Brain age and Alzheimer's-like atrophy are domain-specific predictors of cognitive impairment in Parkinson's disease
Graphical abstract
Introduction
Cognitive impairment is a frequent and debilitating non-motor complication in Parkinson's disease (PD) (Aarsland et al., 2017). Estimates of the prevalence of mild cognitive impairment (MCI) in PD range from 27% to 40 % and up to 80% of patients will eventually develop Parkinson's disease dementia (PD-D) during the course of the disease (Hely et al., 2008; Monastero et al., 2018; Svenningsson et al., 2012). Extensive research has identified multiple demographics, clinical, and genetic conditions associated with cognitive decline in PD, as for example age, male sex, disease duration, presence of depression or visual hallucinations, postural instability-gait difficulty (PIGD) type PD and a positive Apolipoprotein E ε4 carrier status (Schrag et al., 2017; Weintraub and Burn, 2011). From a histopathological perspective, depositions of cortical Lewy bodies and coexisting Alzheimer's disease (AD) pathology are considered to be the most important drivers of dementia (Berg et al., 2014; Compta et al., 2011; Smith et al., 2019a). However, the exact mechanisms and their differential contribution to the development of MCI or dementia in PD are still a matter of debate.
From neuroimaging studies, it is well established that cognitive decline in PD is linked to atrophy in frontal, temporo-parietal, and occipital brain areas, including the hippocampus and the basal ganglia (Hall and Lewis, 2019). While atrophy in patients with manifest PD-D is generally marked (Zarei et al., 2013), the morphometric changes associated with the transition from normal cognition to MCI are subtle and difficult to detect with standard volumetric or voxel-based techniques. Investigations are further complicated by the clinical variability of MCI subtypes (Kalbe et al., 2016). In fact, voxel based imaging studies have produced rather heterogeneous results, with frontal, hippocampal and/or temporo-parietal atrophy being probably the most consistent finding in early stage cognitive decline in PD (Krajcovicova et al., 2019; Mak et al., 2015b; Pereira et al., 2014; Yau et al., 2018).
Pattern classification methods offer the advantage over voxel-based methods in that they consider image information from the entire brain simultaneously. Davitzikos et al. used structural imaging data from AD patients and healthy controls to train a classifier that was shown to discriminate between the two groups with high accuracy (Davatzikos et al., 2008; Davatzikos et al., 2009). As output, the classifier provides a scalar quantity, the SPARE-AD score (Spatial Pattern of Abnormality for Recognition of Early Alzheimer's Disease), expressing the similarity of a sample's structural MRI scan to those of a cohort of AD patients. Higher SPARE-AD scores indicate a more AD-like pattern of brain atrophy, which is primarily characterized by marked temporal atrophy (Davatzikos et al., 2009; Dickerson et al., 2009; Habes et al., 2016b). However, it is important to note that from this similarity alone it is not possible to infer on an underlying AD pathology without further biomarker assessments (Da et al., 2013; Fjell et al., 2010). So far, only two recent studies have applied the classifier on non-demented patients with PD, showing a significant correlation between the SPARE-AD score and future cognitive decline (Tropea et al., 2018; Weintraub et al., 2012a). The results suggest that the SPARE-AD score may be a promising imaging marker for investigating cognitive dysfunction in PD and for assessing a patient's risk of status conversion to PD-MCI or PD-D. Although of pathophysiological and clinical interest, the cognitive profile associated with elevated SPARE-AD scores has not been previously studied in PD patients.
It has become popular over recent years to study changes in brain aging associated with disease (Cole and Franke, 2017). Brain aging in general can be considered a multidimensional process leading to complex changes at molecular, genetic, metabolic, cellular, morphological and cognitive levels that are to date insufficently understood (Cohen et al., 2019; Wyss-Coray, 2016). Using methods from machine learning it is now possible to calculate brain age estimates from structural MRI, the difference between this estimate and the chronological age indicating advanced or resilient brain aging (also termed as “brain age gap” or “relative brain age”, RBA) (Franke et al., 2010; Smith et al., 2019b). The concept is considered promising as it allows to assess whether a specific clinical, therapeutic or environmental condition exerts deleterious or protective effects on physiological brain aging in vivo (Kaufmann et al., 2019). For example, increased RBA was shown to predict future cognitive decline in patients with MCI and AD (Franke and Gaser, 2012) and also outperformed other cognitive and cerebrospinal fluid (CSF) biomarkers such as amyloid beta 1-42, tau and phosphorylated tau in their ability to predict conversion from MCI to AD within an interval of three years (Gaser et al., 2013). To the best of our knowledge, only two recent studies have used this method to investigate brain age in PD so far (Beheshti et al., 2020; Eickhoff et al., 2020). The former study found increased RBA values in PD patients versus healthy controls which amounted to 1.5 ± 6.0 years with respect to gray matter and to 2.5 ± 5.9 years with respect to white matter compartments (Beheshti et al., 2020). The latter study reported increased RBA values of 2.7 and 3.5 years for two different PD cohorts (Eickhoff et al., 2020). However, the relation between MRI based estimates of RBA and cognitive decline in PD has not yet been systematically explored.
