Article Text

Download PDFPDF

Research paper
Serotonin transporter binding and anxiety symptoms in Parkinson’s disease
  1. Merijn Joling1,2,3,4,
  2. Odile A van den Heuvel3,4,5,
  3. Henk W Berendse1,3,
  4. Jan Booij2,3,
  5. Chris Vriend3,4,5
  1. 1 Department of Neurology, VU University Medical Center, Amsterdam, The Netherlands
  2. 2 Department of Nuclear Medicine, Academic Medical Center, Amsterdam, The Netherlands
  3. 3 Research Program Neurodegeneration, Amsterdam Neuroscience, Amsterdam, The Netherlands
  4. 4 Department of Anatomy and Neurosciences, VU University Medical Center, Amsterdam, The Netherlands
  5. 5 Department of Psychiatry, VU University Medical Center, Amsterdam, The Netherlands
  1. Correspondence to Merijn Joling, Department of Neurology, VU University Medical Center, P.O. Box 7057, 1007 MB Amsterdam, The Netherlands; me.joling{at}


Background Anxiety is a common neuropsychiatric symptom in Parkinson’s disease (PD), yet the neural mechanisms have been scarcely investigated. Disturbances in dopaminergic and serotonergic signalling may play a role in its pathophysiology. 123I-N-ω-fluoropropyl-2β-carbomethoxy-3β-(4-iodophenyl)nortropane (123I-FP-CIT) is a single-photon emission CT radiotracer, and its binding in striatal and extrastriatal subcortical brain areas represents predominant binding to the presynaptic dopamine transporter (DAT) and the serotonin transporter (SERT), respectively. Availability of DAT and SERT may thus provide an in vivo measure for the integrity of both dopamine and serotonin neurons.

Methods We studied the association between anxiety symptoms, measured with an affective subscale of the Beck Anxiety Inventory, and (extra)striatal 123I-FP-CIT binding in 127 non-demented patients with PD with a median disease duration of 2.55 (IQR 2.90) years. We conducted the analyses on patients currently on or not on dopamine replacement therapy (DRT).

Results Severity of anxiety symptoms showed a significant negative association with 123I-FP-CIT binding ratios in the right thalamus (β=−0.203, p=0.019; ΔR2=0.040) (multiple testing pcorr <0.020). In the subgroup of patients not on DRT (n=81), we found a significant negative association between anxiety and thalamic 123I-FP-CIT binding ratios bilaterally (right: β=−0.349, p=0.001, ΔR2=0.119; left: β=−0.269, p=0.017, ΔR2=0.071) (pcorr <0.020).

Conclusion This study shows that higher levels of anxiety in patients with PD are associated with lower thalamic 123I-FP-CIT binding, pointing towards a contribution of serotonergic degeneration to anxiety symptoms in PD.

View Full Text

Statistics from


Anxiety is a common neuropsychiatric symptom in Parkinson’s disease (PD). It has a higher prevalence in PD than in the general elderly population,1 increases the psychological burden of disease, and is associated with exacerbation of motor symptoms such as dyskinesia, freezing of gait and on/off fluctuations.2–4 In addition, anxiety appears to occur more frequently during wearing-off5 and can be alleviated by dopaminergic medication.6–8 This suggests an association of anxiety with the waxing and waning of extracellular dopamine levels that results from the degeneration of dopaminergic neurons and the compensatory treatment with dopamine replacement therapy (DRT).

Other studies have suggested that other factors beyond low dopamine levels during the wearing-off phase are involved in the pathophysiology of anxiety of PD.9 In non-PD samples, anxiety has been associated with serotonergic, noradrenergic, γ-aminobutyric acid (GABA)-ergic and cholinergic deficits (see ref 10 for a review). These neurotransmitter systems are also affected in PD. Lower serotonin transporter (SERT) binding, for example, has previously been described in PD,11 12 and it has also been suggested that degeneration of the serotonergic system plays a role in anxiety in PD (see ref 13 for a review).

