Scribd Logo
Upload

Welcome to Scribd!

  • 2 views

    Boon Et Al

    Uploaded by

    Uriel Bühler

    Copyright:

    © All Rights Reserved
    Download as PDF or read online from Scribd
    Flag for inappropriate content

    You might also like

    Boon Et Al

    Uploaded by

    Uriel Bühler
    2 views8 pages

    Original Title

    Boon et al

    Copyright

    © © All Rights Reserved

    Available Formats

    PDF or read online from Scribd

    Did you find this document useful?

    Is this content inappropriate?

    Report this Document

    Copyright:

    © All Rights Reserved
    Download as PDF or read online from Scribd
    Flag for inappropriate content
    Download as pdf
    2 views8 pages

    Boon Et Al

    Uploaded by

    Uriel Bühler

    Copyright:

    © All Rights Reserved
    Download as PDF or read online from Scribd
    Flag for inappropriate content
    Download as pdf
    of 8
    Contents lists availabe at ScienceDirect Neurolmage: Clinical = bin 5 ELSEVIER Journal homepage: www. elseviercomvlocatelynic Functional connectivity between resting-state networks reflects decline in| 38 executive function in Parkinson’s disease: A longitudinal fMRI study Lennard I. Boon“", Dagmar H. Hepp“, Linda Douw”, Noélle van Geenen”, Tommy A. A. Broeders”, Jeroen J.G. Geurts’, Henk W. Berendse", Menno M. Schoonheim * Amara UME, Yt Aseria Deum of eve, Ams Nale, De Ban 117, Ansa, The Meo ' Antara UME, Vie Unveia Amsedam Deere of Anatomy and Nancy, oom Narn, be acon 117, Ada, The epee Defic in cognitive fanctioning are a carimon yet poorly understood sjmpiom in Parkinson's escase (PD). ets Recent sade hive highlighted the imporance of (lyme) intersesons beeen resting rate networks for ogiton, ich emains indented in PD. We Invetigned how atered (yam) finetonl Interactions in hi FMRI stud, 50 PD patents (mean age 65.5 years + 6.27) an dopaminergic medication were studied holga testing was performed a wo Sine point, with follow-up dation of appcouimatly thee eat Functional connectivity within and between seven reting-atate netvorks was ealalted (bath staialy and ynanaly and eoerelted wi our neuropeychologieal ts scores; cmbined sore of (ou) exeeutve tas, motor perseverton, memary, and category Nueey task Cognitive dysfunction was determined based on & Tongiudiel sample of ageanatced helt contels (2 = 13). PD patients showed dysfunction on sl cut of seven cogalve asks when compared to healthy conto Severty of executive dysfunction was corelated with higher sti and lower dynamic functional connectivity betwoen deep gray mate reylons andthe Rentoparctal network (DEMEIPN). Overtime, dedining exceutve funtion was elated to increasing sate DGM-EPN connectivity, together with changes of connectivity involving the dorsal attention newark amongst others with the venta tention network). lati functional conpecdvty between the ventral and dorelatenton network correlated wth motor perseveration, ‘ur findings demonstrate that in PD patents, ystuneconal communication betweea ()subcorcl, ronto- parietal and setention networks mos underlies worsening of exectiveRretionng, (8) tention networks fre involved in motor perseveration Dyna) freon enneiey 1, Introduction task performance (Kchapia etal, 2013), as was demonstrated by phar- macologica I(Gotham ct al, 1988), pharmacogesomies (Coos, 2006), Cognitive impairment is a common non-motor symptom of Parkin- and neuroimaging research (Chrisiopher 4) son's disease (PD) that negatively impacts daly functioning and quality Importantly, executive deficits can occur at any sage ofthe disease Of life (Asrsland et sl, 2017). Cognitive impairment in non-demented and can be progressive, but are not predictive of the development of PD D patients i typically dominated by attentional, visuospatial and ex- dementia (Willians-Gray etal, 2009) Progressive executive and gen- cute defies, with strong heterogeneity between patients (ais eral cognitive decline in PD may reflect complex brain pathology and fa, 2017), An executive syndrome in PD (deficits i