Sala etal. Alzheimer’ s Research & Ther apy (2021) 13:187
https://doi.org/10.1186/s13195-021-00925-1
RESEARCH
I n viv o human molecular neuroimag ing
ofdopaminer gic vulner abilit y
alongtheAlzheimer ’ s disease phases
Arianna Sala 1,2† , Silvia P aola Caminiti 1,2† , Luca P resotto 3 , Andr ea Pilotto 4,5 , Claudio Liguori 6 ,
Agostino Chiara valloti 7,8 , V alentina Garibotto 9 , Giovanni Battista F r isoni 9,10 , Marcello D’ Amelio 11,12 ,
Barbara P aghera 13 , Orazio S chillaci 7,8 , Nicola M ercuri 6,11 , Alessandr o P adovani 4 and Daniela P erani 1,2,3*
Abstr ac t
Backgr ound: P reclinical and pathology evidence suggests an in volvement of brain dopamine (D A) circuitr y in
Alzheimer ’ s disease (AD). W e in vivo inv estigated if , when, and in which target reg ions [123I]FP-CIT -SPEC T reg ional
binding and molecular connectivit y are damaged along the AD course .
Methods: W e retrospectively selected 16 amyloid-positiv e subjects with mild cognitive impairment due to AD (AD-
MCI), 22 amyloid-positiv e patients with probable AD dementia (AD-D), and 74 healthy contr ols, all with a vailable [123I]
FP -CIT -SPEC T imaging . W e tested whether nigr ostriatal vs. mesocor ticolimbic dopaminerg ic targets pr esent binding
potential loss , via MANC O V A, and alterations in molecular connectivity , via par tial correlation analysis. Results wer e
deemed significant at p < 0.05, after Bonf erroni correction for multiple comparisons .
Results: W e found significant r educ tions of [123I]FP-CIT binding in both AD-MCI and AD -D compared to contr ols.
Binding reductions wer e prominent in the major tar gets of the ventr otegmental-mesocorticolimbic pathway , namely
the ventral striatum and the hippocampus, in both clinical g roups , and in the cingulate gyrus , in patients with demen-
tia only . Within the nigr ostriatal projections, only the dorsal caudate nucleus show ed reduced [123I]FP - CIT binding , in
both gr oups. Molecular connectivit y assessment rev ealed a widespread loss of int er-connec tions among subcor tical
and cor tical targets of the mesocor ticolimbic network only (poor overlap with the control gr oup as expressed by a
Dice coefficient ≤ 0.25) and no alterations of the nigr ostr iatal network (high overlap with controls , Dice coefficient =
1).
Conclusion: Local- and system-level alt erations of the mesocor ticolimbic dopaminergic cir cuitr y characterize AD ,
already in pr odromal disease phases . These results might f oster new therapeutic strategies f or AD . The clinical corre-
lates of these findings deser ve to be car efully considered within the emergence of both neuropsy chiatr ic symptoms
and cognitive deficits .
Keywor ds: Biomark er , D opamine, Molecular connectivity , Substantia nigra, V entral tegmental area
© The Author(s) 2021. Open Acc ess This ar ticle is licensed under a Creative Commons A ttribution 4.0 International License, which
permits use, sharing, adaptation, distribution and repr oduction in any medium or format, as long as you giv e appropriate credit t o the
original author(s) and the source, pr ovide a link to the Creative Commons licence, and indicat e if changes were made . The images or
other third part y material in this article are included in the ar ticle ’ s Creative C ommons licence, unless indicated otherwise in a credit line
to the material. If material is not included in the ar ticle ’ s Creative Commons licence and your intended use is not permitted by statut or y
regulation or exceeds the permitted use , you will need to obtain permission directly from the cop yright holder . T o view a copy of this
licence, visit http:// creat iv eco mmons. org/ licen ses/ b y/4. 0/ . The Creative Commons P ublic Domain Dedication waiver ( http:// creat iveco
mmons. org/ publi cdoma in/ z ero/1. 0/ ) applies to the data made available in this article, unless otherwise stated in a credit line to the data.
Backgr ound
The role of dopaminergic neurotransmission circuit s
in the pathoph y siolog y of Alzheimer ’ s dise as e (AD)
is cur ren tly deba ted. E vidence of dopaminergic dys -
func tion in AD traces back to a pivotal binding study ,
repor ting dec rease d bindings of [3H]Spiroper idol in the
Open Ac cess
*Correspondence: perani.daniela@hsr .it
† Arianna Sala and Silvia P aola Caminiti contributed equally to this work.
