|Home | About | Journals | Submit | Contact Us | Français|
While Parkinson’s disease (PD) traditionally has been defined by its characteristic motor hallmarks, non-motor features such as cognitive impairment and dementia are increasingly recognized as part of PD. Mild cognitive impairment is common in non-demented PD patients, occurring in about 20-50%. Frequency estimates and clinical features of mild cognitive impairment in PD (PD-MCI), however, vary across studies due to methodological differences and lack of uniform diagnostic criteria for PD-MCI. Overall, PD-MCI patients exhibit nonamnestic deficits in cognitive domains such as executive function, attention, and visuospatial function; however, the cognitive phenotype of PD-MCI is heterogeneous with some patients demonstrating greater amnestic deficits. PD-MCI patients, particularly those with posterior cortical profiles, may be at high risk for developing dementia. Various biomarkers studied in PD-MCI including cerebrospinal fluid, genetic analyses, and neuroimaging suggest that there may be distinct PD-MCI profiles. Future studies using uniform PD-MCI diagnostic criteria and incorporating biomarkers and longitudinal follow-up of PD-MCI cohorts are needed to understand PD-MCI as a transitional state between normal cognition and dementia.
Parkinson’s disease (PD) is a neurodegenerative disease affecting over 4 million people over age 50 years with rates expected double over the next 2 decades.1 While PD traditionally has been defined by its characteristic motor hallmarks of rest tremor, bradykinesia, rigidity, and gait impairment, non-motor signs and symptoms are increasingly recognized as part of PD. Non-motor features of PD include not only cognitive impairment and dementia, but also mood disorders, psychosis, sleep disturbances, and autonomic dysfunction. These non-motor features have been associated with increased disability and reduced quality of life 2, 3 and are often unresponsive to levodopa or dopaminergic therapies. Non-dopaminergic neurotransmitters such as acetylcholine, norepinephrine, and serotonin are frequently implicated in the pathogenesis of the non-motor features and provide the rationale for several pharmacological interventions for cognition and mood. Moreover, these non-motor features typically increase with PD duration, and longitudinal studies suggest that they are the predominant source of disability at long-term follow-up.4, 5 This review will focus on mild cognitive impairment in PD (PD-MCI), a non-motor complication frequently encountered in the course of PD and often a precursor to dementia in PD.
PD-MCI has been increasingly recognized as a distinct entity and a potential prodromal state to PD dementia (PDD). As such, it is important to first highlight several features of PDD. Epidemiological studies suggest that the point prevalence rate of dementia in PD is about 40%.6 Longitudinal studies report that dementia ensues in the majority of patients at follow-up, occurring in 78% after 8 years 7 and 83% after 20 years.4 PDD has a substantial impact on both patients and caregivers and is associated with increased nursing home placement, morbidity, and mortality. 2, 3, 5 Clinically, the cognitive profile of patients with PDD typically reflects a “subcortical dementia” syndrome with greater impairment in nonamnestic cognitive domains (e.g., executive function, attention, and visuospatial function) and less impairment in declarative memory, language and praxis. The cognitive features of PDD, however, may be heterogeneous, and some patients may exhibit more “cortical” profiles with impaired memory and language 8-12. In 2007, the Movement Disorder Society (MDS) Task Force on dementia in PD published proposed diagnostic criteria for PDD. In contrast to DSM-IV criteria 13, memory impairment is not required. Rather, the MDS-PDD criteria place greater emphasis on deficits in nonamnestic cognitive domains and on the presence of concomitant behavioral features (e.g., apathy, mood disturbances, psychosis). Risk factors for PDD include mild cognitive impairment and cognitive dysfunction at baseline. 14, 15 Other factors such as older age, longer PD duration, older age at PD onset, greater motor severity, akinetic-rigid motor phenotype, psychosis, depression, and genetic factors such as APOE4 and MAPT alleles also have been associated with increased risk of PDD. 16, 17 To date, symptomatic treatments of PDD are limited, and there are no established neuroprotective interventions. Cholinesterase inhibitors and memantine in PDD provide modest benefit in PDD, and only rivastigmine has received approval by the Food and Drug Administration in the United States for PDD. 18 Since PD-MCI may represent the earliest stage of progressive cognitive deterioration and a risk factor for PDD 14, 19, 20, greater understanding the characteristics, progression, and pathogenesis of PD-MCI may lead to improved symptomatic interventions or neuroprotective strategies for PD cognitive decline.
Although the term, mild cognitive impairment (MCI), was introduced in late 1980’s, the concept of MCI has evolved from its initial use representing a stage in a global cognitive measure to a cognitive syndrome with both clinical and research diagnostic criteria and reflecting different underlying etiologies. 19, 21-23 In general, MCI represents a degree of cognitive impairment that is not normal for age. Although not all MCI progresses to a dementia, the construct of MCI implies that there is a continuum from normal cognition to dementia, with MCI representing a transitional or prodromal state. Historically, criteria for MCI developed from longitudinal, epidemiological studies of aging that demonstrated cognitive decline or conversion to dementia in subsets of elderly participants. 21
Early work in MCI emphasized memory impairment in the setting of intact general cognitive abilities and daily functioning and focused on MCI as a prodromal state for Alzheimer’s disease (AD). 19 This section will discuss MCI in general and in relation to incipient AD as this literature has influenced the concept of MCI in PD. Criteria for MCI as put forth by Petersen et al 19 specified the following: 1) a memory complaint, preferably corroborated by an informant, 2) impairment in memory as documented according to appropriate reference values, 3) essentially normal performance in non-memory cognitive domains, 4) generally preserved activities of daily living, and 5) not demented. With the recognition that not all MCI evolved into AD, these criteria were subsequently revised by Winblad et al 22 to incorporate both amnestic and nonamnestic clinical phenotypes of MCI. The revised criteria utilized an algorithmic approach to establish: 1) the diagnosis of MCI (i.e., presence of a cognitive complaint, which was not normal for age and represented a decline in cognitive function, but did not represent dementia or impair functional activities), 2) presence of memory impairment (i.e. categorized as yes or no, and thereby as amnestic or nonamnestic MCI, respectively), and 3) number of domains impaired (i.e., single or multiple), thereby leading to 4 MCI subtypes (i.e., amnestic MCI single domain, amnestic MCI multiple domain, nonamnestic MCI single domain, or nonamnestic MCI multiple domain). These criteria provide a more global approach to the clinical phenotype of MCI and recognize that etiologies of MCI are varied and not solely due to AD. Indeed, MCI can be attributed to degenerative, vascular, psychiatric, or other medical conditions. Studies have suggested that patients with amnestic single and multiple domain MCI subsequently progress to AD, whereas patients in the other MCI categories, particularly the nonamnestic ones, develop conditions such as frontotemporal dementia, vascular dementia, dementia with Lewy bodies, or depression. 21 Thus, MCI represents a heterogeneous clinical syndrome that can be ascribed to different etiologies.
