Methods
Participants
Data of 25 PD-MCI were analyzed from the cohort recruited in the context of a previously published study (18). Patients were recruited at the Parkinson's Disease and Movement Disorders Unit of Neurology Clinic in Padua from 2016 to 2020. Parkinson's disease diagnosis was based on the most recent MDS clinical diagnostic criteria (19), confirmed by abnormal DaTscan SPECT imaging. Exclusion criteria included deep brain stimulation, atypical Parkinsonian disorders, severe psychiatric or neurological comorbidity, presence of pathogenic genetic mutations, and clinically relevant cerebrovascular disease on MRI. All participants underwent a complete neuropsychological evaluation, simultaneous PET/MRI with FMM, and a genetic assessment. A customized genetic panel (more than 90 genes associated to Movement Disorders) was used to analyze patients' DNA, and only individuals without genetic mutations were further included in this study. Regarding genetic variability as a possible confounding factor, we excluded from this study subjects carrying known PD genetic mutations and variants, but we did not screen for apolipoprotein E (APOE) ε4. In particular, mutations of the glucocerebrocidase (GBA) gene have been associated with more rapid cognitive decline in PD-MCI, with subsequent α-synuclein deposition enhancement as well as effects in proteins implicated in dopamine production, metabolism, and signaling (20).
The study was approved by the Ethic Committee of the University of Padua (4340/AO/17). All patients gave written informed consent according to the Declaration of Helsinki.
Clinical and Neuropsychological Examination
Demographic and clinical characteristics were collected by expert neurologists (AA, ACC). The severity of extrapyramidal symptoms was assessed with the motor Unified Parkinson's Disease Rating Scale (21) as well as with the Hoehn and Yahr score “on” medication. Levodopa Equivalent Daily Dose and Dopamine Agonist Equivalent Daily Dose were calculated for each patient (22). Patients' age at disease onset was defined as the age at which they noticed the first motor symptom suggestive of PD. All participants underwent a comprehensive neuropsychological evaluation, in line with the MDS task force level II PD-MCI diagnostic criteria (23) [for further details on cognitive tests adopted, see Fiorenzato et al. (24)]. In all patients the evaluation of functional and instrumental activities of daily living was performed independently of the impairment ascribable to motor or autonomic symptoms. Regarding the behavioral evaluation, the Beck Depression Inventory (BDI-II), Starkstein Apathy Scale, and the State-Trait Anxiety Inventory (STAI-Y1 and Y2) were used to detect the presence of eventual depression, apathy, state and trait anxiety, respectively. Patients were evaluated in “on” medication state. The cognitive tests were administered by trained neuropsychologists, in the morning, on two separate occasions within 3–5 days.
PET/MRI Acquisition and PET/FMM Images Classification
Parkinson's disease patients in accordance with the amyloid imaging procedure guidelines (25) received an intravenous injection of about 185 MBq FMM (performed manually over 10 s and flushed with 30 ml of saline over about 15 ± 5 s) directly in an integrated 3T PET/MRI system (Biograph mMR; Siemens, Erlangen, Germany). Images were acquired between 0–10 and 90–110 min after injection according to Cecchin et al. (26). Anatomical volumetric data via T1-weighted-3D magnetization-prepared rapid acquisition gradient echo sequence (TR 1.900 ms, TE 2.53 ms, slice thickness 1 mm, matrix 256 × 256, FOV 250 mm) were simultaneously acquired. Additionally, a 1 mm-isotropic T2-weighted-3D, and Two-Dimensional Susceptibility-Weighted Imaging, were acquired for clinical evaluation, excluding secondary parkinsonisms, the presence of vascular brain damage, and allowing visual rating scales assessment.
Visual assessment of FMM images is a robust and reliable method for detection of brain neuritic Aβ plaques (25). A binary visual classification was performed by an expert nuclear medicine physician (DC, with both in-person and e-training), blinded to cognitive status and diagnosis, who rated each scan as amyloid-positive (Aβ+) or negative (Aβ-).
For further details about PET reconstruction and PET/FMM images classification procedures see Biundo et al. (18).
