Interactions between functional networks in Parkinson's disease mild cognitive impairment

1
Vol.:(0123456789)
Scien i ic Repo s | (2023) 13:20162 | h ps://doi.o g/10.1038/s41598-023-46991-3
www.na u e.com/scien i ic epo s
In e ac ions be ween unc ional
ne wo ks in Pa kinson’s disease
mild cogni i e impai men
Manuel Delgado‑Al a ado
1,2,3,8, Vicen e J. Fe e ‑Galla do
4,8, Ped o M. Paz‑Alonso
4,5,
Césa Caballe o‑Gaudes
4 & Ma ía C. Rod íguez‑O oz
6,7*
The s udy o mild cogni i e impai men (MCI) is c i ical o unde s and he unde lying p ocesses o
cogni i e decline in Pa kinson’s disease (PD). Func ional connec i i y (FC) dis up ions in PD‑MCI
pa ien s ha e been obse ed in se e al ne wo ks. Howe e , he unc ional and cogni i e changes
associa ed wi h he dis up ions obse ed in hese ne wo ks a e s ill unclea . Using a da a‑d i en
me hodology based on independen componen analysis, we examined di e ences in FC RSNs
among PD‑MCI, PD cogni i ely no mal pa ien s (PD‑CN) and heal hy con ols (HC) and s udied hei
associa ions wi h cogni i e and mo o a iables. A signi ican di e ence was ound be ween PD‑MCI s
PD‑CN and HC in a FC‑ ai comp ising senso imo o (SMN), do sal a en ion (DAN), en al a en ion
(VAN) and on opa ie al (FPN) ne wo ks. This FC‑ ai was associa ed wi h wo king memo y, memo y
and he UPDRS mo o scale. SMN in ol emen in e bal memo y ecall may be ela ed wi h he
FC‑ ai co ela ion wi h memo y de ici s. Meanwhile, wo king memo y impai men may be e lec ed
in he DAN, VAN and FPN in e connec i i y dis up ions wi h he SMN. Fu he mo e, in e ac ions
be ween he SMN and he DAN, VAN and FPN ne wo k e lec he in e wined decline o mo o and
cogni i e abili ies in PD‑MCI. Ou indings sugges ha he memo y impai men s obse ed in PD‑MCI
a e associa ed wi h educed FC wi hin he SMN and be ween SMN and a en ion ne wo ks.
Pa kinson’s disease (PD) is cu en ly iewed as a complex neu odegene a i e disease ha in ol es se e al ce eb al
a eas and neu o ansmi e sys ems. Al hough mo o symp oms a e he hallma k o clinical diagnosis, non-mo o
symp oms play a p ominen ole in he cou se o he disease. Fo many pa ien s, he mos dis up i e non-mo o
mani es a ions o PD a e cogni i e impai men and demen ia, which each a long- e m p e alence o up o 80%1.
Mild cogni i e impai men (MCI) is highly p e alen in PD (PD-MCI) (mean 26.7%; ange 18.9–38.2%)2 and
is known o be a isk ac o o PD wi h demen ia2,3. Ye , PD-MCI is a a he he e ogeneous condi ion, encom-
passing se e al sub ypes o cogni i e decline, hough o con e di e en isks o p og ession o demen ia4. This
he e ogenei y e lec s he di e se neu opa hological p ocesses o PD-MCI5 which ha e been mos consis en ly
associa ed wi h he p og essi e accumula ion o α-synuclein, amyloid-β and pa hologic au species, pa icula ly
in on al, empo al and cingula e a eas6.
PD esea ch has led o he cha ac e iza ion o wo cogni i e de e io a ion p o iles: a on o-execu i e dys unc-
ion mainly d i en by dopamine gic loss and mani es ing as de ici s in lexibili y, planning, wo king memo y
and ein o cemen lea ning; and a choline gic co ical dys unc ion leading o memo y de ici s and demen ia
(i.e., amnes ic)7. Howe e , hese wo p o iles do no always cap u e he he e ogeneous cogni i e de e io a ion o
PD-MCI. Ins ead, a ange o cogni i e impai men s a ises om he p og essi e and he e ogeneous in ol emen
o dis inc neu al ne wo ks, modula ed by dopamine gic and choline gic neu o ansmi e sys ems8.
A wide body o e idence suppo s he exis ence o a hie a chically o ganized es ing s a e ne wo k (RSN)
sys em in which associa ion ne wo ks se e highe -le el cogni i e unc ions9. He e, we use he se en majo
RSNs iden i ied in Yeo e al.10 namely he isual (VN), senso imo o (SMN), do sal a en ion (DAN), en al
a en ion (VAN), limbic (LN), on o-pa ie al (FPN) and de aul mode (DMN) ne wo ks, plus he s ia um. As
OPEN
1Neu ology Se ice, Hospi al Sie allana, 39300 To ela ega, Spain. 2Neu odegene a i e Diso de s Resea ch
G oup, Uni e si y Hospi al Ma qués de Valdecilla-IDIVAL, 39008 Can ab ia, Spain. 3Cen o de In es igación
Biomédica en Red de En e medades Neu odegene a i as (CINERNED), Mad id, Spain. 4Basque Cen e on
Cogni ion B ain and Language (BCBL), 20009 San Sebas ian, Spain. 5Ike basque, Basque Founda ion o Science,
48009 Bilbao, Spain. 6Neu ology Depa men , Clínica Uni e sidad de Na a a, A . de Pío XII, 36, 31008 Pamplona,
Na a a, Spain. 7Na a a Ins i u e o Heal h Resea ch (IdiSNA), 31008 Pamplona, Spain. 8
These au ho s
con ibu ed equally: Manuel Delgado-Al a ado and Vicen e J. Fe e -Galla do. *email: mc o oz@una .es
2
Vol:.(1234567890)
Scien i ic Repo s | (2023) 13:20162 | h ps://doi.o g/10.1038/s41598-023-46991-3
www.na u e.com/scien i ic epo s/
desc ibed below, ou o hese RSNs (SMN, DAN, VAN and FPN) ha e been implica ed in he pa hology seen
in PD-MCI pa ien s.
The FPN is a c i ical componen in execu i e unc ion and wo king memo y11, and i s main hubs su e om
dopamine dene a ion in PD12. Meanwhile, a en ional o ien ing is d i en by bo h he en al a en ion ne wo k
(VAN)13 and he do sal a en ion ne wo k (DAN)14, which ha e also been associa ed wi h PD-MCI15–17. The
o ien ing a en ional sys em is associa ed wi h he choline gic neu ons o he basal o eb ain, and he choline gic
neu o ansmi e loss p esen in PD-MCI subjec s has been linked o a en ion de ici s in his sys em11,18. Finally,
he SMN mainly se es p ima y mo o unc ions, al hough i also coo dina es wi h o he cogni i e ne wo ks.