1 Vita-Salute San Raffaele University, Via Olgettina 60, Milan 20132, I taly
F ull list of author information is available at the end of the ar ticle
P age 2 of 12
Sala etal. Alzheimer’ s Research & Ther apy (2021) 13:187
caudate n ucleus of si x brains of patients with autopsy-
confirmed A D [ 1 ]. Se veral po st -mor tem st udies subs e -
quently s howed alterations in the substantia nigra (SN )
[ 2 – 6 ] and ventral tegmental area ( V T A) [ 3 ] and both
pre- [ 7 – 11 ] and post -s ynaptic [ 5 , 12 – 18 ] neurotransmis -
sion alterations in several dopaminergic t argets [ 5 , 7 , 8 ,
10 , 11 , 14 – 16 , 18 ]. These p at holog y results were com -
plemented, more recent ly , by more limited ne uroimag-
ing e vidence, repor ting altera tions in the striatal [ 19 – 23 ]
and hippo campal [ 24 ] dopaminergic f unction limitedly to
patients with overt A D dementia (AD-D).
Se veral questions remaine d unanswered. First, it is
unclear whether dopaminergic dysf unction repres ents an
early vs . late occurrence along the AD course. Se cond, it
is unknow n whether the nigrostriatal and the mes ocorti -
colimbic dopaminergic path ways are differ ently aff ec ted
in AD and whether sp ecific t argets are more vulnerable
than others. L ack of compr ehensive information a bout
dopaminergic deficit s in AD hinders the discover y and
application of neurotransmission-targeting therapeutic
strategies along the AD course.
In the present study , we aimed to a ssess (i) the ext ent
of pre-sy naptic dopaminergic dysfunc tion in AD , (ii)
when it takes place along the disea se course, and (iii)
which spe cific targets b elonging to the nigrostri atal and
meso corticolimbic dopaminergic pathwa ys are affecte d.
W e use d com plementar y analytic al strat eg ies , namely
the ev aluation of reg ional [123I]FP -C IT binding in each
dopaminergic pathway and t he asse ssment of t heir
molecular connectiv ity alterations [ 25 ].
Methods
Study design
Particip ants with AD-D [ 26 ] and with mild cog nitive
impairment due to AD [ 27 ] (AD-MCI) and he alth y
contr ols (HC ) were retrospe ctively collecte d from
three clinical cent ers : U niversity of Bresci a (Brescia ,
I taly ; 4 AD-D, 7 AD-MCI, 43 HC , acquisition p erio d
2013–2018), G eneve University Hospital (G eneve,
Switzerland; 11 AD-D , 2 A D-MCI, 31 H C, acquisition
per iod 2008–2017), and University Hospital of Rome
T or V er gata (Rome, I taly ; 7 AD-D, 7 AD-MCI, acquisi -
tion per iod 2012–2013).
All patients under went struc tural imaging (MRI or
C T scan) and stand ardized ne urological examinations .
The following conditions were e xc luded: (1) aty pical
parkinsonism/dementia ; (2) frontotem p oral dementia;
(3) prominent c ortical or sub cortic al infarcts ; (4) other
neurologic or p sychiatric dise as es or medic al condi -
tions potentially ass oci at ed w ith cognitive deficit s; (5)
histor y of dr ug or alcohol a buse, us e of antipsychot -
ics ; and (6) use of antidepressant or serotonergic drug s
which can interfere with [123I]FP -C IT SPEC T acquisi -
tion [ 28 ].
W e included only subje cts b elonging to the AD con -
tinuum (i.e., am yloid-p ositive), in ag reement with the
recently pr opo sed N IA-A A research frame work [ 29 ].
Amyloid positiv ity was e stablished ba se d on CSF-Aβ42
in N = 25 cas es and on amyloid-PET in t he remaining N
= 13 ca ses , in accordanc e with the CSF cut -off s use d in
each clinical center or base d on amyloid-PE T positiv ity .
Patien ts under went [123I]FP -C IT SPEC T imaging for
research purp os es (University of Br esc ia) or for clinical
purp ose s (Gene ve University Hospital; University H ospi -
tal of Rome T or V ergata), i.e., to e xclude a dementia with
Le w y bo dies (DLB) di agnosis . In all the include d clinical
ca ses , [123I]FP-CIT - SPEC T scans were rated as ne gative
accor ding to a predefined rank ing scale [ 28 ].
A group of 74 healthy volunt eers w as include d in the
study as HC ( T able 1 ). They pres ent ed a ne gative medical
histor y for neurolog ical dise ase and were not tak ing psy -
choactive me dication. All subjec ts presented w ith a con-
firme d clinical diagno sis of isol at ed ac tion or rest tremor
sy ndromes over a 4-year follow- up and normal [123I]FP-
C IT binding [ 30 , 31 ]. Se e F ig . 1 for a flow diag ram depict -
ing the included/ex cluded individu als .