cognitive exi- involvement of thee non-dopaminergle neurotransmiter systems Diliy, planning, working memory, and learning) is possibly relaced {0 (Basset, 2011), indicating global network disruptions, whieh remain suboptimal dopamine levels: both exceses ad deficits of dopamine in understudied, the frontosriatal pathways have been associated with impairments in Previous research in PD indicated disturbances of connectivity males ib mane (Ll Boon 22i8-1962/0""2020 "The Author, Published by ‘Hhevier Inc. This is an open acess article under the CC BYNCND license patterns of the default mode network (DMN) in relation to global ‘cognitive dysfunction (Tesitore et al, 2012), Domalnspecific cognitive ‘dysfunction in PD has been related to other ctworks, such ar the attention networks (dorsal attention network, DAN; ventral attention network, VAN) and the bilateral fronoparietal networks (FPN) (Basso ct al, 2014, 2015). In addition, recent technologies! advances have allowed for the evaluation of time-varying. fluctuations in functional connectivity, the socalled dynamic connectivity, for instance showing that the recoafiguetion between feontoparietal and feontotemporal Drain networks i related to executive funetioning (Br et 9, 2079), 4s such, there are sirong indications that PD features extensive cortical network dysfunction in addition tothe classical dopaminergic systems, possibly driven by a loss of dynamic interplay betwoen net- ‘works, In this study, we longitudinally investigated cognitive dysfunc tion, in particular exccutve dysfunction, and functional connectivity (FC) within and between RSNs. We hypothesized that more dispersed ‘network interactions, in addition to connections between the deep ray ‘matter and fronte-paretal network, would be involved inthe develop- ‘ment of executive deficits as well as other cognitive domains in PD. 2. Methods 2.1. Pariipants In this etrospeetive study, date of idiopathic PD patients and healthy ‘controls was used, obtained in the context ofa longitdinal study cohort. Exclusion erteria for PD patients were stereotactic surgery inthe past and extensive white matter lesions or other abnormalities at MRI (see als (Stoffers etal, 2007) for details on recruitment and inclusion). At intial inclusion, the patients di not receive fMRI scans; all (MRI and ‘neuro(peycho)logical data used in this manuscript was acquired at 4 (first timepoint") and 7 year (‘second timepoint’) follow-up visits (Getween 2008 and 2012) in the ovtpatient clinic of the Amsterdam UMC, location VUme. For an overview of the timeline of the data Acquistion, see Fig. |. All examinations were performed in the doy mine “ON” state. All patients and healthy controls have been reported in previous analyses on this dataset in which functional coanectviey differences were liked to global cognitive decline and visual hal nations (Hepp et aly 2017; Olde Dubbelink et al, 2014), but these studies id not specially asses (dynamic) RSN connectivity in celation to domain-specific engnitive dysfunction, Unified PD Rating Scale motor ratings were obtained by a trained physician. Global cognitive functioning was asessed using the Cam- bridge Cognitive Examination scale. The total dese of dopainomi- rmeties was converted 9 @ so-alled levodopa equivalent daly dose {(LEDD) using a previously described conversion rate, see (Ode Dubhe- link et al, 2023) for other definitions. ‘All participant gave written informed consent to the research pro- tocol, which was approved by the local medical ethical committee ‘conform the Helsinki declaration, ease Cel 28 2000) 102455 2.2. Neuropeychological evaluation Neuropsychological function was assessed using the Cambridge [Neuropsychological Test Automated Battery (Eclipse 2.0, Cambridge, England). Executive tests included spatial working memory (outcome measure: strategy), stockings of Cambridge, spatial span length (partly a ‘working. memory test, hence executive function), and inta- extra dimensional setshifing (outcome measure: number of stages ‘completed. Memory was