One of the main reasons for the development of MCI criteria has been the early identification of patients who are at risk of converting to dementia and ultimately, advances in therapies that effectively halt or slow the progression of MCI. Although epidemiologic studies over the years have used different MCI criteria and have examined different subject populations, prevalence rates of MCI range span about 14-18% for persons aged 70 years and older. 24-26 Rates of MCI progression range from about 6-10% per year in community-based studies and 10-15% per year in AD clinical care centers, both greater than base rates of 1-2% per year. 21, 25, 27 Higher rates in the AD centers likely reflect referral bias and a greater likelihood of amnestic MCI in the cohort. The rate of progression of MCI to AD may be affected by a variety of factors including degree of cognitive impairment, genetic risk (e.g., apolipoprotein E [APOE] epsilon 4 carrier status), changes on neuroimaging (e.g., volume loss of the mesial temporal lobe or whole brain on structural magnetic resonance imaging [MRI], hypometabolism in temporoparietal areas on fludeoxyglucose F 18-positron emission tomography [PET], or increased amyloid binding using carbon-11 Pittsburgh Compound B scans), or cerebrospinal fluid (CSF) abnormalities (e.g., low beta-amyloid 1-42 peptide or elevated total tau or phosphorylated tau levels).
Recently, the recognition of biomarkers that may be associated with MCI and its conversion to dementia, particularly AD, have led to revised criteria by the National Institute on Aging and Alzheimer’s Association workgroup. 23 The workgroup recommendations include two sets of complementary MCI criteria, one for clinical diagnosis and the other for research purposes in settings with the capability of biomarker studies. The core clinical criteria include the following: 1) concern regarding a change in cognition as reported by the patient, reliable informant, or clinician observation, 2) impairment in one or more cognitive domains with evidence by impaired cognitive performance greater than the patient’s age and educational background (often considered as scores 1 to 1.5 standard deviations [SD] below appropriate normative data, though specific cutoff scores are not stipulated) or a decline in performance over serial evaluations, 3) preservation of independence in functional abilities, and 4) insufficient evidence of dementia. Specific cognitive tests are not prescribed though some examples are given. Formal cognitive testing of memory and non-memory domains is highly recommended, particularly over serial evaluations; several simple, informal bedside tests can be performed but do not adequately assess the full extent and characteristics of cognitive impairment. Once the clinical and cognitive criteria for MCI have been met, then the etiology of MCI is determined and attributed to either AD or non-AD processes. Non-AD processes may include vascular, traumatic, or other medical/neurological causes such as frontotemporal dementia, dementia with Lewy bodies, prion disease, neoplasm, or metabolic disorders. In contrast, the recently proposed MCI research criteria necessitate the use of biomarker studies. These biomarkers reflect different elements of AD pathology (e.g., beta-amyloid or tau deposition, neuronal injury, inflammation, oxidative stress). Using biomarker studies, MCI can be classified as: due to AD with intermediate or high likelihood, unlikely due to AD, or if the biomarker results are uninformative or not tested, MCI by clinical criteria. These clinical and research criteria yield important considerations for the development of operational definitions of mild cognitive impairment in PD.
While the presence of cognitive deficits in non-demented PD has been recognized for many years, it is only more recently that the construct of mild cognitive impairment in PD (PD-MCI) has emerged and operational definitions have been proposed. Increased appreciation of the importance and profound impact of non-motor complications in PD, including the frequent occurrence of PDD in about 80% at later disease stages, as well as MCI criteria (discussed above) have helped pave the way for PD-MCI research. Studies suggest that PD-MCI may represent the earliest stage of cognitive decline and a risk factor for developing dementia in PD. 14, 20 Thus, PD-MCI may represent an intermediate state between normal cognition and dementia in PD, similar in concept to amnestic MCI and the subsequent development of AD. As such, the recognition of PD-MCI has led to recent studies describing its frequency, clinical phenotype, and associated biological markers. Interpretation of these studies, however, has been complicated by differences in the definitions of PD-MCI used, populations studied, neuropsychological tests administered, and recognition of other features that may affect PD cognition (e.g., medications, mood, sleep). Furthermore, to date, whether all PD-MCI progress to dementia or whether there are distinct underlying neuropathologies or genetic susceptibilities are unknown. Studies will be needed to investigate the clinical progression and treatment responses of PD-MCI and its cognitive subtypes. To address some of these issues, operational criteria for the diagnosis of PD-MCI are under development by the MDS Task Force on PD-MCI. These criteria would provide the first steps in identifying uniform PD-MCI cohorts for clinical research studies and therapeutic trials. This review will highlight the epidemiology of PD-MCI, its clinical characteristics, pathogenesis and associated biomarkers, and diagnostic criteria.
