doi:10.1093/brain/awab191 BRAIN 2021: 144; 3114–3125 | 3114 The spatiotemporal changes in dopamine, neuromelanin and iron characterizing Parkinson’s disease Emma Biondetti,1,2,3 Mathieu D. Santin,1,2 Romain Valabrègue,1,2 Graziella Mangone,1,4,5 Rahul Gaurav,1,2,3 Nadya Pyatigorskaya,1,3,6 Matthew Hutchison,7 Lydia Yahia-Cherif,1,2 Nicolas Villain,1,5 Marie-Odile Habert,8,9 Isabelle Arnulf,1,3,10 Smaranda Leu-Semenescu,10 Pauline Dodet,10 Miquel Vila,11 Jean-Christophe Corvol,1,4,5 Marie Vidailhet1,3,5 and Stéphane Lehéricy1,2,3,6 In Parkinson’s disease, there is a progressive reduction in striatal dopaminergic function, and loss of neuromela- nin-containing dopaminergic neurons and increased iron deposition in the substantia nigra. We tested the hy- pothesis of a relationship between impairment of the dopaminergic system and changes in the iron metabolism. Based on imaging data of patients with prodromal and early clinical Parkinson’s disease, we assessed the spatio- temporal ordering of such changes and relationships in the sensorimotor, associative and limbic territories of the nigrostriatal system. Patients with Parkinson’s disease (disease duration 5 4 years) or idiopathic REM sleep behaviour disorder (a pro- dromal form of Parkinson’s disease) and healthy controls underwent longitudinal examination (baseline and 2-year follow-up). Neuromelanin and iron sensitive MRI and dopamine transporter single-photon emission tomography were performed to assess nigrostriatal levels of neuromelanin, iron, and dopamine. For all three functional territories of the nigrostriatal system, in the clinically most and least affected hemispheres separately, the following was performed: cross-sectional and longitudinal intergroup difference analysis of striatal dopamine and iron, and nigral neuromelanin and iron; in Parkinson’s disease patients, exponential fitting analysis to assess the duration of the prodromal phase and the temporal ordering of changes in dopamine, neuromelanin or iron relative to controls; and voxel-wise correl- ation analysis to investigate concomitant spatial changes in dopamine-iron, dopamine-neuromelanin and neuromela- nin-iron in the substantia nigra pars compacta. The temporal ordering of dopaminergic changes followed the known spatial pattern of progression involving first the sensorimotor, then the associative and limbic striatal and nigral regions. Striatal dopaminergic denervation occurred first followed by abnormal iron metabolism and finally neuromelanin changes in the substantia nigra pars compacta, which followed the same spatial and temporal gradient observed in the striatum but shifted in time. In conclusion, dopaminergic striatal dysfunction and cell loss in the substantia nigra pars compacta are interrelated with increased nigral iron content. 1 Sorbonne Université, Institut du Cerveau – Paris Brain Institute – ICM, INSERM, CNRS, 75013 Paris, France 2 ICM, Centre de NeuroImagerie de Recherche – CENIR, 75013 Paris, France 3 ICM, Team “Movement Investigations and Therapeutics” (MOV’IT), 75013 Paris, France 4 Assistance Publique Hôpitaux de Paris, Hôpital Pitié-Salpêtrière, Department of Neurology, Centre d’Investigation Clinique Neurosciences, 75013 Paris, France Received February 03, 2021. Revised April 19, 2021. Accepted May 06, 2021. Advance access publication May 12, 2021 C The Author(s) (2021). Published by Oxford University Press on behalf of the Guarantors of Brain. V This is an Open Access article distributed under the terms of the Creative Commons Attribution-NonCommercial License (https://creativecommons.org/licenses/ by-nc/4.0/), which permits non-commercial re-use, distribution, and reproduction in any medium, provided the original work is properly cited. For commercial re-use, please contact journals.permissions@oup.com Nigrostriatal changes in parkinsonism BRAIN 2021: 144; 3114–3125 | 3115 5 Assistance Publique Hôpitaux de Paris, Hôpital Pitié-Salpêtrière, Department of Neurology, 75013 Paris, France 6 Assistance Publique Hôpitaux de Paris, Hôpital Pitié-Salpêtrière, Department of Neuroradiology, 75013 Paris, France 7 Biogen Inc., Cambridge, MA 02142, USA 8 Assistance Publique Hôpitaux de Paris, Hôpital Pitié-Salpêtrière, Department of Nuclear Medicine, 75013 Paris, France 9 Sorbonne Université, CNRS, INSERM, Laboratoire d’Imagerie Biomédicale—LIB, 75006 Paris, France 10 Assistance Publique Hôpitaux de Paris, Hôpital Pitié-Salpêtrière, Sleep Disorder Unit, 75013 Paris, France 11 Neurodegenerative Diseases Research Group, Vall d’Hebron Research Institute (VHIR)-Center for Networked Biomedical Research on Neurodegenerative Diseases (CIBERNED)-Department of Biochemistry and Molecular Biology, Autonomous University of Barcelona (UAB)-Catalan Institution for Research and Advanced Studies (ICREA), 08010 Barcelona, Spain Correspondence to: Emma Biondetti, PhD Institute for Advanced Biomedical Technologies, Department of Neuroscience Imaging and Clinical Sciences, “G. D’Annunzio University” of Chieti-Pescara Via Luigi Polacchi 11, 66100 Chieti, Italy E-mail: emma.biondetti@unich.it Keywords: Parkinson’s disease; neuromelanin; iron; dopamine transporter; imaging Abbreviations: DaT = dopamine transporter; DaTScanTM = dopamine transporter single-photon emission tomog- raphy; DRS = Dementia Rating Scale; iRBD = idiopathic REM sleep behaviour disorder; MDS-UPDRS = Movement Disorder Society Unified Parkinson’s Disease Rating Scale; QSM = quantitative susceptibility mapping; SBR = specific binding ratio; SN(c) = substantia nigra (pars compacta); SNR = signal-to-noise ratio Introduction may also follow a spatial gradient of degeneration,14,16 although not all studies agree.13 Additionally, nigral changes differed be- Parkinson’s disease is a neurological disorder characterized by pro- tween the neuromelanin-rich SNc, where disease-related iron ac- gressive degeneration of dopaminergic neuromelanin-containing cumulation is expected, and the naturally iron-rich SN pars cells in the substantia nigra pars compacta (SNc) and reduced stri- reticulata.17 Most studies did not specifically investigate where atal dopaminergic function associated with increased iron depos- neurodegeneration occurred nor investigated prodromal ition in the substantia nigra (SN).1,2 Changes occur after a long Parkinson’s disease.12,13 A relationship between changes in these prodromal phase1 preceding motor symptom onset by 10 to three compounds should translate into spatiotemporal relation- 25 years for putaminal dopamine uptake3 and by 5 years for SNc ships between their changes, observable in each functional terri- volume loss (assessed using neuromelanin signal changes).4 Iron tory of the basal ganglia. levels also increase before motor sign manifestation in patients Here, we aimed to investigate the relationship between dopa- with idiopathic REM sleep behaviour disorder (iRBD), a prodromal minergic dysfunction and changes in neuromelanin and iron con- condition of parkinsonism,5 and in both symptomatic and asymp- centrations in the nigrostriatal system in patients with prodromal tomatic carriers of the LRRK2 and parkin mutations,6 but the time and clinical Parkinson’s disease. Particularly, we aimed to assess to onset before motor sign appearance is unknown. the temporal ordering of such changes and their spatial distribu- Dopamine, neuromelanin, and iron concentrations are prob- tion within the nigrostriatal system’s sensorimotor, associative, ably tightly linked. First, they all participate in the dopamine meta- and limbic territories. bolic pathway, whose end-stage product is neuromelanin. In dopaminergic neurons, neuromelanin can chelate excess iron, pro- viding a neuroprotective role.2,7 However, excess intracellular neu- Materials and methods romelanin could also compromise neuronal function and trigger Participants Parkinson’s disease-like pathology.8 Second, previous imaging studies in Parkinson’s disease have partially investigated the The ICEBERG longitudinal cohort was prospectively investigated.4 relationship between neuromelanin, iron, and dopamine with This cohort included healthy control subjects, patients with iRBD, mixed results.5,9–14 Most have reported positive correlations be- and patients with Parkinson’s disease who were assessed at base- tween striatal dopamine transporter (DaT) and nigral neuromela- line (visit 1) and after 2 years (visit 2). Patient inclusion criteria nin levels.9–12 Results on iron assessment are contradictory, as comprised a clinical diagnosis of iRBD or Parkinson’s disease, re- some studies reported negative correlations with nigral neurome- spectively performed by sleep neurologists and movement dis- lanin levels13,14 and striatal DaT levels.12 Others found no correl- order specialists, 18–75 years of age, minimal or no cognitive ation in Parkinson’s disease between nigral iron and disturbances (Mini-Mental State Examination score 4 26/30) and, neuromelanin12,13 or between nigral iron and striatal DaT.12 for patients with Parkinson’s disease, time elapsed from the first Likewise, in iRBDs, nigral iron levels assessed using transcranial appearance of motor symptoms (i.e. disease duration) 5 4 years. sonography did not correlate with striatal DaT, suggesting inde- Patients with Parkinson’s disease met the UK Parkinson’s Disease pendency of dopaminergic dysfunction and iron increase.5 Third, Society Brain Bank criteria.18 Patients with iRBD had a history of the progression of changes is spatially specific, starting in sensori- dream-enacting behaviours with (potentially) injurious move- motor areas of the striatum and the SN.9,15 SN iron accumulation ments and enhanced tonic chin muscle tone or complex 3116 | BRAIN 2021: 144; 3114–3125 E. Biondetti et al. behaviours during REM sleep but did not meet the criteria for values.37 The posterior limb of the internal capsule was delineated Parkinson’s disease or dementia.19 The local ethics committee by aligning the Eve magnetic susceptibility template38 to the sub- approved this study (RCB: 2014-A00725-42) and all participants ject’s QSM image via the corresponding magnitude image. gave written informed consent. Each subject’s QSM and neuromelanin-sensitive images were aligned to an ad hoc symmetrical brain template, enabling Clinical examination analysis in a common anatomical space.4 Particularly, each QSM or neuromelanin-sensitive image was first rigidly aligned to the Clinical examination included the Hoehn and Yahr scale, which T1-weighted anatomical image.39 Then, the T1-weighted anatomic- evaluates the severity of Parkinson’s disease-related symptoms20; al image was aligned to the template by concatenating an affine the Movement Disorder Society Unified Parkinson’s Disease Rating and a non-rigid registration.40 Then, the same spatial transform- Scale (MDS-UPDRS)21 part III in the OFF condition,21 which evalu- ation from the T1-weighted image to template was applied to the ates motor disability 412 h after withdrawing dopaminergic treat- neuromelanin-sensitive and QSM images.39,40 To mitigate the ment; the Mattis Dementia Rating Scale (DRS)22,23 and the negative effect on the statistical analysis of both potential spatial Montréal Cognitive Assessment score (MoCA),24 which evaluate misregistration and data dispersion in multiple voxel-wise tests, cognitive impairment; and Ardouin’s Scale Of Behaviour In before quantitative analysis all coregistered images were spatially Parkinson’s Disease (ASBPD),25,26 which evaluates mood/behav- smoothed via convolution with a 1-mm wide 3D Gaussian kernel. ioural status. In patients with Parkinson’s disease, the body side of This modest kernel width was chosen to enhance the detection of first motor symptoms was defined as the clinically most affected. small-scale effects in the SN, while mitigating partial volume effects with the neighbouring CSF. Imaging data acquisition In template space, signal-to-noise ratios (SNRs) of neuromela- nin-sensitive MRI were calculated relative to a background region All subjects underwent MRI on a Siemens Prisma 3 T system of interest including the tegmentum and superior cerebellar (Siemens Healthcare) using a 64-channel head coil for signal peduncles similar to a previous study.4 reception. The following images were acquired: whole-brain In each subject, striatal DaT specific binding ratios (SBRs) were T1-weighted 3D MP2RAGE (three-dimensional magnetization-pre- calculated as follows. The DaTScanTM image was rigidly aligned to pared two rapid gradient echo) for anatomical reference27; T1- the corresponding T1-weighted anatomical MRI via the corre- weighted 2D TSE (two-dimensional turbo spin echo) with a field of sponding computed tomography image. Partial volume artefacts view restricted to the midbrain for neuromelanin-sensitive imag- were corrected using Yang’s iterative method (seven iterations) as ing28; and whole-brain T2*-weighted multi-echo 3D FLASH (fast follows.41 Based on structural segmentations of the T1-weighted low angle shot) for iron-sensitive quantitative susceptibility map- image calculated using FSL FIRST,42,43 five masks were calculated ping (QSM).29 Supplementary Table 1 lists all the parameters for corresponding to five tissue compartments (white matter, CSF, the MRI acquisition. The magnetic resonance images were visually and three grey matter compartments). These compartments were inspected, and those with severe motion, arterial flow artifacts defined according to their dopaminergic density level, as meas- affecting SN visibility or incomplete SN coverage were excluded. ured post-mortem in humans.44 Particularly, the three grey matter In a subset of participants, DaT single-photon emission tomog- compartments respectively included regions with high (nucleus raphy imaging was performed on a hybrid gamma camera accumbens, caudate nucleus, and putamen), intermediate (globus Discovery 670 Pro system (GE Healthcare) using the 123I-FP-CIT pallidus) and low dopaminergic density levels (thalamus and cor- tracer (DaTScanTM). All subjects received thyroid gland blockade via one dose of Lugol regimen and an intravenous injection of 123I- tex). To normalize the signal intensity of the DaTScanTM image, FP-CIT tracer (185–200 MBq) and underwent DaTScanTM between 3 thus obtaining SBR values, a reference region of interest was and 4 h after tracer injection. The acquisition parameters were cir- defined in the occipital lobe by non-linearly aligning the auto- cular orbit, clockwise rotation, zoom factor = 1.5\1.5, matrix size = mated anatomical labelling template45 to the T1-weighted anatom- 128  128 voxels, number of slices = 128, voxel size = 2.9  2.9 ical image. The partial volume corrected DaTScanTM image was mm2, slice thickness = 2.9 mm, rotations = 1, number of frames in referenced to the average signal intensity in this occipital region of rotation = 120, actual frame duration = 30 s, energy window = interest.46 143.1–174.9 KeV. All DaTScanTM images were reconstructed on the For each subject, in the striatum, the sensorimotor, associative GE Healthcare Xeleris Workstation using an iterative algorithm fol- and limbic putamen, the sensorimotor, associative and limbic lowed by spatial filtering (fourth-order low-pass filter with a cut- caudate nucleus, and the nucleus accumbens were automatically off frequency equal to 3.5 mm–1) and Chang attenuation correction segmented by aligning a digital template of the human basal gan- (m = 0.12 cm–1).30 glia (YeB atlas)47,48 to the subject’s anatomical T1-weighted MRI. In each striatal region of interest, the average DaT-SBR and iron sen- sitive QSM signal were calculated. In the midbrain, a region of Image analysis interest mostly corresponding to the SNc was used, which was When not otherwise stated, all analyses were carried out using previously calculated in template space based on neuromelanin- MATLAB (R2017b, The MathWorks, Natick, MA, USA). sensitive MRI of 61 healthy control subjects.4 Further, this SNc re- For QSM calculation in each subject, the gradient echo signal gion of interest was manually segmented based on the functional magnitude and phase from multiple channels of the receive coil subdivision of the SN in primates obtaining three subregions: were combined using an in-house optimized method.31 Then, the dorsolateral sensorimotor, dorsomedial limbic and ventral asso- local field map was calculated by non-linear fitting the complex ciative.49 In each nigral subregion, both the voxel-wise and average gradient echo signal over echo times,32 background magnetic values of the QSM signal and the neuromelanin SNR were fields were removed using the Laplacian boundary value method33 calculated. and the local field-to-magnetic susceptibility inverse problem was Before