tested using the pattern recognition memory test, motor perseveration using the Vienna perseveration task, and verbal ency sing the one-minute category verbal uency test (ani- ‘mals) from the CAMCOG. Cognitive scores were converted to zscores based on the scores ofthe healthy controls at each time point as (Duteh) ‘norm scares were not available for all tests, and inorder to account for Teaming effects (Crank eal, 1996; Rabbits et al, 2002). Negative 2 scores represent poorer performance on that particular neuropsycho- logical test. The scores were ured as clinical input forthe relationssip| ‘with (dynamic) functional connectivity. Subjects with a value <-15 were considered to be impaired. 23. MRI data acquisition 3 TMR scans (GE Signa HDXT, VISM) included a sagittal three dimensional TH-weighted fast spoiled gradient-echo sequence (repeti- tion time 7.8 ms, ech time 3.0 ms, inversion ime 450 ms, lip angle 12, 10 x 0.9 x 0.9 mm voxe size, 170 slices in sagittal plane) Functional [MRI included 202 volumes of axial eche-planar images, of which the First wo were discarded (repetition time 1800 ms, echo time 35 ms, flip angle 90, 3.8 mm isotropie voxel size, 33 slices in axial plane). Bot in the structural and functional recordings, full-brain coverage was reached. Recordings took place in an’ eyeeclosed, resting-state concition. 24. Image procesing Resting. state (MRI images were pre-processed by MMS, DH and LIB (neuroscientist and medical doctor with respectively 10, 5 and 4 years ‘of experience) using standard PSL-5 software (FMRIB Software Library, Oxiord, England; inip://vew imriboxaculsl) and custom built scripts in bash and Matlab, version 2012a (Mathworks, Natick, MA, USA), We used an independent component analyss-based strategy for ‘Awtomatic Removal of Motion Artifacts (CA-AROMA, vO.4-beta 2017, ‘Nijmegen, the Netherlands) (Pruim etal, 2015). Cortical regions of interest (ROIs) were derived from the Brainnetome atlas (Fan cal 2016) and deep gray matter (DGM) ROIs were derived from FIRST (part ‘of FSL) All cortical ROIs were nos-linearly registered to 3D-T1 space with inverted FNIRT parameters and multiplied with grey matter seg- ‘mentation maps of SIENAX, wile using FIRST ROIs for DGM areas. ‘Subsequently, ll ROIs were combined into one alas (225 ROIS) and this combined atlas as registered to MRI using invered boundary- ‘based registration parameters and masked with @ custom-made MRI Fig. 1. Overview of data analysis Al he fst me pot, a croseaectional correlation was performed between (ypazic) functional connectivity of resingstate networks and newopsjehalogis dala Ter the Pukiaon's deste (PD) group 50), The neropsjchlogial dala f the PD patient were converted Io z-score, Dsed a the Scores of belly coal [Nea fo the 31 PD patete who had undergone a second sty va, longiudalcorclaons were made between the cage ia functional (dynam) cennectity and she change in neuopsjchalogleal perfomance (exposed as cre) ‘mask to remove any residual nonbrain tissue and to reduce the effect of ‘echo planar imaging disortions. After this masking, not alletlar-based, brain regions ad sufficient numbers of voxels in order to be represen: tative. We therefor included only those regions witha eas 30% voxels remaining in a least 90% ofthe subjects (cer eta, 2017). On the basis of these criteria, 29 regions were excluded (bilateral orbital gyrus, nleus accumbens, parshippoesmpal gyrus, and inferior temporal ‘yrs. The final atlas therefore segmented the IMRI sequence into 196 reglons for which mean tme series were obtained To define resting-state networks, AMIRI seans were registered to standard space using aforementioned parameter, where an indepen ‘dent component analysis was run using the MELODIC pipeline (part of FS). The concatenated IMI dataset was decomposed into 30 compe: nents, which led t seven visually idenifed RSNs: DAN, DMN, bilaeral FPN, sensorimotor network (SMI), DCM, VAN, and visual network) ig. % Supplementary Table 1; see Supplementary Table 2 for the location ofthe peak coordinates of