Analyzing longitudinal data from the LANDSCAPE cohort (Balzer-Geldsetzer et al., 2011), the paper addresses the following questions:
- (a)
Are SPARE-AD scores and increased RBA in PD associated with dysfunction in different cognitive domains? In view of the temporal and parieto-occipital focus of brain atrophy associated with AD (Dickerson et al., 2009), we hypothesize that SPARE-AD might be a predictor of cognitive dysfunction in the more ‘posterior’ memory, language and visuospatial domains (Buckner, 2004; Kehagia et al., 2013). On the other hand, because aging is accompanied by frontally accentuated brain atrophy (Salat et al., 2004), we hypothesize that an increased RBA in PD might particularly relate to dysfunction in executive and attentional/working memory domains (Buckner, 2004; Kehagia et al., 2013).
- (b)
Do increased brain age gaps in PD relate to the patients’ current cognitive status or future status conversion? If so, how does this compare to the predictive power of SPARE-AD? Given the property of RBA to predict future cognitive decline and conversion to dementia in patients with AD (Franke and Gaser, 2012; Gaser et al., 2013), we hypothesize that this may also be the case for patients with PD. Along with increasing severity of cognitive disease stages, we further expect larger changes in RBA and SPARE-AD compared to healthy subjects.
Section snippets
Study collective
Longitudinal data from 178 PD patients (83 patients with normal cognition (PD-NC), 78 patients with MCI (PD-MCI), and 17 patients with PD-D) and 84 healthy control subjects (HC) enrolled in the imaging subproject of the LANDSCAPE study were analyzed. The LANDSCAPE study is a prospective, multicenter, observational cohort study for the investigation of cognition in PD and was set up as a continuation of the DEMPARK study (Balzer-Geldsetzer et al., 2011). Six university centers (Dresden,
Baseline data
Demographic and clinical characteristics of patient subgroups and HC are presented in Table 1. Detailed results from neuropsychological test batteries are provided in Supplementary Table 1. Disease duration, UPDRS III motor scores and LEDD were similar between PD-NC and PD-MCI groups (all p-values > 0.11), whereas disease was more advanced in patients with PD-D compared to the two other PD subgroups (disease duration: p< 0.006; UPDRS III: p < 0.001, H&Y stage: p < 0.001, LEDD: p < 0.052). MMSE
Discussion
The study has four key findings:
- (a)
Patients with PD-MCI and PD-D had increased SPARE-AD scores compared to patients with PD-NC and HC.
- (b)
RBA was increased in all PD groups independent of their cognitive status, the brain age gap amounting to 2.2 years in non-demented PD patients and to 3.5 years in patients with PD-D.
- (c)
Elevated SPARE-AD scores were associated with impairment in executive, memory, and language functions, whereas increased RBA correlated with dysfunction in the attention and working
Acknowledgments
The LANDSCAPE study (Representatives: Dr. S. Baudrexel, Prof. Dr. R. Dodel, Prof. Dr. D. Berg, Prof. Dr. R. Hilker-Roggendorf, Prof. Dr. E. Kalbe, Prof. Dr. J. Kassubek, Prof. Dr. Klockgether, Dr. Liepelt-Scarfone, Prof. Dr. B. Mollenhauer, Prof. Dr. J. Schulz, Dr. A. Spottke, Prof. Dr. A. Storch, Prof. Dr. H.-U. Wittchen) is part of the Competence Network Degenerative Dementias (KNDD) and was funded by the German Federal Ministry of Education and Research (project number 01GI1008C). The
Disclosure of statement
All authors have no conflicts of interest to report.
Author contributions
Daniel Charissé: Formal analysis, Investigation, Writing - original draft; Guray Erus: Methodology, Writing - review & editing; Raymond Pomponio: Methodology, Writing - review & editing; Martin Gorges: Investigation, Writing - review & editing; Nele Schmidt: Investigation, Writing - review & editing; Christine Schneider: Investigation, Writing - review & editing; Inga Liepelt-Scarfone: Conceptualization, Funding acquisition, Project administration, Writing - review & editing; Oliver Riedel:
References (93)
- et al.
Bias-adjustment in neuroimaging-based brain age frameworks: A robust scheme
NeuroImage Clin
(2019) - et al.
Working memory and inhibition across the adult life-span
Acta psychologica
(2008) - et al.
Neuropsychology of aging
Handbook Clin Neurol
(2019) - et al.
Predicting age using neuroimaging: innovative brain ageing biomarkers
Trends Neurosci
(2017) - et al.
Grey matter volume correlates of cerebrospinal markers of Alzheimer-pathology in Parkinson's disease and related dementia
Parkinsonism Relat Disord
(2012) - et al.