Al hough he ole o he SMN in PD-MCI is an ongoing discussion. On one side, SMN dis up ion could be a
e lec ion o mo o impai men s in PD-MCI pa ien s, which is e lec ed in hei sco es in he "mo emen diso -
de s socie y uni ied Pa kinson’s disease scale” (MDS-UPDRS)2,19. On he o he side, SMN dis up ions ha e been
ela ed o impai ed senso y in eg a ion o mo o unc ion in PD20. The SMN has also been associa ed wi h he
on o-execu i e dys unc ion cogni i e p o ile obse ed in PD, mainly impac ing he FPN. Fo example, he SMN
plays a c ucial ole in e bal sho - e m memo y, coo dina ing wi h he on o- empo al a eas21. Mo e e idence
o he SMN cogni i e ole a ises om dis up ions in his ne wo k in associa ion wi h cogni i e impai men in
PD and e en in Alzheime ´s disease (AD)22–24.
He e, we ocus on al e a ions in he RS unc ional connec i i y (FC) be ween hese RSNs in PD-MCI. Res ing
s a e unc ional connec i i y (RSFC) can be de ined as a signi ican empo al co ela ion be ween unc ionally
ela ed b ain egions in he absence o any s imulus o ask25. Me hodological app oaches o he analysis o RSFC
ha e employed RS unc ional magne ic esonance imaging ( MRI). S udies o RSFC o he diseased b ain ha e
adop ed a a ie y o me hods, howe e , mos a emp o de e mine whe he he e is an inc ease (i.e., highe
synch onici y) o a dec ease (i.e., lowe synch onici y) wi hin o be ween RSNs. Such s udies a e imp o ing ou
unde s anding o he unc ional b ain changes unde lying PD-MCI. Con en ional me hods o analyzing da a
in RSFC MRI s udies canno e ec i ely cap u e and di e en ia e he sha ed unde lying ac o s o con lic ing
p ocesses ha a ise om dis inc unc ional pa e ns obse ed in he b ains o heal hy and diseased indi idu-
als, using a da a-d i en app oach. In he p esen wo k, we sough o use a ecen ly implemen ed me hodology
ha is ideal o examining he in e ac ion be ween mul iple b ain sys ems o ne wo ks a he FC le el. These FC
pa e ns (i.e., FC- ai s) ep esen di e en unc ional mechanisms be ween ne wo ks. A e wa ds, we associa e
hose be ween-ne wo k FC- ai ea u es wi h he cogni i e pe o mance es s used o p ope diagnosis and
cogni i e sub ype classi ica ion in PD-MCI in line wi h Mo emen s Diso de s Socie y (MDS) c i e ia le el II26.
Al hough ew s udies ha e used MDS c i e ia le el II27, i is hough ha hese c i e ia gi e be e diagnos ic
sensi i i y and speci ici y han c i e ia le el I26. This is why in he p esen wo k we use c i e ia le el I o in e
which MCI cogni i e domain is associa ed wi h he FC- ai s. The da a-d i en me hodology ha we used he e
is e med ConnICA, which implemen s Independen Componen Analysis (ICA) o ex ac obus independen
FC- ai s om a se o indi idual FC ma ices28,29. An FC- ai ep esen s a FC pa e n be ween es ing s a e
ne wo ks in ou FC ma ices wi h a possible ole in cogni i e, mo o , o clinical impai men s.
We hypo hesize ha cogni i e unc ion de ici s in PD-MCI would be d i en by al e ed pa e ns o in e -
connec i i y be ween he b ain ne wo ks. Speci ically, he FC in ne wo ks such as he SMN, FPN, DAN and
VAN would e lec such cogni i e decline and would, he e o e, cap u e he FC di e ences be ween PD-MCI and
PD-CN. He e, we aim o ex ac independen FC- ai s by applying connICA in PD-MCI pa ien s, cogni i ely no -
mal PD pa ien s (PD-CN) and heal hy con ols (HC), o compa e and disen angle key FC- ai s ha a e al e ed
in PD-MCI. Fu he mo e, we expec hese speci ic connec i i y pa e ns o be associa ed wi h pe o mance on
a en ion, execu i e unc ion, memo y, language, and isuospa ial asks.
Resul s
42 PD pa ien s (mean age 70.27 ± 6.32yea s, 15 emales) (23 PD-MCI and 19 PD-CN) and 21 HC (mean age,
67.52 ± 6.97yea s, 9 emales) emained om he 71 ini ial subjec s a e da a quali y con ol o mo ion o a i-
ac s in he es ing-s a e MRI images. Table1 depic s he sociodemog aphic and clinical cha ac e is ics o he
g oups. PD pa ien s ha e highe sco es han HC on he HADS scale, bu his di e ence is no clinically ele an .
PD-MCI pa ien s we e olde han PD-CN pa ien s (P = 0.017) and HC (P = 0.018), had ewe yea s o educa ion
han PD-CN (P = 0.003) and HC (P = 0.001) and highe sco es in he MDS-UPDRS-III han PD-CN pa ien s
(P = 0.028). Cogni i e assessmen s showed PD-MCI pa ien s p esen ed a mul idomain de ici wi h highe de ici s
in a en ion and wo king memo y, execu i e unc ion, and memo y domains han in language and isuospa ial
abili ies (see Table2 o coho cogni i e di e ences and TableS.1 o PD-MCI speci ic domain de ici s).
P obabilis ic-ICA was ca ied ou ex ac ing 65 independen FC- ai s and hei co esponding subjec -
speci ic connICA weigh s. Among hese, only one FC- ai showed a s a is ically signi ican g oup e ec (Table3).
Pos -hoc pai wise compa isons be ween g oups showed ha his FC- ai (Fig.1) di e en ia ed PD-MCI om
PD-CN and HC.
