The results of our systematic literature search and review are that: (1) an average of 26.7% (range 18.9–38.2%) of non-demented PD patients have PD-MCI; (2) cognitive deficits can be detected in some patients even at the time of PD diagnosis; (3) the frequency of MCI increases with age, and duration and severity of PD; (4) impairments can occur in a range of cognitive domains; (5) non-amnestic single domain MCI is more common than amnestic single domain MCI; and (6) PD-MCI appears to be a risk factor for the development of PDD.
Prevalence and Correlates
The majority of PD patients will develop dementia.3, 22
The point prevalence of PDD is approximately 30%, 75
and the cumulative prevalence is at least 75% for PD patients surviving more than 10 years.2, 3
As MCI precedes PDD, the cumulative prevalence of PD-MCI must be at least as high as that of PDD. Consequently, our finding that approximately 27% of PD patients meet criteria for PD-MCI at any given time is not surprising. Our results are similar to those recently reported by Aarsland et al.,76
who used a common methodology for the definition of MCI on pooled data of over 1,000 non-demented PD patients from multiple centers, and found that 25.8% (23.5–28.2%) had MCI.
The lowest proportion of patients with MCI was found in a study of patients at early PD stages,13
and the highest proportion in studies including cases with more advanced disease severity and duration.16, 68, 76
The variation in MCI frequency present in the articles reviewed also reflects differences in methodology (e.g., study settings and populations, recruitment methods, MCI diagnostic criteria, number of cognitive domains assessed, number of tests used for each domain, and how impairment on a test was defined). The fact that cognitive deficits in PD are detectible in some patients even at the time of clinical diagnosis highlights complexities in differentiating PD from dementia with Lewy bodies (DLB).
The correlates of PD-MCI have not been studied extensively. Of the studies included in this review, there was evidence that increasing age, 34, 47
more severe PD, 34, 47
late onset of disease, 13
and lower educational levels 53
were associated with PD-MCI.
Profile of Cognitive Impairment
Although the cognitive deficits in PD have traditionally been classified as being “subcortical” in nature77
(i.e., relatively greater impairments in executive abilities, information processing speed, and working memory compared with episodic memory storage and language), our review showed that a range of cognitive domains are impaired in PD patients without dementia. Other research in PD has demonstrated deficits in executive (i.e., impaired planning and working memory),78
and even language abilities.82–85
In all of the studies reviewed herein, single domain MCI was more common than multiple domain MCI, and non-amnestic MCI was more common than amnestic MCI in patients with impairment in a single domain. Debate exists about the extent to which the mild memory and language deficits in PD are secondary to executive and working memory problems. Additionally, as prospective studies using formal definitions for MCI subtypes are almost non-existent, the usefulness and predictive value of this PD-MCI classification structure is currently hypothetical.
Epidemiology of Progression from PD-MCI to Dementia
The few longitudinal studies of non-demented PD patients find that 20–60% develop PDD over a period of 2–5 years,22, 46, 76, 86–89
and even newly diagnosed PD patients on average experience significant decline in a range of cognitive domains over a several-year period.49,90
The finding that PD-MCI patients are at higher risk for developing dementia is consistent with a clinico-pathologic study reporting cognitive impairment in the earliest stages of clinically manifested PD, possibly related to changes in brainstem monoaminergic nuclei or early involvement of forebrain cholinergic nuclei.91
Preliminary research suggests that the majority of PD-MCI cases convert to PDD over a several-year period.16, 22, 68
The two longitudinal studies included in this review differed in terms of design and methodology. Williams-Gray et al.68
followed an incident cohort of non-demented PD patients in two waves, but did not specifically examine progression from a state of “mild impairment” to PDD. The focus of this research to date has been to determine which demographic (increasing age), neuropsychological (semantic verbal fluency and visuospatial deficits), and genetic factors (MAPT
H1/H1 tau genotype) predicted conversion to PDD at the second wave of follow-up22
. The other study used a cross-sectional sample of survivors from a prevalence sample.16
In this study, Janvin et al. found that 62% of PD-MCI patients converted to PDD over a 4-year period, compared with 20% of PD patients with normal cognition. The frequency of conversion to PDD over a 4-year period was: multiple domain MCI (63%), single, non-memory domain MCI (69%), single domain, amnestic MCI (40%), and normal cognition (20%).
Regarding other risk factors, in one of the longitudinal studies reviewed increasing severity of depression was associated with an increased risk of conversion from PD-MCI to PDD.16
In other research not covered in this manuscript, demographic and clinical correlates or risk factors for PDD development have included older age, male sex, lower educational level, longer duration of PD, and greater motor impairment.22, 64, 83, 86, 90, 92, 93
PD-MCI appears to be a clinically significant syndrome, as even mild cognitive deficits or self-rated cognitive deficits in early PD are associated with functional impairment5, 94
and worse quality of life (QoL).95, 96
Thus, identification and intervention at the earliest stage of PD-MCI is a crucial unmet need for the overall care of PD patients. The high frequency of MCI in PD highlights the need for clinicians to routinely screen for cognitive impairment in PD, as the results, including their prognostic implications, may influence clinical decision-making. However, research in this area is preliminary, and additional studies are needed to validate measures that are sensitive to initial changes in independent activities of daily living (IADLs), QoL and interpersonal relationships that can occur at the stage of PD-MCI.