Cognitive impairment has been frequently reported in non-demented PD using various definitions of MCI and evaluating different cohorts of PD patients. 10, 28-33 The presence of mildly impaired cognition has been noted even in early or newly diagnosed PD. 28, 29, 33 Frequency estimates of PD-MCI, however, vary due to factors such as differences in populations studied (e.g., clinic or community-based; incident or prevalent PD), exclusion of PD dementia (and criteria used), inclusion of normal control groups or use of normative data for neuropsychological tests, clinical and neuropsychological criteria used, and the number and type of neuropsychological tests or domains assessed. In addition, other contributing factors such as PD medications, co-morbid psychiatric disorders (e.g., depression, anxiety, apathy), fatigue and sleep disorders, and the motor demands of neuropsychological tests have not always been considered and merit attention when interpreting neuropsychological test performance in PD and defining PD-MCI. This section will review the epidemiology of PD-MCI in cross-sectional incident and prevalent PD cohorts as well as in longitudinal studies, citing frequency estimates and highlighting the PD-MCI definitions utilized (Table 1).
Even newly diagnosed or early stage PD patients may have impaired cognition, thereby fulfilling definitions of MCI. Early studies recognized mild cognitive deficits in newly diagnosed and untreated PD patients but generally had limited cognitive testing and small sample sizes, ranging from about 15-60 patients. 34, 35 In recent years, several studies have evaluated larger incident PD cohorts in the community, yielding estimates of about 20-35% with PD-MCI. In a population study identifying new cases of parkinsonism in the United Kingdom (CamPaIGN Study), Foltynie et al 28 reported that 57/159 (36%) of their incident PD cases were cognitively impaired but not demented. Cognitive impairment was defined by MiniMental State Examination score (MMSE) ≥ 24 and poor performance on a pattern recognition memory task reflecting temporal lobe function and/or the modified Tower of London task suggesting frontal lobe function; pattern recognition memory task scores were greater than 1 standard deviation (SD) below normative mean scores, and the cut-off score for the modified Tower of London task scores was <8/14 as based on age- and IQ-matched normative data. In this cohort, dementia was present in 13/159 (8%). In a study of 115 newly diagnosed PD patients and 70 cognitively normal, healthy controls in the Netherlands, Muslimovic et al 33 classified 24% of the PD patients, compared to 4% of the controls, as having cognitive impairment but not dementia. Subjects underwent detailed neuropsychological evaluations capturing 6 cognitive domains (psychomotor speed, attention, language, memory, executive functions, and visuospatial/constructive skills). Test scores were first categorized as impaired if they were at least 2 SD below the normative mean scores corrected for age and education; then cognitive impairment was defined by impaired performance on 3 or more neuropsychological tests. Patients were excluded if MMSE scores were < 24, suggesting dementia. In the Norwegian Park West Study, 196 non-demented, newly diagnosed, drug-naïve PD patients and 201 healthy controls were studied with a battery of neuropsychological tests assessing 3 cognitive domains (memory, attention/executive function, visuospatial function). Dementia was excluded on the basis of MDS-PDD criteria, impaired neuropsychological testing in 2 or more cognitive domains or MMSE < 26, and specified cut-off score on the Informant Questionnaire on Cognitive decline in the elderly. Of the non-demented PD patients, 37/196 (18.9%) were classified as MCI as defined by cognitive z-scores -1.5 SD in one of 3 cognitive domains. 29 Although these studies vary in definitions of PD cognitive impairment used, they document that cognitive dysfunction occurs early in the course of PD and even prior to initiation of dopaminergic therapy for motor aspects of the disease.
Other studies of cognitive impairment in PD have focused on patients who have been treated for their PD for typically several years. These prevalent studies include both community and clinic-based cohorts and report somewhat higher frequencies of PD-MCI, ranging from about 20-55%. While several large community studies report frequency estimates of cognitive impairment in PD, not all have excluded demented patients in the analyses 36, 37 and as a result, preclude accurate estimates of MCI in non-demented PD cohorts; therefore, this review will focus on those studies in which a non-demented PD cohort is documented. In a community study of 103 PD patients with disease durations of at least 4 years, 42/76 (55%) of the non-demented PD patients had cognitive impairment. Cognitive impairment was defined by scores < 2 SD below the control group means on one or more neuropsychological tests (i.e., Dementia Rating Scale, Benton Visual Retention Test, Judgment of Line Orientation, and Stroop Word Test) and not fulfilling DSM-III-R dementia criteria. 10 Several studies have reported similar frequencies in clinic-based cohorts. Caviness et al 30 studied 86 PD patients who were enrolled in a brain donation program and reported that 18/86 (21%) met modified Petersen’s criteria 19 for PD-MCI, defined as the presence of subjective cognitive complaints without significant functional decline and deficits of at least 1.5 SD below the age corrected mean scores in one of the 5 cognitive domains; 17% were demented as based on DSM-IV criteria for PDD, and the remaining 62% were cognitively normal. PD-MCI patients had a mean disease duration of 9.2 (5.0) years, which was significantly longer than the cognitively normal PD (5.4 +/- 4.2 years) and shorter than the demented PD groups (16.8 +/- 6.6 years). A retrospective record review of 72 non-demented PD patients demonstrated that 38/72 (52.8%) had MCI; cognitive status of these patients was determined in a consensus conference using Petersen criteria 19 along with a subjective complaint, deficits on at least 2 measures in a cognitive domain and use of published normative data. 38 Dalrymple-Alford et al reported MCI in between 14-89% of the 156 PD patients who underwent neuropsychological testing in their convenience sample, depending on the cut-off scores for number of tests impaired and standard deviations below normative data means used to define MCI. 39 In a clinic-based study of 106 PD patients selected for normal age- and education-adjusted MMSE scores, MCI was detected in 29.2% of patients, defined by scores ≥ 1.5 SD below the standardized mean on specific neuropsychological tests 40; mean PD duration in this cohort was 6.5 (5.8) years. Similarly, a subanalysis of a large community study found that 50 of the 88 PD patients (58%) with impaired cognition on the Scales for Outcomes in Parkinson’s disease-cognition (SCOPA-COG) had normal MMSE scores based on age and education corrected normative data. Although PD patients with dementia were not specifically excluded from the total PD sample, this may represent a group with mild cognitive impairment.