The FC- ai showed a la ge di e ence be ween PD-MCI and bo h PD-CN and HC (Table3). As depic ed
in he co esponding ci cula g aph in Fig.1A, he FC- ai connec i i y is mainly d i en by in a-hemisphe ic
and in e -hemisphe ic connec ions be ween bila e al egions o he SMN (i.e., soma osenso y, p ima y mo o ,
p emo o and supplemen a y mo o a eas) and he DAN and VAN. I also includes, o a lesse ex en , connec ions
be ween egions o he VAN, FPN and DMN ne wo ks, mainly la e alized o he le hemisphe e. As depic ed
in he nodal s eng h map (Fig.1C), soma omo o egions (i.e., bila e al p ecen al and pos cen al gy i, as well
as supplemen a y mo o a eas) a e he hubs wi h he s onges connec ions in his FC- ai . No ably, he co -
esponding connICA weigh s exhibi ed a s ong associa ion wi h a en ion z-sco e and memo y z-sco e and o
a lesse ex end wi h he language z-sco e, and MDS-UPDRS-III as e lec ed in he F- alues (Fig.1B, Table3).
3
Vol.:(0123456789)
Scien i ic Repo s | (2023) 13:20162 | h ps://doi.o g/10.1038/s41598-023-46991-3
www.na u e.com/scien i ic epo s/
Discussion
The p esen s udy aimed o in es iga e FC changes in PD-MCI ha occu in he majo RSN. Wi h he p esen
app oach we ha e been able o iden i y he cha ac e is ic di e ences o b ain connec i i y be ween PD-MCI and
cogni i ely no mal g oups (HC and PD-CN). Using he connICA amewo k we we e able o ex ac independen
connec i i y ai s om a se o indi idual unc ional connec omes ha ep esen in e ac ions be ween mul iple
b ain sys ems o ne wo ks a he FC le el. Ou indings e ealed ha PD-MCI pa ien s exhibi ed a dis inc i e
FC- ai cha ac e ized by unc ional coupling o he SMN wi h he DAN, VAN and FPN. We also linked he FC-
ai o cogni i e and beha io al es s used o p ope diagnosis and cogni i e sub ype classi ica ion in PD-MCI
Table 1. Sociodemog aphic and clinical cha ac e is ics o he coho . HADS Hospi al anxie y and dep ession
scale, MDS-UPDRS Mo emen diso de s-Uni ied Pa kinson’s Disease Ra ing Scale, LEDD Le odopa
Equi alen Daily Doses, i In e qua ile ange. As e isks indica e signi ican di e ences be ween g oups.
PD-CN (n = 19) PD-MCI (n = 23) HC (n = 21) P- alues PD-CN s. PD-MCI PD-CN s. HC PM-MCI s. HC
Age (yea s) 66.58 ± 7 72.22 ± 5.43 68.65 ± 3.9 1.5 × 10−2 * (ANOVA) 5.4 × 10−3*3.3 × 10−1 5.0 × 10−3*
Gende (male) 16 (80%) 12 (52.2%) 11 (52.4%) 1.1 × 10−1 (χ2)
Educa ion (yea s) 13.84 (i 9.5–20) 8.96 (i 6.5–10.5) 12.85 (i 9–16) 1.8 × 10−3 (K uskal
Wallis) 3.0 × 10−3*5.2 × 10−1 5.0 × 10−3*
Disease du a ion (yea s) 6.789 (i 4–10) 9 (i 4–13) – 3.4 × 10−1 (Mann Whi -
ney U)
LEDD 947.77 ± 433.33 1089.86 ± 507.31 –3.5 × 10−1 (T)
MDS-UPDRS III 18.63 ± 8.51 26.36 ± 12.07 –2.8 × 10−2* (T)
MDS-UPDRS-III Axial
sco e 1.684 ± 1.765 2.565 ± 1.805 1.1 × 10−1 (T)
PDQ-39 20.74 (i 10.5–26.5 33.391 (i 20.5–45.00) – 2.2 × 10−2* (Mann Whi -
ney U)
HADS dep ession 2.316 (i 1–3.5) 4.217 (i 2 – 6) 1.7 (i 1 – 2.25) 3.0 × 10−3* (ANOVA) 3.3 × 10−1 2.0 × 10−3 * 3.4 × 10−2*
HADS anxie y 3.053 (i 1.5–4) 4.869 (i 2 – 7.5 3.8 (i 1–5) 1.2 × 10−1 (ANOVA)
HADS o al 5.368 (i 2–8.5) 9.087 (i 5 – 12) 5.5 (i 2 – 7) 1.6 × 10−2* (ANOVA) 9.1 × 10−1 2.3 × 10−2*2.4 × 10−2*
H&Y s age 2.0526 (i 2 –2.75 2.413 (i 2– 3) – 6.3 × 10−2 (Mann Whi -
ney U)
Table 2. Cogni i e cha ac e is ics o he coho . Tes s z-sco e g oup means and s anda d de ia ions. Z-sco es
om es s wi hin he same cogni i e domain a e combined in one z-sco e ha ep esen s he whole domain
(span in e se + digi and symbols = A en ion and wo king memo y). ANOVAs we e compu ed o g oup mean
compa isons, and - es s we e used o be ween-g oup compa isons. As e isks indica e signi ican di e ences
be ween g oups. RAVLT Rey Audi o y Ve bal Lea ning Tes , ROCF Rey–Os e ie h Complex Figu e, VOSP
Visual Objec and Space Pe cep ion Ba e y, MMSE Mini Men al S a e Examina ion, MOCA Mon eal
Cogni i e Assessmen .
