Parkinsonism and Related Disorders 88 (2021) 68–75 Contents lists available at ScienceDirect Parkinsonism and Related Disorders journal homepage: www.elsevier.com/locate/parkreldis In vivo cholinergic basal forebrain degeneration and cognition in Parkinson’s disease: Imaging results from the COPPADIS study Michel J. Grothe a, b, *, 1, Miguel A. Labrador-Espinosa a, b, 1, Silvia Jesús a, b, Daniel Macías-García a, b, Astrid Adarmes-Gómez a, b, Fátima Carrillo a, Elena Iglesias Camacho a, Pablo Franco-Rosado a, Florinda Roldán Lora c, Juan Francisco Martín-Rodríguez a, b, Miquel Aguilar Barberá d, Pau Pastor d, Sonia Escalante Arroyo e, Berta Solano Vila f, Anna Cots Foraster f, Javier Ruiz Martínez g, Francisco Carrillo Padilla h, Mercedes Pueyo Morlans h, Isabel González Aramburu i, Jon Infante Ceberio i, Jorge Hernández Vara j, Oriol de Fábregues-Boixar j, Teresa de Deus Fonticoba k, Berta Pascual-Sedano b, l, m, COPPADIS Study Group, Jaime Kulisevsky b, l, Pablo Martínez-Martín b, Diego Santos-García n, Pablo Mir a, b, ** a Unidad de Trastornos del Movimiento, Servicio de Neurología y Neurofisiología Clínica, Instituto de Biomedicina de Sevilla (IBiS), Hospital Universitario Virgen del Rocío/CSIC/Universidad de Sevilla, Seville, Spain b Centro de Investigación Biomédica en Red Sobre Enfermedades Neurodegenerativas (CIBERNED), Instituto de Salud Carlos III, Madrid, Spain c Unidad de Radiodiagnóstico, Hospital Universitario Virgen del Rocío, Seville, Spain d Hospital Universitari Mutua de Terrassa, Terrassa, Barcelona, Spain e Hospital de Tortosa Verge de la Cinta (HTVC), Tortosa, Tarragona, Spain f Institut Catalá de la Salud (Girona) - Institut d’Assisténcia Sanitaria (IAS), Spain g Instituto de Investigación Biodonostia, Hospital Universitario Donostia, San Sebastián, Spain h Hospital Universitario de Canarias, Santa Cruz de Tenerife, Spain i Unidad de Trastornos del Movimiento, Servicio de Neurología, Hospital Universitario Marqués de Valdecilla, Santander, Spain j Neurology Department and Neurodegenerative Diseases Research Group. Vall D’Hebron Universitary Campus, Barcelona, Spain k Complejo Hospitalario Universitario de Ferrol (CHUF), Hospital Arquitecto Marcide, Ferrol, Spain l Unidad de Trastornos del Movimiento, Servicio de Neurología, Hospital de Sant Pau, Barcelona, Spain m Faculty of Health Sciences, Universitat Oberta de Catalunya (UOC), Barcelona, Spain n Complejo Hospitalario Universitario de A Coruña (CHUAC), A Coruña, Spain A R T I C L E I N F O A B S T R A C T Keywords: Introduction: We aimed to assess associations between multimodal neuroimaging measures of cholinergic basal Nucleus basalis Meynert forebrain (CBF) integrity and cognition in Parkinson’s disease (PD) without dementia. Substantia innominata Methods: The study included a total of 180 non-demented PD patients and 45 healthy controls, who underwent MRI structural MRI acquisitions and standardized neurocognitive assessment through the PD-Cognitive Rating Scale DTI Diffusion (PD-CRS) within the multicentric COPPADIS-2015 study. A subset of 73 patients also had Diffusion Tensor Parkinson’s disease Imaging (DTI) acquisitions. Volumetric and microstructural (mean diffusivity, MD) indices of CBF degeneration were automatically extracted using a stereotactic CBF atlas. For comparison, we also assessed multimodal indices of hippocampal degeneration. Associations between imaging measures and cognitive performance were assessed using linear models. Results: Compared to controls, CBF volume was not significantly reduced in PD patients as a group. However, across PD patients lower CBF volume was significantly associated with lower global cognition (PD-CRStotal: * Corresponding author. Unidad de Trastornos del Movimiento, Servicio de Neurología y Neurofisiología Clínica, Instituto de Biomedicina de Sevilla (IBiS), Campus Hospital Universitario Virgen del Rocío, Avda. Manuel Siurot, s/n, 41013, Sevilla, Spain. ** Corresponding author. Unidad de Trastornos del Movimiento, Servicio de Neurología y Neurofisiología Clínica, Instituto de Biomedicina de Sevilla (IBiS), Hospital Universitario Virgen del Rocío/CSIC/Universidad de Sevilla, Seville, Spain. E-mail addresses: mgrothe@us.es (M.J. Grothe), pmir@us.es (P. Mir). 1 These authors contributed equally. https://doi.org/10.1016/j.parkreldis.2021.05.027 Received 28 November 2020; Received in revised form 25 May 2021; Accepted 25 May 2021 Available online 31 May 2021 1353-8020/© 2021 The Author(s). Published by Elsevier Ltd. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/). M.J. Grothe et al. Parkinsonism and Related Disorders 88 (2021) 68–75 r = 0.37, p < 0.001), and this association remained significant after controlling for several potential confounding variables (p = 0.004). Analysis of individual item scores showed that this association spanned executive and memory domains. No analogue cognition associations were observed for CBF MD. In covariate-controlled models, hippocampal volume was not associated with cognition in PD, but there was a significant association for hippocampal MD (p = 0.02). Conclusions: Early cognitive deficits in PD without dementia are more closely related to structural MRI measures of CBF degeneration than hippocampal degeneration. In our multicentric imaging acquisitions, DTI-based diffusion measures in the CBF were inferior to standard volumetric assessments for capturing cognition- relevant changes in non-demented PD. 1. Introduction ongoing prospective, multi-center, non-interventional, long-term study on PD progression that includes detailed clinical evaluations as well as Although Parkinson’s disease (PD) is primarily perceived as a assessments of serum biomarkers, genetics, and neuroimaging data [16]. movement disorder, progressive cognitive deficits typically emerge as The participants selected for the current cross-sectional study corre­ the disease progresses, eventually resulting in a characteristic PD- spond to a subset of COPPADIS-2015 participants from five different associated dementia syndrome [1]. In recent years, structural neuro­ centers who underwent a 3D T1-weighted structural MRI scan at the imaging proxies of cholinergic system integrity could provide indirect in baseline evaluation. A subsample of 73 PD patients also underwent a vivo evidence for the