One reason for the wide prevalence estimates of PD-MCI has been the use of different PD-MCI definitions. To address this, Aarsland et al 32 evaluated data from a large multi-center cohort of PD patients using standardized analytic methods and a common definition of PD-MCI. A total of 1346 PD patients from 8 centers were studied. Although the cohorts differed in their setting (4 community-based, 4 clinic-based), type of cohort (3 cross-sectional, 3 prevalent, 1 incident, and 1 de novo), dementia criteria used (3 DSM-II-R or DSM-IV, 3 DSM-IV plus other tests, and 1 MDS-PDD), and individual neuropsychological tests used to assess cognition and depression, a common PD-MCI definition was applied. In order to achieve this definition, the different neuropsychological tests used were first converted to cognitive domain z-scores representing 3 cognitive areas (memory, executive function/attention, and visuospatial function) and then patients were classified as having MCI if their age- and education-corrected z-score on one or more cognitive domains was at least 1.5 SD below the mean of either control subjects or normative data. With the multi-center pooled data analyses, a total of 25.8% of PD subjects had MCI (range 18.9-39.4%; 95% confidence interval 23.5-28.2). This study illustrates that PD-MCI is common, even across different cohorts, and that standardized diagnostic criteria are needed to capture uniform populations for observational and interventional studies.
Concomitantly, the MDS Task Force on PD-MCI was commissioned to critically evaluate the literature on PD-MCI, determine its frequency and characteristics, and develop formal diagnostic criteria. 31 As a result, a comprehensive review of the literature on PD and cognitive impairment was conducted with MedLine searches, pre-specified inclusion/exclusion criteria, and structured data extraction and quality assessment procedures. From the literature reviewed, the Task Force focused on 8 articles that represented a total of 776 PD patients from 6 cross-sectional studies and 198 from 2 prospective studies. These studies, despite variations in methodology, study design, population, and criteria used, demonstrated a mean cross-sectional prevalence rate of 26.7% (range 18.9-38.2%) for PD-MCI. Two longitudinal studies included in this review also reported frequency estimates of PD-MCI after 3-5 year follow-up. Janvin et al 14 reported that at 4 year follow-up, dementia developed in 18/29 patients with PD-MCI at baseline (62%) whereas in only 6/30 with intact cognition at baseline (20%). At 3-5 year follow-up of those participants in the CamPaIGN study, 13/126 (10%) were demented as defined by MMSE < 24 and DSM-IV criteria; an additional 57% had cognitive impairment, mainly frontostriatal deficits, based on specified cut-off values including selected neuropsychological test scores < 1 SD below normative means. 20
Cognitive dysfunction in non-demented PD encompasses a broad spectrum of clinical deficits and severity, affecting both nonamnestic and amnestic domains. There is a long history documenting impaired neuropsychological performance in these patients, even prior to the more recent application of PD-MCI definitions. Cognitive complaints in non-demented PD typically include slowed processing, difficulty with multi-tasking or planning, decreased attention and concentration, and word finding disturbances. This section will first highlight impairment in cognitive domains in non-demented PD patients in general, and then focus on those studies in which PD-MCI is specifically defined.
Impairment in a variety of cognitive domains such as executive function, psychomotor speed, visuospatial abilities, language, and memory has been noted in non-demented PD patients. Executive function, which encompasses the ability to plan, organize, initiate, and regulate goal-directed behavior and relies on frontal-striatal circuitry including prefrontal regions such as the dorsolateral prefrontal cortex and its connections to the basal ganglia 41, is a prominent feature not only in the “subcortical dementia” syndrome in PD but also in non-demented PD. 8, 42 Non-demented PD patients exhibit deficits in tasks that require cognitive sequencing 34 planning 35, or set maintenance or shifting to novel stimuli. 43 Similar to the motor feature of bradykinesia, some PD patients exhibit bradyphrenia or slowing of psychomotor speed on tests, independent of motor demands. 44 Visuospatial, visuoperceptual, or visuoconstructive abilities may be impaired in non-demented PD, though this has not been consistently reported across studies. 10, 45,46 Language function is relatively spared in non-demented PD patients; however, decreased information content of spontaneous speech, impaired comprehension of complex sentences, and impaired verbal fluency may occur, thereby reflecting frontal lobe involvement in concept formation, set shifting, attention, and working memory. 47-50 Regarding memory, PD patients exhibit difficulty learning novel information as demonstrated by impaired free recall performance but may improve with semantic cues or recognition tasks. 11, 51, 52 Impaired registration and retrieval may relate to inattention or executive dysfunction rather than encoding deficits, though some studies demonstrate problems with encoding. 53
The clinical phenotype of PD-MCI is heterogeneous with both nonamnestic and amnestic cognitive domains affected and single and multiple-domain impairment. Nonamnestic, single domain impairment predominates in PD-MCI, with executive function as the most frequent cognitive domain affected, though some PD-MCI patients exhibit greater amnestic or cortical-type profiles. 20, 32 It should be noted that the PD-MCI cognitive profiles are linked to issues such as the PD population studied, neuropsychological tests selected, cut-off scores used, sources of normative data, and PD-MCI definitions employed. Moreover, further categorization of already small PD-MCI groups leads to very small numbers of patients in the cognitive domain subtypes, in some cases. This section will review the cognitive profiles of PD-MCI subjects reported in those studies for which prevalence estimates were discussed above (Table 2).
In studies evaluating PD-MCI in incident PD cohorts, deficits were evident in executive function, attention, psychomotor speed, memory, and visuospatial skills. 28, 29, 33 Of the PD-MCI subjects in the CamPaIGN study, 58% had single domain MCI with frontostriatal deficits in 34% and temporal lobe deficits in 24%; multiple-domain MCI occurred in 42%. 28 Muslimovic et al 33 reported PD-MCI in about 24% with impaired performance particularly on neuropsychological tests measuring executive function, memory and psychomotor speed, although specific MCI subtypes are not presented. In the ParkWest incident PD cohort 29, evaluation of PD-MCI subtypes revealed that the majority (23/37, 62.2%) had nonamnestic single domains affected. Amnestic MCI was less common, occurring as single-domain impairment in 9/37 (24.3%) and as part of multiple-domain MCI in 4/37 (10.8%).
