Introduction

Cognitive impairment and dementia in synucleinopathies

Parkinson disease (PD) is the most common human neurodegenerative movement disorder, and is overall the second most common neurodegenerative disorder after Alzheimer disease (AD). PD is characterized by the intraneuronal deposition of α-synuclein (α-Syn) aggregates (Lewy bodies) and degeneration of monoaminergic brainstem nuclei (particularly the substantia nigra pars compacta). Parkinson disease affects approximately 1 % of the population over the age of 60 [1]. Cognitive impairment is common in PD, and the risk of developing dementia in patients with PD is 2–6-fold higher than in the general population, corresponding to lifetime risk estimates of 30–80 % [2–4]. Clinical diagnosis of dementia in patients with PD is dichotomized into two conditions. Parkinson disease with dementia (PDD) is related closely to an increasingly recognized syndrome, dementia with Lewy bodies (DLB). DLB is in fact the second most common neurodegenerative dementia, accounting for up to 20 % of all neurodegenerative dementia cases [5–7]. The key similarity between PDD and DLB is the requirement for PD neuropathologic changes, together with dementia. Several expert consensus meetings on DLB, its diagnosis, and its management have been convened by investigators at Newcastle upon Tyne [8–10]. Recommendation from the first DLB workshop [8] was that diagnostic classification of DLB be applied to patients in whom dementia preceded the clinical onset of PD, or when dementia occurred within 1 year after PD diagnosis. Classification as PDD was recommended when dementia developed more than 1 year after PD diagnosis.

Overlapping features of PDD and DLB

Although it was hoped that the PDD vs. DLB clinical classification rule would improve diagnostic homogeneity and identify distinctions between these two syndromes, subsequent studies have shown substantial overlap of PDD and DLB clinical phenotypes in numerous ancillary, biomarker and neuropathology studies. Thus, neuropsychological, neuropathological and neuroimaging studies do not show reliable differences between subjects classified clinically as PDD vs. those with DLB. Neuropsychometric evaluations may identify prominent visual dysfunction in pure DLB, however, there is no reliable neuropsychometric distinction between DLB and PD [11–13]. Similar visual psychometric abnormalities are found in PDD patients [14]. Features that may distinguish DLB form AD, including parkinsonism, visual illusions/hallucinations, fluctuating cognitive deficits, sensitivity to dopaminergic medications and striatal dopaminergic deficits on molecular neuroimaging studies [10] are manifest also in PDD. Most importantly, neuropathological studies of both PDD and DLB reveal similar admixtures of protein deposition abnormalities. By definition, both DLB and PDD are characterized by intraneuronal α-Syn aggregates and nigrostriatal degeneration. The degree and nature of the basal ganglia pathology, however, does not distinguish the disorders [15, 16]. Neuropathological studies focusing on cortical pathology identify advanced stages of α-Syn pathology (corresponding to Braak PD stages 5 and 6 [17] or equivalent [18] as well as AD neuropathologies including extracellular Aβ-amyloid plaques and intracellular neurofibrillary tangles (NFTs) in significant subsets of both PDD and DLB cases [19–28]. “Pure” α-Syn changes (without Aβ deposits or NFTs) are observed in approximately 25–35 % of cases, with additional Aβ-plaques in the remainder. Amongst those subjects with Aβ-plaques, up to half demonstrate additional NFT pathology sufficient for diagnosis of AD. In several series, the presence and severity of dementia are not predicted on the basis of the α-Syn stage or Braak AD stage [29] alone, but are best characterized by combinations of α-Syn and AD pathologies, particularly NFT extent [27, 28]. Neuroimaging studies reveal severe nigrostriatal dopamine projection losses in both PDD and DLB [30, 31], and the presence of Aβ-amyloid marker binding is reported variably in PET studies of both disorders (see below).

