cognitive reserve have slower rates of cognitive decline but are not protected from PDD64) and the genetic profile of each person with PD (figure 3).

Figure 3

The complex factors governing cognitive dysfunction in each individual with PD. Aβ, amyloid β plaques; α-syn, α synuclein; APOE, apolipoprotein E; CAA, cerebral amyloid angiopathy; COMT, catechol-O-methyltransferase; GBA, glucocerebrosidase; MAPT, microtubule associated protein τ; NFT, neurofibrillary tangles; PD, Parkinson's disease.

Executive function, that is the ability of a person to solve a problem, sequence and modify behaviour in response to a changing situation, is mediated by the prefrontal cortex.65 Impaired executive function is common in PD and is thought to be driven primarily by dopaminergic frontostriatal cortical loops.66 The relationship between dopamine levels and executive function in those with PD is complex because of the characteristic sequence in which dopaminergic neurons within the SNpc degenerate.67 Whether or not facets of executive function improve68 ,69 or deteriorate70 ,71 with dopaminergic stimulation depends on how advanced degeneration within the SNpc is. This complex relationship is explained by the ‘dopamine overdose hypothesis’,72 originally proposed by Gotham et al in 1988.73 Genetic factors also influence the role of dopamine on executive function. Specifically, a polymorphism in the catechol-O-methyltransferase gene causes variability in its capability to remove dopamine48 and this affects dopamine levels within the dorsolateral prefrontal cortex (DLPC), in turn influencing performance on tasks of executive function.16

Acetylcholine (ACh) is found throughout the brain and there is growing evidence to suggest that it may also play a fundamental role in the cognitive deficits seen in PD. The cortex is supplied by the basal forebrain nuclei, particularly the nucleus basalis of Meynert (nbM) whereas the thalamus, cerebellum and various brainstem nuclei are supplied by the pedunculopontine nucleus (PPN).7 ,74 ACh is thought to be particularly important in regulating attention via projections from the nbM to the DLPC in humans.75 ,76 It has long been established that the nbM and the PPN undergo degeneration in PD77 ,78 and the caudorostral propagation of α-syn proposed by Braak et al56 suggests that LBP infiltrates the nbM and PPN at approximately the same time as deposition within the SNpc, that is, the two major ACh systems in the brain are affected by the time that the first motor symptoms of PD emerge.

Subjects with PDD have a more profound reduction of ACh markers in the brain than those with PD or AD79 and the pathological changes seen at postmortem are supported by nuclear medicine studies.80 ,81 Poor attention at baseline and the PIGD phenotype are independent predictors of cognitive decline in PD82 and ACh deficiency is proposed as the unifying link.7 At least initially, PPN degeneration with subsequent reduction of ACh in the thalamus and brainstem could contribute to gait disturbance and falls whereas nbM degeneration and subsequent reduction of ACh in the DLPC could be the driver of attentional deficit, as supported to some extent by positron emission tomography (PET) studies.80 However, this is likely to be too simplistic given the heterogeneous pathological process occurring in every individual with PD.

Further supportive evidence that reduced ACh is associated with cognitive decline in PD comes from drugs that alter ACh levels in the brain. Cholinesterase inhibitors (ChEIs), that is, drugs that indirectly increase the availability of ACh by inhibiting its breakdown, are associated with improvements in cognition in PDD,83 ,84 whereas drugs with anticholinergic properties, for example amitriptyline, have been associated with more rapid cognitive decline in PD.85

Just as with ACh, loss of noradrenergic neurons (from the locus coeruleus) and serotonergic neurons (from the dorsal and median raphe nuclei) occurs in PD,62 and damage to these neurotransmitter symptoms begins before the development of the motor symptoms of PD. The role that depletion of these neurotransmitters plays in the development of cognitive dysfunction is not fully established.