PET Quantification
The PET frames were realigned, averaged, and co-registered with their respective MRI scans with the Freesurfer v7.01 (http://surfer.nmr.mgh.harvard.edu/). Since realignment might bias standardized uptake value ratios (SUVRs) comparison and partial volume correction (PVC) estimation (27), in order to improve the reliability of the realignment process, both T1w3d and PET were first manually z-cropped including medulla and cerebellum and rigidly realigned to the anterior commissure-posterior commissure line (AC-PC line) using the Freeview Freesurfer tool. We performed voxel-based three-compartment PVC to the MRI coregistered PET images using the PETSurfer tool in FreeSurfer (28, 29). Standardized uptake value ratios were computed to address intersubject effects using cerebellar gray matter (GM) as the reference region.
PET SUVR Subcortical ROIs
Fourteen subcortical striatal and extra striatal regions ROIs (nucleus accumbens, caudate, putamen, amygdala, globus pallidus, thalamus, and hippocampus) were extracted in the native MRI space using Freesurfer v7.01 Desikan/Killiany atlas 11 segmentations (30). Standardized uptake value ratio maps were then projected to the subcortical regions to extract the mean value for each PD patient.
PET SUVR Cortical Surface-Based Analysis
Vertex-wise general linear model (GLM) (between group comparisons) comparing cerebral cortical SUVR of the PD-MCI Aβ+/– was run using Freesurfer. Standardized uptake value ratio maps of each subject were sampled onto the left and right surfaces via the individual subject's surface, and a surface-based smoothing of 8-mm full width at half maximum was applied. Surface areas of significant amyloid load, which survived a cluster-wise Monte Carlo correction for multiple comparisons after running 10,000 permutations, were considered for the following steps.
PET SUVR Subcortical Analysis
Mean subcortical SUVR values were compared between hemispheres using the Wilcoxon test and tested for Left-Right hemisphere correlation using the non-parametric Spearman test. Subcortical regions with an inter-hemisphere difference p > 0.2 and a correlation significance of p < 0.001 were averaged, to reduce the degrees of freedom in multiple comparisons testing as well as to increase the statistical power. The obtained values were compared between PD-MCI-Aβ+ and Aβ- with a non-parametric Mann–Whitney U test for amyloid load.
Voxel Based MRI Cortical Analyses
The Brain Anatomical Analysis using Diffeomorphic deformation (BAAD 4.31—http://www.shiga-med.ac.jp/~hqbioph/BAAD/Welcome_to_BAAD.html) (31) and Statistical Parametric Mapping tool (SPM12, https://www.fil.ion.ucl.ac.uk/spm/software/spm12/) were used to calculate the cortical alteration patterns. This tool includes a Computational Anatomy Toolbox (CAT12)-based (http://www.neuro.uni-jena.de/cat/), T1-weighted-3D diffeomorphic segmentation after inhomogeneity correction, T1-weighted-3D quality check assessment, and Total Intracranial Volume (TIV) estimation. T2-weighted-3D was included in the multimodal segmentation to correct brain atrophy estimation for the presence of white matter lesions. Brain Anatomical Analysis using Diffeomorphic deformation integrates CAT12 tool for the normalization to the standard MNI space (31). Moreover, it provides at subject level a voxel-wise non-parametric statistical map of GM and white matter alterations, comparing each participant's normalized and segmented brain MRI images to the age- and sex-matched normative data with the SnPM13 tool (http://warwick.ac.uk/snpm), as previously described (31). See first level maps comparison, PD-MCI subgroups vs. healthy population, in Supplementary Figure 1; in addition, further methodological details are reported in our previous work (18).
Statistical maps of GM alterations were then included in a second level GLM analysis comparing PD-MCI-Aβ+ vs. PD-MCI-Aβ- subgroups to assess for possible GM differences. Areas of shared alterations were tested with a conjunction analysis including the two subgroups as factors. To reduce false positives (32), a Bayesian Probabilistic Threshold-Free Cluster Enhancement (TFCE) method was used to define GM patterns differences in PD-MCI subgroups. A significant threshold family-wise error (FWE) corrected of p < 0.001 was adopted. The atlas (AAL3v1, http://www.gin.cnrs.fr/en/tools/aal/) provided with the automated anatomical parcellation tool was used for results.