Amyloidopathy in Alzheimer disease and normal aging

Recently developed molecular imaging approaches now permit the detection of pathological accumulations of Aβ–amyloid plaques [32], and are a subject of several manuscripts in this issue of Clinical and Translational Imaging. PET studies of fibrillary Aβ–amyloid deposition with the thioflavin derivative ligand [11C]Pittsburgh compound-B (PiB) and related clinically-approved radiofluorinated tracers including [18F]Florbetapir, [18F]Florbetaben and [18F]Flutemetamol, reveal important, new biomarker evidence for the role(s) of amyloid deposition in AD and aging (see [33] for review). To summarize the current understanding, Aβ–amyloid deposition is imaged with PET in virtually all patients with autopsy-confirmed AD. The few subjects diagnosed clinically with probable AD with “negative” amyloid scans are believed to be due to clinical misdiagnoses of frontotemporal dementia or of pure DLB as probable AD [30]. Comparative amyloid imaging and pathological confirmatory studies support the image-based classification of individual subject Aβ–amyloid status [34, 35]. The amyloid cascade hypothesis, that extracellular amyloidosis precedes the development of tau-based intracellular NFT pathology in AD, is supported by PiB-PET findings from several investigations. First, the intensity of amyloid deposition does not correlate with the severity of dementia amongst AD patients with “positive” amyloid-PET scans [36]. Second, there is high frequency of AD-range “PiB-positive” findings in pre-AD patients diagnosed with mild cognitive impairment (MCI). Amyloid-positive MCI subjects have very high rates of progression to AD over 3 years [37]. Finally, there is a substantial age-related prevalence of AD-range amyloid tracer binding amongst asymptomatic normal elderly subjects [38–40], and increased risk for development of MCI and AD in these amyloid-positive individuals [41]. These findings in normal elderly and sporadic AD subjects are recapitulated in a recent study of dominant, monogenetic AD. This report indicates the onset of amyloid deposition more than a decade before symptomatic memory deficits, supporting a long asymptomatic lag period between AD-range amyloid deposition and the onset of significant neurodegeneration [42]. Once cognitive impairment and decline develop, there was little further amyloid binding in PET imaging with [18F]Florbetapir. On this construct background, amyloid-PET findings in subjects with synucleinopathy were researched and are summarized below.

Literature review

In this manuscript, we report the results of literature searches for published amyloid-PET studies in patients with parkinsinism, including PD, PD with mild cognitive impairment (PD-MCI), PDD and DLB. We conducted an OVID-Medline search of the published literature, seeking articles identified by the intersection of {Parkinson disease.mp OR exp Parkinson Disease OR exp Lewy Body Disease} AND {florbetapir.mp OR florbetaben.mp OR flutemetamol.mp OR AV-1.mp OR AV-45.mp OR F-PIB.mp OR PIB.mp}. This search strategy identified 48 references. The authors reviewed each of these by title and abstract content, seeking original reports of PET amyloid tracer binding in patients with synucleinopathies. Papers reporting fewer than three subjects per parkinsonian diagnosis group were excluded, as were those failing to provide detailed description of the tracer binding levels that could be related to established normal and probable AD subject groups. Review articles identified in this search were not included in the analysis, but the citations therein were reviewed for possible additional original reports of amyloid binding in parkinsonian subjects. Reports that were identified as overlapping in subject content were excluded. After inclusion of these additional citations and application of exclusion criteria, 19 reports met our screening [30, 43–60].

Amyloidopathy in PD, PD-MCI, PDD and DLB

A number of prior studies have examined Aβ–amyloid deposition imaging, each involving limited numbers of PD, PDD and DLB subjects [30, 43–60]. Studies employing [11C]PiB predominate in the reported literature. We extracted from the reports the incidence of pathological (AD-range intensity) Aβ–amyloid deposition in these groups (Table 1). Most papers described the numbers of subjects with this intensity range of cortical amyloid tracer binding. However, at least one early publication reported subjects with vs. without any cortical binding above background (see Fig. 2 in [45]). In this case, we revised the authors’ estimates of amyloid incidence by imposing a cutoff threshold of 1.4:1 (cerebral cortex-to-cerebellar cortex ratio), as identified in other studies using this same tracer methodology.