123I-N-ω-fluoropropyl-2β-carbomethoxy-3β-(4-iodophenyl)nortropane (123I-FP-CIT) is a single-photon emission CT (SPECT) radiotracer that binds with high affinity to both the presynaptic dopamine transporter (DAT) and to the presynaptic SERT, although with lower affinity.14 Previous studies have shown that striatal 123I-FP-CIT binding (putamen and caudate nucleus) predominantly represents binding to DAT, whereas extrastriatal binding in subcortical brain areas (eg, thalamus, hypothalamus, midbrain, pons) predominantly represents binding to SERT.15 16 Consequently, striatal 123I-FP-CIT binding to DAT and extrastriatal 123I-FP-CIT binding to SERT could serve as proxy for the integrity of the dopaminergic and serotonergic system, respectively, and allow us to study their involvement in PD-related anxiety symptoms.

Molecular imaging studies that investigated the association between striatal DAT availability and PD-related anxiety have shown mixed results. Some studies reported a positive association,17 18 while others reported a negative association.19 20 These studies included patients with PD at different disease stages and medication states, which may be a potential explanation for the inconsistency in the reported findings. To the best of our knowledge, the association between SERT availability and anxiety in PD has not yet been studied.

Brain regions that are highly innervated by serotonergic projections have previously been associated with anxiety in PD in a structural brain imaging study (eg, the amygdala),21 and are also implicated in anxiety in non-PD samples (eg, the thalamus)22 (see ref 23 for a review). Based on the aforementioned results we hypothesised that extrastriatal SERT in the amygdala, hippocampus and thalamus would show a negative association with the severity of anxiety symptoms in PD. Based on the close relationship between anxiety and depression,4 the involvement of DAT in depression,24 and the aforementioned results, we also analysed the association between anxiety and DAT binding in the striatum.

Patients and methods


For this cross-sectional study, we selected patients with PD from a database of consecutive cases who presented between May 2008 and July 2015 to the outpatient clinic for movement disorders of the neurology department of the VU University Medical Center (VUmc) in Amsterdam, The Netherlands. Both an 123I-FP-CIT SPECT scan and a T1-weighted MRI brain scan had to be available for a patient to be eligible for participation. In addition, the availability of a Beck Anxiety Inventory (BAI) score was an inclusion criterion. We excluded patients on selective serotonin reuptake inhibitors (SSRIs), since these drugs can influence 123I-FP-CIT binding to the SERT.15 Based on the Mini-Mental State Examination score, we excluded patients scoring below 25 from this study to exclude patients with signs of dementia (see flow chart in figure 1 for inclusion and exclusion criteria). Movement disorder specialists clinically established a diagnosis of PD according to the UK PD Society Brain Bank criteria,25 supported by an abnormal 123I-FP-CIT SPECT scan in 124 (97.6%) patients. Three patients (2.4%) had a scan that was visually abnormal, but still normal with quantification. All included patients gave written informed consent to use their clinical and neuroimaging data for scientific purposes, and the study was approved by the local medical ethics committee.

Figure 1

Patients included in the study. 123I-FP-CIT, 123I-N-ω-fluoropropyl-2β-carbomethoxy-3β-(4-iodophenyl)nortropane; BAI, Beck Anxiety Inventory; MMSE, Mini-Mental State Examination; PD, Parkinson’s disease; SSRI, selective serotonin reuptake inhibitors.

Clinical characteristics

We evaluated the severity of the motor symptoms with the Unified Parkinson’s Disease Rating Scale – motor section (UPDRS-III)26; of the 46 patients on DRT, 31 patients (67.4%) were in the ‘on’ state, 5 (10.9%) were in the ‘off’ state, and of another 10 patients (21.7%) state was unknown. Levodopa equivalent daily dose (LEDD) was calculated for patients with DRT as described previously.27 On the same day as the UPDRS-III, anxiety symptoms were assessed with the BAI28 and depressive symptoms with the Beck Depression Inventory (BDI).29 Eight patients were on anxiolytics (four on benzodiazepines, two on a tricyclic antidepressant and one on zopiclone).