Studies of prevalent PD cohorts reveal similar cognitive profiles with predominant single domain MCI and greater nonamnestic subtypes of PD-MCI. 30-32, 38, 40 Caviness et al 30 described single domain MCI in 67% of their PD-MCI patients, compared to 33% with multiple-domains affected. Frontal/executive function was the most common domain affected, in either group, followed by amnestic deficits. The distribution of PD-MCI subtypes was as follows: nonamnestic single domain > amnestic single domain = nonamnestic multiple domain > amnestic multiple domain. Visuospatial skills and attention did not meet the specified cut-off scores used in defining PD-MCI. In the multi-center pooled analysis of PD-MCI patients, Aarsland et al 32 reported that single domain MCI was most frequent, affecting 19.5% compared to 6.3% with multiple-domains affected. Of the PD-MCI subtypes, the majority had nonamnestic single domain (attention/executive or visuospatial dysfunction) > amnestic single domain > amnestic multiple domain > nonamnestic multiple domain.
Few studies have evaluated the progression of PD-MCI and its subtypes. In one study, 18/29 of the PD-MCI subjects (62%) who completed follow up at 4 years had converted to PDD. 14 Single domain nonamnestic MCI was associated with conversion to PDD whereas predominant amnestic deficits and multiple domains were not, although the sample size was small and the neuropsychological battery, limited in scope. In this cohort, patients who developed dementia also had higher Beck Depression Inventory scores. Follow-up of the CamPaIGN cohort at 3-5 years revealed that 13/126 (10%) converted to PDD and cognitive impairment in 57%, predominantly affecting frontostriatal functions. 20 Predictors of global cognitive decline included a more posterior cortical profile (e.g., impaired semantic fluency or pentagon copying) and non-tremor dominant motor phenotype at baseline.
Some, but not all, studies demonstrate other clinical differences in PD-MCI or PD-MCI subtypes. PD-MCI has been associated with older age at evaluation, older age at PD onset, male gender, depression, worse motor symptoms, and advanced disease. 32 In an incident PD cohort, however, there were no significant differences in demographic or motor features in the PD-MCI compared to PD without MCI. 29 Larger sample sizes of PD-MCI will enable further subtype comparisons, and longitudinal follow-up PD-MCI patients will permit examination of MCI subtype progression and risk factors for conversion to PDD.
The diagnosis of PD-MCI rests upon the concept that MCI, in general and specific to PD, refers to a clinical syndrome of cognitive impairment in the absence of dementia. However, as the studies of PD-MCI illustrated in this review demonstrate, the diagnosis of PD-MCI has been made in many different ways and using various definitions; these factors likely influence our current knowledge of frequency estimates and characteristics of PD-MCI. Operational definitions of PD-MCI are currently underway by the MDS Task Force on PD-MCI. This section will explore issues pertaining to diagnostic criteria, effects of associated co-morbidities in PD, challenges in the cognitive assessment of PD patients, and screening tests (Table 3).
In keeping with established MCI criteria, many of the studies of MCI in PD have utilized established criteria 19, 22 or modified versions of these. Items often modified pertain to the presence of memory complaints, preservation of activities of daily living, and documentation of memory (or other cognitive) impairment according to different reference values. Few PD-MCI studies have required the presence of a memory (or cognitive) complaint by the patient. Since nonamnestic deficits such as executive dysfunction are more common than memory ones, subjective complaints of memory in PD may not be the most appropriate requirement for the diagnosis of PD-MCI; this item may need to be expanded to broadly address complaints in all cognitive areas. Furthermore, estimates of cognitive impairment by patients and their caregivers may not be reliable due to either over or under-reporting 30, 54, 55 or to difficulty separating cognitive from motor problems. Thus, information from several sources (e.g., patient, caregiver, and clinician) may be needed to determine if a patient has declined cognitively. In addition, PD-MCI studies vary in the manner in which the preservation of activities of daily living is evaluated. This issue is particularly challenging in PD since daily activities may be profoundly affected by PD motor symptoms; specific measures that distinguish the cognitive from motor effects on daily functioning and sensitive cognitive measures to identify functional impairment are needed. 56 Although the established MCI criteria do not specify how many neuropsychological tests or which tests should be used, what cognitive domains and how many should be examined, what normative data should be used, or which cut-off scores should be employed, these are important considerations in the diagnosis of PD-MCI and the development of operational definitions. The PD-MCI studies previously described have handled these issues pertaining to neuropsychological tests (e.g., number of neuropsychological tests, specific tests, number of cognitive domains, sources of normative data, and cut-off scores) in various ways. Studies illustrate that the use different cut-off scores (ranging from 0.5 to 2 SD below normative data) greatly influence frequency estimates of PD-MCI. When only a single neuropsychological test, rather than 1 of 5 cognitive domains, was required to be ≤ -1.5 SD, the reported number of PD-MCI cases doubled. 30 Different cut-off scores change the classification of not only PD-MCI cases but also controls 57; when 2 scores in 1 cognitive domain fell 2 SD below normative values, rates of MCI were 14% in PD and 0% in controls, but when only 1 score in 1 domain was 1 SD below normative values, MCI occurred in 89% of PD and 70% of controls.
Motor and non-motor features of PD may affect cognitive function in PD and thereby, the diagnosis of PD-MCI. Cognitive performance may differ in “on” versus “off” motor states, particularly in tests of executive function and also in those PD patients with motor fluctuations. 58, 59 Neuropsychological tests with timed components or significant motor demands may be difficult for PD patients to complete. Non-motor features such as depression, anxiety, apathy, psychosis, fatigue, and sleep disturbances are common in PD, particularly in those patients with impaired cognition or dementia. 9, 11 These associated non-motor features may interfere with cognitive testing and interpretation of neuropsychological test results. Thus, motor and non-motor co-morbidities in PD pose unique challenges in the evaluation of PD patients and classification of cognitive status.