PD-CN (n = 19) PD-MCI (n = 23) HC (n = 21) ANOVAP- alues PD-CN s. PD-MCI PD-CN s. HC PD-MCI
s.HC
MOCA 25.947 ± 3.407 20.348 ± 3.4 26.4 ± 2.891 2.0 × 10−7*2.0 × 10−5*6.6 × 10−1 1.5 × 10−6*
MMSE 29.053 ± 1.129 27.13 ± 1.792 29.3 ± 0.865 1.7 × 10−6*2.3 × 10−4*4.5 × 10−1 1.2 × 10−5*
A en ion and wo k-
ing memo y − 0.405 ± 0.560 − 1.536 ± 0.334 0.050 ± 0.905 5.0 × 10−11*5.0 × 10−10*6.8 × 10−2 1.0 × 10−8*
Span in e se − 0.343 ± 0.598 − 1.099 ± 0.645 0.037 ± 1.040 4.0 × 10−5* 3.0*10−3*1.7 × 10−1 8.3 × 10−5*
Digi and symbol − 0.346 ± 0.704 − 1.973 ± 0.552 − 0.063 ± 1.073 1.2 × 10−11*2.3 × 10−10*1.7 × 10−1 7.0 × 10−10*
Execu i e unc ion − 0.310 ± 0.731 − 2.755 ± 1.363 0.037 ± 0.920 1.0 × 10−12*1.0 × 10−8*2.0 × 10−1 1.0 × 10−9*
T ail making es − 0.284 ± 1.207 − 4.660 ± 2.533 0.030 ± 0.980 1.0 × 10−12*5.0 × 10−8*3.8 × 10−1 1.0 × 10−9*
Phonemic luency − 0.317 ± 0.687 − 1.053 ± 0.682 0.045 ± 1.074 2.0 × 10−4*1.0 × 10−3*2.2 × 10−1 2.0 × 10−4*
Memo y − 0.329 ± 0.914 − 1.35 ± 0.752 − 0.179 ± 0.712 2.0 × 10−6*3.0 × 10−4*2.9 × 10−1 8.0 × 10−7*
RAVLT − 0.766 ± 1.288 − 1.708 ± 1.009 − 0.018 ± 1.038 2.0 × 10−5*1.1 × 10−2*5.0 × 10−2 3.0 × 10−6*
ROCF 0.107 ± 0.782 − 0.992 ± 0.924 − 0.075 ± 0.857 1.0 × 10−4*1.0 × 10−4*4.9 × 10−2 2 × 10−3*
Language 0.089 ± 0.838 − 1.175 ± 1.140 − 0.175 ± 0.872 1.0 × 10−4*3.0 × 10−4 3.41 × 10−1 3.0 × 10−3*
Seman ic luency 0.053 ± 0.950 − 0.748 ± 1.035 − 0.179 ± 1.010 3.3 × 10−2
Bos on naming 0.125 ± 1.116 − 1.603 ± 2.024 − 0.171 ± 0.957 6.0 × 10−4*2.0 × 10−3*3.8 × 10−1 6.0 × 10−3*
Visuospa ial abili ies − 0.656 ± 1.319 − 1.589 ± 1.875 0.032 ± 0.845 2.0 × 10−3*7.5 × 10−2 5.8 × 10−2 9.0 × 10−4*
VOSP objec s − 0.377 ± 0.894 − 1.433 ± 1.297 0.065 ± 0.835 5.0 × 10−5*4.0 × 10−3*1.2 × 10−1 7.0 × 10−5*
VOSP nume ic − 0.936 ± 2.305 − 1.745 ± 3.033 − 0.002 ± 1.119 5.8 × 10−2
4
Vol:.(1234567890)
Scien i ic Repo s | (2023) 13:20162 | h ps://doi.o g/10.1038/s41598-023-46991-3
www.na u e.com/scien i ic epo s/
in line wi h he MDS c i e ia le el II. This FC- ai was associa ed wi h high-cogni i e unc ions such as memo y
and wo king memo y as well as mo o unc ion. This ex ac ion was done in a comple ely da a-d i en app oach
by maximizing subjec s’ iden i iabili y29, and applying ConnICA28.
This dis inc i e FC- ai indica es ha al e ed unc ional coupling o he SMN in coo dina ion wi h he
DAN, he VAN and he FPN may be a majo indica o o PD-MCI. The di e en ial FC pa e n o his FC- ai
was associa ed wi h a en ion and wo king memo y, memo y, and language asks, sugges ing cogni i e de ici s
in mul iple domains. Impo an ly, hese we e he domains ha we e mos nega i ely a ec ed in ou PD-MCI
popula ion (Table S.1), ein o cing he ele ance o his ai in he ea ly phases o cogni i e decline in PD. These
esul s a e consis en wi h p e ious s udies showing ha PD-MCI pa ien s displayed dis up ed in e -connec i i y
in he DAN o he SMN and connec i i y educ ions be ween he SMN and he cogni i e con ol ne wo k14,23,30,31.
Al hough he SMN is no conside ed o be an associa ion ne wo k25, egions belonging o his ne wo k, such as
he soma osenso y co ex, a e c ucial o highe o de p ocesses (e.g., e bal c ea i i y)32. Indeed, he inciden al
memo y es (i.e., ROCF) and he delayed ecall sub es o he RAVLT, bo h used o assess he memo y cogni-
i e domain, ha e a s ong senso imo o componen . The ROCF asks subjec s o eplica e a p e iously obse ed
igu e, and he RAVLT is a e bal sho - e m memo y es . The e o e, he co ela ion be ween ou FC- ai and
pa icipan s’ memo y sco es on hese es s could e lec memo y ecall de ici s ela ed o SMN dis up ions21,33,34.
The a en ion and wo king memo y assessmen s ha also co ela e wi h he obse ed FC- ai a e ep e-
sen a i e o wo king memo y35–37. FPN plays an impo an ole in coo dina ion wi h SMN in spa ial wo king
memo y ask pe o mance35,38. Mo eo e , al hough he FPN is in ol ed in bo h execu i e unc ion and wo k-
ing memo y, cogni i e neu oscience s udies sugges ha PD mainly a ec s he on opa ie al a eas in ol ed in
wo king memo y whe eas execu i e a eas a e p ese ed39,40. Likewise, he obse ed FC- ai was associa ed wi h
he wo king memo y es s co esponding o he a en ion and wo king memo y domains bu was no associa ed
wi h es s examining execu i e unc ion. The in ol emen o he DAN and VAN is also consis en wi h p e ious
esea ch indings. Fi s , a nega i e connec i i y be ween he SMN and DAN has been showed o ad e sely a ec
heal hy subjec s’ pe o mance on asks equi ing a en ion and wo king memo y41. Second, he DAN and VAN
ne wo ks, in which choline gic neu ons play a c ucial ole, ha e been implica ed in a en ion o ien ing asks11.
Bo h ne wo ks a e ypically comp omised in PD-MCI and PD wi h demen ia42–44.
Ou FC- ai con ains b ain egions also ound in MCI wi h AD pa ien s. In AD-MCI he main a ec ed b ain
a eas include empo al, pos e io pa ie al and hippocampus, which ypically show educ ions in FC o me abolic
consump ion45,46. The PD-MCI FC- ai ound in he p esen s udy includes he DAN and pos e io pa ie al
a eas linked o memo y, language and isuospa ial de ici s ela ed o he choline gic sys em which, as indica ed,
is also a ec ed in AD12,47,48. Finally, he FPN in PD-MCI was assumed o con ibu e o he dysexecu i e cogni-
i e p o ile49. Howe e , some s udies ha e also linked he FPN o memo y de ici s, mo e in line wi h i s ole in
AD-MCI, whe e he FPN has been obse ed o be hypoac i a ed12,50. No wi hs anding, he main componen
o ou FC- ai is he SMN, de e io a ion in which is exclusi e o PD-MCI subjec s since i s o igin comes om
dopamine gic dene a ion esul ing in execu i e unc ion and wo king memo y de ici s12,40,51.