Beck anxiety inventory

Patients were asked to fill out the BAI,28 a 21-item questionnaire asking patients to report anxiety symptoms over the last week, ranging from 0 (not at all) to 3 (severe). A total score of >12 is considered to represent clinically relevant anxiety in PD.30 For this study, up to three missing items were accepted, and in that case the values were imputed with the mean score of the available items. Patients with >3 missing values were excluded. Since many symptoms of anxiety overlap with the motor symptoms of PD, we used the BAI affective subscale (BAIaffective) to assess ‘affective symptoms’. BAIaffective is a subset of BAI items covering the affective aspects of anxiety that are not directly associated with PD-related motor symptoms.31 This subscale was also previously applied to study the volumetric brain correlates of anxiety symptoms in PD.21

123I-FP-CIT SPECT image acquisition and preprocessing

123I-FP-CIT was intravenously administered in a dose of approximately 185 MBq (specific activity >185 MBq/nmol; radiochemical purity >99%; produced as DaTSCAN according to good manufacturing practices criteria at GE Healthcare, Eindhoven, The Netherlands). Static images were obtained for 30 min after 3–4 hours using a dual-head gamma camera (E.Cam; Siemens, Munich, Germany) with a fan-beam collimator. Images were reconstructed as described earlier,24 and reoriented to an anterior-posterior commissure plane in Statistical Parametric Mapping V.12 software (SPM12; Wellcome Trust Centre for Neuroimaging, London, UK).

MRI Timage acquisition

Structural images were acquired using a three-dimensional T1-weighted sequence on different MRI systems at the VUmc (see online supplementary material for all scan parameters).

Supplementary file 1

Regions of interest

We used the striatal caudate nucleus, putamen and nucleus accumbens, and extrastriatal amygdala, hippocampus and thalamus as regions of interest (ROIs). All ROIs were individually constructed using FreeSurfer V.5.3 (Athinoula A Martinos Center for Biomedical Imaging, Boston, Massachusetts, USA) with default settings. The putamen was divided into an anterior and posterior putamen by a line perpendicular to the anterior commissure in the mid-sagittal plane. The caudate nucleus and nucleus accumbens FreeSurfer segmentations were combined (left and right separately) to avoid spill-over effects when calculating the DAT binding ratios.

All ROIs were visually inspected for segmentation errors and if necessary excluded from analysis; this resulted in one pairwise exclusion for the bilateral anterior and posterior putamen, and two pairwise exclusions for the bilateral caudate/accumbens.

123I-FP-CIT SPECT and T1 coregistration

Because 123I-FP-CIT SPECT contains insufficient anatomical details, coregistering the SPECT scan to a T1-weighted image is often challenging. We therefore devised a method to optimise the process. Voxel intensity in the 123I-FP-CIT SPECT scan is highest in the striatum. Using the tools from the FMRIB Software Library (FSL V.5.0.8;, we added the FreeSurfer segmentations of the striatal regions to each patient’s T1-weighted image. The intensity of the striatal regions in the T1-weighted image was increased to obtain an image in which—like in an 123I-FP-CIT SPECT scan—the striatal regions were easily distinguishable from the background. Coregistration was subsequently successfully performed in SPM12 using the hyperintense striatal regions as a common landmark in both images (see online supplementary figure S1 for a graphical representation of the processing pipeline).

123I-FP-CIT SPECT image analysis

We calculated binding ratios per subject for the ROIs in the DAT-rich striatum and the SERT-rich extrastriatal areas. The cerebellum was used as the reference region (REF; WFU Pickatlas, Wake Forest University, Winston-Salem, North Carolina, USA; automated anatomical labelling atlas; bilateral Crus 2). We converted the REF mask from Montreal Neurological Institute (MNI) space to subject space by using the inverse normalisation parameters that were obtained when converting the T1-weighted MRI scan to the MNI space. Binding ratios were calculated according to the following formula: ((ROI−REF)/REF).

Voxel-based analysis

We performed voxel-based multiple regression analyses with age as covariate in SPM12 to corroborate the findings of our ROI analyses. Masks were applied the same way as in an earlier study.32 Statistical threshold was set to p<0.050, family-wise error, corrected for multiple comparisons.