There are many neuropsychological tests available to evaluate cognitive functions (e.g., executive function, attention, psychomotor speed, language, visuospatial abilities, and memory). As such, it is likely beyond the scope of any MCI (or dementia) criteria to prescribe specific neuropsychological tests. It is useful, however, for a comprehensive evaluation to include tests that capture each cognitive domain, have been well-studied and validated, and have appropriate normative data; diagnostic procedures for PD-MCI and PDD provide examples of tests frequently used in evaluating cognitive impairment in PD. 60, 61
Several global cognitive scales have been proposed as screening tests for PD-MCI and/or PDD, including the MoCA, SCOPA-COG, Parkinson Neuropsychometric Dementia Assessment (PANDA), PD-Cognitive Rating Scale (PD-CRS), and Mattis DRS among others; most of these tests take about 10-30 minutes to administer. 62, 63 The MoCA, which was originally developed to screen for MCI in the general population, assesses orientation, executive function, attention/concentration, naming, verbal abstraction, and visuoconstructive abilities. Although the MoCA performed better in detecting PDD, it had an acceptable sensitivity (0.83), but low specificity (0.53) using a cutoff of 26/27 in detecting MCI in one PD study. 64 In another study of PD patients, of whom 21/114 (18%) had PD-MCI, the MoCA, SCOPA-COG, and standardized MMSE demonstrated excellent discrimination of PDD from non-demented PD and PD-MCI from cognitively normal PD; the MoCA in particular performed well, with an optimal screening cut-off for PD-MCI of <26/30 having 90% sensitivity, 75% specificity, and a negative predictive value of 95%. 65 The SCOPA-COG, which includes tests of nonverbal and verbal memory, attention, executive function, verbal fluency, and visuospatial function, is reliable and sensitive to PD cognitive deficits. A large study of 400 PD patients and 150 healthy controls replicated validation studies and also detected cognitive impairment in 88 PD patients as well as in 50/88 (58%) of those PD with normal MMSE scores. 36 The PANDA which includes tests of immediate and delayed recall memory, alternating verbal fluency, visuospatial abilities, and working memory/attention has been reported to discriminate between PDD, PD-MCI, non-demented PD, and controls; however, in its current form, the scale lacks assessment of language function; data on clinimetric scale properties, construct validity, and test-retest reliability, assessment; and clearly defined PD-MCI population, which may necessitate additional study. 66 The PD-CRS scale assesses not only frontal-subcortical but also posterior cortical deficits in PD with tasks of sustained attention, working memory, alternating and action verbal fluency, clock drawing, immediate and delayed free-recall verbal memory, confrontation naming, and clock copying. 12 The PD-CRS differentiated between PD-MCI and PDD patients and also between PD-MCI and cognitively intact PD patients; items of alternating verbal fluency and delayed verbal memory independently differentiated the PD-MCI group from both controls and cognitively intact PD. The Mattis DRS, a measure of global cognitive function but with an emphasis on frontal-lobe dysfunction, has been widely used in PDD with cut-off scores of < 123 frequently specified. Compared to the other screening tests, the DRS takes longer to administer, requiring about 20-30 minutes. 67 In a recent study of 40 PD patients, the DRS demonstrated a sensitivity of 72% and specificity of 86% to correctly classify 80% of PD-MCI, using a normality cut-off of 138. 68
Cognitive deficits in non-demented PD have been frequently attributed to neurochemical alterations in dopaminergic, cholinergic, and other systems and neuropathological contributions of limbic and cortical Lewy bodies and neurites, amyloid deposition, neurofibrillary tangles, and cerebrovascular disease. Extensive literature, beyond the scope of this review, has highlighted the relationships among executive function and dopamine. Executive dysfunction may reflect frontal lobe impairment, particularly affecting the dorsolateral prefrontal cortex, resulting from disruptions of frontostriatal loops, due to either degeneration of dopaminergic nigrostriatal or mesocortical pathways. 42 Hypodopaminergic states, such as prior to initiation of dopaminergic therapies, after withdrawal of levodopa or other dopaminergics, or induced by dopamine blocking medications or MPTP exposure, have been associated with executive dysfunction, decreased mental flexibility, and impaired working memory.69-71 Moreover, dopaminergic medications may improve some neuropsychological measures in PD patients such as non-specific effects on alertness/arousal, working memory, planning tasks, cognitive flexibility, apathy. 58, 72,73 The cognitive effects of dopaminergic therapies, however, are highly variable in PD, also leading to either worsened probabilistic reversal learning, decision-making, or choice reaction times or showing no effect on cognitive slowing or reward associated learning in others. 72-74 Besides dopamine, the cholinergic system has been implicated in PD-related cognitive impairment with degeneration of the nucleus basalis of Meynert, decreased cholinergic activity in the cortex, and reduced choline acetyltransferase activity in the frontal and temporal lobes in PD. 75,76 These deficits may be linked to impairment in attention, learning and memory in PD.
To date, neuropathological studies of PD-MCI are limited. The neuropathological staging of PD by Braak and colleagues, however, supports changes in brain regions involved in cognition in stages in early symptomatic PD (neuropathological stage 3). 77 The main characteristics of this stage are changes in melanoneurons in the substantia nigra and projection neurons in the magnocellular nuclei of the basal forebrain. Cholinergic deficits in the basal forebrain and prefrontal cortex may be responsible for some of the cognitive deficits in PD. 75, 78 Neuropathological stage 4 is represented by involvement of limbic structures including the anteromedial temporal mesocortex, which contribute to memory function. In an autopsy series of PD patients, the majority of individuals with either intact cognition or marginal impairment (MMSE scores 25-30) or mildly impaired cognition (MMSE scores 21-24) were classified as neuropathological stage 3 to 4. 79 Adler et al reported heterogeneous neuropathological findings in 8 PD-MCI cases, which included 4 patients who had amnestic single domain MCI, 3 with nonamnestic single domain MCI (executive dysfunction), and 1 with nonamnestic multiple domain MCI (executive/visuospatial dysfunction). 80 The 8 cases had varied distributions of Lewy bodies, with 5/8 demonstrating limbic and/or neocortical involvement; regarding AD neuropathology, 2 cases, both of whom had amnestic MCI subtypes were Braak AD stage IV and 4 other cases were stage III. Cerebrovascular pathology occurred in 3 cases. Future studies will be needed to examine the underlying neuropathology of PD-MCI and its relationship to MCI subtype.