Demog raphic differ ences be tween g roups were ev alu -
ated by means of AN OV A and chi-square d tests .
T able 1 Descr iptive demog raphic and clinical featur es of the study groups
Abbreviations : AD-D Alzheimer ’ s disease dementia, CDR Clinical Dementia R ating, HC health y controls, AD-MCI MCI due to Alzheimer’ s disease, MMSE Mini-Mental State
Examination
*Neither MMSE nor CDR was available in 3 AD-D patients, who w ere test ed by means of the Montreal C ognitive Assessment t est (total MoCA score = 11; 12; 19,
respectively)
AD-D AD-MCI HC T est value; p
N 22 16 74 –
Sex , N (M/F) 10/12 11/5 31/43 –
Age , mean ± SD 72.18 ± 6.27 71.38 ± 10.53 67.15 ± 13.87 F = 1.84; p = 0.16
MMSE, mean ± SD* 17.92 ± 6.70 26.88 ± 1.20 – T = 4.75; p < 0.001
CDR, mean ± SD* 1.84 ± 0.77 0.50 ± 0.00 – T = − 7.65; p < 0.001
P age 3 of 12
Sala etal. Alzheimer’ s Research & Ther apy (2021) 13:187
[123I]FP-CIT SPECT
Intra venous administration of 110–185 MB q of [123I]FP -
C IT was p erforme d 30 min after thyroid blockade (800
mg of KClO4) in all subject s . Brain SPEC T acquisitions
were performe d 3 to 4 h after inje ction with the follow -
ing protocols : (i) at t he University of Br esci a and (ii) at
the U niversity of Rome T or V er gata using a dual-he ad
gamma camera e quippe d with a LE HR parallel hole colli -
mator (Discover y 630, General Ele ctr ic, Milw auk ee, W I)
accepting events in a 159-Ke V photop eak ± 10% energ y
window . Data were reconstructe d by filtered back pro -
je ction, with Butter worth 3-dimensional (3D) p ost -fil-
ter ( order 10.0; cut -off 0.50 c ycle/cm) and correcte d for
att enuation (Chang ’ s method coeffic ient 0.15 cm − 1 )); (iii)
at Gene va University Hospitals acquisitions were per -
formed b ase d on a three head c amera (T o shiba GC A-
9300A) with fan b eam collimators using a triple energy
window for sc at ter correction. Data were reconstruc ted
using filtered back proje ction with a Shepp-L ogan fil -
ter at t he Nyquist fre quenc y and correcte d for a tten ua-
tion (Chang ’ s me thod, uniform att enuation c oefficient
of 0.12/cm − 1 , accoun ting for the scatt er corrections). A
single reconstruc tion protocol was u sed for centers 1 and
2, in order to pr oduce comparable data. F or the Gene va
cent er , this was not p ossible a s the gamma camera use d
had differen t (and improved) physical proprieties . The
reconstruc tion method wa s introduced a s a covariate
in the following analysis . A s this cen ter provide s both
patients and con trol subjec ts , this is not expe cted to
affect the result s of the analysis .
Image pre-processing and qu antification were c en -
tralize d at San Raff aele Hospital (Milan). Molec ular
images were normali zed ba se d on a high-resolution
[18F ]D OP A templa te ( ht tp:// ww w . nitrc. org/ proje ct s/
spmte m pl at es ) [ 32 ]. Patients’ images were spatially
normali zed using Statistic al Parametric Mapping 12
(SPM12, htt p:// w ww . fil. ion. ucl. ac . uk/ spm/ sof tw are/
spm12 ). Parametric images were generated for each
subje ct using the Image Calcul at or (ImCalc) function
in SPM12. S pe cific ally , sp ec ific binding ratios (SBRs)
were calculated using the following formula :
where [123I]FP -CIT binding counts of three o ccipital
lobe slices wa s use d as the reference r egion [ 28 ].
F or the dopaminergic sy stem analysis , we considered
regions of interest (ROIs) pert aining to the nigrostriatal
and the meso corticolimbic dopaminergic path w ays , as
desc ribe d else where ( c fr [ 33 ].). The V T A and SN wer e
not included in the analysis , due to the limited spatial
resolution of SPEC T imaging . The meso corticolimbic
targets consiste d of the ven tral striatum, ant erior and
middle cingul at e cortices , and ventral and medial f ron -
tal area s , as well as the amygdala and parahipp ocampal
cortex; the nig rostriatal target s consisted of the dorsal
caudate n ucleus and dorsal putamen, f ron tal premo -
tor , mot or , ex e cutive dorsolateral frontal regions , and
somatosensor y cortex. W e limited our analyse s ex clu -
sively to r eg ions belonging to the nigrostr iatal and mes -
ocorticolimbic str uctures and path w ays, ver y rich in
dopamine transp orters [ 34 ].