template alignment, the magnetic resonance images solved using the L1-Morphology Enabled Dipole Inversion and DaT-SBR values of Parkinson’s disease patients were lateral- method.32,34–36 Each QSM image was referenced by subtracting the ized so that the clinically most affected side was aligned with the average QSM value measured in the posterior limb of the internal left hemisphere. In iRBD, there is no ‘most affected side’, thus capsule, a region characterized by stable magnetic susceptibility here, to perform lateralization, the brain hemisphere with the Nigrostriatal changes in parkinsonism BRAIN 2021: 144; 3114–3125 | 3117 lowest mean of global striatal DaT-SBR was selected as the poten- procedure. For all correlation analyses, baseline data from iRBD tial side of future appearance of Parkinson’s disease-related motor patients and Parkinson’s disease patients were considered to ex- symptom. Healthy control subject data were not lateralized. plore changes along a pseudo-temporal axis of disease, but visit 2 For each patient and each striatal region, the DaT-SBR per cent data were excluded to avoid introducing intra-subject correlation ratio (%) was calculated as the ratio of the patient’s DaT-SBR to the bias. average DaT-SBR in healthy control subjects. Similarly, the QSM and neuromelanin SNR per cent ratios were calculated in each SN Data availability subregion. For each striatal region, the DaT-SBR per cent ratio of individual The data supporting the findings of this study are available from Parkinson’s disease patients was modelled against disease dur- the corresponding author upon reasonable request. ation using an exponential fitting curve as, in Parkinson’s disease, degeneration of the nigrostriatal system has been shown to occur Results at a faster rate in the early phases of disease than in later phases.1,3,4,15 Based on the exponential fitting curve, the pro- Demographic and clinical characteristics dromal phase of disease and the DaT-SBR loss at the time of diag- Patient characteristics are summarized in Table 1. Between May nosis were respectively calculated as the time closest to t = 0 2015 and September 2019, 55 healthy control subjects, 43 patients where DaT-SBR = 100%, and the zero-intercept of the curve. Based with iRBD, and 135 patients with idiopathic Parkinson’s disease on inverse mapping of the exponential fitting, the temporal place- were enrolled and underwent MRI at visit 1. Monogenic cases of ment of the average DaT-SBR ratio in iRBDs at visit 1 was esti- Parkinson’s disease were not included in this study. A subset of mated. Similar to the striatal DaT-SBR, for each SN subregion, the these subjects underwent MRI at visit 2, and DaTScanTM at visits 1 QSM and neuromelanin SNR per cent ratios of individual and 2 (Table 1). All subjects underwent MRI on the clinical assess- Parkinson’s disease patients over disease duration were modelled using exponential fitting functions. Then, the prodromal phase of ment day and DaTScanTM within 6 months of MRI. Only four disease and the variation in QSM signal or neuromelanin SNR at patients with iRBD returned within 3 months after clinical assess- the time of disease diagnosis were calculated. ment due to unavailability of the MRI scanner. No subjects were entirely excluded following MRI quality control, as they all contrib- uted at least one measurement to the analysis (Supplementary Statistical analysis material). Statistical analyses were performed using MATLAB and R At visit 1, patients with iRBD were older than healthy control (v3.6.1).50 All statistical tests were two-tailed and P 5 0.05 was con- subjects (P = 0.004) and Parkinson’s disease patients (P 5 0.001). sidered to determine significance. The male/female ratio was higher among iRBD patients compared At visit 1, adjusted between-group differences in age and clinic- to both healthy control subjects (P = 0.002) and Parkinson’s disease al scores were estimated using ANOVA and Tukey’s post hoc test patients (P = 0.005). The three groups differed in Hoehn and Yahr for all-pairwise comparisons. Adjusted sex differences were esti- stage (all P 5 0.001) and MDS-UPDRS-III OFF score (all P 5 0.001). All mated using Pearson’s chi-square test with Yates’ continuity cor- groups had similar MoCA scores, whereas iRBD patients had a rection and the false discovery rate (FDR) procedure. lower Mattis DRS score than healthy control subjects (P = 0.04). In all the following analyses, sex and age were included as Parkinson’s disease patients had higher hypodopaminergic ASBPD covariates of no interest. Indeed, sex and age influence the pheno- than healthy control subjects, and higher hyperdopaminergic typical expression of Parkinson’s disease51; healthy ageing induces ASBPD scores than both healthy control subjects and iRBD patients neuromelanin and iron accumulation in the nigrostriatal sys- (all P 5 0.001). Dopaminergic medication was administered to 121 tem52–54 and DaT decrease in the striatum,55 and sex influences Parkinson’s disease patients at visit 1, and to four iRBD patients both DaT uptake and neuromelanin accumulation with healthy and 77 Parkinson’s disease patients at visit 2. ageing.54,56 For each region of interest, visit and hemisphere, intergroup Group differences in DaT, neuromelanin and iron differences in DaT, iron and neuromelanin were estimated using linear models plus Tukey’s post hoc test, with the imaging-based At visit 1, Parkinson’s disease patients had lower DaT levels than measurement (the average DaT-SBR, neuromelanin SNR or QSM) healthy control subjects and iRBD patients in all striatal regions bi- as the continuous dependent variable, group and sex as categorical laterally (Fig. 1B and C). In the clinically most affected hemisphere, variables and age as a linear effect. Similarly, for each region of all territories of the caudate nucleus and putamen were similarly interest and hemisphere, inter-visit differences in DaT, iron and affected, whereas the limbic striatum was less affected (Fig. 1B). In neuromelanin were estimated using linear models plus Tukey’s the contralateral hemisphere, the putamen was the most affected post hoc test, with the inter-visit difference of the imaging-based with all territories similarly involved, followed by the caudate nu- measurement as the continuous dependent variable and the inter- cleus and the nucleus accumbens (Fig. 1E and F). In the caudate visit time as a linear effect. nucleus, sensorimotor subregions were more affected, followed, in To evaluate spatially specific concomitant changes between order, by the associative and the limbic subregions for both striatal DaT and nigral neuromelanin or iron, Pearson’s correla- Parkinson’s disease patients compared to healthy control subjects tions were calculated between the average DaT-SBR in each striat- (Fig. 1B and E) and Parkinson’s disease patients compared to iRBD al region of interest and the voxel-wise values of nigral patients (Fig. 1C and F). In iRBD patients compared to healthy con- neuromelanin SNR or QSM in the SNc, simultaneously adjusting trol subjects, DaT loss was significant in the sensorimotor puta- for multi-voxel (SNc) and multi-region of interest (striatum) com- men bilaterally and less in the associative putamen in the most parisons using an approximate multivariate permutation affected hemisphere (Fig. 1D and G). In Parkinson’s disease method.57 Similarly, to estimate spatially specific concomitant patients at visit 2, a similar pattern was observed bilaterally changes in nigral neuromelanin and iron, voxel-wise Pearson’s (Fig. 1H and K). At visit 2, DaT loss in iRBD patients compared to correlations were calculated between neuromelanin SNR and QSM healthy control subjects progressed in all striatal subregions ex- in the SNc, adjusting for multi-voxel comparisons using the FDR cept the bilateral nucleus accumbens (Fig. 1J) and the caudate 3118 | BRAIN 2021: 144; 3114–3125 E. Biondetti et al. Table 1 Demographic and clinical characteristics of the cohort Healthy control subjects Patients with iRBD Patients with Parkinson’s disease Visit 1 Visit 2 Visit 1 Visit 2 Visit 1 Visit 2 n 55 28 43 21 135 83 Age, years 62.0 ± 9.2 62.3 ± 8.9 67.6 ± 5.1* 69.2 ± 5.3 61.6 ± 9.4# 63.9 ± 9.1 Sex, male/female 30/25 9/19 38/5* 17/4 85/50# 52/31 Time from visit 1, n/a 2.0 ± 0.1 n/a 2.1 ± 0.2 n/a 1.9 ± 0.4 years Disease duration, n/a