Finally, MDS-UPDRS-III was also associa ed wi h his dis inc i e FC- ai . Gi en ha PD pa ien s we e “ON
medica ion” du ing he s udy, his ela ionship migh be due o he mo o p og ession o he disease, p obably
in aspec s ha show less imp o emen ollowing dopamine gic ea men 52. Mo eo e , PD-MCI is associa ed
wi h wo se mo o sco es53, he e o e his FC- ai could e lec an in e ac ion be ween he mo o impai men s,
ep esen ed by he SMN, and he cogni i e de e io a ion, e lec ed in DAN, VAN and FPN. Hence, his FC- ai
could be in e p e ed as an abe an FC pa e n be ween he SMN and he DAN, he VAN and he FPN in PD-
MCI pa ien s, e lec ing mo o and cogni i e de ici s ela ed o choline gic p oblems a he han dopamine gic
deple ion.
Al hough his s udy has e ealed new insigh s in o FC pa e ns in PD-MCI h ough he applica ion o an inno-
a i e analy ic echnique (connICA), some limi a ions should be no ed. Fi s , all pa ien s we e “ON medica ion”,
which has been epo ed o modi y RSFC54. Ne e heless, mo ion in “OFF medica ion” PD pa ien s unde going
Table 3. Summa y o he signi ican FC- ai . FC ne wo k con ibu ions, main ne wo ks, and associa ions
wi h cogni i e and mo o unc ions. FC-T ai F-s a is ics epea ed-measu es ANOVA ( mANOVA) model,
and he co esponding pos -hoc be ween-g oup compa isons and associa ions wi h cogni i e unc ions a e
epo ed. Ne wo k p esence o he FC- ai is ep esen ed by − o ne wo k no p esen , + o ne wo k p esen
wi h medium s eng h, and + + o ne wo k p esen wi h high s eng h. As e isks indica e he compa isons ha
we e s a is ically signi ican . P- alues below 0.05 a e FDR co ec ion a e conside ed s a is ically signi ican
in he mANOVA model. Be ween-g oup pos -hoc es s and co ela ion es s wi h p- alues below 0.01 a e
conside ed s a is ically signi ican .
FC- ai SMN + + DAN + + LIMBIC- VAN + + DMN + FPN + VISUAL- S ia um-
Main hubs SMN: Bila e al p ecen al gy us, Bila e al pos cen al gy us, Bila e al Soma omo o
S a is ics
GROUP mANOVA F(8,64) = 8.63, p = 5.41 × 10–4*
PD-MCI s. PD-CN, F(4,41) = 23.09, p = 6.0 × 10–6*
PD-MCI s. HC, F(4,42) = 29.14, p = 6.0 × 10–7*
PD-CN s. HC, F(4,40) = 1.03, p = 3.1 × 10–1, n.s
Associa ions
A en ion and wo king memo y, F(6,119) = 13.244, p = 4.0 × 10–4*
Memo y, F(6,119) = 14.129, p = 2.7 × 10–4*
Language, F(6,119) = 8.276, p = 4.8 × 10–3*
MDS-UPDRS-III, F(6,119) = 8.561, p = 4.1 × 10–3*
5
Vol.:(0123456789)
Scien i ic Repo s | (2023) 13:20162 | h ps://doi.o g/10.1038/s41598-023-46991-3
www.na u e.com/scien i ic epo s/
Figu e1. Cha ac e iza ion o he FC-T ai ha showed signi ican di e ences be ween he g oups. (A) Ci cula
plo showing he 1% s onges connec ions in he le and igh hemisphe es, o de ed by unc ional ne wo k.
(B) The F- alues ob ained by compa ing a baseline ANOVA model o he non-in e es a iables (i.e., age, MR
sequence, a e age ENORM, o al g ey ma e (TGM), o al in ac anial olume (eTIV)) wi h models adding one
cogni i e es . (C) B ain map showing he nodal s eng h o he 1% s onges connec ions calcula ed using he
no malized FC alues o he FC- ai .

6
Vol:.(1234567890)
Scien i ic Repo s | (2023) 13:20162 | h ps://doi.o g/10.1038/s41598-023-46991-3
www.na u e.com/scien i ic epo s/
MRI has pe asi e and con ounding e ec s ha i is p e e able o a oid. Second, in he p esen wo k we used a
s a ic FC connec i i y app oach. Dynamic FC me hods can be use ul o ob ain addi ional in o ma ion such as
dwell ime o swi ching be ween ne wo ks55. Thi d, we analyzed da a collec ed om wo es ing s a e uns wi h
di e en empo al esolu ion (monoband wi h TR = 2s s. mul iband wi h TR = 800ms, and adap ing he lip
angle acco dingly), which can aise conce ns ega ding measu emen - o-measu emen epea abili y. Howe e ,
hese uns we e acqui ed in he same session on each subjec , all o he sequence pa ame e s we e ma ched as
closely as possible. In addi ion, we applied wo s eps o minimize di e ences be ween each subjec FC ma ices:
(1) The signal was il e ed o keep he same spec al con en . (2) we maximized he subjec ’s iden i iabili y29
which elimina es he di e ences be ween he FC ma ices o he wo uns. In addi ion we pe o med a epea ed
measu es ANOVA ( mANOVA) analysis, which accoun s o double measu emen s as demons a ed Ge e al.
2017 whe e mANOVA models di e en ia ed in a- and in e -subjec a ia ion and compu ed g oup di e -
ences, and obse ed la gely consis en spa ial pa e ns o es – e es eliabili y be ween he human connec ome
p ojec (HCP) and Genomics Supe s uc P ojec (GSP) samples, despi e hei di e ences in scanning si e and
image p o ocols (HCP TR = 0.7s; GSP TR = 3s)56–58. Mo eo e , in his s udy he a ia ion be ween sequences
was minimized as images we e ob ained wi h he same MR scanne .
In conclusion, using a da a-d i en me hodology app oach (connICA) ha is op imal o examining FC
in e ac ions be ween RSN, we demons a ed ha PD-MCI is associa ed wi h subs an ial RSFC changes in c i ical
ne wo ks implica ed in cogni i e and mo o de ici s. A FC- ai was eliably ex ac ed which e lec ed a dis inc-
i e unc ional connec i i y be ween ne wo ks in he PD-MCI g oup comp ising he SMN, DAN, VAN and FPN
ne wo ks and was associa ed wi h cogni i e pe o mance on memo y and wo king memo y asks, as well as wi h
mo o symp om se e i y. These esul s sugges ha PD-MCI impai men s can be induced by a p og ession o FC
abno mali ies p esen since he onse o he disease, which p obably e lec dopamine gic de ici s and o he ea ly
e en s, as well as impai men in do sal and en al a en ion egions which a e likely mo e ela ed o choline gic
deple ion42,44,59. The iden i ica ion o hese di e en ial FC pa e ns con ibu es o imp o ing ou unde s anding
o cogni i e decline in PD and pa es he way o u he examina ions FC di e ences unde lying o he clinical
and cogni i e changes in sub ypes o PD-MCI pa ien s which could help iden i y isk o p og ession o demen ia.