Each ROI mask w as convolved with an 8-mm F WHM
Gaussian kernel in order to minimize the parti al vol -
ume effect .
SBR
=
voxel
i
occipital lobe
−
1
Fig . 1 F low diagram showing the number of subjects/patients who under went [123I]FP - CIT SPEC T imaging and were initially scr eened for the
present study . Red arrows indicate those individuals that wer e excluded
P age 4 of 12
Sala etal. Alzheimer’ s Research & Ther apy (2021) 13:187
Statistical analy sis
Mean [123I]FP-CIT SBR within each ROI wa s extracted
from each normali zed p arametric image. The sub corti -
cal/cortic al dopaminergic targets that showed sig nificant
tracer binding , as compared to the ref erence r egion, were
selec ted for fur ther analysis ( one-sample T - test, p < 0.05,
B onferroni-corrected for multiple com pari sons). This
procedure resulted in a p o ol of N = 10 ROIs belonging
to the mesocor ticolimbic dopaminer gic p at hway (L/R
ventral striatum, L/R hippoc ampus , L/R am ygdala , L/R
anterior cingulate cortex, L/R middle cing ulate corte x)
and N = 4 ROIs b elonging to the nigrostriatal dopamin -
ergic path way (L/R dorsal caudate n ucle us , L/R dorsal
putamen).
Univ ariate analysis
Regional differences in [123I]FP-CIT SBR be tween par-
ticipants w ith AD-D , A D-MCI, and H C were t ested v ia
M AN COV A . The me an SBRs obtaine d from each ROI
were included as de pendent var iables ; age, gender , and
reconstruc tion method were included a s nuisance c ov ari -
ates. Re sults of the M AN COV A were deeme d signific ant
at p < 0.05, following B onferroni correction for multiple
comparisons ( N = 14). Pair w ise p ost hoc analy ses were
subse quently run on signific ant MA NC OV A r esult s , set -
ting the significance threshold at p < 0.05, Bonferroni-
correcte d for m ultiple comparisons ( N = 3). Statistical
analyse s were performe d using Statistical Package for the
So cial S ciences (SPSS19). In order to fur ther character -
ize the results of the reg ional analysis , a complemen t ar y
vox el-ba sed analy sis wa s run, a ssessing the vox el-w ise
distribution of [123I]FP-CIT SBR differences obtained
in the previou s analytic al step. [123I]FP -CIT SBR p ara -
metric image s of AD-D vs . HC and AD-MC I vs . H C were
compared by means of a two-sample T - test in SPM12,
running in M A TL AB (Math work s Inc., Sherb orn, Ma ss .,
USA). Resulting T -maps were converted into Cohen ’ s d
effect si ze maps by the following formula:
Multiv ariate analysis
Molecul ar connectiv ity was e stimat ed follow ing the prin-
ciple that neurotransmitt er release is correlated among
territorie s receiving dopaminergic proje ctions from the
same afferen ts (cf. [ 25 ]). Investigation of t hese p at terns
of molec ular connectiv ity has provide d, in vivo, result s
consisten t with the known bio chemical architecture of
the dopaminergic system in normal subj ec ts [ 25 ]. Ass ess -
ment of molecular conne ctivity b etwe en targets of e ach
D
=
2 T
df
dopaminergic pathwa y (nigrostr iatal; meso corticolim-
bic ) wa s per formed v ia par tial correla tion analysis (cfr
[ 25 ]). This proce dure r esulted in the e stimation of t he
molec ular struc tur e of each dopaminergic p at hwa y in
each group. In order to estimate the molecular connec -
tivity strength b etwe en dopaminergic node s , a parti al
correla tion matrix w as computed for each clinical group
by means of M A TL A B ’ s parrcor r f unction. Partial cor -
relations allow t o investigat e the rela tionship betwe en
two regions , while f actoring out the con tributions of all
other regions (cf. [ 25 ]). Gender , age, and reconstruc -
tion method were included a s nuisance covariates . The
resulting dopaminergic network s were formed by no des ,
represented by the aforemen tioned ROIs , and by e dges ,
represented by the estimated par tial correlation c oef -
ficients . Parti al corre lation coefficients were de emed
signific ant a t p < 0.01, uncorre cted for multiple com pari -
sons , and at p < 0.05, Bonferroni-correc ted for multiple
comparisons . In order to provide an estimate of the over -
all degre e of integrity of e ach molecular dop aminergic
network , we computed an index of similarity b etween
clinical groups and H C , by means of the Dice c oeffic ient.