n/a n/a 1.1 ± 0.5 (4 iRBD) 1.5 ± 1.1 3.7 ± 1.1 years Hoehn and Yahr stage 0.1 ± 0.5 0.1 ± 0.4 0.7 ± 1.0* 1.2 ± 0.9 2.0 ± 0.2 *,# 2.0 ± 0.2 LEDD, mg n/a n/a n/a 210.5 ± 181.7 (4 iRBD) 350.6 ± 252.6 (121 PD) 490.8 ± 267.5 (77 PD) MDS-UPDRS III OFF 5.5 ± 5.2 6.8 ± 5.4 11.5 ± 6.3* 19.0 ± 8.3 30.3 ± 7.8 *,# 33.9 ± 7.8 MoCA 28.0 ± 1.7 28.6 ± 1.8 27.3 ± 2.3 26.7 ± 3.2 27.4 ± 2.2 27.2 ± 3.6 Mattis DRS 139.6 ± 3.9 141.3 ± 3.0 137.3 ± 6.2* 139.5 ± 4.5 138.9 ± 4.4 136.7 ± 15.8 ASBPD 0.8 ± 1.0 0.7 ± 1.2 1.7 ± 2.5 2.4 ± 3.1 2.0 ± 1.9* 2.1 ± 1.9 hypodopaminergic ASBPD 0.1 ± 0.4 0.2 ± 0.5 0.3 ± 0.7 0.4 ± 1.0 1.1 ± 1.5*,# 1.3 ± 1.8 hyperdopaminergic Laterality of first motor n/a n/a n/a n/a 78/57 46/37 symptoms in the body, right/left DaTScanTM delay rela- 0.2 ± 0.2 0.2 ± 0.3 0.3 ± 0.2 0.1 ± 0.1 0.2 ± 0.2 0.1 ± 0.2 tive to MRI, years Values are presented as mean ± SD. ASBPD = Ardouin’s Scale Of Behaviour In Parkinson’s Disease; LEDD = levodopa equivalent daily dose; Mattis DRS = Mattis dementia rat- ing scale; MoCA = Montréal Cognitive Assessment Score; n/a = not available. *P 5 0.05 denote significant differences between patients with Parkinson’s disease (PD) or patients with iRBD and healthy control subjects at visit 1. # P 5 0.05 denote significant differences between patients with Parkinson’s disease and patients with iRBD and healthy control subjects at visit 1. nucleus in the least affected hemisphere (Fig. 1M). However, in the most affected hemisphere and to progress, albeit less severely, most affected hemisphere, inter-visit DaT loss was only significant contralaterally. in the sensorimotor and the associative putamen (Fig. 1P). In the Scatter box plots of DaT-SBR measurements in the striatum least affected hemisphere, inter-visit DaT loss was only significant and neuromelanin SNR and QSM measurements in the SNc are in the associative caudate nucleus of patients with Parkinson’s shown in Supplementary Figs 1–13. disease compared to healthy control subjects (Fig. 1Q). Overall, similar spatial gradients were found bilaterally, with increased se- Temporal progression of global changes in DaT, verity in the most affected hemisphere. Conversely, no striatal re- neuromelanin and iron gion exhibited intergroup differences in iron. In the most affected hemisphere at visit 1, neuromelanin loss Similar to the spatial group differences, DaT loss in patients with was significant in the whole SNc of patients with Parkinson’s dis- Parkinson’s disease progressively involved the sensorimotor, asso- ease compared to both healthy control subjects and iRBD patients ciative and limbic putamen, the sensorimotor, associative and lim- (Fig. 2B and C). At visit 2, neuromelanin loss was significant in the bic caudate nucleus, and the nucleus accumbens bilaterally (Fig. 3 whole SNc of Parkinson’s disease patients compared to healthy and Supplementary Table 3). The estimated prodromal phase of control subjects (Fig. 2F) and in the sensorimotor SNc of patients DaT loss lasted between 20.7 years (sensorimotor putamen) and with Parkinson’s disease compared to iRBD patients (Fig. 2G). In 1.1 years (nucleus accumbens) (Supplementary Table 3). At visit 1, the least affected hemisphere, at visit 1 there were no changes. At estimated DaT losses in iRBD patients were comparable to those in visit 2, neuromelanin loss was significant in the sensorimotor and patients with Parkinson’s disease 16.5 years before disease diagno- limbic SNc of patients with Parkinson’s disease compared to sis (sensorimotor putamen), whereas values in the nucleus accum- healthy control subjects (Fig. 2H) and in the sensorimotor SNc of bens were within normal ranges (Supplementary Table 3). patients with iRBD compared to healthy control subjects (Fig. 2I). In the most affected hemisphere of patients with Parkinson’s Overall, neuromelanin changes followed the expected gradient: disease, estimated SNc iron accumulation began 9.6 years (sensori- the dorsolateral sensorimotor SNc was more impacted in motor), 6.2 years (limbic) or 4.1 years (associative) before diagnosis, Parkinson’s disease patients in the most affected hemisphere at whereas estimated neuromelanin loss began 5.9 years (sensori- visit 1 (Fig. 2B) and the least affected hemisphere at visit 2 (Fig. 2H) motor) or 3.6 years (limbic) before diagnosis (Fig. 3). In both the as- and in iRBD patients in the least affected hemisphere at visit 2 sociative and the whole SNc in the least affected hemisphere, (Fig. 2I). Iron accumulation was only significant in the whole SNc neuromelanin signal changes were minimal. of patients with Parkinson’s disease compared to healthy control subjects without gradients in the most affected hemisphere at visit Spatiotemporal DaT-neuromelanin-iron correlation 1 (Fig. 2B) and in the limbic SNc of patients with Parkinson’s dis- ease compared to healthy control subjects in the least affected patterns hemisphere at visit 2 (Fig. 2H). There were no inter-visit group dif- In the most affected hemisphere, DaT-neuromelanin (Fig. 4A–G), ferences in neuromelanin or iron in the bilateral SNc. Similar to neuromelanin-iron (Fig. 4K) and DaT-iron (Fig. 4L–R) correlated sig- the striatum, SNc degeneration appeared to begin in the clinically nificantly in the SNc. Both the neuromelanin-DaT and iron-DaT Nigrostriatal changes in parkinsonism BRAIN 2021: 144; 3114–3125 | 3119 Figure 1 Group differences in striatal DaT levels. (A) Functional subregions in the striatum. (B–S) Left and right panels show the results of the statistic- al comparison between groups in the clinically most and least affected brain hemispheres, respectively. Each panel has three columns, representing the intergroup comparison between Parkinson’s disease patients (PD) and healthy control subjects (HCs) (left), Parkinson’s disease patients and iRBD patients (middle) or iRBD patients and healthy control subjects (right). Rows correspond to the group differences at (B–G) visit 1 (V1); (H–M) visit 2 (V2); and (N–S) the inter-visit differences in longitudinal subjects. Results were adjusted for age and sex and corrected for multiple comparisons using Tukey’s test at the P 5 0.05 level. For improved clarity in the visual representation of statistical differences, P-values were converted into z-scores. Cd = caudate nucleus; NAc = nucleus accumbens; Pu = putamen correlation patterns appeared as longitudinal bands extending Discussion through the SNc full anteroposterior length and occupying large fractions of the SNc width (Fig. 4A–G and L–R). In the SNc, the neu- This study reveals that changes in neuromelanin and iron in the romelanin-DaT correlation patterns of all putaminal territories SNc and DaT in the striatum followed similar spatial and time- overlapped considerably (Fig. 4H). In the most affected hemi- shifted temporal patterns along the sensorimotor, associative and sphere, neuromelanin-DaT correlations always extended over a limbic territories. First, synaptic dopaminergic dysfunction occurred in the striatum, then iron changes and neuromelanin larger SNc area than in the least affected hemisphere (Fig. 4A–G). loss occurred in the SNc. Changes correlated in the SNc. These Despite the considerable overlap in the SNc, the territories of the results suggest an interrelationship between dopaminergic striatal caudate nucleus exhibited an anterolateral-to-posteromedial gra- dysfunction, nigral cell loss and increased nigral iron content. dient extending from the sensorimotor to the limbic caudate nu- cleus with the associative caudate nucleus in between (Fig. 4I). Similar gradients were found for the limbic putamen (anterolat- Topography of changes eral), the limbic caudate nucleus (intermediate) and the nucleus Parkinson’s disease patients exhibited the expected gradient of accumbens (posteromedial) (Fig. 4J). The iron-DaT correlations DaT loss, progressively involving the sensorimotor, anterodorsal overlapped considerably for all striatal regions (Fig. 4L–R). Unlike associative and limbic ventral striatal regions,58 and greater neuro- the neuromelanin-DaT correlations, the iron-DaT correlations melanin loss in the posterolateral sensorimotor SNc than in the involved the bilateral SNc for all striatal regions except the nucleus anteromedial associative and limbic SNc.4 Nigral iron was evenly accumbens (Fig. 4L–R). In patients, neuromelanin-iron correlations increased in