Ma e ials and me hods
Pa icipan s
Se en y-one igh -handed pa icipan s including 20 PD-CN pa ien s, 23 PD-MCI pa ien s and 28 heal hy con-
ols (HC) we e ec ui ed a he Mo emen Diso de s Uni a he Hospi al Uni e si a io Donos ia (Donos ia-San
Sebas ián, Spain).
PD was diagnosed acco ding o he UK B ain Bank c i e ia60. Exclusion c i e ia included his o y o head
auma, psychia ic o neu ological diso de s o he han PD, o he majo medical como bidi ies, alcohol o
d ug dependence o abuse, and being le -handed. PD wi h demen ia was diagnosed acco ding o he MDS Task
Fo ce c i e ia61, and pa ien s ul illing hese c i e ia we e excluded as well. All pa icipan s we e sc eened o MRI
compa ibili y acco ding o s anda d p ocedu es. Expe imen al p ocedu es we e explained o pa icipan s, and
w i en in o med consen was ob ained p io o s udy pa icipa ion, acco ding o he Decla a ion o Helsinki.
The s udy was app o ed by he Gipuzkoa Clinical Resea ch E hics Commi ee.
Clinical and neu opsychological e alua ion
Diagnosis o PD-MCI was made acco ding o he MDS Task Fo ce c i e ia (le el II ca ego y)26 when he ollowing
wo c i e ia we e ul illed: (1) cogni i e decline epo ed by ei he he pa ien o in o man , o obse ed by he
neu ologis , ha did no in e e e signi ican ly wi h he unc ional independence o he pa ien ; (2) he pa ien
sco ed mo e han 1.5 s anda d de ia ions below con ol alues in a leas wo es s in he neu opsychological ba -
e y, ei he wi hin a single cogni i e domain o ac oss di e en cogni i e domains. No ma i e neu opsychological
es alues we e aken om 32 heal hy con ols ec ui ed among accompanying pe sons o PD pa ien s. Z sco es
o he es s o each cogni i e domain we e calcula ed as ollows: ( es sco e—mean sco e o con ol sample)/
(s anda d de ia ion o con ol sample). Z sco es we e used o diagnose PD-MCI. These we e hen a e aged o e
bo h es s e alua ing each domain o p o ide composi e z sco es ha will be used la e o co ela ion analysis.
The Hoehn and Yah 62 and MDS-UPDRS pa III scales63 we e used o e alua e mo o ea u es. As gai ,
eezing o gai and pos u al s abili y decline a e o en seen along wi h he cogni i e decline, sco es o hese
i ems om MDS-UPDRS III we e e alua ed sepa a ely (i ems 3.10, 3.11 and 3.12 o he MDS-UPDRS scale)63,
and hen summed up in a new a iable “MDS-UPDRS-III axial sco e”. The Hospi al Anxie y and Dep ession
Scale (HADS) was used o assess he p esence o anxie y o dep ession symp oms and he 39-I em Pa kinson’s
Disease Ques ionnai e (PDQ-39) was used o e alua e heal h- ela ed quali y o li e. No e ha o HC subjec s
MDS-UPDRS-III, Hoehn and Yah , and HADS scales we e no acqui ed and he e o e ha e a alue o ze o in
all s a is ical analyses.
Neu opsychological e alua ion was pe o med using he Mini Men al S a e Examina ion (MMSE) and Mon-
eal Cogni i e Assessmen (MOCA) o global cogni ion and a comp ehensi e neu opsychological ba e y
including wo alida ed es s o each o he i e cogni i e domains (see Table2). A en ion, comp ising ale ness
as well as wo king memo y main enance and manipula ion, was measu ed by he in e se digi span memo y
and symbol digi modali y es s64. Execu i e unc ion, comp ising cogni i e lexibili y, isual scanning, mo o
unc ion and e bal wo king memo y, was measu ed by he ail making es B and he phonemic luency es 65,66.
Memo y, comp ising he abili y o e ie e i ems bo h immedia ely (sho - e m) and a e a delay (long- e m),
was measu ed by he immedia e ecall sub es o he Rey–Os e ie h Complex Figu e (ROCF) es 33 o inciden al
memo y, as well as he delayed ecall sub es o he Rey Audi o y Ve bal Lea ning Tes (RAVLT)34. Language,
comp ising seman ic memo y, lexical access and dysnomia, was measu ed using he seman ic luency and Bos on
7
Vol.:(0123456789)
Scien i ic Repo s | (2023) 13:20162 | h ps://doi.o g/10.1038/s41598-023-46991-3
www.na u e.com/scien i ic epo s/
naming es s67. Visuospa ial unc ion, comp ising s uc u al knowledge and space pe cep ion, was measu ed by
he objec decision and numbe loca ion es s om he Visual Objec and Space Pe cep ion Ba e y (VOSP)68.
MRI acquisi ion
All pa icipan s we e scanned in a Siemens T io 3T MR-scanne wi h a 32-channel head coil a he Basque Cen e
on Cogni ion, B ain and Language (Donos ia-San Sebas ián, Spain). Pa ien s we e unde he i s mo ning dose
o an ipa kinsonian medica ion du ing he MRI scanning session o minimize discom o and mo emen 54.