critical role of cholinergic degeneration in baseline DTI acquisition. Patients were excluded if they met screening PD-associated cognitive decline [2]. Although the cholinergic nuclei are criteria for dementia (Mini Mental State Examination (MMSE) < 26 and not directly visible on structural MRI contrasts, volumetric analysis of inability of performing basic activities of daily living as determined by regions-of-interest (ROI) focussed on the cholinergic basal forebrain clinical interview). Detailed inclusion and exclusion criteria for the (CBF) revealed marked atrophy of this region in both PD dementia COPPADIS-2015 study have been detailed before [16] and are summa­ (PDD) and pathologically related dementia with Lewy bodies (DLB) rized in the supplement. [3–6]. However, CBF degeneration in the predementia phase of PD and This study was approved by the local ethics committees on human its relevance for the emergence of subtle cognitive deficits is less well experimentation of the different participating centers. Written informed explored. A previous manual volumetry study found reduced CBF vol­ consent was obtained from all subjects participating in this study. umes in PD patients with mild cognitive impairment (PD-MCI) to be predictive of progression to dementia [7], but others found no differ­ 2.2. Clinical and neuropsychometric evaluation ences between PD-MCI and PD with normal cognition at the group level [4]. Disease stage and motor symptom severity were evaluated by the More recent studies used automated CBF morphometry techniques Hoehn and Yahr (H&Y) scale and the Unified Parkinson’s Disease Rating based on stereotactic mappings of the cholinergic nuclei that may pro­ Scale-Part III (UPDRS-III), both assessed in “off” condition. vide a more comprehensive, and potentially more sensitive, in vivo Neuropsychological performance was evaluated in “on” condition analysis of CBF degeneration [5,8–11]. Using this approach it could be using the Parkinson’s Disease – Cognitive Rating Scale (PD-CRS) [17], demonstrated that although CBF volumes were not significantly reduced including individual item scores for tests of sustained attention, working in de novo PD patients as a group, reduced CBF volumes could be memory, alternating and action verbal fluency, immediate and delayed detected in a subset of patients [9,10]. These patients also showed a verbal memory, naming, as well as drawing and copy of a clock. significantly increased risk for imminent cognitive decline, suggesting Although all patients were non-demented according to clinical screening that cognition-relevant changes in CBF volume are already present at criteria, eight patients had PD-CRS total scores ≤64 indicative of this early disease stage. dementia-level cognitive impairment [17] and were excluded from In addition to the recent in vivo research on CBF degeneration in PD, further analyses. previous neuroimaging studies had implicated other cognition-relevant brain structures in PD-related cognitive impairments, particularly the hippocampal memory system [12–14], and the relative contributions of 2.3. MRI acquisition these different neuroanatomic systems to cognitive decline in PD remain largely unknown. Moreover, recent multimodal MRI studies indicated Structural MRI data were acquired on different clinical MRI scanners that microstructural diffusion alterations as measured by diffusion (one 1.5 T and four 3 T machines) using scanner-specific 3D T1- tensor imaging (DTI) may be more sensitive markers of weighted structural imaging sequences with approximately 1 mm cognition-relevant neurodegenerative processes in PD compared to isotropic spatial resolution. For a subsample of the PD patients scanned volumetric measurements on structural MRI [11,14,15]. on 3 T MRI scanners additional diffusion-weighted images were ac­ In the present study we aimed to comprehensively assess the relation quired using scanner-specific echo-planar imaging sequences. Detailed of MRI-measured CBF and hippocampal volumes with cognitive differ­ center-specific acquisition parameters for the T1-and diffusion-weighted ences in a well-characterized multicentric cohort of non-demented PD scans are listed in Supplementary Table 1. Average time between MRI patients, and to further explore the utility of DTI-based diffusion alter­ acquisition and cognitive evaluation was 75 ± 67 days (IQR: 24–104 ations for detecting cognition-relevant tissue changes in these days). structures. 2.4. Processing and analysis of structural MRI and DTI data 2. Methods Automated CBF and hippocampus volumetry on T1-weighted struc­ 2.1. Participants tural MRI scans followed procedures that have been described in detail previously [5,9,18,19], and were implemented using statistical para­ This study included a total of 180 non-demented PD patients and 45 metric mapping software (SPM12, Wellcome Trust Center for Neuro­ healthy controls from the COPPADIS-2015 (Cohort of Patients with imaging) and the CAT-toolbox (http://www.neuro.uni-jena.de/cat/) Parkinson’s Disease in Spain, 2015) cohort. COPPADIS-2015 is an implemented in Matlab R2018a (MathWorks, Natick, MA, USA). Briefly, MRI scans were segmented into gray matter (GM), white matter, and 69 M.J. Grothe et al. Parkinsonism and Related Disorders 88 (2021) 68–75 cerebrospinal fluid partitions and high-dimensionally registered to corrected using the False Discovery Rate (FDR). Montreal Neurological Institute (MNI) standard space using the highly accurate DARTEL algorithm [18,19]. GM volumes of the CBF and hip­ pocampus were then automatically extracted by summing up the modulated GM voxel values within respective ROIs in MNI standard space. The total intracranial volume (TIV), as a measure of head size, was calculated as the sum of total volumes of the GM, white matter, and cerebrospinal fluid partitions [19]. The CBF ROI was based on our previous methodological study Table 1 characterizing functionally homogeneous subdivisions within the Sample characteristics. human CBF as defined by combined information from existing stereo­ HC PD Cohen’s Group tactic atlases of basal forebrain cholinergic nuclei in MNI space [20]. In (N = 45) (N = 180) d Difference our primary analyses we used the total anatomic CBF space as an overall Statistic value (P value) measure of CBF