Table 1 Amyloid PET Imaging in subjects with Parkinsonism (PD, PD-MCI, PDD and DLB)
Full size table

Although there is a wide range of positive Aβ–amyloid incidences and relatively few subjects in most individual studies, there is general agreement that the incidence of AD-range amyloid positivity in PD with normal cognition is very low (individual study range 0–13 %; average 6 %). Within 7 individual reports including cognitively normal, non-parkinsonian subjects, the average incidence of Aβ–amyloid positivity in PD is lower than that in the normals (normal 15 %; PD 6 %). In subjects with PD-MCI, the incidence of Aβ–amyloid positivity ranges between 0 and 11 % (average 8 %), and in 3 of these 4 reports that include also cognitively normal PD groups, the incidence of amyloid positivity is no higher than in PD. Subjects with PDD have higher incidence of Aβ–amyloid positivity than do normals, PD and PD-MCI, ranging from 0 to 80 % in individual reports (average 27 %), while the incidence in DLB ranges between 33 and 100 % in individual reports (average 59 %).

Owing to the small sample sizes in some individual reports, and also potentially to differing image analysis approaches and definitions of AD-range Aβ–amyloid deposition, the individual study ranges within diagnostic groups across studies are relatively broad. However, there is general agreement both within and across virtually all reports on some aspects. First, PD and PD-MCI have similar incidences of Aβ–amyloid positivity, and these are lower than age-comparable normal subjects. Large studies of elderly subjects indicate an average of 30 % of normal subjects between ages 60 and 90 are amyloid-positive [61–63]. Parkinson disease with dementia subjects have higher than normal incidence of amyloid positivity, and DLB subjects have even higher incidence. Dementia with Lewy body subjects, however, demonstrate lower incidence of Aβ–amyloid positivity than do other large studies of probable Alzheimer disease [36, 40, 41, 43, 53, 63].

An apparent association in PD and PD-MCI subjects, not predicted by parallel studies in non-parkinsonian subjects, is the relationship of below-AD-range cortical amyloid tracer binding with measures of cognitive function. In non-parkinsonian subjects, there is no significant relationship found between intensity of amyloid tracer binding and cognitive measures, including amongst MCI and AD subjects; amyloid burden estimated in PET imaging correlates neither with severity of cognitive impairment, nor with its progression. However, a recent study in PD-MCI suggests that cortical PiB binding may correlate with global cognitive function [57].

Discussion

There are several important limitations of our Aβ–amyloid imaging literature summary in PD and related dementias. The most obvious problem is the lack of a standardized approach to data collection, processing and analysis across the many laboratories reporting original data. Most investigators to date have employed [11C]PiB as the amyloid-binding radiotracer. However, there is a dichotomy of scanning procedures, with some investigators employing tracer kinetic approaches to the estimation of tracer distribution volume (DV) measures; some with use of arterial plasma reference and others with the use of a reference brain region such as the cerebellar cortex. Other laboratories use the simple late tracer distribution (SUV) as the primary imaging outcome, again scaling the cerebral values to a reference region, most often the cerebellar cortex. While this diversity of approaches often leads to very similar binary classification of individual subject amyloid burden (positive vs. negative), there are systematic biases and distinctions that preclude our ability to impose a quantitative threshold across all studies. For this reason, we have relied heavily on each reporting laboratory to characterize the parkinsonian subjects in relation to normal and probable AD studies performed at their site and in similar fashion. Nevertheless, there are very likely classification inconsistencies between laboratories when individual subjects have scan results near to the pathologic AD threshold. An additional source of variability across studies concerns the method of cerebral cortical VOI definition. Some approaches rely on anatomic imaging (usually MRI) for stereotaxic direction of cortical and cerebellar volumes, while others may rely on PET image data alone. Some investigators report an average cerebrocortical binding measure, while others focus on cortical regions with highest and most prevalent amyloid binding in probable AD such as the precuneus and posterior cingulate cortex. Finally, there is likely a significant recruitment bias, both within and between laboratories, related to identification of PDD vs. DLB subjects. The former are most often seen in movement disorders clinics, while the latter are likely to arise in cognitive disorders clinics. Differing thresholds for determination of the movement and cognitive clinical abnormalities in the patients identified and referred for imaging may certainly contribute to the apparent imaging differences observed.