Results
W e included 22 patients diag nose d with probable AD-D;
16 subje cts w ith AD-MCI, in accordance with current
clinical/rese arch criteria [ 26 , 27 ]; and 74 age-matc hed
HC , all with available [123I]FP-CIT SPEC T imaging .
AD-D and AD-MC I subje cts were comparable for age
and sex distribution, differing for MMSE and C DR ( p <
0.001) ( T able 1 ).
Univariate analy sis
Result s of the MA NC OV A , comparing [123I]FP -C IT
SBR in the ma jor sub cortical/cortic al targets of the dopa -
minergic path ways , among AD-D , AD-MC I, and HC ,
are shown in F ig s . 2 A , and 3 A and T a ble A . 1 . Mole cular
imaging a ssessment demonstrated a signific ant r educ -
tion of [123I]FP-CIT SBR in both AD-MC I and AD-D ,
predominantly in the r eg ions belonging to the meso cor -
ticolimbic dopaminergic path way .
As for the meso corticolimbic path w ay , AD-D showe d
a dec rease d [123I]FP-CIT SBR in the main subcortic al
and cortical target s of the V T A , namely the ventral stria -
tum and hippo campus , bilaterally , and the right middle
cingul at e g yr us , com pared to control subject s . Similar
alterations wer e repor ted in AD-MC I compared to HC ,
with de creas ed [123I]FP-CIT SBR in the ventral striatum,
bilaterally , and the right hippo campus . No sig nificant
differenc es in [123I]FP-CIT SBR were detecte d be tween
AD-D and AD-MC I in any meso corticolimbic target ( p
< 0.05, B onferroni-corrected for multiple com pari sons).
P age 5 of 12
Sala etal. Alzheimer’ s Research & Ther apy (2021) 13:187
Fig . 2 Regions of inter est within the mesocor ticolimbic pathway sho wing significant decreases in pr e -synaptic dopaminergic activity in AD.
A Violin plots represent the distribution of SBR in ROIs with significantly decr eased DA T density ( p < 0.05, Bonferroni-corrected for multiple
comparisons). Asterisks denote post hoc comparisons, Bonf erroni correction for multiple comparisons, at p < 0.05 (*), p < 0.01 (**), and p < 0.001
(***). Brain renderings wer e obtained from the BrainNet Viewer toolbo x [ 35 ]. B Brain renderings showing the distribution of vox el-wise diff erences
in [123I]FP-CIT BP s in each clinical gr oup, r esulting from statistical comparison with HC. The magnitude of the diff erence is r epor ted by means of
Cohen ’ s d effect size . Only ROIs showing significantly decreased [123I]FP - CIT BP in association with Alzheimer ’ s disease are sho wn. Abbreviations:
AD-D, Alzheimer ’ s disease dementia; AD -MCI, mild cognitive impairment due to Alzheimer ’ s disease; HC, healthy controls
P age 6 of 12
Sala etal. Alzheimer’ s Research & Ther apy (2021) 13:187
As for the nigrostr iatal path w ay , only the dorsal cau-
date n ucleus showed de crea se d [123I]FP -CIT SBR , bilat -
erally , in b oth the AD-D and AD-MC I com pared to HC .
No differenc es in DA T density wer e detec ted in any ot her
nigrostr iatal target .
V o x el-wi se analysis , indep endently assessing differ -
ences be tween each clinical g roup vs . HC , showed that
the strongest altera tions in DA T density (Cohen ’ s d >
1) were locali zed in the ventral striatum (pe ak MNI
coordinates: x = − 6; y = 8; z = − 6 [A D-MCI]; x = 8; y
= 14; z = − 4 [AD-D]) f or the meso corticolimbic path w ay
and in the caudat e head (p eak MNI co ordina tes : x = − 8;
y = 8; z = 0 [AD-MC I]; x = − 12; y = 12; z = 12 [AD-D])
for the nigrostriatal pathway (F igs . 2 B and 3 B).