degeneration, but in secondary analyses we also considered the functionally-defined anterior-medial (aCBF) and Demographics posterior-lateral (pCBF) subdivisions separately (Supplementary Sex (M/F) 30/15 113/67 X2 = 0.10 (0.755) Fig. S1). The hippocampus ROI was based on a recently developed MNI Age, years 59.4 (5.7) 61.7 (9.6) − 0.26 t = − 2.07 standard space template of international consensus criteria for hippo­ (0.041) campus outlines on structural MRI [18]. Education 14/19/12 95/44/41 X2 = 7.78 Processing of DTI data was implemented using the DTI toolbox of FSL (primary/ (0.020) secondary/ (Version 5.0, FMRIB, Oxford, UK), and included correction for eddy university) current effects and head motion, skull stripping, and fitting of diffusion tensors to the data to derive scalar maps of mean diffusivity (MD). MD Clinical variables Age at onset, years 57.6 maps were spatially normalized to MNI standard space using the (10.1) deformation fields derived from normalization of the co-registered T1- Disease duration, 4.5 (4.3) weighted MRI scans. Individual CBF and hippocampal MD indices were years calculated by extracting average voxel values within the respective ROIs UPDRS-III “Off” 21.1 score (10.1) from the normalized MD maps [11]. Hoehn & Yahr stage 2 (2–2) All images passed initial visual quality assessments for gross image Anticholinergic 175/5 artefacts, as well as subsequent assessments of overall tissue type seg­ treatment (No/Yes) mentation and global spatial normalization accuracy. Mean absolute Total levodopa 517 (430) and relative inter-volume head displacement during the DTI acquisitions equivalent doses were 1.76 ± 2.04 mm and 0.61 ± 0.71 mm, respectively. Cognitive characteristics Total MMSE 29.7 (0.5) 29.5 (0.8) 0.33 t = 2.57 (0.012) Total PD-CRS 100.1 94.4 0.45 t = 3.57 (0.001) 2.5. Statistical analysis (8.3) (13.2) Fronto-subcortical items Differences in ROI volumes between groups were assessed with Immediate verbal 8.3 (1.4) 7.6 (1.9) 0.36 t = 2.57 (0.012) analysis of covariance (ANCOVA) models controlled for acquisition site, memory Sustained 9.3 (0.8) 8.8 (1.6) 0.37 t = 3.15 (0.002) TIV, age, sex, and education. Associations of volumetric and diffusivity attention indices of CBF degeneration with cognitive deficits in PD were assessed Working memory 8.4 (1.5) 8.0 (1.8) 0.26 t = 1.70 (0.093) using linear regression models with PD-CRS total score (PD-CRStotal) as Clock copy 9.9 (0.3) 9.6 (0.8) 0.40 t = 3.71 the primary outcome. Separate linear regression models were calculated (<0.001) for CBF volume and CBF MD as predictor variables, and for significant Delayed verbal 5.2 (1.9) 4.9 (2.6) 0.14 t = 0.98 (0.328) memory models the added value of combining volumetric and microstructural Alternating 12.9 (3.5) 11.8 (3.8) 0.31 t = 1.94 (0.057) information was assessed in combined multimodal regression models. verbal fluency Secondary analyses assessed the domain specificity of CBF-cognition Action verbal 16.4 (4.2) 15.2 (5.5) 0.22 t = 1.53 (0.129) associations across individual PD-CRS item scores, as well as the fluency Posterior-cortical items regional specificity of the cognition associations across aCBF and pCBF Naming 19.8 (0.5) 19.3 (1.6) 0.37 t = 3.75 subdivisions. All regression models were controlled for acquisition site, (<0.001) and TIV in the case of volumetric measures, and the robustness of sig­ Clock drawing 9.8 (0.4) 9.3 (1.6) 0.38 t = 4.19 nificant CBF-cognition associations was assessed in additional models (<0.001) further controlling for several potential confounding variables, Regional MRI Volumes including age, sex, education, disease duration, motor symptom severity Cholinergic basal 585.9 587.9 − 0.04 F = 0.0002 (UPDRS-III scores), total levodopa equivalent doses, and offset between forebrain (45.9) (49.4) (0.988) Hippocampus 1151.1 1150.6 0.003 F = 0.14 (0.71) MRI acquisition and cognitive evaluation. Analogous models were (121.1) (118.4) calculated using volumetric and diffusivity indices of hippocampal degeneration. In a complementary analysis we additionally assessed The descriptive values presented are: number for categorical variables; median associations between neuroimaging markers and cognition separately (IQR) for Hoehn & Yahr stage; and mean (standard deviation) for all other for each acquisition site, followed by a meta-analysis of the center- continuous variables. Statistic values correspond to chi-squared test for cate­ gorical variables, t-test for continuous variables, and ANCOVA models specific outcomes (see supplementary material). Statistical analyses controlled for acquisition site, TIV, sex, age and education for regional MRI were performed using the computing environment R (version 3.5.1). volumes. Disease duration was defined as the time passed since initial symptoms Following recommendations described in the statistical literature [21], were noticed by the patient, irrespective of the time of clinical diagnosis. we primarily report uncorrected p-values in this hypothesis-driven study Abbreviations: HC = Healthy controls; PD = Parkinson’s disease; with a limited number of planned comparisons focused on the associa­ TIV = Intracranial volume; MMSE = Mini Mental State Examination; UPDRS-III tion between neuroimaging measures of CBF degeneration and cognitive = Unified Parkinson’s Disease Rating Scale-Part III; PD-CRS = PD-Cognitive deficits in PD. However, for transparency we also report p-values Rating Scale. 