Complexities in the Assessment of Cognition in PD
Assessment of cognition in PD can be complicated by disease or medication-related effects, such as bradykinesia, fatigue, sleepiness, and mood disorders, which can adversely impact test results regardless of cognitive abilities. Specifically, motor slowing (i.e., bradykinesia) and resting or intentional tremor may lead to impaired performance on any timed test, while tremor can interfere with performance on any test requiring motor abilities (e.g., use of a pencil to complete a task). However, there has been very little research examining the impact of these factors on cognitive performance specifically.
Another issue is that the definition of MCI forces a dichotomization (present-absent) of a continuous variable (cognitive test performance), and debate continues regarding the appropriate cut-off score and number of tests used to define PD-MCI. Caviness et al.32
reported that 21% of subjects had PD-MCI if an abnormality on multiple tests within a domain was required, but this rose to 42% if an abnormality on only one test was required. There is concern that the commonly-used definitions of MCI may lack sensitivity to detect early cognitive decline (rather than impairment) in high functioning persons.97
Persons functioning at a high level (i.e., above average) premorbidly have to experience sizeable declines before scoring at least 1.5 SD below normative means. Consequently, a considerable proportion of such patients with cognitive decline would be classified as having “normal cognition.” While it can be argued that high premorbid functioning protects against MCI in much the same manner as it does against dementia, this argument ignores the fact that high functioning persons may be in more demanding work or social settings in which even small cognitive declines translate into subtle functional impairments. On the other hand, a key feature in diagnosing MCI pertains to a change in cognition. Thus, MCI is not just a value on a cognitive test relative to the mean; rather, it is critical that the person has experienced a change in cognition compared with baseline.
Additional unresolved issues are whether cognitive decline in PD is linear and whether different profiles of cognitive deficits may have a different evolution and prognosis. Previous studies have noted that time to PDD diagnosis is highly variable98
and that cognitive changes after relatively long-term follow-up are not too consistent.97
Although a distinct evolution of different cognitive domains cannot be inferred from a study using only the MMSE, the re-analysis of data from a long-term prevalence study identified a variation in the slope of decline, specifically a rapid cognitive deterioration after a relatively stable period.99
These data should be examined in light of neuroimaging18, 19
and clinical52, 68
data showing that the transition from MCI to dementia in PD is characterized by the addition of posterior cortical deficits upon frontal-subcortical ones.
None of the reviewed papers examined biomarkers specifically as they pertain to PD-MCI. One of the incident cohorts included in this review underwent a second wave follow-up, 22
and in that research the tau MAPT
H1/H1 genotype (but not COMT genotype) was found to be a risk factor for PDD, a finding recently confirmed.22
Other genetic polymorphisms (e.g., COMT polymorphisms and BDNF Val66Met genotype 100, 101
) have been shown to be associated with impairments in specific cognitive domains or abilities in PD. Several CSF biomarkers for cognitive decline in PD have been proposed.102
Recent research suggests decreased CSF β-amyloid (Aβ) 1–42 is associated with the early stages of cognitive decline in PD.24, 26
This decrease might be due to specific Aβ plaque pathology, but it may be nonspecific, as Aβ 1–42 has been shown to be decreased in neurodegenerative disorders lacking distinct plaque pathology103–105
and in vivo
plaque imaging (PET imaging with the Aβ-binding Pittsburgh Compound B) shows no correlation between the plaque load and cognition in PD. 106
Instead, the findings suggest a different mechanism of Aβ processing, perhaps due to synaptic α-synuclein pathology. 107, 108
Structural neuroimaging, including diffusion tensor imaging, have reported white matter abnormalities in non-demented PD patients.108
In another study17
that classified patients as PD normal cognition (PD-NC), PD-MCI, or PDD and used two different imaging analyses, PD-MCI patients compared with PD-NC either had reduced gray matter in the prefrontal cortex and temporal lobes,17
or were found to have similar regional brain volumes. Using FDG-PET, a PD-related cognitive pattern in non-demented PD patients has been reported, characterized by metabolic reductions in frontal and parietal association areas, and relative increases in the cerebellar vermis and dentate nuclei.109
The pattern predicted memory and visuospatial performance, and in a subsequent study single domain MCI patients had a PD-related cognitive pattern expression intermediate (but not statistically significantly different) between normal and multiple domain MCI patients.18
Clearly there is a need for the prospective and longitudinal assessment of accessible biomarkers (including CSF, blood, and neuroimaging) and neuropathological examination to further address this issue.
Conclusions and Future Directions
There has been limited research on the epidemiology and prognostic utility of PD-MCI as a clinical syndrome. Nonetheless, the studies selected for review here, and other studies of PD-MCI not meeting inclusion criteria, yield relatively consistent prevalence estimates of MCI and its subtypes. They also show that single domain MCI is more common than multiple domain MCI, and that non-amnestic, single domain MCI is more common than amnestic, single domain MCI.
The Task Force has used this critical review of the literature and consensus of experts to formulate PD-MCI diagnostic criteria that will be published separately. Once diagnostic criteria for PD-MCI are proposed, prospective studies enrolling subjects with a wide range of pre-morbid ability will be needed to examine the predictive value of both MCI overall and MCI subtypes. Intervention trials can target MCI as a clinical syndrome, perhaps stratifying MCI by subtype to determine if particular interventions (pharmacological vs. behavioral) have differential acute or long-term effects. Although it is not known if or how a diagnosis of PD-MCI should impact clinical management, at a minimum these patients should be carefully monitored for ongoing cognitive decline.