Multivariate analy sis
Result s of the m ultivar iat e analysis are summari zed in
F ig . 4 . Molecul ar connectivity a ssessment demonstrated
Fig . 3 Regions of inter est within the nigrostriatal path way showing significant decr eases in pre-synaptic dopaminergic activity in AD. A V iolin plots
repr esent the distribution of SBR in ROIs with significantly decreased DA T activity ( p < 0.05, Bonferroni-cor rected f or multiple compar isons). Asterisks
denote post hoc comparisons, Bonf erroni correction for multiple comparisons, at p < 0.05 (*), p < 0.01 (**), and p < 0.001 (***). Brain r ender ings
were obtained fr om the BrainNet V iewer toolbox [ 35 ]. B Brain r enderings show the distribution of vox el-wise differ ences in [123I]FP-CIT BP in each
clinical group , resulting from statistical comparison with healthy contr ols. The magnitude of the diff erence is repor ted by means of C ohen ’ s d effect
size. Only ROIs showing sig nificantly decreased [123I]FP - CIT BP in association with Alzheimer ’ s disease are shown. Abbr eviations: AD-D, Alzheimer ’ s
disease dementia; AD-MCI, mild cognitive impairment due to Alzheimer ’ s disease; HC, healthy contr ols
P age 7 of 12
Sala etal. Alzheimer’ s Research & Ther apy (2021) 13:187
a widespre ad loss of inter -connections b etwe en subcor -
tical and cortic al targets of the meso corticolimbic path-
way , in AD-D and A D-MCI ( p < 0.01, uncorre cted for
multiple com p arisons ; p < 0.05, B onferroni-correcte d for
multiple com p arisons). No conne ctivity alterations were
found within the nigrostriatal p at hway , with preser ved
caudate and putaminal int er -conne ctions in both clinical
groups ( p < 0.01, uncorre cted for multiple com paris ons).
Quantitative a ssessment of the overall in tegr ity of each
dopaminergic pathway , by means of Dice coefficient , con -
firme d tha t molec ular connectiv ity wa s se verely alt ered
within the meso corticolimbic network (Dice c oefficient
= 0 [AD-MC I]; 0.25 [AD-D], indicating po or to fair simi -
larity): Mole cular connec tivity appe ared to be preser ved
in AD-D and AD-MC I within the nigrostr iatal network
(Dice coefficient = 1 [AD-MCI]; 1 [A D-D], indicating
ex cellent overla p) ( p < 0.01, uncorrecte d for multiple
comparisons).
Discussion
In vivo rese arch studies on neurotransmission altera -
tions are cruci al to provide biologic al-bas ed e vidence
of molec ular alterations in neurodegenerative disea ses ,
suppor ting available symptomatic thera p eutic strate -
gie s and fostering drug dis cover y . In this study , we pro -
vide unique in vivo e vidence for spe cific alterations of
the dopaminergic path ways along the AD stages in a
well-characterize d sample of am yloid- p ositive individ -
uals w ith AD-D and AD-MC I. W e demonstra te d that
the dopaminergic projec tions arising f rom t he V T A are
the most vulnerable in AD, with a significant reduction
of [123I]FP-CIT SBR in the ma jor mes ocorticolimbic
targets , since the prodromal dise ase pha ses (Fig . 2 ). W e
also found ext ensive alterations of t he molec ular archi -
tecture of the meso corticolimbic path w ay (F ig . 4 ). The
dopaminergic proje ctions arising f rom the S N were
instead spared, w ith loss of [123I]FP-CIT SBR limited
Fig . 4 Molecular connectivit y of dopaminergic networks in healthy contr ols and AD. Brain r enderings show the molecular structure of the
nigrostriatal and mesocorticolimbic dopaminergic networks at p < 0.05, Bonferroni-corrected for multiple comparisons (red edges), and p < 0.01,
uncorrected for multiple comparisons (orange edges). Overlap between molecular connec tivity networks in healthy controls and each clinical
group is also sho wn on the right, as computed by means of the Dice coefficient. Brain renderings wer e obtained from the BrainNet Viewer toolbo x
[ 35 ]. Abbreviations: AD-D , Alzheimer ’ s disease dementia; AD-MCI, mild cognitive impairment due to Alzheimer ’ s disease; HC, healthy controls
P age 8 of 12
Sala etal. Alzheimer’ s Research & Ther apy (2021) 13:187
to the head of the caudat e nucleus (F ig . 3 ), and no ne t -
work altera tion in its mole cular circuitr y (F ig . 4 ).
The first finding of this study pert ains to the pr es -
ence of significant reductions of [123I]FP-CIT SBR in
se veral targets of the dopaminergic p at hways in AD
(F igs . 2 and 3 ). These results are supp orted by re cen t
results obt ained on a v alidated mouse mo del of AD
[ 36 , 37 ] and corroborated by pre vious p ost-mor tem [ 2 ,
7 – 11 ] and in v ivo imaging e vidence [ 19 , 38 ], repor ting
pre-sy naptic dopaminergic dysfunc tion in AD-D . W e
also provide a ne w remarkable finding of system-le vel
dopaminergic alterations, w ith disruption of the pat -
tern of regional conne ctivity ob ser ve d in HC (F ig . 4 ) .