each component). RSNs identified ‘were visually inspected (by MMS, LIB) to match with previous literature G 2005) Each brain region was assigned to one network ‘only, based on maximum overlap. The localization of RSNs was based on an independent component analysis of50 PD patients (based only on the first ime polt of the stu). For each participant, Pearson correlation coefficients were calcu lated between time series of all 196 bran region to construct cones tivity matrices. Ngative correlations were converted to absolute values. Next, we calculated the average connectivity ofeach of the seven RSNs with he rest of the brain, as well ax within- and between-RSN connectivity 2.6. Dynamic PC anal In addition to static analyses, Pearson correlation coeicients were ‘calculated per window per subject, with a window length of $8.6 (27% repetition time), resulting in 34 sliding windows, using 2 shift of 10s ‘The choie ofthe window length and # windovts was based onan earlier study (Engels et al, 2018) The standard deviation overtime for each functional connection wes calculated and normalized forthe average FC ‘cross time ofthat connection, which resulted inthe calculation ofthe ‘coefficients of variation ofeach connection. Subsequently, within- and berween-RSN dynamic FC was calculated, Dynamic FC interactions that were significantly correlated with newropsychological test performance were compared with null-models to assess whether the effets were due to random noise. These models were ereted using phaseandomization of our (Fourier-ransformed) data (richard snd The'ler, 1994), Next, we averaged dynamic FC over 50 randomization rans and compared empiccal- with randomized dy. namic FC values 27. Staisical analysis Statistical analyses were performed in IBM SPSS version 22 (Chicago, IL, USA), p< 0.05 was considered significant. Clinical characteristics of Parscipents were compared using independent samples ‘tess. Normality of al variables was assessed with Kolmogorov-Smirnov tests and histogram inspection. As stale FC didnot meet this criterion, It was Transformed using an inversion transformation (1/2), resulting in normality (the sign of resulting beta valves was Mipped because ofthis transformation). Longitudinal changes in nevropsychologiea! perfor ‘mance and functional connectivity were assessed wsing paried tests. To test the association between (dynamie) BC and cognition, a hierarchical linear regression using a backward elimination method was performed er cognitive outcome measure (in which averaged executive performance ‘was treated as one outcome measure). Note that multiple functional ema: ina 8 (020) 102468 een Sagi, coronal and taneverte views are shin. The allocation of repans of Interest to RSW ca loo be fond in Supplementary Ta connections were tested within the same regression model Covariates in ‘the regression models included age, sox, education leve! (dichotomized at level 3), disease duration and LEDD. Residuels were normally distributed for all regression analyses, justifving the use of parametric regression models, To limit the numberof networks analyzed, FC berween one RSN and the rest of the brain was used as input forthe frst Grase-setional) regression model and only thase RSNs showing a main effet with the rest ofthe brain were explored further tn the longitudinal regression models, longitudinal cognitive change was elated to change in FC. Longitudinal FC scores were normalized by ‘the individual mean overall FC, to exclude the possibility that global connectivity changes overtime (eg a physiologial and/or technical explanation) affected the results Dynamic FC values were compared to null models using paired 3. Results 3.1. Study parttpants ‘A total of 59 patients were included atthe rst time point, who all Ihad undergone imaging and neuropsychologiel assessments, 15 Healthy controls were included with aeuropsychological assessments our patients had severe motion artifacts during MR registration and for 5 patients neuropsychological data was not available (due to