the SNc, as previously reported using iron-sensitive appeared in the bilateral anterior SNc, corresponding to associa- MRI in Parkinson’s disease,59,60 but unlike previous findings sug- tive and limbic areas (Fig. 4K). gesting higher iron in the lateral SN.13 A recent meta-analysis 3120 | BRAIN 2021: 144; 3114–3125 E. Biondetti et al. Figure 2 Group differences in nigral neuromelanin and nigral iron levels. (A) Functional subregions in the substantia nigra. (B–I) Left and right panels show the results of the statistical comparison between groups in the clinically most and least affected brain hemispheres, respectively. Rows corres- pond to the group differences at (B–E) visit 1 (V1), and (F–I) at visit 2 (V2). Non-significant between-group comparisons were omitted. Results were adjusted for age and sex and corrected for multiple comparisons using Tukey’s test at the P 5 0.05 level. For improved clarity in the visual representa- tion of statistical differences, P-values were converted into z-scores. HCs = healthy control subjects; NM = neuromelanin; PDs = patients with Parkinson’s disease. showed greater iron increase in the lateral SNc than in the medial disease, measurements of neuromelanin residual volume and SNc SNc when using iron-sensitive R2* measurements but not when neuromelanin SNR offer complementary information on Parkinson’s using QSM.60 Others found that QSM changes spanned larger areas disease-induced temporal changes in neuromelanin. of the bilateral SNc than R2* changes,61 or were restricted to ventral At baseline, in the clinically least affected hemisphere, SN regions not isolated here.16 Thus, spatial gradients of iron de- Parkinson’s disease patients presented no significant neuromela- position need further investigation. nin loss or iron accumulation in the SNc. Given that the disease is In iRBD patients, striatal DaT loss exhibited the same spatial gradi- probably present bilaterally by the time it manifests unilaterally, ent but with a lower magnitude than in Parkinson’s disease patients, these results could have a biological explanation. After dopamin- consistent with observations in larger striatal areas.5 Moreover, the ergic neuron death, neuromelanin granules can remain in the presence of significant DaT loss without significant neuromelanin extracellular space for long periods of time due to their insolubil- loss reflected the known earlier degeneration of striatal terminals in ity,66 as has been observed in patients with Parkinson’s disease,67 Parkinson’s disease.62 Patients with iRBD had no significant changes normally ageing subjects,68 and in the first rodent model of age- in nigral neuromelanin SNR at visit 1,4 but only in the sensorimotor dependent human-like neuromelanin production in the SNc.69 SNc at visit 2.4 Similarly, iRBD patients had no significant nigral iron Thus, extracellular neuromelanin could still be MRI visible for changes as in a study using R2*,63 but in contrast with another study some time after neuron death, before it is phagocytosed. As neuro- using QSM (which measured iron in the whole SN).64 Inter-visit melanin loss and iron accumulation in the SNc are tightly linked, changes in neuromelanin and iron accumulation were not significant. the delayed degradation of extracellular neuromelanin could also On neuromelanin-sensitive MRI, neuromelanin SNR was less sensi- explain the delayed observation of iron accumulation on iron-sen- tive than SNc volume measurements (as used here) for detecting sitive MRI. Alternatively, we cannot rule out that the MRI methods changes early in the disease,65 suggesting that, in early Parkinson’s employed in the present study might have insufficient sensitivity Nigrostriatal changes in parkinsonism BRAIN 2021: 144; 3114–3125 | 3121 Figure 3 Temporal changes in striatal DaT, nigral neuromelanin and nigral iron in patients relative to healthy controls. (A–C and F–H) For all Parkinson’s disease patients relative to healthy control subjects (HCs), in each striatal region and each brain hemisphere, panels shows the mono-exponential func- tions fitting the regional DaT-SBR per cent change against disease duration. (D and E) For each nigral region in the most affected hemisphere, the figure also shows the mono-exponential functions fitting iron (QSM) or neuromelanin (NM SNR) per cent changes against disease duration. Cd = caudate nu- cleus; NAc = nucleus accumbens; NM = neuromelanin; Pu = putamen; QSM = quantitative susceptibility mapping. to detect smaller-scale changes at this stage of the disease (for fur- the significant DaT-neuromelanin correlation was congruent with ther discussion on methodological limitations, see the ‘Limitations previous findings in Parkinson’s disease,9–12 using dopaminergic and considerations’ section). tracers in the whole striatum,10,11 in the whole caudate nucleus and putamen,12 between specific striatal and nigral territories9 Temporal ordering of changes and between striatal DaT and post-mortem SNc cell counts.70 Conversely, DaT and neuromelanin were uncorrelated in more In the most affected hemisphere, striatal dopaminergic dysfunc- advanced Parkinson’s disease, suggesting that DaTScanTM is lin- tion was estimated to begin decades before disease diagnosis de- early related to nigral cell losses up to 50%.71 In the SNc, DaT loss pending on the striatal territory: longer in the sensorimotor putamen (–20.7 years) than in the associative striatum (–12.7 to correlated with neuromelanin loss in large overlapping areas par- –4.4 years) and shorter in the limbic striatum (–6.1 to –1.1 years). ticularly for the putamen. However, correlations with regions of Previous PET studies reported a 6–17-year prodromal range in the the caudate nucleus from limbic to associative to sensorimotor entire most affected putamen.3 In our cohort, the estimated pro- progressively extended further posteriorly and medially in the dromal phase in the most affected putamen lasted 13 years, as for SNc. The limbic putamen, limbic caudate nucleus and nucleus methylphenidate PET tracers.3 accumbens exhibited the same rostrolateral-to-caudomedial pat- In the most affected hemisphere, nigral iron accumulation pre- tern. Thus, the shape and location of the DaT-neuromelanin cor- ceded neuromelanin loss by 2–3 years in the sensorimotor and limbic relation patterns partially reflected the functional organization of territories. In the associative SNc, the prodromal phase appeared lon- striatonigral projections reported in non-human primates, al- ger, probably because of imprecise estimations due to minimal neuro- though with a larger overlap,49 potentially caused by: greater over- melanin changes in this region. For neuromelanin loss, the average lap in humans than in primates; loss of neuronal specificity as onset of the prodromal phase in the sensorimotor SNc (4.8 years) reported in Parkinson’s disease72; or technical issues deriving from agreed with our previous study (5.3 years).4 the coarser resolution of DaTScanTM and neuromelanin-sensitive MRI compared with histological methods. In the SNc, DaT-neuro- melanin correlations were mainly lateralized in the most affected Spatial relationships between dopaminergic and hemisphere, as previously shown.9–11 Probably, the least affected iron changes hemisphere was progressively affected along the same time- DaT loss in all striatal areas correlated with both SNc neuromela- shifted spatial gradient as the most affected hemisphere. nin loss and iron accumulation. In the most affected hemisphere, Alternatively, this marked lateralization could indicate different 3122 | BRAIN 2021: 144; 3114–3125 E. Biondetti et al. Figure 4 Spatial correlation patterns between nigral neuromelanin or nigral iron and striatal DaT in the SNc. For each striatal region (putamen, caud- ate nucleus and nucleus accumbens) and each functional territory (sensorimotor, associative and limbic), the DaT-neuromelanin correlation patterns (A–G), the neuromelanin-iron correlation patterns (K) and the DaT-iron correlation patterns (L–R) in the SNc are shown overlaid on the brain tem- plate. The DaT-neuromelanin regional overlaps are shown for the putamen, caudate and limbic areas (H–J) along with Dice scores quantifying the relative overlap between pairs of correlation patterns. In each image, the brain template is displayed using the same intensity range in arbitrary