The ‘dual syndrome hypothesis’, proposed by Kehagia et al in 2010,17 ,18 links neuropsychological profiles seen in PD-MCI and PDD with neurotransmitter changes. One of the major drivers of the theory is that executive dysfunction at baseline was not a predictor of PDD after 5 years of follow-up in the CamPaIGN cohort study.15 ,16 Neither was phonetic fluency, a cognitive task mediated by the frontal lobes.86 In contrast, semantic fluency and pentagon copying, mediated by the temporal lobe and parieto-occipital cortex, respectively, were strongly identified as risk factors for developing PDD.15 ,16 As already stated, the dual syndrome hypothesis proposes that executive dysfunction in PD is predominantly caused by catecholaminergic deficits (the role of dopaminergic change is currently better understood than the role of noradrenergic depletion) with a lesser contribution from cholinergic deficit.17 ,18 Those subjects who demonstrate early evidence of posterior cortical dysfunction (visuospatial, memory and language deficits) and those with the PIGD phenotype are at the greatest risk of rapid cognitive decline and PDD.17 ,18 In these subjects it is proposed that ACh depletion is the primary cause of the deficits, driven by infiltration of the cholinergic basal forebrain nuclei and PPN by ascending LBP pathology, as well as a variable contribution from AD-related pathology. However, overlap between predominantly catecholaminergic-based executive dysfunction and the predominantly cholinergic-based posterior cortical dysfunction occurs.17 ,18 For example, ACh depletion has been implicated in attention deficit, often seen in PD-MCI and mediated by the DLPC.75 The dual syndrome hypothesis is being investigated by the Incidence of Cognitive Impairment in Cohorts with Longitudinal Evaluation – Parkinson's Disease (ICICLE-PD) study,12 a UK based cohort study of incident PD, and a recent publication from that group supports the validity of the hypothesis.87

Key points

  • Changes in executive function in PD appear to be driven by dopamine levels in frontostriatal cortical loops and the DLPC.

  • Reduction in ACh, caused by infiltration of the major ACh nuclei by LBP, is related to attention deficits and is also proposed to drive changes in memory, language and visuospatial function.

  • PD-MCI with prominent executive dysfunction may not be a risk factor for developing PDD.

Management of cognitive decline in PD

The general approach to treatment of those with PD-MCI and PDD should begin with exclusion of other potential contributors to cognitive dysfunction such as infection, constipation, sleep disruption, delirium and depression. All medications should be reviewed and rationalised because a number of drugs commonly prescribed in PD, for example anticholinergics and dopamine agonists, can have a negative impact on cognition and mental state.88 The degree of fixed cognitive dysfunction in a patient with PD can be overestimated if other potential causes are not identified and addressed. PDD is associated with falls7 ,8 and therefore access to appropriate physiotherapy and occupational therapy is important. There are no studies regarding non-pharmacological treatments such as cognitive training in those with PDD.89

It is important to communicate clearly with patients and family members about cognitive impairment so that they are appropriately informed of what might be expected, although the uncertainty regarding progression of PD-MCI to PDD11 makes this more difficult. Issues such as driving capacity, advanced directives, lasting power of attorney and end-of-life care should be discussed where appropriate and support from PD nurse specialists and palliative care teams can be invaluable in this regard. We believe that earlier access to, and better interaction with, palliative care teams would improve quality of life for those with advanced PD as well as their families.90

In relation to pharmacological therapy, ChEIs have been shown to improve cognition in those with PDD. Although there are a number of smaller open label studies, only two large randomised-controlled trials have been performed.91 ,92 One trial compared donepezil with placebo over 24 weeks91 and the other compared rivastigmine with placebo over 24 weeks92 with a further 24-week extension in which all recruits received rivastigmine.93 Both trials show a modest benefit in cognition using a tool called the Alzheimer's Disease Assessment Scale—Cognitive Subscale with better evidence for rivastigmine than donepezil. Meta-analyses of all ChEI studies support their use for cognitive issues83 and behavioural disturbance in PDD.84

A major disadvantage of the current pharmacological approach to treating PDD is that significant damage to the neurotransmitter systems targeted by ChEIs has already occurred by the time the symptoms of PDD occur.56 ,60 ,62 For this reason, early and accurate prediction of those most likely to develop dementia represents a major challenge for future research, emphasising the importance of biomarker studies.12 ,53 Early and accurate identification of at-risk individuals is a prerequisite for optimal response to disease-modifying medications, once they become available.

Memantine is an N-methyl-D-aspartate receptor antagonist that blocks the activity of the excitatory neurotransmitter glutamate. It has been studied in nearly 300 PDD subjects at a dose of 20 mg per day, with variable results. A recent meta-analysis of these studies concluded that there was no evidence that it improves cognition in PDD.83

Atomoxetine, a selective noradrenaline reuptake inhibitor, improved cognition in 55 non-demented PD subjects over 8 weeks94 but this was a secondary outcome measure. More recently a small randomised-controlled trial demonstrated that a single dose of atomoxetine reduces impulsivity and risk-taking in non-demented PD subjects.95 Further research into the role of noradrenaline reuptake inhibitors in the management of cognitive impairment in PD is needed.