70 M.J. Grothe et al. Parkinsonism and Related Disorders 88 (2021) 68–75 3. Results Table 2 Linear model stats for the association of volumetric imaging markers with 3.1. Patient characteristics cognitive scores. Cognitive Volume Demographic and clinical characteristics of the participants are scores Cholinergic Basal Forebrain Hippocampus detailed in Table 1. Although all PD patients were non-demented by Part. T value (P P(FDR) Part. r T value (P P inclusion criteria, they showed a large variability in global cognitive r value) value) (FDR) performance (PD-CRStotal: mean = 94 ± 13, range = 68–135). According to a previously established cut-off of PD-CRStotal ≤ 81 [17], 34 patients PD-CRStotal 0.37 5.20 <0.001 0.24 3.20 0.002 (<0.001) (0.002) (19%) would classify for a categorisation as PD-MCI. 0.22 2.91 0.012 0.05 0.61 0.541 In covariate-controlled comparisons to the healthy controls, neither (0.004) (0.541) CBF nor hippocampal volume was significantly reduced in the PD group Fronto-subcortical items as a whole (p > 0.71). Sustained 0.19 2.48 0.016 0.13 1.70 0.102 attention (0.014) (0.091) 3.2. MRI-measured cholinergic basal forebrain and hippocampus volume 0.10 1.25 0.244 (0.214) in relation to cognition in PD Working 0.33 4.56 <0.001 0.18 2.41 0.028 memory (<0.001) (0.017) Table 2 summarizes the results of linear regression models for the 0.19 2.49 0.054 − 0.01 − 0.08 0.94 association of CBF and hippocampus volume with PD-CRS scores. After (0.014) (0.94) accounting for TIV and acquisition site, lower CBF volume was signifi­ Clock 0.2 2.75 0.01 0.18 2.41 0.028 drawing (0.007) (0.017) cantly associated with lower PD-CRStotal (rpartial = 0.37, p < 0.001), and 0.10 1.33 0.244 0.06 0.81 0.488 this association also remained significant after controlling for con­ (0.184) (0.417) founding variables (p = 0.004). Among individual PD-CRS item scores, Alternating 0.27 3.75 0.001 0.18 2.37 0.028 CBF volume was most strongly associated with immediate memory, but verbal (<0.001) (0.019) covariate-controlled models were also significant for delayed memory, fluency 0.15 2.00 0.094 0.06 0.81 0.488 (0.047) (0.418) working memory, and alternating verbal fluency. In subregion-specific Action verbal 0.20 2.70 0.01 0.02 0.32 0.747 analyses all associations were slightly more pronounced for the pCBF fluency (0.008) (0.747) than for the aCBF, but none of the differences were statistically signifi­ 0.09 1.14 0.257 cant (p > 0.83; Supplementary Table S2). (0.257) Immediate 0.37 5.27 0.27 3.66 0.001 In the basic regression models (only controlling for TIV and acqui­ <0.001 verbal (<0.001) (<0.001) sition site) hippocampal volume was also associated with PD-CRStotal memory 0.28 3.69 0.002 0.15 1.96 0.26 (rpartial = 0.24, p = 0.002) and particularly with delayed and immediate (<0.001) (0.052) verbal memory scores, but none of these associations remained signifi­ Delayed 0.31 4.29 <0.001 0.29 3.95 0.001 cant in the fully covariate-controlled model (Table 2). verbal (<0.001) (<0.001) memory 0.17 2.22 0.073 0.14 1.8 0.26 For comparison, we also assessed associations of CBF and hippo­ (0.027) (0.074) campus volumes with cognitive performance in the healthy control Posterior-cortical items group using identical linear regression models. In analyses controlled for Naming 0.05 0.59 0.554 0.16 2.07 0.052 TIV and acquisition center, neither PD-CRS total score nor any of the (0.554) (0.04) 0.08 1.03 0.488 individual item scores were significantly associated with CBF or hip­ (0.306) pocampal volume in this group (p > 0.18). Clock copy 0.23 3.16 0.003 0.23 3.16 0.006 (0.002) (0.002) 3.3. DTI-based diffusion measures of cholinergic basal forebrain and 0.10 1.32 0.244 0.10 1.24 0.488 hippocampal integrity in relation to cognition in PD (0.19) (0.218) Linear model stats are presented as (partial) r and T value (P value). For each In the subsample with DTI data, CBF volume showed a comparable cognitive item score, results from single predictor linear regression models association with PD-CRStotal as in the full sample (Fig. 1A; Supplemen­ (controlled for acquisition site and TIV) are indicated in the first row, and for tary Table S3), but CBF MD was not significantly associated with PD- significant models fully covariate-controlled regression results are indicated in CRStotal (Fig. 1C, Table 3; rpartial = − 0.10, p = 0.42) or any individual the second row. Abbreviations: PD-CRS = PD-Cognitive Rating Scale. item scores (Table 3). Combining both volumetric and diffusivity mea­ surements of CBF integrity into a multimodal regression model for PD- CRStotal did not increase the fit compared to the model based on CBF 4. Discussion volume alone (ΔR2 =0.003, p = 0.60). Associations between hippocampal volume and PD-CRS scores In this study we analyzed multimodal neuroimaging measures of CBF remained nonsignificant in the subsample with available DTI data degeneration in relation to cognitive performance in a well- (Fig. 1B; Supplementary Table S3), but hippocampal MD did show a characterized multicentric sample of PD patients without dementia, significant association with PD-CRStotal (rpartial = − 0.43, p < 0.001; and further assessed the specificity of the CBF-cognition associations in Fig. 1D), which also remained significant in fully covariate-controlled relation to imaging markers of hippocampal degeneration. We found models (p = 0.02), and was mainly driven by associations with alter­ that lower cognitive performance among PD patients was robustly nating verbal fluency scores (Table 3). Combining volumetric and associated with structural MRI-based volume measures of CBF degen­ diffusivity measurements of hippocampus integrity into a multimodal eration, but not with volumetric measures of hippocampal degeneration. regression model for PD-CRStotal did not significantly increase the fit In a subset analysis of DTI data, diffusivity measures of the CBF were not compared to the model based on hippocampal MD alone (ΔR2 =0.034, associated with cognition, but diffusivity measures of the hippocampus p = 0.23). were more sensitive to cognition-relevant tissue changes than hippo­ campal volume. While severe CBF degeneration is an established neuropathologic feature of dementia in PD that can be readily detected on structural MRI 71 M.J. Grothe et al. Parkinsonism and Related Disorders 88 (2021) 68–75 Fig. 1. Associations of volumetric and diffusion-based imaging markers with PD-CRS total scores. (A) Residualized cholinergic basal forebrain volume accounted for TIV and acquisition site plotted against PD-CRStotal score. (B) Residualized cholinergic basal forebrain mean diffusivity accounted for acquisition site plotted against PD-CRStotal score. (C) Residualized hippocampal volume accounted for TIV and acquisition site plotted against PD-CRStotal score. (D) Residualized hippocampal mean diffusivity accounted for acquisition site plotted against PD-CRStotal score. For visualization purposes mean diffusivity scores were scaled by a factor of 1000. For better comparability between volumetric and diffusion-based imaging markers, all imaging-cognition associations were plotted for the subset of PD patients who had both imaging modalities