units. All images are displayed showing the most affected brain hemisphere on the left side of the image. For improved image visualization quality, the correlation patterns (A–G, K and L–R) were spatially upsampled by a factor of four using nearest-neighbour interpolation. AS = associative; Cd = caudate nucleus; LI = limbic; NAc = nucleus accumbens; Pu = putamen; SM = sensorimotor. trajectories of dopaminergic neuron dysfunction and neuromela- and neuromelanin correlated in the anterior but not the postero- nin-containing neuron loss in early disease stages.9 lateral SNc, which had greater neuromelanin loss. Possibly, the Striatal DaT loss and nigral iron accumulation correlated in the significant DaT-iron loss reflected different aspects of Parkinson’s bilateral SNc contradicting previous negative results,12 potentially disease progression.2 In the anteromedial SNc with no or minimal because of different nigral region of interest delineations (SNc here neuromelanin loss, significant neuromelanin-iron correlations as opposed to the entire SN in previous studies), the larger sample could reflect a redox-active iron overload in neuromelanin organ- size examined here, or different mean disease durations (1.5 years elles preceding neuronal death.2 In the posterolateral SNc with here as opposed to 6 years in previous studies).12 greater neuromelanin loss, absent neuromelanin-iron correlations Lastly, neuromelanin and iron only correlated in the anterome- could additionally reflect activated microglia and perivascular dial SNc as shown recently,14 suggesting a correspondence of this macrophage contents.2 SNc region with nigrosome-2. Contrary to previous findings, we could not identify a second cluster corresponding to nigrosome- Limitations and considerations 1.14 As nigrosome-1 is affected earlier in Parkinson’s disease, possibly, neuromelanin loss and iron accumulation in this region Patients with iRBD had a relatively high MDS-UPDRS-III OFF score followed different trajectories in our patient cohorts compared (i.e. 11.5 ± 6.3; Table 1). This aspect, as already discussed in a previ- with the one investigated by Langley et al.14 ous study on the ICEBERG cohort,4 could result from most patients The specific relationship between changes in SNc iron and being close to phenoconversion. Approaching phenoconversion those in the nigrostriatal dopaminergic pathway, spatiotemporally was also suggested by the fact that four iRBDs had started receiv- consistent (same temporal pattern in the three nigrostriatal terri- ing dopaminergic medication by the time of visit 2 (Table 1). In the tories beginning with striatal dopaminergic dysfunction, followed presence of phenoconversion, previous studies on iRBD have by SNc iron accumulation preceding SNc neuromelanin loss and, reported relatively high MDS-UPDRS-III scores73 (i.e. 44) and shown hence, dopaminergic cell loss) suggested interdependent dopa- a faster progression of motor examination signs 1–2 years before minergic and iron changes. This suggests that DaT changes phenoconversion.74 Moreover, a study stratifying iRBD patients occurred first without dopaminergic neuron loss, congruent with based on the outcome of DaTScanTM has reported that, iRBD with recent findings in a Parkinson’s disease animal model.69 Iron de- or without dopaminergic denervation, had MDS-UPDRS-III scores position was not detected in early prodromal Parkinson’s disease respectively equal to 10.5 ± 7.5 (indicative of prodromal Parkinson’s or in Parkinson’s disease at visit 2 but was estimated to occur be- disease) and 6.0 ± 4.8, whereas healthy control subjects had a score fore the onset of neuromelanin changes. Possible explanations in- equal to 3.2 ± 3.2.75 clude insufficient QSM sensitivity for detecting small increases in The present study had a methodological limitation, as reverse iron or a non-linear iron accumulation over time. Moreover, iron extrapolating DaT-SBR values in the iRBD group based on the Nigrostriatal changes in parkinsonism BRAIN 2021: 144; 3114–3125 | 3123 exponential fitting performed in the Parkinson’s disease group in nigral neuromelanin both assessed using MRI. Neuromelanin assumed that iRBDs had a similar temporal trajectory as patients changes followed the same (time-shifted) spatiotemporal gradient with Parkinson’s disease although this assumption may not apply as in the striatum. Correlations between the three measures in the across all individuals of either cohort. Moreover, this assumption SNc suggested a relationship between dopaminergic changes and has limited accuracy as some patients with iRBD appear to have iron deposition in Parkinson’s disease. normal striatal dopaminergic function on DaT-SPECT5,76 or 18F- DOPA PET.77,78 Thus, the results of reverse extrapolation should be Acknowledgements interpreted with caution as they reflect a potentially heteroge- neous iRBD group, as also suggested by the larger standard devia- The authors would like to thank all the subjects who participated tions in the DaT-SBR scatter box plots in this group compared to in this study. the Parkinson’s disease group (Supplementary Figs 1–7). There were also some technical limitations in the present study. First, the sensitivity of the imaging methods employed here, name- Funding ly, 3 T MRI and DaTScanTM, intrinsically limited the maximum spa- The ICEBERG study was funded by grants from the tial resolution achievable and, in turn, the accuracy of Investissements d’Avenir, IAIHU-06 (Paris Institute of spatiotemporal variations in dopamine, neuromelanin and iron that Neurosciences – IHU), ANR-11-INBS-0006, Fondation d’Entreprise could be detected. Based on the results of the present study, EDF, Biogen Inc., Fondation Thérèse and René Planiol, Fondation DaTScanTM had a higher sensitivity in detecting earlier and more Saint Michel, Unrestricted support for Research on Parkinson’s dis- subtle denervation changes in the striatum than neuromelanin-sen- ease from Energipole (M. Mallart), M.Villain and Société Française sitive or iron-sensitive 3 T MRI in the SNc. However, it is possible de Médecine Esthétique (M. Legrand). that more subtle spatiotemporal variations in nigral iron or neuro- melanin could be detected by performing MRI at higher field strengths (e.g. 7 T). MRI at 7 T could outperform MRI at 3 T, because Competing interests of both the known shortening of tissue relaxation times and higher E.B. reports grants from France Parkinson and Biogen Inc during image resolution achievable by increasing the static magnetic field the conduct of the study. R.G. and S.L. received grant funding from which, combined, could help detect more subtle changes in the SNc. Biogen Inc. M.H. is an employee of and owns stock/stock options in Alternatively, increased sensitivity to neuromelanin changes in the Biogen Inc. N.V. reports grants from Fondation Recherche SNc could be achieved using 18F-AV-1451 PET instead of neuromela- Alzheimer, grants from Département Médical Universitaire APHP, nin-sensitive MRI.79 Finally, even higher sensitivity to striatal Sorbonne Université, non-financial support from GE HEALTHCARE changes could be attained by using PET with DaT tracers instead of SAS, non-financial support from MERZ PHARMA France, non-fi- DaTScanTM because of the higher spatial resolution achievable with nancial support from UCB Pharma SA, non-financial support from PET. However, both 7 T MRI and PET are technically more challeng- MEDTRONIC France S.A.S, non-financial support from Laboratoire ing to use because of the stronger magnetic field inhomogeneities AGUETTANT, outside the submitted work. M.-O.H. reports receiv- induced in the brain at higher fields (MRI) or because a cyclotron is ing fees as a speaker from Lilly and as a consultant from Blue needed to produce the radiotracer on site (PET). Earth company. I.A. received honoraria from Roche Pharma, A second technical limitation was represented by the Idorsia Pharma and Ono Pharma, and speaking engagement from DaTScanTM measurements, which, in addition to being sensitive to UCB Pharma, unrelated to this study. J.