Statistical Analyses
Pairwise differences between the two subgroups (PD-MCI-Aβ+ vs. PD-MCI-Aβ-) in demographic, motor, clinical, cognitive, and behavioral characteristics were assessed with Mann–Whitney U test or Fisher's exact test for categorical variables. Each cognitive raw score was converted to z-score using the Italian normative data, considering as pathologic a performance below the −1.5 SD cut-off. The z-compound (mean z-score among tests of each cognitive domain) was also calculated. Moreover, we tested the hypothesis that increased amyloid load might alter a cognitive domain performance, by running a univariate linear regression analysis between the significantly affected cognitive domains and the mean SUVR, extracted from each cluster/subcortical region with a marked amyloid deposition. Statistical analyses were run with “R software” (R version 3.6.2 (2019-12-12)—Copyright (C) 2019).
Discussion
In the present study, we used simultaneous PET/MRI imaging study to investigate how Aβ burden affects cognitive performance in PD-MCI patients, diagnosed with a comprehensive neuropsychological assessment, allowing a level II diagnosis.
We found that in our PD-MCI cohort, 8/25 (31%) patients were Aβ+, corroborating previous data on brain amyloid prevalence in non-demented PD (20). While all demographic, clinical, behavioral, and motor variables did not differ, PD-MCI-Aβ+ showed a trend for higher age of motor's symptoms onset, suggesting that extracellular β-amyloid deposition is an age-related pathological marker (33).
From a neuropsychological point of view, PET amyloid positivity was associated with worse performance in the executive domain, supporting previous PD literature on the contribution of Aβ on attentive and executive dysfunctions (10, 34, 35).
Interestingly, brain Aβ did not impact other cognitive domains such as visuo-spatial and semantic memory, which are considered highly sensitive in detecting the deterioration and progression to severe impairment and dementia (1, 2, 18, 23, 36–38). These results seem to suggest that PD-MCI-Aβ+ cognitive pattern does not share the typical cognitive profile of MCI due to Alzheimer's disease (AD).
The observation of a significant proportion of PD-MCI-Aβ- can possibly suggest that Aβ is not the main pathogenetic cause of cognitive deterioration in PD (10, 11, 18, 34). Instead, amyloidosis may enhance early deficits in those cognitive domains already vulnerable to PD primary pathology (2, 18), possibly indicating that in PD concomitant conditions may aggravate cognition and herald dementia (39–41). In this regard, in a recent longitudinal study performed in a large PD de novo cohort, we concluded that amyloid burden together with asymmetric dopaminergic loss (in the left hemisphere) and aging exhibited independent and interactive contributions to PD-cognitive progression (namely, involving attentive/executive alterations) (42).
One of the strengths of our study is the quantification of cortical Aβ load using a surface-based approach and SUVR measures, not reported in our previous work (18).
We found increased Aβ uptake in the cortical areas, mostly located in the frontal (i.e., caudal and rostral middle frontal cortices and pars triangularis) as well as in the parietal regions (i.e., paracentral lobule and precuneus), which overall correlated with a poor executive performance.
From a cognitive perspective, the involvement of these areas as well as their key role within the fronto-striatal circuit is well-established. They subserve a wide range of cognitive tasks, such as executive functions and working-memory abilities, involving information monitoring and manipulation, planning and organization of complex behaviors, and attention shifting (43), with frontal striatal network alterations underlying early changes in PD, and predicting cognitive decline as well as dementia (44, 45). Noteworthy, the present findings corroborate our previous PET-amyloid evidence in a PD de novo cohort (10), showing that increased cortical and subcortical amyloid was associated with a worse performance in attentive/executive domains. Yet, this reinforces the concept of cortical and subcortical amyloid accumulation as an additional biomarker contributing to cognitive impairment in PD (10, 12, 46).