available. scans, the degree to which CBF degeneration also relates to more subtle imaging studies of cholinergic degeneration in non-demented PD found cognitive deficits that emerge in the predementia phase of PD is less well that cortical cholinergic denervation was associated with reduced per­ known [2–5,22]. A previous study on the de novo PD cohort of the formance in several cognitive domains, including attention, executive Parkinson’s Progression Marker Initiative (PPMI) did not observe sig­ function, and memory, but not with visuospatial function [23,24]. It nificant differences in MRI-measured CBF volume between healthy could be possible that the relative influence of CBF degeneration on controls and the PD group as a whole, but found reduced CBF volume in domain-specific cognitive functions may change over the course of a subset of PD patients, and these were at a significantly higher risk for disease, where effects on typical cognitive measures of visuospatial future cognitive decline [9]. A subsequent study further showed that performance may only become apparent at more severe degrees of lower CBF volume was also associated with worse cognitive perfor­ cholinergic degeneration and closer to the time of dementia diagnosis. mance in non-demented PD patients at a cross-sectional level [10]. Here A striking finding of our study is that, in contrast to CBF volume, we add further evidence for a robust association between MRI-measured hippocampal volume did not show robust associations with cognition CBF volume and cognitive performance in PD without dementia by after controlling for confounding variables that are likely to mediate analysing an independent, multicentric sample of well-characterized PD such an association, particularly age and disease severity. Several pre­ patients and using strict control of several possible confounding vious structural MRI studies had reported hippocampal atrophy in PD variables. patients with cognitive impairment and particularly in PDD [12,13]. Analysis of the individual item scores of the PD-CRS battery revealed However, studies investigating the implication of the hippocampus in that CBF volume was associated with performance in several neuro­ cognitive impairments among non-demented PD patients have produced psychological functions including both executive and memory domains. mixed results, and a recent meta-analysis of structural MRI findings This is partly in contrast to predictions from the “dual syndrome hy­ could confirm significant hippocampus atrophy in PDD but not in pothesis”, which states that the cholinergic deficit in PD may be more PD-MCI [25]. Our combined assessment of hippocampal and CBF vol­ closely associated with impairments in mnemonic and visuospatial ume in the same study sample of non-demented PD patients indicates functions, as compared to more dopamine-dependent executive dys­ that cognitive deficits at this stage more closely relate to CBF atrophy functions [1]. Similar to our present findings, previous PET-based than to hippocampal atrophy. Macroscopic hippocampal atrophy as 72 M.J. Grothe et al. Parkinsonism and Related Disorders 88 (2021) 68–75 Table 3 CBF volume and MD to be significant predictors of prospective cognitive Linear model stats for the association of diffusivity imaging markers with decline. Similarly, another recent DTI study found significant cognitive scores. cross-sectional correlations between diffusivity changes in CBF ROIs and Cognitive Mean Diffusivity cognitive performance in non-demented PD patients, but CBF volumes scores Cholinergic Basal Forebrain Hippocampus were not assessed in that study [26]. These different findings could at least partly relate to differences in patient characteristics, as the pro­ Part. r T value P Part. r T value (P P(FDR) portion of PD-MCI patients was notably higher (20 out of 52 patients; (P (FDR) value) value) 38%) in this previous study compared to our current study sample (16 out of 73 patients with PD-CRStotal ≤ 81; 22%). However, they may also PD-CRStotal − 0.10 − 0.82 0.418 − 0.43 − 3.98 <0.001 (0.418) (<0.001) reflect a generally higher variability in DTI-based diffusion measures, − 0.29 − 2.40 0.029 which are known to be more susceptible to noise compared to standard (0.02) volumetric measures on structural MRI [27]. Fronto-subcortical items This study has several limitations. First, although the multicentric Sustained 0.01 0.05 0.958 0.02 0.20 0.843 design of the study allowed us to assess CBF-cognition associations in a attention (0.958) (0.843) relatively large and clinically diverse sample of PD patients, the extracted imaging indices may have been affected by the multicentric Working − 0.10 − 0.84 0.95 − 0.26 − 2.22 0.049 memory (0.404) (0.030) acquisition of the imaging data, which is particularly a concern for the − 0.01 − 0.08 0.94 DTI signal measurements [27]. In addition to a generally higher sus­ (0.939) ceptibility to inter-scanner related variance, DTI signal measurements Clock − 0.08 − 0.63 0.95 − 0.26 − 2.27 0.049 may potentially also be more prone to other signal confounders, such as drawing (0.528) (0.026) for example contamination by CSF signal that borders the ventral limit − 0.15 − 1.17 0.371 (0.247) of the basal forebrain. Thus, besides modality-specific sensitivities for Alternating 0.04 0.37 0.958 − 0.49 − 4.72 <0.001 distinct neuropathological processes, the observed differences in verbal (0.710) (<0.001) cognition associations between these imaging modalities may also relate fluency − 0.44 − 3.90 0.001 to technical differences in the MRI- and DTI-based measurements. (<0.001) Action verbal − 0.01 − 0.10 0.958 − 0.30 − 2.58 0.036 However, it has to be noted that we controlled all our statistical analyses fluency (0.924) (0.012) for the different acquisition sites, and the multicentric design also − 0.20 − 1.63 0.323 allowed us to demonstrate the robustness of the CBF-cognition associ­ (0.108) ations across centers (see center-specific meta-analysis in the supple­ Immediate − 0.24 − 2.01 0.216 − 0.18 − 1.53 0.147 mentary material), which is a key requirement of potential imaging verbal (0.048) (0.131) memory − 0.14 − 1.10 0.274 markers that are to be used in wider clinical setting. Another limitation (0.274) is the cross-sectional nature of our study. It remains to be