-C.C. has served in advisory striatal dopaminergic uptake, could also reflect changes in boards for Air Liquide, Biogen Inc., Denali, Ever Pharma, Idorsia, pharmacological compensation mechanisms occurring in Prevail Therapeutic, Theranexus, UCB; has received grants from Parkinson’s disease, such as downregulation of DaT, especially in Sanofi and the Michael J Fox Foundation. All other authors report the early phases of disease.80 Thus, further large scale longitudinal no competing interests. studies using different tracers to monitor striatal presynaptic dopamine function (e.g. PET with VMAT2, DaT or 18F-fluorodopa) and MRI are needed to assess the potentially varying spatiotempo- Supplementary material ral relationship between dopamine function, neuromelanin loss and iron deposition in the nigrostriatal system. Supplementary material is available at Brain online. In general, further studies are needed to indicate which imag- ing-based biomarker or combination of biomarkers could provide comprehensive information on longitudinal changes in the nigros- References triatal system linked to Parkinson’s disease. In particular, longitu- 1. Greffard S, Verny M, Bonnet AM, et al. Motor score of the Unified dinal data spanning longer periods of time could enable the curves Parkinson Disease Rating Scale as a good predictor of Lewy described in Fig. 3 to be updated to reflect potential plateaus in body-associated neuronal loss in the substantia nigra. Arch striatal DaT loss and nigral iron accumulation or neuromelanin Neurol. 2006;63(4):584–588. loss. Notably, the cohort investigated here was enrolled in a pro- 2. Zucca FA, Segura-Aguilar J, Ferrari E, et al. Interactions of iron, spective 5-year longitudinal study with imaging data collected at dopamine and neuromelanin pathways in brain aging and three time points (Year 1, Year 3, and Year 5). Only data from the Parkinson’s disease. Prog Neurobiol. Aug 2017;155:96–119. first two time points were considered in the present study, as data 3. de la Fuente-Fernandez R, Schulzer M, Kuramoto L, et al. Age- collection is ongoing. Thus, future work will involve updating the specific progression of nigrostriatal dysfunction in Parkinson’s present analyses by also incorporating data from Year 5. disease. Ann Neurol. May 2011;69(5):803–810. In conclusion, the temporal evolution of changes in the nigros- 4. Biondetti E, Gaurav R, Yahia-Cherif L, et al. Spatiotemporal triatal dopaminergic system was characterised, which progressed changes in substantia nigra neuromelanin content in following the known spatial pattern from sensorimotor to associa- Parkinson’s disease. Brain. Aug 28 2020;143(9):2757–2770. tive and limbic striatal regions. Striatal dopaminergic dysfunction 5. Iranzo A, Lomen ~ a F, Stockner H, et al. Decreased striatal dopa- evaluated using DaTScanTM occurred first, subsequently associ- mine transporter uptake and substantia nigra hyperechogenic- ated with abnormal iron metabolism and finally with a reduction ity as risk markers of synucleinopathy in patients with 3124 | BRAIN 2021: 144; 3114–3125 E. Biondetti et al. idiopathic rapid-eye-movement sleep behaviour disorder: A 26. Rieu I, Martinez-Martin P, Pereira B, et al. International valid- prospective study. Lancet Neurol. 2010;9(11):1070–1077. ation of a behavioral scale in Parkinson’s disease without de- 6. Pyatigorskaya N, Sharman M, Corvol JC, et al. High nigral iron mentia. Mov Disord. 2015;30(5):705–713. deposition in LRRK2 and Parkin mutation carriers using R2 27. Marques JP, Kober T, Krueger G, van der Zwaag W, Van de relaxometry. Mov Disord. 2015;30(8):1077–1084. Moortele PF, Gruetter R. MP2RAGE, a self bias-field corrected se- 7. Hare DJ, Double KL. Iron and dopamine: A toxic couple. Brain. quence for improved segmentation and T1-mapping at high 2016;139(Pt 4):1026–1035. field. Neuroimage. 2010;49(2):1271–1281. 8. Vila M. Neuromelanin, aging, and neuronal vulnerability in 28. Sasaki M, Shibata E, Tohyama K, et al. Neuromelanin magnetic Parkinson’s disease. Mov Disord. 2019;34(10):1440–1451. resonance imaging of locus ceruleus and substantia nigra in 9. Martin-Bastida A, Lao-Kaim NP, Roussakis AA, et al. Parkinson’s disease. Neuroreport. 2006;17(11):1215–1218. Relationship between neuromelanin and dopamine terminals 29. Wang Y, Liu T. Quantitative susceptibility mapping (QSM): within the Parkinson’s nigrostriatal system. Brain. 2019;142(7): Decoding MRI data for a tissue magnetic biomarker. Magn Reson 2023–2036. Med. 2015;73(1):82–101. 10. Kuya K, Shinohara Y, Miyoshi F, Fujii S, Tanabe Y, Ogawa T. 30. Chang L-T. A method for attenuation correction in radionuclide Correlation between neuromelanin-sensitive MR imaging and computed tomography. IEEE Trans Nuclear Sci. 1978;25(1):638–643. (123)I-FP-CIT SPECT in patients with parkinsonism. 31. Santin MD. Optimised Generation of MRI Images by a Multi- Neuroradiology. 2016;58(4):351–356. Antenna MRI System. International patent application 11. Kuya K, Ogawa T, Shinohara Y, et al. Evaluation of Parkinson’s WO2019077246. 2019. disease by neuromelanin-sensitive magnetic resonance imag- 32. Liu T, Wisnieff C, Lou M, Chen W, Spincemaille P, Wang Y. ing and (123)I-FP-CIT SPECT. Acta Radiol. 2018;59(5):593–598. Nonlinear formulation of the magnetic field to source relation- 12. Isaias IU, Trujillo P, Summers P, et al. Neuromelanin imaging ship for robust quantitative susceptibility mapping. Magn Reson and dopaminergic loss in Parkinson’s disease. Front Aging Med. 2013;69(2):467–476. Neurosci. 2016;8:196. 33. Zhou D, Liu T, Spincemaille P, Wang Y. Background field re- 13. Reimao S, Ferreira S, Nunes RG, et al. Magnetic resonance correl- moval by solving the Laplacian boundary value problem. NMR ation of iron content with neuromelanin in the substantia nigra of Biomed. 2014;27(3):312–319. early-stage Parkinson’s disease. Eur J Neurol. 2016;23(2):368–374. 34. de Rochefort L, Brown R, Prince MR, Wang Y. Quantitative MR 14. Langley J, Huddleston DE, Crosson B, Song DD, Factor SA, Hu X. susceptibility mapping using piece-wise constant regularized Multimodal assessment of nigrosomal degeneration in inversion of the magnetic field. Magn Reson Med. Oct 2008;60(4): Parkinson’s disease. Parkinsonism Relat Disord. 2020;80:102–107. 1003–1009. 15. Nandhagopal R, Kuramoto L, Schulzer M, et al. Longitudinal pro- 35. Liu T, Liu J, de Rochefort L, et al. Morphology enabled dipole in- gression of sporadic Parkinson’s disease: A multi-tracer positron version (MEDI) from a single-angle acquisition: Comparison emission tomography study. Brain. 2009;132(Pt 11):2970–2979. with COSMOS in human brain imaging. Magn Reson Med. Sep 16. Bergsland N, Zivadinov R, Schweser F, Hagemeier J, Lichter D, 2011;66(3):777–783. Guttuso T Jr, Ventral posterior substantia nigra iron increases over 36. Liu J, Liu T, de Rochefort L, et al. Morphology enabled dipole in- 3 years in Parkinson’s disease. Mov Disord. 2019;34(7):1006–1013. version for quantitative susceptibility mapping using structural 17. Snyder AM, Connor JR. Iron, the substantia nigra and related consistency between the magnitude image and the susceptibil- neurological disorders. Biochim Biophys Acta. Jul 2009;1790(7): ity map. Neuroimage. 2012;59(3):2560–2568. 606–614. 37. Straub S, Schneider TM, Emmerich J, et al. Suitable reference 18. Hughes AJ, Daniel SE, Kilford L, Lees AJ. Accuracy of clinical tissues for quantitative susceptibility mapping of the brain. diagnosis of idiopathic Parkinson’s disease: A clinico-patho- logical study of 100 cases. J Neurol Neurosurg Psychiatry. 1992; Magn Reson Med. 2017;78(1):204–214. 38. Lim IA, Faria AV, Li X, et al. Human brain atlas for automated re- 55(3):181–184. 19. American Academy of Sleep Medicine. International classification gion of interest selection in quantitative susceptibility map- of sleep disorders. 3rd ed. American Academy of Sleep Medicine; ping: Application to determine iron content in deep gray matter 2014. structures. Neuroimage. 2013;82:449–469. 20. Hoehn MM, Yahr MD. Parkinsonism: Onset, progression and 39. Ourselin S, Roche A, Subsol G, Pennec X, Ayache N. mortality. Neurology. 1967;17(5):427–442. Reconstructing a 3D structure from serial histological sections. 21. Goetz CG, Tilley BC, Shaftman SR, et al. Movement Disorder Image and Vision Computing. Jan 2001;19(1-2):25–31. Society-sponsored revision of the Unified Parkinson’s Disease 40. Modat M, Ridgway GR, Taylor ZA, et al. Fast free-form deform- Rating Scale (MDS-UPDRS): scale presentation and clinimetric ation using graphics processing units. Comput Methods Programs testing results. Mov Disord. 2008;23(15):2129–2170. Biomed. 2010;98(3):278–284. 