We assessed normality of data by plotting histograms, examining Q-Q plots and using Kolmogorov-Smirnov tests. After checking for multicollinearity, homoscedasticity and independence of variables, we performed hierarchical multiple regression analyses with BAIaffective as the independent factor, and age as nuisance covariate. For the regression analyses, we calculated multiple comparison-corrected p values with Simple Interactive Statistical Analysis (, a tool that uses the mean association between variables that are mutually correlated (binding ratios in three different bilateral striatal ROIs and three different bilateral extrastriatal ROIs) for the alpha correction (r=0.7 striatal ROIs, r=0.5 extrastriatal ROIs), and allows a less stringent correction than the Bonferroni method for multiple comparisons. For the striatal ROIs this resulted in a statistical threshold (pcorr) of pcorr<0.030 and for extrastriatal ROIs a pcorr<0.020. We considered a p value between 0.050 and 0.100 for clinical data, or between pcorr and p=0.050 for the binding ratios as a trend. Analyses were performed on the total group and on groups stratified for use of DRT (n=46 with medication, n=81 without). Post hoc analyses were performed to check for influence of volume on thalamic findings. For this, ROI volumes obtained with FreeSurfer were added to a hierarchical multiple regression analysis as dependent variable. To avoid possible scanner effects on volume measures, we used only scans that were acquired on one particular scanner (GE Signa HDxT 3T, General Electric, Milwaukee, Wisconsin, USA) (n=92) (see online supplementary material for all scan parameters). All analyses of clinical characteristics and ROIs were performed on SPSS V.22.


Group characteristics

The characteristics of the 127 patients with PD are summarised in table 1. The BAIaffective subscale correlated positively with the BDI score (r=0.703, p<0.001). This was true for both patients with and without DRT (r=0.758, p<0.001; r=0.654, p<0.001, respectively). BAIaffective subscale did not correlate with age, disease duration, UPDRS-III or LEDD.

Table 1

Patient characteristics

ROI-based 123I-FP-CIT SPECT analyses

Striatal DAT binding

The multiple regression analysis on (1) the total group and (2) separately for the patients with or without DRT did not show any significant associations between striatal 123I-FP-CIT binding ratios and BAIaffective score. In patients without DRT, however, we observed a trend-significant negative association between 123I-FP-CIT binding ratios in the right anterior putamen and the BAIaffective (β=−0.236, p=0.038, ΔR2=0.054) (pcorr <0.030).

Extrastriatal SERT binding

In the total group of patients with PD, BAIaffective scores showed a statistically significant negative association with 123I-FP-CIT binding ratios in the right thalamus (β=−0.203, p=0.019, ΔR2=0.040), and a trend-significant negative association in the left thalamus (β=−0.186, p=0.039, ΔR2=0.034) and right amygdala (β=−0.200, p=0.026, ΔR2=0.039) (pcorr <0.020).

In patients with PD without DRT, we observed a significant negative association between BAIaffective and 123I-FP-CIT binding in the right and left thalamus (right: β=−0.349, p=0.001, ΔR2=0.119; left: β=−0.269, p=0.017, ΔR2=0.071) (pcorr <0.020) (see figure 2). This association was not evident in patients using DRT (right: β=0.016, p=0.913, ΔR2=0; left: β=−0.017, p=0.912, ΔR2=0). In addition, we saw a trend-significant negative association between BAIaffective and 123I-FP-CIT binding ratios in the right hippocampus for patients without DRT (β=−0.231, p=0.040, ΔR2=0.052) (pcorr <0.020).

Figure 2

Scatter plots showing association between serotonin transporter binding ratios in the thalamus and the affective subscale of the Beck Anxiety Inventory (BAIaffective) in Parkinson’s disease (PD) patients without dopamine replacement therapy (DRT).

Voxel-based 123I-FP-CIT SPECT analysis

Striatal DAT binding

The voxel-based multiple regression analysis of the striatal areas did not show any significant associations between striatal 123I-FP-CIT binding ratios and BAIaffective scores corrected for age.

Extrastriatal SERT binding

In line with the results of the ROI analysis, the BAIaffective score showed a significant negative association with 123I-FP-CIT binding ratios in the left posterior thalamus, corrected for age (see table 2, online supplementary figure S2). For none of the other ROIs did we find a significant association.

Table 2

Voxel-based analysis

Post hoc analysis: ROI volume

Volume of the bilateral thalamus was not associated with BAIaffective: neither in patients with (n=37; left: β=0.209, p=0.151, ΔR2=0.042; right: β=0.124, p=0.368, ΔR2=0.015) nor without DRT (n=55; left: β=−0.128, p=0.175, ΔR2=0.016; right: β=−0.031, p=0.745, ΔR2=0.001).