There is a growing interest in the identification of biomarkers that characterize MCI states and predict conversion to dementia. As previously discussed, biomarkers have been recently incorporated into research MCI criteria with an emphasis on the early detection of and interventions for AD.23 Biomarkers for PD-MCI including CSF, genetics and neuroimaging are currently under study, though at present, many studies are limited by small samples and a lack of uniform PD-MCI definitions, detailed neuropsychological evaluation, and longitudinal follow-up. Future studies will be needed to explore these biomarkers, their use in combination, and their incorporation in PD-MCI criteria. This section will review several of these studies focused on CSF, genetics, and neuroimaging.
Proposed CSF biomarkers for PD cognitive decline include those CSF biomarkers that have been associated with AD (i.e., low beta-amyloid 1-42 peptide and elevated total tau or phosphorylated tau levels), though few studies have specifically examined PD-MCI cohorts. Decreased CSF beta-amyloid 1-42 has been demonstrated in PDD patients compared to healthy controls; some, but not all, non-demented PD demonstrate intermediate reductions of CSF beta-amyloid 1-42. 81-83 Newly diagnosed PD patients from the Norwegian ParkWest cohort had decreased CSF beta-amyloid levels, though not as reduced as in AD; CSF beta-amyloid 1-42 levels were significantly associated with memory impairment but not with attention/executive or visuospatial dysfunction. 84 CSF total tau or phosphorylated tau levels, which did not differ between PD patients and controls, were not correlated with the cognitive measures. Although PD-MCI has been reported within this patient cohort in other studies, specific relationships between the CSF levels and diagnosed PD-MCI patients were not included here. Siderowf et al 85 examined CSF beta-amyloid 1-42, total tau, and phosphorylated tau levels in 45 PD patients who had at least 1 year longitudinal follow-up evaluation including Mattis DRS-2 scores. Reduced CSF beta-amyloid 1-42 independently predicted cognitive decline in PD patients, though those with PD-MCI were not specified. Montine et al studied 86 CSF beta-amyloid 1-42, total tau and phosphorylated tau in PDD and PD patients with cognitive impairment but not dementia (termed PD-CIND), which likely equates with “PD-MCI.” CSF beta-amyloid 1-42 was reduced in PD-CIND and PDD, whereas total tau levels were unchanged and phosphorylated tau was significantly decreased in PD-CIND. Approximately one-third of PD-CIND patients had abnormally increased CSF phosphorylated tau/beta-amyloid 1-42 ratios, in contrast to the >90% of amnestic MCI or AD who exhibited this profile. These findings suggest that validated CSF biomarkers may be helpful in identifying PD patients who may have co-morbid AD.
Several genetic biomarkers have been explored in PD patients with cognitive dysfunction including polymorphisms related to dopamine regulation and tau proteins. Since a functional polymorphism in the catechol-O-methyl transferase gene (COMT Val158Met) alters the activity of the dopamine-regulating enzyme in the prefrontal cortex, gene variants may affect executive function tasks. Val/Val polymorphisms lead to increased dopamine catabolism and thus, decreased post-synaptic dopaminergic stimulation, whereas Met/Met polymorphisms lead to decreased COMT enzyme activity and increased dopamine levels. In the CamPaIGN cohort, the COMT genotype was not associated with subsequent cognitive decline and dementia over the 5 years of follow-up.87 However, the microtubule-associated protein tau gene (MAPT) H1/H1 genotype was found to be a risk factor for PDD in this cohort and also associated with greater posterior cortical cognitive impairments. This finding was replicated in another study that demonstrated stronger associations of the MAPT H1/H1 genotype with PDD patients compared to controls. 17 These findings suggest that these two distinct cognitive syndromes in PD (frontostriatal versus posterior cortical deficits) may have different prognoses or risk for developing dementia. Although APOE-epsilon 4 has been associated with AD, conflicting results have been found in PD and PDD. Longitudinal follow-up of the CamPaIGN cohort did not reveal any significant effect of APOE-epsilon 4 carrier status on risk of dementia or rate of cognitive decline, although there were more APOE-epsilon 4 carriers in the PDD compared to non-demented PD cases, and larger studies are needed. 16 These biomarker findings merit replication in larger samples, with particular attention to PD-MCI cohorts, and in due course, longitudinal follow-up of the patients to establish risks of cognitive decline.