determined Delayed − 0.27 − 2.37 0.186 − 0.25 − 2.18 0.049 whether the observed cognition-relevant differences in CBF volume verbal (0.021) (0.033) among PD patients also translate to a higher risk for progression to de­ memory − 0.14 − 1.14 0.274 − 0.03 − 0.24 0.94 (0.257) (0.812) mentia. Furthermore, the COPPADIS cohort represents a rather selective Posterior-cortical items research cohort that may not be reflective of the general PD population, Naming − 0.02 − 0.20 0.958 − 0.21 − 1.80 0.099 and thus our results should be replicated in less preselected cohorts from (0.846) (0.077) community-based studies. Another important consideration is that the employed MRI markers of CBF integrity can only serve as indirect Clock copy − 0.09 − 0.77 0.95 − 0.32 − 2.85 0.026 (0.445) (0.006) markers of cholinergic degeneration. The analyzed CBF region corre­ − 0.18 − 1.42 0.323 sponds to the localization of the cholinergic nuclei in the human brain as (0.161) determined by cytoarchitectonic mappings from combined histology Linear model stats are presented as (partial) r and T value (P value). For each and post-mortem MRI [20], and imaging-pharmacologic [6] and cognitive item score, results from single predictor linear regression models imaging-neurophysiologic [28] studies provide evidence for a link be­ (controlled for acquisition site) are indicated in the first row, and for significant tween CBF integrity on MRI and cholinergic system functioning. How­ models fully covariate-controlled regression results are indicated in the second ever, one cannot exclude that the imaging indices might also reflect row. change in other neuronal or glial components. Finally, the correlation Abbreviations: PD-CRS = PD-Cognitive Rating Scale. coefficients for the association between CBF volume and cognition are of rather moderate effect size, which, however, lies well in the range of measured by volumetric changes on MRI may only occur at a later stage previous studies assessing associations between regional imaging mea­ of PD-related cognitive decline. sures and cognition in PD [4,10,14]. Associations between neuro­ However, more subtle neuroimaging measures may increase the imaging measures and clinical outcomes can rarely be expected to be sensitivity to detect early neurodegenerative tissue changes in the hip­ strictly linear and will generally depend on the functional consequences pocampus already at a predementia stage in PD. Thus, a previous of the neurodegenerative process that is indexed by the macro- and multimodal MRI/DTI imaging study found that hippocampal diffusivity microstructural neuroimaging alterations. A recent changes correlated with memory performance among non-demented PD imaging-neuropathology association study found that CBF volume on patients, but no significant associations were observed for hippocampal structural MRI correlated with the extent of Lewy body pathology in the volume [14]. In our DTI substudy we could confirm that hippocampal nucleus basalis of Meynert, but not with semi-quantitative assessments diffusivity measures were more closely associated with cognition in of cell loss in this region [29]. This indicates that the neuroimaging non-demented PD compared to volume, although this association measures may also pick up relatively subtle tissue changes in response to showed a rather unexpected specificity for verbal fluency instead of neurodegenerative pathology [30]. memory measures. In conclusion, we demonstrate in a well-characterized multicentric Interestingly, in contrast to the findings for the hippocampus, DTI- sample of PD patients without dementia that cognitive differences in this based diffusivity measures of the CBF were not associated with cogni­ non-demented spectrum are more closely related to structural MRI tion in our sample. This is partly in contrast to a previous MRI/DTI study measures of CBF degeneration than hippocampal degeneration. DTI- in de novo PD patients from the PPMI cohort [11], which found both based diffusion measures could increase the sensitivity to detect 73 M.J. Grothe et al. Parkinsonism and Related Disorders 88 (2021) 68–75 cognition-relevant tissue changes in the hippocampus, but at least in the Rodríguez Méndez L11, Rodríguez Pérez AB17, Roldán F1, Ruíz De Arcos context of unharmonized multicentric imaging acquisitions appear to be M27, Sánchez Alonso P37, Sánchez-Carpintero M11, Sánchez Díez G15, inferior to standard volumetric measures for capturing cognition- Sánchez Rodríguez A9, Santacruz P22, Seijo M29, Serarols A29, Segundo relevant CBF changes. These findings may have important implica­ Rodríguez JC34, Sierra Peña M9, Suárez-Castro E11, Tartari JP4, Valero tions for the development of clinically useful imaging biomarkers aiding C39, Vargas L1, Vázquez Gómez R29, Vela L35, Villanueva C23, Vives B32, in the early detection of PD patients at increased risk for cognitive Villar MD8. decline. 15. Hospital Universitario Ramón y Cajal. Madrid. 16. Hospital Meixoeiro, Complejo Hospitalario Universitario de Authors’ contribution Vigo. Vigo. 17. Hospital General Universitario de Elche, Elche, Spain. Michel J. Grothe: conception of the manuscript, study design, anal­ 18. Complejo Asistencial Universitario de Burgos, Burgos, Spain. ysis, writing of the first draft of the manuscript, review and critique. 19. Complejo Hospitalario de Navarra. Pamplona. Navarra. Miguel A. Labrador-Espinosa: conception of the manuscript, study 20. Hospital Universitario Central de Asturias, Oviedo, Spain. design, analysis, review and critique. Silvia Jesús: review and critique; 21. Hospital Infanta Sofía, Madrid, Spain. recruitment and/or assessment of participants. Daniel Macías-García: 22. Hospital Clınic de Barcelona, Barcelona, Spain. review and critique; recruitment and/or assessment of participants. 23. Hospital Clínico San Carlos, Madrid, Spain. Astrid Adarmes-Gómez: review and critique; recruitment and/or 24. Hospital del Mar, Barcelona, Spain. assessment of participants. Elena Iglesias Camacho: review and critique. 