Our literature review and synthesis suggests unpredicted aspects of Aβ–amyloidopathy in PD and associated cognitive impairment syndromes. First, it appears that intense Aβ–amyloid deposition in PD patients with MCI may not anticipate by many years the development of PDD, while it does predict conversion to AD in non-PD-MCI patients. Second, there is suggestion that there may be concomitant behavioral effects of Aβ–amyloid deposition in PD as compared to the asymptomatic and behaviorally silent accumulations in non-PD elderly subjects. Both of these observations raise the possibility that the neurobiological salience of cortical Aβ–amyloid deposition differs in conjunction with neuronal α-Syn accumulation, although cellular mechanism(s) for this interaction are not yet established. This suggests an area in need of further, direct investigation, as mechanism(s) involved in this hypothetical interaction could offer novel therapeutic targets. A final unusual aspect of the Aβ–amyloid studies in parkinsonism suggests an overall lower prevalence in cognitively normal PD and in PD-MCI than in elderly subjects without neurologic abnormalities. This might suggest that synucleinopathy is in some manner protective against the mechanism(s) underlying cerebral Aβ–amyloid deposition. Although the available data do not exclude this possibility, we think it unlikely. Rather, we favor the construct that neurologically normal subjects begin asymptomatic accumulation of cerebral Aβ–amyloid deposits with increasing incidence at advancing age. If a subject with existing Aβ–amyloid deposition develops α-Syn pathology, they rapidly manifest both cognitive impairment and motor parkinsonism, and are classified clinically as DLB. Whereas, subjects who develop α-Syn pathology without antecedant Aβ–amyloid deposition are more likely to develop cognitive decline much later in the course of PD, and are more likely to have Aβ–amyloid depositions less intense than is typical of probable AD. Future studies focusing on the presymptomatic molecular imaging of DLB would be particularly informative in testing this construct.

An important descriptive result of most individual studies and of our summary is that the known pathological heterogeneity within patients classified clinically as DLB vs. PDD is reflected in the apparent heterogeneity of Aβ–amyloid positivity in these groups. It was originally hoped that clinical distinction according to the timing of movement and cognitive deficits would lead to homogeneity in underlying pathophysiologies in parkinsonian dementias. However, it has emerged that there is little evidence for clinical, behavioral or other routine diagnostic assessment that supports distinction of DLB and PDD. Results of molecular endophenotype imaging, as summarized herein, suggest an alternative classification approach: Patients could be classified according to the presence or absence of protein misfolding and aggregate deposition characteristics. At the present time, there are sensitive and specific imaging biomarkers for fibrillary Aβ–amyloid deposition, but not yet specific probes for the other potential targets in DLB and PDD. Misfolded protein aggregates containing α-Syn (in Lewy bodies and Lewy neurites), could theoretically be depicted in biomarker imaging of novel ligands, yet to be introduced and validated [64]. In addition, probes targeting the accumulations of misfolded tau proteins (including NFT tau accumulations) have been recently pursued, and several promising ligands are under active investigation [65]. Biomarker imaging with these multiple approaches could result in future imaging distinctions more closely related to underlying pathophysiologic disease mechanisms. Overall, the search for effective disease-modifying therapy for dementia in PD is likely to require this classification for appropriate selection of patients for clinical trials.

Conclusions

The pathophysiologic processes underlying dementia in patients with concomitant parkinsonian neurodegenerations are heterogeneous, involving cortical α-Syn, Aβ–amyloid and tau NFTs in varied combination. The currently established clinical classification of DLB vs. PDD is insufficient to identify and separate processes that may contribute to dementia. It is conceivable that PDD and DLB represent phenotypic variations of the same underlying process. Another possibility is that the neuropathologic resemblance of PDD and DLB represents convergence of different initial pathophysiologic pathways at advanced stages in the disease processes. A final and more likely consideration, however, may be that the heterogeneity (even at end-stage) among patients with either PDD or DLB reflects distinct neuropathologic processes that are present in some individuals across both clinical diagnostic groups—that is; the “clinical one-year rule” may be suboptimal or even inappropriate for distinguishing subtypes of dementia pathologies in synucleinopathies. Many interesting and potentially important aspects of amyloidopathy suggested in the parkinsonian dementias will require a focused, more homogeneous approach for confirmation and extension that addresses technical inconsistencies in future studies. It is hoped that molecular imaging-based classifications will fill a critical gap in our present understanding, and provide a basis for future development of novel-targeted interventions and their assessments.