Structural and metabolic neuroimaging may provide another potential biomarker for PD-MCI. Using voxel-based morphometry (VBM) analyses of brain MRI, Beyer et al found that PD-MCI patients, defined using Petersen criteria, had reduced gray matter in the left frontal and bilateral temporal lobe regions, compared to PD without MCI; however, these differences did not remain significant after corrections for multiple comparisons and patient groups were small in size. 88 A small group of PD-MCI patients, defined as having at least one of 3 cognitive domain scores 1.5 SD below normal control data, exhibited anterior caudate atrophy and posterior ventricular enlargement on MRI using a radial distance 3D statistical mapping technique. 89 Using VBM techniques, Lee et al 90 compared a large sample of PD patients with amnestic MCI (n=41) to amnestic MCI patients (without PD, n=78); multiple domains were affected in 26/41 (63%) PD-MCI with impaired visuospatial function. The amnestic MCI PD group had decreased gray matter in the precuneus, left frontal, left primary motor cortex compared to controls and right temporal and anterior prefrontal areas compared to amnestic MCI (without PD). The affected regions were similar but more extensively involved in those PD-MCI with multiple domain impairment. Several studies have focused on MRI correlates of PD-MCI in the incident Norwegian ParkWest PD cohort. Dalaker et al 91 reported that VBM analyses did not reveal differences in gray matter atrophy in newly diagnosed PD patients, of whom 11/42 had MCI, compared to healthy controls. In the same cohort, PD-MCI patients had enlarged left inferior lateral ventricle and third ventricles, measured on MRI scan, compared to PD without MCI and healthy controls; fourth ventricle size was highly correlated with memory in PD-MCI patients. 92
Neuroimaging studies using single photon emission computed tomography (SPECT) or PET generally suggest impaired metabolism in posterior cortical regions, similar to regions that are frequently abnormal in PDD patients. Huang et al 93 studied PD-MCI patients with single and multiple-domains affected with 18F-flurodeoxyglucose (FDG) PET scans, compared to PD without MCI. PD-MCI was diagnosed using consensus criteria 22 and at least one of the 4 cognitive domains assessed was 1.5 SD below an age-corrected normative sample. PD-MCI patients with multiple domains impaired had decreased glucose metabolism in not only prefrontal regions but also parietal areas; PD-MCI patients with only a single domain affected showed a similar pattern but to a lesser degree than in the multiple domain subtype. Nobili et al 94 compared PD-MCI patients, who reported memory complaints and were defined as having an objective memory impairment (i.e., scores for immediate or delayed recall tasks below 1.5 SD mean normative data), to PD patients without subjective or objective cognitive deficits, amnestic MCI patients (without PD) and healthy controls. The PD-MCI group (of whom 11 had an isolated memory deficit and 4, a mild deficit in verbal fluency) demonstrated relative hypometabolism in bilateral posterior parietal lobe and right occipital lobe compared to the controls and parietal, temporal, and occipital hypoperfusion compared to the cognitively intact PD. Interestingly, compared to the amnestic MCI patients (without PD), the PD-MCI had greater hypoperfusion in parieto-occipital regions, whereas the amnestic MCI patients had greater hypoperfusion in medial temporal lobe regions. In addition, FDG-PET demonstrated extensive areas of hypometabolism in parietal, temporal, and occipital cortex in PD-MCI patients, defined by a Clinician’s Dementia Rating Scale (CDR) score of 0 (normal) or 0.5 (questionable dementia) and MMSE ≥ 24, compared to those PD without cognitive impairment and healthy controls. 95 Although a CDR score of 0.5 traditionally equates with “questionable dementia,” the authors considered that in the setting of MMSE ≥ 24, this designation represents a cognitively impaired but not demented PD group. Structural and metabolic neuroimaging studies support regional atrophy patterns or altered metabolism in PD-MCI, with notable abnormalities in posterior cortical areas and possible shared features with PDD and/or AD.
To date, clinical research trials and therapeutic interventions specifically for PD-MCI are limited. Medications such as cholinesterase inhibitors or memantine have been studied in PDD, but not PD-MCI. Dopaminergic medications such as levodopa or dopamine agonists may have variable effects on cognition, with improvement in executive function tasks in some PD patients and in others, worsening or no effect. 72 A recent randomized, double-blind, placebo-controlled, multi-center study assessed the effects of rasagiline, a monoamine oxidase type-B inhibitor, on cognitive deficits in non-demented PD patients; 48 patients completed the study. 96 In this trial, PD patients were included if they had subjective cognitive complaints, preserved daily living activities, impairment in at least 2 of 4 cognitive domains (defined as performance 1.5 SD below age and education adjusted normative data), and did not fulfill DSM-IV criteria, thus meeting PD-MCI definitions described in this review. The rasagiline-treated group showed significant improvement in some cognitive tasks relevant to executive function and attention, compared to the placebo group (Digit span – backwards, p=0.04, verbal fluency, p=0.038, composite cognitive domain z-score, p<0.005). Atomoxetine, a norepinephrine reuptake inhibitor indicated for attention deficit hyperactivity disorder, has been studied in an 8-week, pilot trial in 12 non-demented PD patients with moderate executive dysfunction. 97 In this open-label study, atomoxetine improved scores on the Clinical Global Impression-Change Scale in 75% of the patients and on behavioral measures of executive dysfunction; side effects included hypomania, reduced sleep, and gastrointestinal effects. Another study with atomoxetine, focused on depression in PD rather than cognition, reported significant improvement in global cognition and daytime sleepiness. 98 Treatment of co-morbid conditions such as depression, anxiety, apathy, and sleep dysfunction may play a role in treating MCI in PD. In addition, non-PD medications with the potential to impair cognition or cause confusion (e.g., centrally-acting medications for pain, bladder function, sleep) should be avoided or used sparingly. There is an emerging literature on cognitive rehabilitation in PD with preliminary studies suggesting that cognitive training with computerized programs improves neuropsychological performance on tests of executive function, attention, visuospatial function, and memory, though more rigorous studies are needed. 99, 100
Mild cognitive impairment in PD is frequent, occurring in about 25% of PD patients, though studies differ in methodologies used. Cognitive impairment in PD is not only a late-stage feature of the disease, but may be evident in about 20-30% of early or newly diagnosed PD patients. Although single domain impairment with nonamnestic deficits predominates in PD-MCI, the cognitive profile of PD-MCI is heterogeneous with a mix of nonamnestic and amnestic deficits and single and multiple-domains affected. Studies suggest that not only frontostriatal dysfunction contributes to PD cognitive impairment, but that posterior cortical deficits may contribute to risk of conversion to PDD. Studies incorporating CSF, genetics, and neuroimaging may provide clues to the pathogenesis of PD-MCI and have potential utility as biomarkers of PD-MCI and PDD, but further research is needed. Diagnostic criteria for PD-MCI are underway by the MDS Task Force, and uniform definitions of PD-MCI will facilitate comparisons across multiple centers and PD populations, provide well-characterized groups for clinical research trials, and identify cognitive changes at early stages. Further understanding of PD-MCI, its clinical subtypes, underlying neurobiology, related risk factors, and rates of progression is needed as are advances in symptom-targeted therapies for PD-MCI and PDD and neuroprotective interventions for cognitive decline.
Movement Disorder Society
Funding: NIH (5R01AG024040-05, K23NS060949), Parkinson’s Disease Foundation, Parkinson Disease Support Center of Kentuckiana, Movement Disorder Society
Conflicts of interest: None