25. Hospital La Princesa, Madrid, Spain. Pablo Franco-Rosado: review and critique. Florinda Roldán Lora: review 26. Centro Neurologico Oms 42, Palma de Mallorca. and critique; acquisition of neuroimaging data. Juan Francisco Martín- 27. Hospital Universitario Virgen Macarena, Sevilla. Rodríguez: review and critique. Miquel Aguilar Barberá: review and 28. Hospital Universitario Lucus Augusti (HULA), Lugo, Spain. critique; recruitment and/or assessment of participants. Pau Pastor: re­ 29. Complejo Hospitalario Universitario de Pontevedra (CHOP), view and critique; recruitment and/or assessment of participants. Sonia Pontevedra, Spain. Escalante Arroyo: review and critique; recruitment and/or assessment of 30. Consorci Sanitari Integral, Hospital Moises Broggi, SantJoan participants. Berta Solano Vila: review and critique; recruitment and/or Despı, Barcelona, Spain. assessment of participants. Anna Cots Foraster: review and critique; 31. Hospital Ruber Internacional, Madrid, Spain. recruitment and/or assessment of participants. Javier Ruiz Martínez: 32. Hospital Universitario Son Espases, Palma de Mallorca, Spain. review and critique; recruitment and/or assessment of participants. 33. Hospital Da Costa de Burela, Lugo, Spain. Francisco Carrillo Padilla: review and critique; recruitment and/or 34. Complejo Hospitalario de Toledo. Toledo. assessment of participants. Mercedes Pueyo Morlans: review and 35. Fundacion Hospital de Alcorcon, Madrid, Spain. critique; recruitment and/or assessment of participants. Isabel González 36. University of Pennsylvania, Philadelphia, PA, USA. Aramburu: review and critique; recruitment and/or assessment of par­ 37. Hospital Universitario Puerta del Hierro. Madrid. ticipants. Jon Infante Ceberio: review and critique; recruitment and/or 38. Hospital Rey Juan Carlos, Madrid, Spain, Madrid, Spain. assessment of participants. Jorge Hernández Vara: review and critique; 39. Hospital Arnau de Vilanova, Valencia, Spain. recruitment and/or assessment of participants. Oriol de Fábregues- Boixar: review and critique; recruitment and/or assessment of partici­ Declaration of competing interest pants. Teresa de Deus Fonticoba: review and critique; recruitment and/ or assessment of participants. Berta Pascual-Sedano: review and None. critique; recruitment and/or assessment of participants. Jaime Kulisev­ sky: review and critique. Overall supervision. Pablo Martínez-Martín: Acknowledgments review and critique. Overall supervision. Diego Santos-García: concep­ tion, organization, and execution of the project; recruitment and/or This work was supported by the Alzheimer Forschung Initiative e.V. assessment of participants; review and critique. Pablo Mir: conception of (AFI International Training Grant to MJG), the Instituto de Salud Carlos the manuscript, study design, review and critique. III-Fondo Europeo de Desarrollo Regional (ISCIII-FEDER) [PI14/01823, PI16/01575, PI18/01898, PI19/01576, PI20/00613], the Consejería de Collaborators (COPPADIS SUDY GROUP) Economía, Innovación, Ciencia y Empleo de la Junta de Andalucía [CVI- 02526, CTS-7685], the Consejería de Salud y Bienestar Social de la Junta Almeria M4, Alonso-Cánovas A15, Alonso-Losada MG16, Álvarez de Andalucía [PI-0471-2013, PE-0210-2018, PI-0459-2018, PE-0186- Sauco M17, Arnáiz S18, Arribas S4, Ascunce Vidondo A19, Bernardo N5, 2019], the Fundación Alicia Koplowitz and the Fundación “Curemos el Bejr-Kasem H12, Blázquez Estrada M20, Botí M4, Borrue C21, Cabello Parkinson” (https://www.curemoselparkinson.org). MJG is supported González C19, Cámara Lorenzo A22, Casas E18, Catalán MJ23, Clavero by the “Miguel Servet” program [CP19/00031], MALE by the University Ibarra P19, Cortina Fernández A11, Crespo Cuevas A24, Cubo E18, Díez- of Seville [USE-20046-J], JFM by the “Sara Borrell” program [CD13/ Fairen M4, Erro E19, Fernández Guillán N23, Gámez P18, Gallego M25, 00229] and VI-PPIT-US from the University of Seville [USE-18817-A], García Caldentey J26, García Campos C27, García Moreno JM27, Gastón SJ by the “Acción B-Clínicos-Investigadores” program [B-0007-2019], I19, Gómez Garre MP1,2, González Aloy J6, González Ardura J28, and DMG by the “Río Hortega” program [CM18/00142]. González García B25, González Palmás MJ29, González Toledo GR8, Golpe Díaz A11, Grau Solá M30, Guardia G4, Horta-Barba A12, Infante J9, Appendix A. Supplementary data Kurtis M31, Labandeira C16, Lacruz Bescos F19, Lage Castro M29, Legarda I32, López Díaz LM33, López Manzanares L25, López Seoane B11, López Supplementary data to this article can be found online at https://doi. Ariztegui N34, Macías Y35, Martí Andres G19, Martí MJ22, Martínez- org/10.1016/j.parkreldis.2021.05.027. Castrillo JC15, Mata M21, McAfee D36, Meitín MT33, Menéndez González M20, Méndez Del Barrio C1,2, Miranda Santiago J18, Moreno Diéguez A11, Morales MI34, Nogueira V33, Novo Amado A11, Novo-Ponte S37, References Ordás C38, Pagonabarraga J12, Pascual-Sedano B12, Paz González JM14, [1] P. Svenningsson, E. Westman, C. Ballard, D. Aarsland, Cognitive impairment in Pareés I31, Pérez Fuertes A11, Pérez Noguera R27, Planellas L22, Prats patients with Parkinson’s disease: diagnosis, biomarkers, and treatment, the MA6, Prieto Jurczynska C38, Puente V24, Redondo Rafales N25, Lancet, Neurology 11 (8) (2012) 697–707. 74 M.J. Grothe et al. Parkinsonism and Related Disorders 88 (2021) 68–75 [2] N.I. Bohnen, M.J. Grothe, N.J. Ray, M. Muller, S.J. Teipel, Recent advances in [17] J. Pagonabarraga, J. Kulisevsky, G. Llebaria, C. Garcia-Sanchez, B. Pascual-Sedano, cholinergic imaging and cognitive decline-Revisiting the cholinergic hypothesis of A. Gironell, Parkinson’s disease-cognitive rating scale: a new cognitive scale dementia, Current geriatrics reports 7 (1) (2018) 1–11. specific for Parkinson’s disease, Mov. Disord. : official journal of the Movement [3] H. Hanyu, Y. Tanaka, S. Shimizu, H. Sakurai, T. Iwamoto, K. Abe, Differences in Disorder Society 23 (7) (2008) 998–1005. MR features of the substantia innominata between dementia with Lewy bodies and [18] D. Wolf, M. Bocchetta, G.M. Preboske, M. Boccardi, M.J. Grothe, Reference Alzheimer’s disease, J. Neurol. 