Combined dementia-risk biomarkers in Parkinson's disease: a prospective longitudinal study
Parkinsonism Relat Disord
(2013) - et al.
Detection of prodromal Alzheimer's disease via pattern classification of magnetic resonance imaging
Neurobiol Aging
(2008) - et al.
MUlti-atlas region Segmentation utilizing Ensembles of registration algorithms and parameters, and locally optimal atlas selection
NeuroImage
(2016) - et al.
Estimating the age of healthy subjects from T1-weighted MRI scans using kernel methods: exploring the influence of various parameters
NeuroImage
(2010) - et al.
Brain molecular aging, promotion of neurological disease and modulation by Sirtuin5 longevity gene polymorphism
Neurobiol Dis
(2011)
Neural correlates of cognitive impairment in Parkinson's disease: a review of structural MRI findings
Int Rev Neurobiol
Neuropathological and genetic correlates of survival and dementia onset in synucleinopathies: a retrospective analysis
Lancet Neurol
Cerebral structure on MRI, Part I: Localization of age-related changes
Biol Psychiatry
Compensatory dopaminergic-cholinergic interactions in conflict processing: Evidence from patients with Parkinson's disease
NeuroImage
Harmonization of large MRI datasets for the analysis of brain imaging patterns throughout the lifespan
NeuroImage
Eta squared and partial eta squared as measures of effect size in educational research
Educ Res Rev
Aging and measures of processing speed
Biol Psychol
Clinical variables and biomarkers in prediction of cognitive impairment in patients with newly diagnosed Parkinson's disease: a cohort study
Lancet Neurol
Estimation of brain age delta from brain imaging
NeuroImage
Cognitive impairment in patients with Parkinson's disease: diagnosis, biomarkers, and treatment
Lancet Neurol
Topography of cortical thinning in the Lewy body diseases
NeuroImage Clin
Risk of dementia in Parkinson's disease: a community-based, prospective study
Neurology
Cognitive decline in Parkinson disease
Nat Rev Neurol
The correlative triad among aging, dopamine, and cognition: current status and future prospects
Neurosci Biobehav Rev
Parkinson's disease and dementia: a longitudinal study (DEMPARK)
Neuroepidemiology
T1-weighted MRI-driven brain age estimation in Alzheimer's disease and Parkinson's disease
Aging Dis
MDS research criteria for prodromal Parkinson's disease
Mov Disord
Time to redefine PD? Introductory statement of the MDS Task Force on the definition of Parkinson's disease
Mov Disord
Frequency of cholinergic and caudate nucleus dopaminergic deficits across the predemented cognitive spectrum of Parkinson disease and evidence of interaction effects
JAMA Neurol
Cortical cholinergic function is more severely affected in parkinsonian dementia than in Alzheimer disease: an in vivo positron emission tomographic study
Arch Neurol
Cognitive correlates of cortical cholinergic denervation in Parkinson's disease and parkinsonian dementia
J Neurol
Memory and executive function in aging and AD: multiple factors that cause decline and reserve factors that compensate
Neuron
Synergistic Interactions between Abeta, tau, and alpha-synuclein: acceleration of neuropathology and cognitive decline
J Neurosci
Statistical power analysis for the behavioral sciences
Aging and Parkinson’s disease: different sides of the same coin?
Mov Disord
Lewy- and Alzheimer-type pathologies in Parkinson's disease dementia: which is more important?
Brain J Neurol
Integration and relative value of biomarkers for prediction of MCI to AD progression: spatial patterns of brain atrophy, cognitive scores, APOE genotype and CSF biomarkers
NeuroImage Clin
Longitudinal progression of Alzheimer's-like patterns of atrophy in normal older adults: the SPARE-AD index
Brain J Neurol
The cortical signature of Alzheimer's disease: regionally specific cortical thinning relates to symptom severity in very mild to mild AD dementia and is detectable in asymptomatic amyloid-positive individuals
Cereb Cortex (New York, N.Y. 1991)
Analysis of longitudinal data
Executive dysfunction in Parkinson's disease: a review
J Neuropsychology
Multi-atlas skull-stripping
Acad Radiol
Incidence and remaining lifetime risk of Parkinson disease in advanced age
Neurology
An R2 statistic for fixed effects in the linear mixed model
Stat Med
Clinical diagnostic criteria for dementia associated with Parkinson's disease
Mov Disord
Cited by (12)
Advanced brain aging in Parkinson’s disease with cognitive impairment
2024, npj Parkinson's DiseasePathobiology of Cognitive Impairment in Parkinson Disease: Challenges and Outlooks
2024, International Journal of Molecular SciencesStumbling Blocks in the Investigation of the Relationship Between Age-Related Hearing Loss and Cognitive Impairment
2024, Perspectives on Psychological ScienceCognitive Profile in Parkinson's Disease Dementia Patients with Low versus Normal Cerebrospinal Fluid Amyloid Beta
2023, Dementia and Geriatric Cognitive Disorders Extra