The ana omical MRI images included a T1-weigh ed MPRAGE (TR = 2.53ms, TE = 3.97ms, lip angle
(FA) = 7°, ield o iew (FoV) = 256 × 256 mm2, 176 axial slices, oxel size = 1 × 1 × 1 mm3) and a T2-weigh ed
Tu bo Spin Echo (TR = 3.2ms, TE = 425ms, FA = 120°, FoV = 256 × 256 mm2, 176 axial slices, oxel size = 1 × 1 × 1
mm3). Two uns o T2*-weigh ed MRI da a we e acqui ed du ing es ing s a e, each wi h 10min du a ion, wi h
1) a s anda d g adien -echo echo-plana imaging sequence (monoband) (TR = 2000ms, TE = 29ms, FA = 78°,
ma ix size = 64 × 64, oxel size = 3 × 3 × 3 mm3, 33 axial slices wi h in e lea ed acquisi ion, slice gap = 0.6mm)
and 2) a simul aneous mul islice g adien -echo echo plana imaging sequence (mul iband ac o = 3) de eloped
by he Cen e o Magne ic Resonance Resea ch (Uni e si y o Minneso a, USA) (TR = 800ms, TE = 29ms,
FA = 60°, ma ix size = 64 × 64, oxel size = 3 × 3 × 3 mm3, 42 axial slices wi h in e lea ed acquisi ion, no slice
gap). Single-band e e ence images we e also collec ed be o e he mul iband es ing s a e acquisi ion o head
mo ion ealignmen . Du ing bo h acquisi ions, pa icipan s we e ins uc ed o keep hei eyes open and ixa e on
a whi e c oss ha hey saw h ough a mi o loca ed on he head coil, and no o hink abou any hing speci ic.
Field maps we e also ob ained o co ec ield dis o ions.
Da a p ep ocessing and quali y con ol
Au oma ed oxel based subco ical segmen a ion and co ical pa cella ion we e ex ac ed om he T1-weigh ed
and T2-weigh ed images using he F eeSu e image analysis so wa e ( 6.0, Ha a d, MA, h ps:// su e . nm .
mgh. ha a d. edu)69. Ana omical pa cella ions we e aligned o he unc ional space using he single-band e e -
ence image. The T1-weigh ed images we e also wa ped o he Mon eal Neu ological Ins i u e (MNI) empla e
MNI152_2009, and all he spa ial ans o ma ions om he subjec ’s space o he MNI space we e collapsed in
a single spa ial ans o ma ion. Res ing-s a e MRI (RS- MRI) da a was p e-p ocessed using AFNI70. Fi s , he
olumes co esponding o he ini ial 10s we e emo ed o allow he signal o achie e s eady s a e magne iza-
ion. Subsequen ly, he oxel ime se ies we e despiked o educe la ge ampli ude de ia ions and slice- ime
co ec ed. Inhomogenei ies caused by magne ic suscep ibili y we e hen co ec ed using FUGUE (FSL) using
he ield map images. Nex , he unc ional images we e ealigned o a base olume, which was he olume wi h
he lowes head mo ion o he monoband da ase s and he single-band e e ence image o he mul iband
da ase s. A e wa ds, a single simul aneous nuisance eg ession s ep was pe o med, he eg esso s we e: 6 h-
o de Legend e polynomials; Low-pass il e ing wi h cu o equency o 0.25Hz; Six ealignmen pa ame e s
plus hei empo al de i a i es; he i s i e p incipal componen s o oxels in deep whi e ma e and he i s
i e p incipal componen s o oxels in he la e al en icles (i.e. ana omical CompCo )71; he i s i e p incipal
componen s o he b ain’s edge oxels72. The masks o whi e ma e , la e al en icles and edges o he b ain we e
ob ained based on he F eeSu e issue and b ain segmen a ions, wa ped o he unc ional space73. In addi ion,
scans po en ially co up ed by a i ac s we e iden i ied and censo ed when he Euclidean no m o he empo al
de i a i e o ealignmen pa ame e s (ENORM) was la ge han 0.4 o he p opo ion o oxels adjus ed in he
despiking s ep exceeded 10%. Fu he mo e, ime cou ses o he de i a i e o oo mean squa e a iance o e
oxels (DVARS) we e also compu ed o each da ase 74.
Image quali y was assessed using mo ion plo s ha included g ayplo s be o e and a e nuisance eg ession
conside ing he p e iously explained mo ion measu es75. Based on he censo ing s ep, subjec s wi h mo e han
20% o he olumes emo ed in any o he wo RS- MRI da ase s we e excluded. This esul ed in a inal sample
o 23 PD-MCI, 19 PD-CN, and 21 HC pa icipan s (i.e. 63 subjec s in o al) whose da a we e included in all he
analyses epo ed below. The ENORM and DVARS me ics a e censo ing did no di e s a is ically be ween he
HC (ENORM mean = 0.1 ± 0.045mm, DVARS mean = 0.025 ± 0.006), PD-CN (ENORM mean = 0.098 ± 0.041mm;
DVARS mean = 0.026 ± 0.004mm) and PD-MCI g oups (ENORM mean = 0.099 ± 0.048mm; DVARS
mean = 0.025 ± 0.005mm); (ENORM p = 0.98, DVARS p = 0.91).
De ini ion o unc ional connec i i y ma ices
FC ma ices (Fig.2A) we e ob ained o each subjec and each ype o unc ional acquisi ion sequence (i.e.
monoband and mul iband) by compu ing he pai wise Pea son’s co ela ion be ween he a e age ime se ies o
ROIs de ined om he 400-pa cels Schae e unc ional co ical a las76 plus he 8 bila e al subco ical egions
om he F eeSu e b ain pa cella ion77. This pa cella ion was chosen since we we e in e es ed in RSN such as
SMN, DAN, VAN and FPN. The Schae e unc ional has a one- o-one associa ion be ween pa cels and 17 la ge
scale unc ional ne wo ks10. The Schae e a las is de ined in he MNI olume ic space and was wa ped back
o he subjec ’s unc ional space using he co esponding spa ial MNI- o- unc ional spa ial ans o ma ions,
whe eas he subco ical segmen a ions we e compu ed in he na i e T1-weigh ed ana omical space and hen
also co egis e ed o he subjec ’s unc ional space.
Independen FC ai s
We aimed o disen angle he la en independen unc ional connec omes embedded in he se o subjec -speci ic
global FC ma ices using a using ConnICA28. ConnICA is wo-s ep p ocess (see Fig.2B), i s a dimensionali y
educ ion wi h PCA, second, an ICA o ex ac spa ially independen pa e n o unc ional connec i i y also
e e ed as FC- ai .
8
Vol:.(1234567890)
Scien i ic Repo s | (2023) 13:20162 | h ps://doi.o g/10.1038/s41598-023-46991-3
www.na u e.com/scien i ic epo s/
Fi s , P incipal Componen Analysis (PCA) was pe o med o educe he dimensionali y o he space spanned
by he FC ma ices. Since he wo FC ma ices compu ed pe subjec we e ob ained om MRI da a wi h wo
di e en sequences (monoband and mul iband), we ook his u he s ep o ensu e hey we e compa ible and o
maximize he iden i iabili y o ou pa icipan s´ FC ma ices by inding he PCA econs uc ion ha bes iden i-
ied each indi idual’s FC ma ix o bo h he mul iband and monoband sequences29. This app oach es ima es he
di e en ial iden i iabili y (Idi ), which quan i ies he di e ence be ween he a e age wi hin-subjec FC simila i y
and he a e age be ween-subjec s FC simila i y in each session as a unc ion o an inc easing numbe o p incipal
componen s. In his assessmen , PCA was applied o he conca ena ed FC ma ices o all K possible numbe s o
p incipal componen s; hen he FC ma ices we e econs uc ed. Finally, we selec ed he numbe o K p incipal
componen s ha maximized his di e en ial iden i iabili y; in ou case, K = 65 (Fig.3).