In this study we investigated the relationship between anxiety symptoms and 123I-FP-CIT binding in striatal and extrastriatal brain regions in patients with PD. We found that 123I-FP-CIT binding ratios in the thalamus, predominantly representing SERT availability,15 were negatively associated with the severity of anxiety symptoms, supporting our hypothesis. This effect was mainly driven by the group of patients not using DRT.

To the best of our knowledge, this is the first study to show an association between SERT binding in the thalamus and anxiety symptoms in a PD population. Using 11C-3-amino-4-(2-dimethylaminomethylphenylsulfanyl)-benzonitrile (11C-DASB), a selective SERT positron emission tomography (PET) tracer, Reimold and coworkers have previously reported an association between SERT binding and anxiety in non-PD samples: in patients with unipolar major depression and in patients with obsessive-compulsive disorder.22 33 Using both 11C-DASB and the selective DAT PET tracer 11C-N-(3-iodoprop-2E-enyl)-2β-carbomethoxy-3β-(4-methylphenyl)nortropane (11C-PE2I), Maillet et al compared apathetic and non-apathetic patients with PD, and reported lower SERT availability in the thalamus, pallidum and mesocorticolimbic and mesostriatal pathways in apathetic patients who were more depressed and more anxious. However, they did not observe a direct relationship between anxiety symptoms and SERT binding in the thalamus.34 Others have reported a relationship between thalamic SERT availability and fatigue in PD,11 and evidence of reduced thalamic SERT availability in PD versus healthy controls.35 36

Lower SERT availability can be interpreted as (1) downregulation of SERT and/or (2) degeneration of serotonergic projections. Downregulation would result in a higher serotonin availability in the synaptic cleft, while with neurodegeneration there would be lower serotonin availability. Much evidence points to dysfunction of the serotonergic system in the pathophysiology of anxiety; it has been thought that anxiety originates from a lack of serotonin (see ref 23 for a review). Moreover, evidence of loss of SERT binding in patients with PD has been reported,11 12 as well as a link to anxiety in PD (see for a review ref 13). Consequently, we interpret degeneration of the serotonergic system to be the most likely cause of the lower SERT binding in our patients with PD.

Studies on the neural correlates of anxiety mainly involve the sensory and limbic circuits, and their reciprocal interactions.37 These networks comprise anatomical regions including nuclei in the thalamus and amygdala (for a review see refs 38 39). The amygdala is associated with both regulation and production of anxiety (see for reviews refs 40 41), and has been shown to be hyperactive in several anxiety disorders (for a review see ref 37). In our study, however, we only observed a trend-significant association of SERT availability in the right amygdala with anxiety. This may be related to the modest affinity of 123I-FP-CIT for SERT, and the small size of the amygdala ROI, resulting in low specific to non-specific binding. Research with more selective SERT tracers (eg, 11C-DASB) is needed for more detailed data on the role of amygdalar SERT in PD-related anxiety.

A growing body of evidence suggests that the thalamus is able to regulate parts of the amygdala. According to a study performed in mice, the thalamus regulates fear processing in the lateral division of the central amygdala, thus coordinating conditioned fear.42 Also in healthy human subjects, a functional connection between the thalamus and the central amygdala has been demonstrated. Using functional MRI (fMRI), this connection was found to be disturbed in patients with generalised anxiety disorder.43 Moreover, Planetta et al showed with diffusion tensor imaging that the integrity of thalamic fibres projecting from the dorsomedial nucleus of the thalamus to the amygdala is reduced in de novo patients with PD compared with healthy controls.44 Taken together, these results seem to suggest that the thalamus has modulatory effects on the amygdala.

The SPECT data of this study do not offer information about the functional deficit in anxiety, but as argued above we assume lower SERT availability to imply lower serotonin presence. A functional study in ferrets has shown that serotonin has an inhibitory effect on the activity of nuclei in the dorsal thalamus.45 In addition, an fMRI study performed in patients with social anxiety disorder showed that the SSRI paroxetine reduced the activation in the thalamus, compared with placebo.46 Although this currently remains speculation, this may imply that reduced serotonin levels, for example, due to PD-related neurodegeneration, lead to dysregulation of the thalamus and dysfunctional coupling between the thalamus and amygdala, possibly resulting in increased vulnerability to anxiety.