Key points

  • Medication needs to be rationalised in those with PD who have cognitive dysfunction

  • Informing patients and families about what to expect in PDD is important because it allows planning for the future to occur

  • Cholinesterase inhibitors have a modest positive effect on cognition in PDD.


Longitudinal cohort studies of incident PD have shown that PDD occurs in approximately half of all those who survive for 10 years14 ,39 ,40 and over 80% at 20 years.2 MCI is common in PD, even at the time of PD diagnosis,12 and increases the risk of developing PDD.11 ,46 However, the degree of fluidity between normal cognition, PD-MCI and PDD is not yet fully established and a proportion of those with PD-MCI remain stable or can even revert back to normal cognition.11

The 2012 MDS diagnostic guidelines10 have formalised the definition of PD-MCI and allow subtyping of the number and type of cognitive domains affected. The guidelines await large-scale validation31 and some aspects require clarification.32 ,33 Whether or not the diagnostic criteria have prognostic significance will be established by future longitudinal studies.

The pathological correlates of PDD are complex but infiltration of the limbic system and neocortex with α-syn in the form of LBP is the main driver.23 ,54 Only a small number of pathological studies of PD-MCI have been performed so far, with limited conclusions.10 LBP and other pathological processes, primarily those associated with AD,55 ,60 lead to dysfunction in multiple neurotransmitter systems within the brain, producing behavioural change.

It has been suggested that executive dysfunction in PD is caused predominately by catecholaminergic change including frontostriatal dopaminergic deficit whereas deficiency of ACh primarily contributes to posterior cortical problems namely memory, language and visuospatial dysfunction, although overlap between neurotransmitter abnormalities occurs.17 ,18 Those with posterior cortical dysfunction and PIGD may be at higher risk of developing PDD15 ,16 and this hypothesis is currently being tested in longitudinal cohort studies.12 An important requirement of future research is to find biomarkers (table 2) that identify those with PD who are most at risk of developing PDD as these individuals may benefit from targeted intervention, ideally with disease-modifying agents once available, rather than the current ‘one-size fits all’ pharmacological approach.

Table 2

Radiological, cerebrospinal fluid (CSF) and blood biomarkers are being studied in order to identify those most at risk of cognitive decline

Current research questions

  • Do the cognitive domains affected in mild cognitive impairment in PD (PD-MCI) determine risk of progression to Parkinson's disease dementia (PDD)?

  • How reversible is PD-MCI and is this dependent on the cognitive domains affected?

  • What is the role of noradrenaline and serotonin deficiency in PD-MCI and PDD?

  • Are non-pharmacological treatments such as cognitive training of benefit in PD-MCI and PDD?

Key references

  • Irwin D, Lee V, Trojanowski J. Parkinson's disease dementia: convergence of α-synuclein, tau and amyloid-β pathologies. Nat Rev Neurosci 2013;14:626–36.

  • Kehagia AA, Barker RA, Robbins TW. Neuropsychological and clinical heterogeneity of cognitive impairment and dementia in patients with Parkinson's disease. Lancet Neurol 2010;9:1200–13.

  • Kempster P, O'Sullivan S, Holton J, et al. Relationships between age and late progression of Parkinson's disease: a clinico-pathological study. Brain 2010;133:1755–62.

  • Svenningsson P, Westman P, Ballard C, et al. Cognitive impairment in patients with Parkinson's disease: diagnosis, biomarkers, and treatment. Lancet Neurol 2012;11:697–707.

  • Williams-Gray CH, Evans JR, Goris A, et al. The distinct cognitive syndromes of Parkinson's disease: 5 year follow-up of the CamPaIGN cohort. Brain 2009;132:2958–69.

Self-assessment questions

Please answer true or false to the below statements,

  1. A level 2 diagnosis of PD-MCI can be made if a person with PD has abnormal results in two neuropsychological tests within the same core cognitive domain?

  2. A person diagnosed with PD in their 50s is likely to have a longer dementia-free survival time than someone diagnosed in their 70s?

  3. Lewy body and Lewy neurite deposition in the limbic system and neocortex is always associated with PDD?

  4. Deficits in executive function in those with PD are thought to be primarily driven by acetylcholine deficiency?

  5. Anticholinergics are a recommended treatment for PDD?



  1. True

  2. True

  3. False

  4. False

  5. False


View Abstract


  • Contributors JC: conception, manuscript production. JEA: manuscript revision and approval of submitted manuscript. SJ: manuscript revision and approval of submitted manuscript.

  • Competing interests None.

  • Provenance and peer review Not commissioned; externally peer reviewed.

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.