Figu e2. ConnICA FC- ai s ex ac ion p ocess. (A) Mul iband and monoband FC ma ices we e compu ed
o each subjec and he FC coe icien s in he uppe - iangula ma ix we e ec o ized and conca ena ed
in o a g oup FC ma ix. (B) The dimensionali y o he g oup FC ma ix was educed o 65 componen s based
on op imal subjec iden i iabili y. Then, he ConnICA decomposi ion was applied so ha (C) he g oup FC
ma ix was decomposed in o spa ially independen unc ional connec i i y componen s (a.k.a. FC- ai s) and
hei co esponding weigh s o each FC ma ix. (D) A e es ing o signi ican di e ences be ween g oups,
each unc ional ai was cha ac e ized by a ci cula g aph, a nodal s eng h map and ANOVA eg essions o
neu ophysiological assessmen s.
9
Vol.:(0123456789)
Scien i ic Repo s | (2023) 13:20162 | h ps://doi.o g/10.1038/s41598-023-46991-3
www.na u e.com/scien i ic epo s/
Nex , we applied connICA28 (Fig.2) o decompose he g oup o subjec -speci ic unc ional connec omes in o
65 independen unc ional connec omes, also e e ed o as FC- ai s, using P obabilis ic ICA78. Each FC- ai
has wo pa s: (1) An FC pa e n ep esen ing di e en unc ional mechanisms (2) a ec o o weigh s indica -
ing how p esen he FC- ai is in each FC ma ix o he inpu , which can be used o see g oup di e ences and
co ela ions wi h cogni i e and clinical es in s a is ical analyses.
S a is ical analyses
One-way ANOVAs we e conduc ed o es o be ween-g oup (HC, PD-CN and PD-MCI) di e ences in demo-
g aphic and clinical a iables. Chi-squa e es s we e used o he ca ego ical a iables, and Mann Whgi ney es s
we e used o non-no mally dis ibu ed a iables. Signi ican di e ences in age and yea s o educa ion we e ound
o he PD-MCI g oup and, consequen ly, hese we e en e ed as co a ia es in all subsequen s a is ical analyses.
To ule ou any in e ac ions be ween FC and g ay-ma e (GM) olume loss we an F eesu e ’s m i_glm i o
check g oup di e ences in GM olume and co ical hickness. This analysis e ealed no signi ican di e ences
in he g oup compa isons (PD-MCI s. PD-CN, PD-MCI s. HCs, PD-CN s. HCs).
To examine g oup di e ences in he FC- ai s, a mANOVA was de ined o explain he wo weigh s o each
FC- ai in each subjec , wi h g oup (HC, PD-CN, PD-MCI) as he be ween-subjec s ac o , acquisi ion sequence
(monoband, mul iband) as a wi hin-subjec ac o , and age, gende and TGM as co a ia es. Di e ences in he FC-
ai s be ween he g oups we e de e mined as s a is ically signi ican a e con olling o mul iple compa isons
using he False Disco e y Ra e (FDR) o a q- alue < 0.05. Simple e ec s pos -hoc - es s we e hen pe o med
o examine he e ec s be ween pai s o g oups.
In addi ion, eg ession one-way ANOVA analyses we e pe o med o assess he associa ion o he FC- ai s,
ha eached signi icance in mANOVA, wi h he neu opsychological and clinical e alua ions. These one-way
ANOVA models aimed o assess he weigh s o he FC- ai s in e ms o each o he ollowing explana o y
a iables ( es ed indi idually): z-sco es o he 5 cogni i e domains assessed, he MOCA, he MMSE, he MDS-
UPDRS-III sco e and he MDS-UPDRS III axial sco e. They we e compa ed agains a baseline model wi h i e
a iables o non-in e es : age, ype o MR sequence (i.e., monoband s. mul iband), a e age alue o Euclidean
no m o displacemen pa ame e s (a g. ENORM), o al g ey ma e (TGM), and o al in ac anial olume (eTIV).
F-s a is ics we e compu ed o each explana o y a iable sepa a ely in compa ison wi h he i e a iables o non-
in e es , and s a is ical signi icance was se a P < 0.05.
Da a a ailabili y
Da a suppo ing he indings o his s udy can be eques ed o he co esponding au ho Ma ía C. Rod íguez-O oz.
Recei ed: 25 Augus 2023; Accep ed: 7 No embe 2023
Re e ences
1. Hely, M. A., Reid, W. G. J., Adena, M. A., Halliday, G. M. & Mo is, J. G. L. The Sydney mul icen e s udy o Pa kinson’s disease:
The ine i abili y o demen ia a 20 yea s. Mo . Diso d. 23, 837–844 (2008).
2. Li an, I. e al. MDS ask o ce on mild cogni i e impai men in Pa kinson’s disease: C i ical e iew o PD-MCI. Mo . Diso d. 26,
1814–1824 (2011).
3. Gasca-Salas, C. e al. Longi udinal assessmen o he pa e n o cogni i e decline in non-demen ed pa ien s wi h ad anced Pa -
kinson’s disease. J. Pa k. Dis. 4, 677–686 (2014).
4. Delgado-Al a ado, M., Gago, B., Na alpo o-Gomez, I., Jiménez-U bie a, H. & Rod iguez-O oz, M. C. Bioma ke s o demen ia
and mild cogni i e impai men in Pa kinson’s disease. Mo . Diso d. 31, 861–881 (2016).
5. Knox, M. G. e al. Neu opa hological indings in Pa kinson’s disease wi h mild cogni i e impai men . Mo . Diso d. 35, 845–850
(2020).
Figu e3. Iden i iabili y assessmen based on P incipal componen analysis (PCA) decomposi ion (A) O iginal
di e en ial iden i iabili y (Idi ) o ou da ase was 26.69; (B) Idi assessmen compu es di e en ial iden i iabili y
o each possible inc easing numbe o p incipal componen s (K); (C) A e applying subjec iden i iabili y
assessmen , he bes econs uc ion was 65 componen s, which p o ided an Idi o 41.11.