Noticeably, the presence of a bilaterally significant association of lower thalamic SERT availability with anxiety was restricted to patients with PD who were not using DRT. Several studies have reported a relief of anxiety symptoms after administration of DRT.6–8 In our study, however, we observed higher average anxiety scores in patients on DRT. It has been suggested that the mechanisms underlying neuropsychiatric symptoms like anxiety may differ between disease stages, with early PD stages exhibiting stronger serotonergic involvement and later stages of PD exhibiting predominant dopaminergic involvement.34 Indeed, in our population, the average disease duration in patients on DRT was longer. An alternative explanation is a phenomenon that has been shown in a rodent model of PD, where L-dopa is converted to dopamine in serotonergic neurons. This vesicle-stored dopamine displaces vesicles containing serotonin, resulting in dysregulated serotonin secretion, possibly increasing SERT levels.36 47

123I-FP-CIT binding ratios in DAT-rich striatal regions did not show any association with severity of anxiety symptoms that survived correction for multicomparisons. Nevertheless, in patients without DRT, we observed a trend-significant negative association in the right anterior putamen. Previous analyses of the relationship between anxiety and striatal DAT availability in patients with PD have provided mixed results. In one SPECT study, using 123I-FP-CIT as a radiotracer, anxiety was positively associated with DAT binding,17 whereas in another study DAT and anxiety were negatively correlated.19 Similarly, in SPECT studies using 99mTc-2β-((N,N'-bis(2-mercaptoethyl)ethylene diamino)methyl), 3β-(4-chlorophenyl)tropane (99mTc-TRODAT-1)—another DAT tracer—both positive18 and negative associations20 between striatal DAT binding and anxiety have been found.

The main strengths of this study are the large number of participants who were scanned using the same SPECT camera and the relatively large number of patients with PD who were drug-naïve. Another strength is the use of individual MRI brain scans to more accurately determine 123I-FP-CIT binding, which was particularly helpful in the extrastriatal brain areas. The present study is however not without limitations. In this study we used a single radiotracer, 123I-FP-CIT SPECT, to simultaneously study the integrity of the striatal dopaminergic and the extrastriatal serotonergic system. Furthermore, the resolution of SPECT scans compared with PET precludes a thorough determination of the exact thalamic subareas involved. Another limitation is that the questionnaire we used assesses anxiety-associated symptoms and does not allow a formal clinical diagnosis of an anxiety disorder. Furthermore, the amount of variance in anxiety symptoms that could be explained by thalamic SERT was relatively small, suggesting that other neurobiological factors likely contribute to the pathophysiology of PD-related anxiety.

In conclusion, this study shows a significant negative association between severity of anxiety symptoms and serotonergic integrity in the thalamus of patients with PD, particularly in patients not using DRT. This observation may help in unravelling the pathogenesis of anxiety symptoms in PD, opening up the possibility of future improvements in the management of PD-related anxiety.


We would like to thank Dr Anouk Schrantee from the Academic Medical Center, Amsterdam, The Netherlands, for her valuable insights and advice in developing the123I-FP-CIT SPECT–MRI coregistration method.


View Abstract


  • Contributors MJ: planning, conducting, analysis and writing. OAvdH: planning, analysis and writing. HWB: planning, consulting, reading and correcting manuscript. JB: planning, consulting, reading and correcting manuscript. CV: planning, conducting, analysis and writing.

  • Funding MJ’s salary was paid by a research grant from GE Healthcare (paid to the institution). HWB is coapplicant of research grants obtained from GE Healthcare and Roche (paid to the institution). JB is consultant at GE Healthcare and received research grants from GE Healthcare (paid to the institution). GE Healthcare did not play a role in the design of the current study, collection and analysis of the data, and the decision to publish. CV’s salary was in part paid by a research grant from GE Healthcare (paid to the institution).

  • Competing interests None declared.

  • Ethics approval Local medical ethics committee.

  • Provenance and peer review Not commissioned; externally peer reviewed.

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.