incomplete cognitive evaluations), leading to an analysis of 50 patients (as opposed to 55 PD patients in previous studies on this cohort (He>p etal, 2017; Olde Dubbelik etal, 2014), Forthe second time point (~3 ‘yeas later), imaging data was avalable for 37 patients. The reason for not partcipating inthe follow-up evaluation varied and included deep brain stimulation placement, death, strong clinical worsening, and refusal to undergo an MII scan. OF the 27 patients, six were excluded due to motion artefacts or the absence of neuropsychologteal data, leaving 31 PD patients with longitudinal data. n adltion, 13 healthy conttols had longitudinal neuropsychological measurements, Table summarizes the clinical charaeteristis ofall participants. At the frst time pot, PD patients didnot dif for the healthy conteols on age (t (68) ~ 0.403; p = 0.688) and sex OCC, 65) ~ 0.098; p ~ 0.761), Furthermore, PD patients shoved lower global cognitive performance than healthy controls (63) = 2.62; p = 0.012). The groups studied «rose-sectionally and longitudinally (both in case of the controls and PD patients) didnot difer significantly with respect to age, se, disease etme: ina 282020) 102465 duration, LEDD, global cognitive function, and neuropsychological function (all seven tests were compared separately statistics no shown) PD patients scored significantly lower than healthy controls on all but one of the specific neuropsychological tests (Stockings of Cam- bridge, an executive test, which was retained inthe average executive = seore) (Table | and Fig. 34). Longitudinal assessment of clinical paremeters in PD patients showed a decrease in global cognitive performance (1(30) = 2.25; p 0.034) and in motor performance ((30) = 5.31; p< 0.001), and In crease in LED (30) = 7.28; p = 0.001). PD patients longitudinally declined on three neuropsychological tests, spatial working memory (between errors (30) ~ 5:43: p < 0.001), spatial span length (30) 2.95;p = 0.006) and pattern recognition memory test (correct responses, £(30) = 2.475 p = 0,020) (Table | and Pig. 38). The raw scores of neu- ‘opsychological performance can be found in Supplementary Table 3 2.2, State funconal connectiviy Global funtional connectivity did not significantly change between baseline and follow-up (Pg. 4A; 0.270 at baseline 0.281 at FUL (30) 0.734; p = 0.469)). However, longitudinal changes in (normalized) functional connectivity between RSNs did show significant changes, a5, summarized in Fig, 4B, Overtime, functional connectivity with the rest of the brain significantly increased for the DMN, DGM and VAN, and significantly decreased for the visual network, No changes were observed for DAN, FPN and SMN. 33. Bective cognition and static FC ‘The erosssectional regression analysis with ecutive tests average of four executive function tests) as dependent variable identified FC of DGM withthe rest ofthe bain as important, as well as DAN with the rest Comin 15) Pola =80) oe pla ie plat 2 ‘Ase (yea) 4.4 (65, range (65.5 (6.27, range sO) (6.2 (562, range 54-77) 69.1 (604, ange 79) 37-0) ‘ras. ms 2a eo) 27 (858) saa eazy EDD ct done me 351543) srsca29) son 485) Wg menery ow cass osta29 6 2v0r we : ars) ” Spat pling oo ossiso) 9 osocon 3 eowse0) 4 Mesery oo -os86 cus Hoss 5 cise) 12 oor peeves soain 6 sas 10 Laas oat heey ow osstay 1s cseasn 8 os? os) ‘alors are expressed as mca (SD) ule otherwie indicted, PD, Pakinon's disease; ISCED, International Standard Clastifiation of Education; UPDRS, Unified Parkinson's Dteae Rating Sale motor rings, LED, Levedope Equivalent Dally Dose; Working memory, Spal Working Memory (between err): Spatial plannlng, Stockings of Cambdge Set sig, lta Extra Dimensional Sth, Memory Pater Recognition Memoey (correc sponses} Moca: perseveration, Vienna perseveration ask edundancy; Fluency, Seman ene (no. of words): n/, no appllable. * sguflcaat dference la cos-seeuona analyss(@ 0.05), slgaiant difereace ip longtuiza anaiyss(@ = 0.05).2 15) numberof subjects with a Z-acore <1, below average 1 Bon etme: ina 282020) 102465 A) Cross-sectional analysis 8B) Longitudinal analysis sm Fire ne pin 1 Second tie point epee LEG o Me od ae 8) Crontaetional analysis of neurapeychologclpesformance of 50 Pakinuon’s eas (PD) patients, compared with a ference group of healthy controls (8 ~ 15). ‘PD patente sored sgificantly lower on sx out of even nearopsychologicl tests p< 0,05). 