Abnormal patterns of dopaminergic connectiv ity were
prev iously repor ted in neuro degenerative disea se
characterize d by know n dopaminergic deficit s , such
as Parkinson ’ s di sea se and DLB, showing prominent
alterations within t he nigrostr iatal system (cf. [ 25 ]). The
present new in vivo finding s demonstrate alt erations of
the neurotransmission archit ec ture in AD as well, but
within the meso corticolimbic system.
The se cond main finding of t his study per tains to the
dise ase st age in which dopaminergic deficits o ccur in AD .
A v ailable in v ivo studies provided e vidence for dopamin -
ergic alterations e xclusively in advanced dis eas e stages
[ 19 – 23 ]. No in v ivo studies are a v ailable directly a ssess -
ing dopamine pathophysiology in A D-MCI. Here, we
found, already at t he stage of MC I, a signific ant loss of
DA T density , at a degre e com p arable to t hat obser ve d in
patients with dementia , in se veral dopaminergic target s
(F ig . 3 ). The lack of signific ant diff erences in regional
DA T density b etwe en par ticipants with AD-MC I vs .
AD-D suggest s that dopaminer gic dysf unction is an e arly
e vent, also pl at eauing early along the dise ase course. Sub -
je cts w ith AD-MCI showe d ex tensive altera tions in the
molec ular archit ec ture of t he meso corticolimbic dopa -
minergic circuitr y , with loss of inter -conne ctions at the
le vel of t he ventral striatum, am ygdala , hippo campus ,
and anterior and middle cingulate g yri. The w idespread
derangement of molecular conne ctivity , ex cee ding the
repor ted [123I]FP-CIT SBR reduc tions (F ig . 5 ), indicates
an early dysf unction of the dopaminergic circuitr y in A D ,
Fig . 5 Summar y repr esentation of dopaminergic dysfunction in early AD . Brain renderings show dopaminergic targets pr esenting with decreased
DA T activity (orange), loss of molecular connec tivity (yellow), or both (red) in pr odromal AD . Alterations in these target r egions can be deemed
indicative of pre-synaptic dopaminergic dysfunction in specific affer ents from the ventral and dorsal V T A and from the medial SN, pars compacta.
F or the pur pose of ensuring a clear visualization, dopaminergic pr ojec tions are r epresented only f or each target region on the ipsilateral side . Brain
renderings wer e obtained from the BrainNet Viewer toolbo x [ 35 ]. Abbreviations: SNc, substantia nig ra pars compacta; V T A, ventral tegmental area
P age 9 of 12
Sala etal. Alzheimer’ s Research & Ther apy (2021) 13:187
po ssibly con tributing to its p at hophysiology , also at con-
sistenc y with pre vious str uctural and functional conne c -
tivity studies [ 33 , 39 , 40 ].
The third main finding of t his study per tains to the
differen t vulnerability of the two main dopaminer -
gic pathways in AD . Pre vious in vivo and po st-mor tem
studies focu sed mainly on the nig rostriatal dopaminer -
gic pathway , repor ting alterations in t he SN [ 2 – 6 ] and
in its major sub cortical target s , i.e., putamen [ 8 , 11 , 15 ,
16 , 23 , 41 ] and c audat e nucleus [ 7 , 11 , 16 , 23 , 41 ]. There
are however s everal studies rep orting l ack of altera tions
in the afor ementioned reg ions [ 1 , 5 , 8 – 10 , 12 , 13 , 17 , 23 ,
38 ]. Of note, most of the a vailable e vidence rep ort s dopa -
minergic alterations without partitioning the striatum
into its ventral and dorsal c omponents , with fe w ex cep -
tions [ 21 ]. Our finding s parti ally suppor t the in volvement
of the nigrostri atal pat hway in AD , with [123I]FP-CIT
SBR reductions limited to the caudate nuc leus , but no
alterations in molecular connec tivity (Figs . 3 and 4 ). The
caudate n ucleus , par t of the “ asso ciative-cognitive lo op, ”
receives it s major dopaminergic input from the dorso -
medi al por tions of the SN pars compacta [ 42 ]. Notably ,
prev ious po st -mor tem ev idence r ep orted alterations in
pre-sy naptic dopaminergic function sp ec ifically in the
dorsal tier of SN pars com pact a in AD , at differenc e to
what is obser ve d in Parkinson ’ s dis eas e, where the ven -
tral por tion of the SN pars com p acta (and relative pro-
je ctions) are mainly affecte d [ 2 ]. Altogether , t his e vidence
suggest s a certain deg ree of v ulnerability of t he nigros -
triatal pathway in AD , with a ver y differ ent topography
compared to Par kinson ’ s di sea se.