252 (4) (2005) 482–484. Standard Space hippocampus Labels According to the EADC-ADNI Harmonized [4] S.H. Choi, T.M. Jung, J.E. Lee, S.K. Lee, Y.H. Sohn, P.H. Lee, Volumetric analysis of Protocol: Utility in Automated Volumetry, Alzheimers Dement, 2017. the substantia innominata in patients with Parkinson’s disease according to [19] M.J. Grothe, I. Kilimann, L. Grinberg, H. Heinsen, S. Teipel, In vivo volumetry of cognitive status, Neurobiol. Aging 33 (7) (2012) 1265–1272. the cholinergic basal forebrain, in: R. Perneczky (Ed.), Biomarkers for Preclinical [5] M.J. Grothe, C. Schuster, F. Bauer, H. Heinsen, J. Prudlo, S.J. Teipel, Atrophy of the Alzheimer’s Disease, Springer Science+Business Media, LLC, 2018, pp. 213–232. cholinergic basal forebrain in dementia with Lewy bodies and Alzheimer’s disease [20] H.J. Fritz, N. Ray, M. Dyrba, C. Sorg, S. Teipel, M.J. Grothe, The Corticotopic dementia, J. Neurol. 261 (10) (2014) 1939–1948. Organization of the Human Basal Forebrain as Revealed by Regionally Selective [6] H. Hanyu, S. Shimizu, Y. Tanaka, K. Hirao, T. Iwamoto, K. Abe, MR features of the Functional Connectivity Profiles, Hum. Brain Mapp. 40 (3) (2019 Feb 15) 868–878, substantia innominata and therapeutic implications in dementias, Neurobiol. Aging https://doi.org/10.1002/hbm.24417. Epub 2018 Oct 11. PMID: 30311315. 28 (4) (2007) 548–554. [21] R.A. Armstrong, When to use the Bonferroni correction, Ophthalmic Physiol. Opt. : [7] J.E. Lee, K.H. Cho, S.K. Song, H.J. Kim, H.S. Lee, Y.H. Sohn, P.H. Lee, Exploratory the journal of the British College of Ophthalmic Opticians (Optometrists) 34 (5) analysis of neuropsychological and neuroanatomical correlates of progressive mild (2014) 502–508. cognitive impairment in Parkinson’s disease, J. Neurol. Neurosurg. Psychiatr. 85 [22] A.K. Liu, R.C. Chang, R.K. Pearce, S.M. Gentleman, Nucleus basalis of Meynert (1) (2014) 7–16. revisited: anatomy, history and differential involvement in Alzheimer’s and [8] M. Grothe, H. Heinsen, S.J. Teipel, Atrophy of the cholinergic Basal forebrain over Parkinson’s disease, Acta Neuropathol. 129 (4) (2015) 527–540. the adult age range and in early stages of Alzheimer’s disease, Biol. Psychiatr. 71 [23] N.I. Bohnen, M.L. Müller, V. Kotagal, R.A. Koeppe, M.A. Kilbourn, R.L. Albin, K. (9) (2012) 805–813. A. Frey, Olfactory dysfunction, central cholinergic integrity and cognitive [9] N.J. Ray, S. Bradburn, C. Murgatroyd, U. Toseeb, P. Mir, G.K. Kountouriotis, S. impairment in Parkinson’s disease, Brain : J. Neurol. 133 (Pt 6) (2010) 1747–1754. J. Teipel, M.J. Grothe, In vivo cholinergic basal forebrain atrophy predicts [24] S. van der Zee, M.L.T.M. Müller, P. Kanel, T. van Laar, N.I. Bohnen, Cholinergic cognitive decline in de novo Parkinson’s disease, Brain : J. Neurol. 141 (1) (2018) denervation patterns across cognitive domains in Parkinson’s disease, Mov. Disord. 165–176. : official journal of the Movement Disorder Society 36 (3) (2021 Mar) 642–650, [10] M.J. Barrett, S.A. Sperling, J.C. Blair, C.S. Freeman, J.L. Flanigan, M.E. Smolkin, C. https://doi.org/10.1002/mds.28360. Epub 2020 Nov 2. PMID: 33137238. A. Manning, T.J. Druzgal, Lower volume, more impairment: reduced cholinergic [25] A.S. Mihaescu, M. Masellis, A. Graff-Guerrero, J. Kim, M. Criaud, S.S. Cho, basal forebrain grey matter density is associated with impaired cognition in C. Ghadery, M. Valli, A.P. Strafella, Brain degeneration in Parkinson’s disease Parkinson disease, J. Neurol. Neurosurg. Psychiatr. 90 (11) (2019) 1251–1256. patients with cognitive decline: a coordinate-based meta-analysis, Brain imaging [11] J. Schulz, G. Pagano, J.A. Fernández Bonfante, H. Wilson, M. Politis, Nucleus and behavior 13 (4) (2019) 1021–1034. basalis of Meynert degeneration precedes and predicts cognitive impairment in [26] F. Gargouri, C. Gallea, M. Mongin, N. Pyatigorskaya, R. Valabregue, C. Ewenczyk, Parkinson’s disease, Brain : J. Neurol. 141 (5) (2018) 1501–1516. M. Sarazin, L. Yahia-Cherif, M. Vidailhet, S. Lehéricy, Multimodal magnetic [12] E.J. Burton, I.G. McKeith, D.J. Burn, E.D. Williams, J.T. O’Brien, Cerebral atrophy resonance imaging investigation of basal forebrain damage and cognitive deficits in Parkinson’s disease with and without dementia: a comparison with Alzheimer’s in Parkinson’s disease, Mov. Disord. : official journal of the Movement Disorder disease, dementia with Lewy bodies and controls, Brain : J. Neurol. 127 (Pt 4) Society 34 (4) (2019) 516–525. (2004) 791–800. [27] S.J. Teipel, S. Reuter, B. Stieltjes, J. Acosta-Cabronero, U. Ernemann, A. Fellgiebel, [13] T.R. Melzer, R. Watts, M.R. MacAskill, T.L. Pitcher, L. Livingston, R.J. Keenan, J. M. Filippi, G. Frisoni, F. Hentschel, F. Jessen, S. Klöppel, T. Meindl, P.J. C. Dalrymple-Alford, T.J. Anderson, Grey matter atrophy in cognitively impaired W. Pouwels, K.H. Hauenstein, H. Hampel, Multicenter stability of diffusion tensor Parkinson’s disease, J. Neurol. Neurosurg. Psychiatr. 83 (2) (2012) 188–194. imaging measures: a European clinical and physical phantom study, Psychiatr. Res. [14] G.A. Carlesimo, F. Piras, F. Assogna, F.E. Pontieri, C. Caltagirone, G. Spalletta, 194 (3) (2011) 363–371. Hippocampal abnormalities and memory deficits in Parkinson disease: a [28] J. Peter, J. Lahr, L. Minkova, E. Lauer, M.J. Grothe, S. Teipel, L. Köstering, C. multimodal imaging study, Neurology 78 (24) (2012) 1939–1945. P. Kaller, B. Heimbach, M. Hüll, C. Normann, C. Nissen, J. Reis, S. Klöppel, [15] F. Sampedro, S. Martínez-Horta, J. Marín-Lahoz, J. Pagonabarraga, J. Kulisevsky, Contribution of the cholinergic system to verbal memory performance in mild Longitudinal intracortical diffusivity changes in de-novo Parkinson’s disease: a cognitive impairment, J. Alzheim. Dis. : JAD 53 (3) (2016) 991–1001. promising imaging biomarker, Park. Relat. Disord. 68 (2019) 22–25. [29] S.J. Teipel, H.C. Fritz, M.J. Grothe, Neuropathologic features associated with basal [16] D. Santos-Garcia, P. Mir, E. Cubo, L. Vela, M.C. Rodriguez-Oroz, M.J. Marti, J. forebrain atrophy in Alzheimer disease, Neurology 95 (10) (2020) e1301–e1311. M. Arbelo, J. Infante, J. Kulisevsky, P. Martinez-Martin, COPPADIS-2015 (COhort [30] S. Teipel, A. Drzezga, M.J. Grothe, H. Barthel, G. Chetelat, N. Schuff, P. Skudlarski, of Patients with Parkinson’s DIsease in Spain, 2015), a global–clinical evaluations, E. Cavedo, G.B. Frisoni, W. Hoffmann, J.R. Thyrian, C. Fox, S. Minoshima, O. Sabri, serum biomarkers, genetic studies and neuroimaging–prospective, multicenter, A. Fellgiebel, Multimodal imaging in Alzheimer’s disease: validity and usefulness non-interventional, long-term study on Parkinson’s disease progression, BMC for early detection, the Lancet, Neurology 14 (10) (2015) 1037–1053. Neurol. 16 (2016) 26. 75