3B) Longitudinal nays of neuropayehalgical performance of 51 PD pases Zacores were based on erossetonal compass between PD patient and a group of ely conto fist time point ~ 15; second time point, n— 13). Overtime, performance ofthe PD. group sigiiany woreened forth out of seven test pe 005, Working memory, Spatial Working Memory (between eos); Spatial lansing, Stockings of Cambridge; Set siting, Inca Extra Dimensional Seshiting: Memory, artera Recogaiion Memory (orrect responses); Motor perseveration, Vienna perseveration tsk redundancy; Heo Semantic Flueney (00. of word A B mm First = Second Functional connectivity First Second DAN DMN FPN SMN DGM VAN Visual ig. 4 Longitudinal changes in fonconal connectivity. 1) Global neon connectivity overtime between theft nl second ime point (a ~ 31 Parkinson's disease (PD) patents), whic was not significa ilferent 3) Panctionlconnetvity (emai) between restingstate networks andthe et ofthe bran. DANGtest 30) = 1}2p 0.272; DMNGret 30) ~ 293} ~ 0.006 FPN et 30) 0.825) O415;SMN-rest (20) = 1.37 p 0.180; DGMes 30) —2.33p- 0.027; VAN-es 30) ~L013}p = G05; Vinuliest 30) 2683 p= 0001. (p< 005), DDAN, dorsal attention network; DMN, default-mode network; FPN, frontoparetal network; SMN, sensorimotor neswork;DGM, dee gray matte; VAN, vena atter- of the brain, The subsequent regional analysis revealed a negative cor- correlated with DAN-AN conneetivity changes (Table 2). relation between exective functioning at baseline and DGM-FPN FC indicating higher FC in patients with dysfunction (see Tabie 24 for Beecubve cogrdton end dynam: F statistical values) However, patients with impaired executive fanee rion and manic Honing (e-value < 15) did not show significantly higher FC than non-_nly connections showing effects of static connectivity were also impted patients (rvalue > 1.5; Pig. 5). The DAN did not show addi-exojored using dynamic connectivity, chus only RSN interactions con- tonal erose-network corelations with exeeutive functioning. cerning the OGM and the DAN. A regresion analysis with executive tests ‘Next, longitudinal changes in cognitive funerion were correlated 35 dependent variable showed @ positive correlation with DGM-FPN ‘withlongituinal BSN FC measurement. Oversime, change inexecutive Yynanie #C and @ negative conration with DAN-VAN dynamic Fe funetioning eorelated positively with DGM-DMN connectivity changes, Buh imtrations shoned sgnieasty higher dynamic connectivity wheces it correlated negatively with DGM-FPN and DGMLDGM con- tare nll models (DCMT eal dna Oot = 0082, surged nectivity changes CIsble 2). In addition, exeetive cognitive decline f'ggg't .036; #8) 347; > 0.008, DAN-VAN. rel dete 0.593 was negatively correlated with DAN-SMN FC, while it was positively 9.932, surugete data 0.588 2 0.033; 48) ~ 267, p= 0.010) 1 Bon etme: ina 282020) 102465 near Herarchen regression analyses were performed witha backward elimination method pr copiveovicome memire, correo for agh sx edicaton, dress duration and LEDD, First, lobalfnetonalconnetvity betwee linia REN and the resto the bran was coneated with cogitve funtion (scan conn). Next nteretions between indivi RSW were asta (basd on the significant coreations on wthoe-ran interactions) and corcated wih cognitive fneton (hr column. Note tha mtiple feetional connections were tested within the same regression mde values ae dapaye inthe econ column asa ingle ae SN, reatngatate network DGM, Deep Gry Matter; DAN, Dorsal Attention Network; FPN, Fontoprital Network; PRM, Patter Recognition Memory; VP, Vina Peseveration Tes; VAN, Venza tenon Network. "average of Sptisl Working Menory, Spatial Span, Stockings of Cambridge and ita Extra Dimensional Setting * significant corlation (p< 0.05). cute tunton otr perseveration Fig. 5: Pos show ean foal connec on rotifers Sond evo of ee os * Sey eer bh As depied o e ee i teen he ep yer (DH) a ee : : Cute ce PR) ns mse Foz S az higher for PD patents impated(evalue © 13) on =z 2 ‘executive tests ((0(48) ~ 1.651 p = 0.106). As depic- g Sele tal on the righ FC between the der ateton Por 2 retwork (DAN) and venta attention network (VAN) ox | wre auitanty owe fr Po pens pied sas “ eso te peverton 2383p Nonimpares impaired Non-impaind impaled oats. "near Herarchea regression analyses were performed witha backward elimination method per comiveovicome menre, correo forages edicaton, Heese ution, and LEDD. Selection ffnetonal connectivity intractons between RSNs was bacon RSNsdeemes toe important based on ah, Note ha mole Funetional connections were tested within the sme repression model R? values are dlepayed nthe second column a ingle vale for eneh separate regression No significant longitudinal coreelation with dynamic RSN in tersetions was found, regression analysis, dysfunctional motor perseveration correlated with Tower DAN-VAN FC only (Tsbie 24). In accordance, DAN-VAN FC was significantly different between patients with motor perseveration (2- value < -15, hence impaired) and nomimpaired patients (value > apn colons wee observe for mot eneveron motor perseveration were investigated. Inthe cross-sectional analysis, in, only connections showing effets of state connectivity were vislospatalfenetlon and semanti Tuency dd not coreate with FCof algo Saplved me a also explored using dynamic connectivity, thus only RSN interactions any of the RSNs and these neuropsychological tests were excluded for further analysis. Dysfunctional motor perseveration correlated with lower FC betvcen the DAN and the rest ofthe brain, In the subsequent concerning the DAN were correlated with motor perseveration. Dysfunctional motor perseveration correlated with higher DAN.SMN, etme: ina 282020) 102465 [near Hlerarchenl regression analyses were performed with backward elimination method pes cogaiv outcome mesure ration, nd LEDD, Selection offentonal connectivity incracion ten RSNs was eran RSW deemed obeimportant based on Tie 2, Note tha mpc come forage, ex, education, disease Functional connections were tested within the sme repression model R? valves are cepayed inthe scond column a ingle value fr each separate regression dynamic FC (Table 2c), DAN-SMN showed significantly higher dynamle (0.058, surrogate connectivity than the null model (real data 0.555 data 0.545 4 0.051; (48) = 2.76; p tion was found between mot perseveration and dynamic RSN 4, Discussion ur results showed that PD patients displayed deficits in almost all cognitive domains, which worsened overtime. Furthermore, we found relative increases in functional connectivity with te rest ofthe brain for the DGM, DMN and VAN and decreases forthe visual network ‘Next, we observed a cross-sectional coreation berween executive dysfunction and increased stati, but decreased dynamic DGM-FPN FC. Longitudinaly, only tate PG changes were related to decline In exec- utive function, and this ineluded RSN-Interations in adtion to DGM- FPN, especialy highlighting the role of the DAN. The DAN was also ‘elated to cros-setional deficits in motor perseveration, We found that higher stati connectivity between DGM and the FPN was related to poorer executive functioning, possibly reflecting fron- tostriatal dysfunction with a dopaminergic basis. However, executive dysfunction didnot correlate withthe LEDD (9 =-0.179;p — 0.179), and DGM-FPN connectivity did not correlate with motor dysfunction () 0.049; p= 0.688), nor withthe LEDD (9 =-0.059; p= 0.666). Inspiteof the lack of these correlations, and the fact that we added LED as co- rein our analyses, in line with previous studies, we hypothesize ‘that dopaminergic therapy could still have impacted the corelations served, by lowering frontostriatal FC (Kwai tal, 2010) and simul- ‘taneously having beneficial effects on executive functioning (G «3, 1988). Sinee individual patents may have different baseline levels of dopamine they may exhibit a diferential sensitivity to the postive and negative effects of dopaminergic medication, which may interfere With LEDD eorreations (Cools, 2006). Interestingly, increases in FC within the basal ganglia (Srewerylckrolikowsk! et al, 2014) and sensorimotor network (Esposio et 91, 201) have been observed upon a dopaminergic challenge in PD patients. Hence, it remains to be deter- mined whether a Ayperdopaminergic stat, in line with the socalled ‘dopamine overdose hypothesis’ (Coos, 2006), ora hypodopaminergic state explains the correlations observed. This knowledge would be a rerequiste for better clinical management of executive dysfunction Our longitudinal regression models suggest that not only FC of the [BGM oF FPN, but also FC of the DAN is involved in executive dysfunc- Hon. Over time, performance on executive tests correlated with FC changes not restricted tothe DGMCFPN, involving a negative coreltion with DAN-SMN connectivity and a postive correlation with DAN-VAN, connectivity. The DAN snd VAN are attention networks, involved in reorienting attention towards salient still inthe healthy brain (Fox (al, 2006), Previous work has shown that as PD progresses, such networks may become involved to maintain optimal performance on executive tests (Bosse, 2011), as alo suggested by studies on (healthy) aging (Reutertorens, 2002/08/30). A recent study has confirmed that reduced FC between the DAN and insular bea regions (the latter being ‘rt of the VAN) is associated wih worse performance on attention/ execative tasks (Saggio et al, 2015), Furthermore, in another study cognitively impaired PD patients had reduced insular dopaminergic D2 receptors, which was associated with worse exceutive functioning (Christopher etal, 2014), Apart from executive dysfunction, our study investigated several additional cognitive tess, of which only motor perseveration showed relations with FC of brain setworks. Motor perseveration (when one is instructed to demonstrate random motor behavior) has been demon- strated to occur in early stages of PD (Stoffers et a, 2001) and was related to lower DAN FC, especially DAN-VAN connectivity. The ge eration of random motor behavior is considered to involve retention of| information, suppression of habitual responses, and switehing of peo- dvetion strategies (Ns 2008), al of which are attention demanding, processes, ‘Therefore, lower connectivity between the aention necworks may play a roe in perseverative tendencies such as ‘motor perseveration. The primary cause of attentional network dysfunetion remains unclear hower, but is thought to revolve around cholinergic deficits (Sarter ets, 201), In addition to static FC, we have added dynamic FC, which has ‘recently been shown to be important for cognitive functioning in PD (Engels ct a, 2018; Fiorenzato ct al, 2019). Accumulating evidence indicates that functional connectivity becween brain regions is onsta tionary and aliemates berween periods of low and high functional coupling overtime (Hutchison ct al, 2013) In our stud, dynamic FC was expressed as the variability (coefficient of variation) of funetional connectivity over a numberof time windows, in which more varability/ fhuctuaion represents higher dynamic FC. The results of ou dynamie FC analysis didnot simply mirror static FC results, as demonstrated by «correlations involving DAN-VAN (executive dysfunetion) nd DAN.SMN, (Wienna perseveration task) dynamic FC. At che same time, the role of DGNLFPN dynamic FC in executive dysfunetion was confiemed, although in this case higher DGMFPN dynamics seems to be beneficial rather than disadvantageoos for cognitive functioning. Together, these results support the notion that the balance of excitation and inhibition, a fundamental feature of brain network activity (Dehghan et al, 2026) ‘may be disturbed in PD, thereby leading to dynamic FC changes. Our longitudinal analysis adds to the recent (crose-

    You might also like