22. Mattis S. Dementia Rating Scale. Professional manual. 41. Yang J, Huang SC, Mega M, et al. Investigation of partial volume Psychological Assessment Resources; 1988. correction methods for brain FDG PET studies. IEEE Trans 23. Mattis S, Bellak L, Karasu TB. Mental status examination for or- Nuclear Sci. 1996;43(6):3322–3327. ganic mental syndrome in the elderly patient. In: Bellak L, 42. Patenaude B, Smith SM, Kennedy DN, Jenkinson M. A Bayesian Karasu TB, eds. Geriatric psychiatry a handbook for psychiatrists model of shape and appearance for subcortical brain segmenta- and primary care physicians. Grune & Stratton; 1976:77–121. tion. Neuroimage. 2011;56(3):907–922. 24. Nasreddine ZS, Phillips NA, Bédirian V, et al. The Montreal 43. Jenkinson M, Beckmann CF, Behrens TE, Woolrich MW, Smith Cognitive Assessment, MoCA: A brief screening tool for mild SM. Fsl. Neuroimage. 2012;62(2):782–790. cognitive impairment. J Am Geriatr Soc. 2005;53(4):695–699. 44. Gerlach M, Gsell W, Kornhuber J, et al. A post mortem study on 25. Ardouin C, Chereau I, Llorca PM, et al. Evaluation des troubles neurochemical markers of dopaminergic, GABA-ergic and glu- comportementaux hyper- et hypodopaminergiques dans la tamatergic neurons in basal ganglia-thalamocortical circuits in maladie de Parkinson. [Assessment of hyper- and hypodopami- Parkinson syndrome. Brain Research. 1996;741(1–2):142–152. nergic behaviors in Parkinson’s disease]. Rev Neurol (Paris). 2009; 45. Tzourio-Mazoyer N, Landeau B, Papathanassiou D, et al. 165(11):845–856. Automated anatomical labeling of activations in SPM using a Nigrostriatal changes in parkinsonism BRAIN 2021: 144; 3114–3125 | 3125 macroscopic anatomical parcellation of the MNI MRI single- 63. Pyatigorskaya N, Gaurav R, Arnaldi D, et al. Magnetic resonance subject brain. Neuroimage. 2002;15(1):273–289. imaging biomarkers to assess substantia nigra damage in idiopath- 46. O’Brien JT, Colloby S, Fenwick J, et al. Dopamine transporter ic rapid eye movement sleep behavior disorder. Sleep. 2017;40(11). loss visualized with FP-CIT SPECT in the differential diagnosis 64. Sun J, Lai Z, Ma J, et al. Quantitative evaluation of iron content of dementia with Lewy bodies. Arch Neurol. 2004;61(6):919–925. in idiopathic rapid eye movement sleep behavior disorder. Mov 47. Bardinet E, Bhattacharjee M, Dormont D, et al. A three-dimen- Disord. 2020;35(3):478–485. sional histological atlas of the human basal ganglia. II. Atlas de- 65. Gaurav R, Yahia-Cherif L, Pyatigorskaya N, et al. Longitudinal formation strategy and evaluation in deep brain stimulation for changes in neuromelanin MRI signal in Parkinson’s disease: A Parkinson disease. J Neurosurg. Feb 2009;110(2):208–219. progression marker. Mov Disord. 2021;36(7):1592–1602. 48. Yelnik J, Bardinet E, Dormont D, et al. A three-dimensional, 66. Sulzer D, Cassidy C, Horga G, et al. Neuromelanin detection by histological and deformable atlas of the human basal ganglia. I. magnetic resonance imaging (MRI) and its promise as a bio- Atlas construction based on immunohistochemical and MRI marker for Parkinson’s disease. NPJ Parkinsons Dis. 2018;4:11. data. Neuroimage. 2007;34(2):618–638. 67. McGeer PL, Itagaki S, Boyes BE, McGeer EG. Reactive microglia 49. Haber SN. The place of dopamine in the cortico-basal ganglia are positive for HLA-DR in the substantia nigra of Parkinson’s circuit. Neuroscience. Dec 12 2014;282:248–257. and Alzheimer’s disease brains. Neurology. 1988;38(8): 50. R Core team. R: A Language and Environment for Statistical 1285–1291. Computing. Version 3.6.1. R Foundation for Statistical 68. Beach TG, Sue LI, Walker DG, et al. Marked microglial reaction Computing, Vienna, Austria; 2017. https://www.R-project.org/ in normal aging human substantia nigra: Correlation with 51. Prange S, Danaila T, Laurencin C, et al. Age and time course of extraneuronal neuromelanin pigment deposits. Acta long-term motor and nonmotor complications in Parkinson Neuropathol. 2007;114(4):419–424. disease. Neurology. 2019;92(2):e148–e160. 69. Carballo-Carbajal I, Laguna A, Romero-Gimenez J, et al. Brain 52. Zecca L, Stroppolo A, Gatti A, et al. The role of iron and copper tyrosinase overexpression implicates age-dependent neurome- molecules in the neuronal vulnerability of locus coeruleus and lanin production in Parkinson’s disease pathogenesis. Nat substantia nigra during aging. Proc Natl Acad Sci U S A. 2004; Commun. 2019;10(1):973. 101(26):9843–9848. 70. Kraemmer J, Kovacs GG, Perju-Dumbrava L, Pirker S, Traub- 53. Ward RJ, Zucca FA, Duyn JH, Crichton RR, Zecca L. The role of Weidinger T, Pirker W. Correlation of striatal dopamine trans- iron in brain ageing and neurodegenerative disorders. Lancet porter imaging with post mortem substantia nigra cell counts. Neurol. 2014;13(10):1045–1060. Mov Disord. 2014;29(14):1767–1773. 54. Xing Y, Sapuan A, Dineen RA, Auer DP. Life span pigmentation 71. Karimi M, Tian L, Brown CA, et al. Validation of nigrostriatal changes of the substantia nigra detected by neuromelanin-sen- positron emission tomography measures: Critical limits. Ann sitive MRI. Mov Disord. Nov 2018;33(11):1792–1799. Neurol. Mar. 2013;73(3):390–396. 55. van Dyck CH, Seibyl JP, Malison RT, et al. Age-related decline in 72. Bronfeld M, Bar-Gad I. Loss of specificity in Basal Ganglia related dopamine transporters: analysis of striatal subregions, nonlin- movement disorders. Front Syst Neurosci. 2011;5:38. ear effects, and hemispheric asymmetries. Am J Geriatric 73. Postuma RB, Iranzo A, Hu M, et al. Risk and predictors of de- Psychiatry. 2002;10(1):36–43. mentia and Parkinsonism in idiopathic REM sleep behaviour 56. Varrone A, Dickson JC, Tossici-Bolt L, et al. European multi- disorder: A multicentre study. Brain. Mar 1 2019;142(3):744–759. centre database of healthy controls for [123I]FP-CIT SPECT 74. Fereshtehnejad SM, Yao C, Pelletier A, Montplaisir JY, Gagnon (ENC-DAT): Age-related effects, gender differences and evalu- JF, Postuma RB. Evolution of prodromal Parkinson’s disease and ation of different methods of analysis. Eur J Nucl Med Mol Imaging. 2013;40(2):213–227. dementia with Lewy bodies: A prospective study. Brain. 2019; 57. Nichols TE, Holmes AP. Nonparametric permutation tests for 142(7):2051–2067. functional neuroimaging: A primer with examples. Hum Brain 75. Dusek P, Ibarburu V, Bezdicek O, et al. Relations of non-motor Mapp. 2002;15(1):1–25. symptoms and dopamine transporter binding in REM sleep be- 58. Lee CS, Samii A, Sossi V, et al. In vivo positron emission tomo- havior disorder. Sci Rep. 2019;9(1):15463. graphic evidence for compensatory changes in presynaptic 76. Iranzo A, Santamaria J, Valldeoriola F, et al. Dopamine trans- dopaminergic nerve terminals in Parkinson’s disease. Ann porter imaging deficit predicts early transition to synucleinop- Neurol. 2000;47(4):493–503. athy in idiopathic rapid eye movement sleep behavior disorder. 59. Langley J, Huddleston DE, Sedlacik J, Boelmans K, Hu XP. Ann Neurol. Sep. 2017;82(3):419–428. Parkinson’s disease-related increase of T2-weighted hypoin- 77. Knudsen K, Fedorova TD, Hansen AK, et al. In-vivo staging tensity in substantia nigra pars compacta. Mov Disord. 2017; of pathology in REM sleep behaviour disorder: A multimo- 32(3):441–449. dality imaging case-control study. Lancet Neurol. 2018;17(7): 60. Pyatigorskaya N, Sanz-Morere CB, Gaurav R, et al. Iron imaging 618–628. as a diagnostic tool for Parkinson’s disease: A systematic review 78. Stokholm MG, Iranzo A, Østergaard K, et al. Assessment of neu- and meta-analysis. Front Neurol. 2020;11:366. roinflammation in patients with idiopathic rapid-eye-move- 61. Du G, Liu T, Lewis MM, et al. Quantitative susceptibility map- ment sleep behaviour disorder: a case-control study. Lancet ping of the midbrain in Parkinson’s disease. Mov Disord. Mar Neurol. 2017;16(10):789–796. 2016;31(3):317–324. 79. Coakeley S, Cho SS, Koshimori Y, et al. [(18)F]AV-1451 binding to 62. Bernheimer H, Birkmayer W, Hornykiewicz O, Jellinger K, neuromelanin in the substantia nigra in PD and PSP. Brain Struct Seitelberger F. Brain dopamine and the syndromes of Parkinson Funct. Mar 2018;223(2):589–595. and Huntington Clinical, morphological and neurochemical 80. Stoessl AJ. Central pharmacokinetics of levodopa: Lessons from correlations. J Neurol Sci. 1973;20(4):415–455. imaging studies. Mov Disord. 2015;30(1):73–79.