Compared to the nigrostri atal pat hway , the meso cor -
ticolimbic path w ay has be en far less studie d in AD . Pre-
vious p ost-mor tem and in vivo neuroimag ing ev idence
suggest s an in volvement of the V T A [ 3 ] and of its target s ,
i.e., nucleus accumbens [ 5 , 10 , 16 ], limbic stri at um [ 21 ],
hippo campus [ 18 , 23 ], amygdala [ 8 , 11 , 18 ], and cingulate
g yr us [ 11 ], with some studies rep orting howe ver negative
results [ 1 , 8 , 10 , 12 , 23 , 41 ]. Alterations of the meso corti -
colimbic targets have be en asso ciated w ith neuropsychi-
atric symptoms in AD [ 21 , 33 ], pres ent in up t o 90% of
patients with dementia [ 43 ]. Depression and apathy ar e
the most fre quent ly repor ted sy mptoms in both AD-MCI
and AD-D [ 44 ]. Result s from the current study further
and strongly suppor t the in volvement of the mesocorti -
colimbic pat hway in AD , with loss of DA T activity and
molec ular connectiv ity in its major target s . Spe cifically ,
we obser ved lo ss of D A T activity in the ventral striatum
and hippo campus , and in the middle cingul at e g yr us ,
receiving dopaminergic input from the ventral and dorsal
por tions of V T A , respe ctively [ 45 ] (F ig . 5 ). W idespread
molec ular connectiv ity altera tions affect b oth subcorti -
cal and cortic al targets , indicating the involvemen t of
the major V T A projec tions , including the meso-limbic,
meso-cortic al, meso-hipp oc ampal, and meso-amygdala
routes ( c f. [ 33 ]) (see Fig . 5 ). The finding of a prominent
involvemen t of the meso corticolimbic system in AD is
par ticularly relevant , as s ome of the V T A targets , and
most prominently t he ventral striatum, r epresent cr ucial
hubs in the c omplex f ee d-for ward organiz ation of the
dopaminergic circuitr y . Not ably , the ventral striat um,
while receiving inputs f rom a re latively small population
of neurons in the V T A , s ends back wide spread projec -
tions to both the V T A and the SN, henc e being able to
affect dop aminergic activity in w idespread p ortions of
the dorsal striatum, particul arly in its ass oci ative r egions
( cf. [ 42 ]). This e vidence rais es the question of whether
our findings of altered dopaminergic ac tivity in the cau -
date n ucleus might be conse quential to the finding of
altered dopaminergic f unction in the ventral striatum.
While our findings show a prominent in volvement of
the meso corticolimbic system and a ver y limited involve -
ment of the nigrostriatal target s , namely t he caudatum,
the pathogenic mechanisms underlying dopaminergic
vulnerability in AD remain unknow n. Some authors have
suggested that the pe culiar physiolog y of midbrain ven -
trategmental dopaminergic neurons , p er se, might render
those dopaminergic ne urons particul arly vulnerable to
amyloid pat holog y in AD [ 36 ].
A link be tween dopaminergic dy sfunction and t au
pathology ha s also b een hy pothesi zed (cf. [ 39 ]). In this
direction, a re cen t po st -mortem study on human brain
tissue rep orted an a sso ciation betwe en tau patholog y and
dysregul ation of genetic path w ays linked to dopaminer -
gic neurotransmission [ 46 ]. Not ably , the earliest tau site
in AD , namely the locus coer uleus (LC ), send s substan -
tial noradrenergic afferen ts to midbrain dopaminergic
nuclei, pr oviding trophic suppor t to both the V T A and
the medial p ortions of SN pars compacta [ 47 ]. While the
hypothesi s tha t tau pathology in the LC is at the basi s
of the dopaminergic deficit s obser ve d in AD remains
spe culative, it is wor th noting tha t the topo graphy of
dopaminergic alteration detecte d in our cohorts , sp e -
cifically encompassing proje ctions f rom t he V T A and the
medi al por tions of the SN pars c ompacta (Fig. 5 ), clo sely
matc hes that of the noradr energic projec tions from the
LC to the midbrain dopaminer gic nuclei [ 48 ]. Alterna -
tively , it can al so be hy pothesi zed a direc t effect of LC on
the dopaminergic func tion of meso corticolimbic targets ,
such as the hippo campus , that receive substantial dopa -
minergic input directly from LC ’ s ty rosine hydro xyla se-
immunoreactive neurons [ 49 ].
Limitations
This study has some limitations . While all subje cts
(MC I and AD) were within the AD spe ctrum and