Increased FMM uptake in PD-MCI-Aβ+ was also present in the right precuneus, which is anatomically and functionally connected with subcortical striatal areas and implied in functional abnormalities in PD-MCI (47). Our findings support previous PET imaging studies in non-demented PD, despite the fact that different Aβ-tracers were used (46, 48), showing a greater amyloid burden particularly in the precuneus and frontal regions, as well as their inverse correlation with cognitive performance (46). Other evidence did not corroborate this pattern, reporting no differences between PD-MCI and PD with normal cognition, possibly due to the small sample size (34). Evidence of amyloid deposition in the parietal regions seems to be consistent across PET imaging studies (12, 48, 49). Of note, one reported an association with visuospatial impairments (49) but no significant correlations between amyloid and executive functions; however, the entire Lewy bodies spectrum was analyzed.
Overall, previous studies in PD population yielded mixed results, and as highlighted in the Petrous et al. review (13), there is only a few published evidence about the role of Aβ in PD-MCI, due to the heterogeneous nature of this clinical syndrome as well as the high variability in the MCI-assessment between studies.
Further, PD-MCI-Aβ+ presented increased FBB uptake in the putamen, but without significant correlation with executive functions. In a previous analysis of the PPMI dataset, we had reported a correlation between executive tests and putamen Aβ accumulation, but this included only early de novo PD with limited cognitive characterization, not allowing to diagnose MCI (10). Another study reported a correlation but using a different PET tracer and in a cohort with various cognitive alterations (50). It is possible that lack of correlation in our cohort was related to the limited number of PD-MCI-Aβ+, but future studies should focus on this specific marker of PD-related amyloidosis exploring also its relevance on the magnitude of levodopa response and on motor complications.
Finally, we also evaluated GM volumetric changes at MRI. Results showed similar atrophy patterns in right fronto-temporal regions in both PD-MCI subgroups, replicating previous findings adopting different structural analysis methods (5, 7). Interestingly, cortical VBM patterns were similar in PD-MCI with and without Aβ deposition, and excluded the presence of middle-temporal atrophy, a characteristic of early AD (51). This supports our previous findings of absence of AD-atrophy pattern in the LBD patients with dementia (18).
Our current work has limitations. Each subgroup (Aβ+ and Aβ-) included a relatively small number of PD-MCI subjects, resulting in lower statistical power to detect differences. However, results are aligned with previous evidence (18) of an independent role of Aβ in shaping MCI-profile and accelerating cognitive deficits in PD. Moreover, regarding the neuroimaging analyses we acknowledge that the small sample size can possibly affect the results; however, to reduce the false positive, we adopted TFCE or Monte Carlo FWE corrections, while to reduce the false negative an explorative uncorrected p < 0.001 was adopted.
Greater numerosity is needed to further explore the relationship between Aβ deposition and increased dementia-risk profiles. In particular, it would be worth exploring if low concentration of Aβ, below AD-range thresholds, may also have clinical relevance (13) in presence of an ongoing multisystem neurodegenerative process like PD. In this regard, a further limitation is that we did not screen our PD-MCI for APOE genotype which is now considered the main risk factor for cognitive decline in elderly patients with PD and Dementia with Lewy Bodies (34, 52, 53).
Finally, we did not include a control group of healthy controls and PD without cognitive deficits to assess age-related brain amyloid distribution. However, a recent publication of elderly individuals without cognitive deficits demonstrated areas of cerebral amyloid deposition (i.e., prefrontal, precuneus) similar to those observed in our Aβ+PD-MCI patients, suggesting that PD-related amyloidosis somehow overlaps age-related amyloid accumulation (54).
In conclusion, our study suggests that amyloid burden in the fronto-striatal network may play a role in worsening executive abilities in PD-MCI patients. Furthermore, the observation of more than 60% Aβ- in our PD-MCI patients led us to speculate about the independent role of amyloid load for PD-MCI. Notably, amyloid accumulation in our cohort was not associated with typical AD cognitive and brain atrophy pattern nor with specific clinical features and demographic characteristics, highlighting its potential unspecific contribution. Considering that up to 80% of PD patients may ultimately develop dementia with negative consequences on life quality and expectancy, a better understanding of the involved predictors is of key importance. We believe clarifying Aβ role in cognitive impairment and progression is clinically relevant, especially in the context of emerging applicability of amyloid-related treatments.