In vivo serotonergic system with [11C]DASB PET scan in patients with GTP-cyclohydrolase-deficient dopa-responsive dystonia
Participants and clinical measures
Ten DRD patients with a confirmed mutation in the GCH1 gene were included. Data from two control groups, whose data were published before12, were used: (1) 14 patients with clinically diagnosed idiopathic cervical dystonia (CD), (2) 12 controls without movement disorder. No pilot PET data on the availability of SERT receptors in patients with dystonia was available, therefore, we based our sample size on previous studies of SERT status in (neuro)psychiatric disorders that used 10 to 15 attendees13.14. None of the participants used serotonergic drugs that influence serotonin, including antidepressants, or had any relevant neurological comorbidity.
The clinical and demographic characteristics of the patients were obtained by means of a standardized interview. The severity of dystonia was assessed with the Burke Fahn Marsden Dystonia Rating Scale (BFMDRS) and the Clinical Global Impression Scale (CGI). Non-motor symptoms were assessed using validated questionnaires. The Mini International Neuropsychiatric Interview—PLUS (MINI-PLUS) was used to assess the presence of psychiatric disorders. Severity of depression, anxiety, fatigue, excessive daytime sleepiness and sleep quality were assessed with the Beck Depression Inventory (BDI), Beck Anxiety Inventory (BAI), Fatigue Severity Scale (FSS), the Epworth Sleepiness Scale (ESS) and the Pittsburgh Sleep Quality Index (PSQI), respectively.
Written informed consent was obtained from all subjects in accordance with the Declaration of Helsinki. This study was approved by the UMCG Medical Ethics Committee (METc 034/14).
Three different alleles in the polymorphic SERT-related region (5-HTTLPR) are known to be associated with different transcriptional activity and, therefore, influence the number of SERTs expressed.15. A set of specific single nucleotide polymorphisms in 5-HTTLPR were tested as previously described12.
Individual 3D axial T1-weighted gradient echo images (3T Intera, Philips, The Netherlands) of the brain were acquired from all participants. PET imaging of DRD patients was performed using the same protocol as previously published for CD patients and controls with a Biograph 40-mCT or 64-mCT (Siemens Healthcare, USA). The protocol consisted of a 60-minute dynamic acquisition scan beginning simultaneously with an intravenous bolus of [11C]DASB (mean dose 382 ± 41 MBq)16. PET images were reconstructed from list mode data in 23 frames (7 × 10 s, 2 × 30 s, 3 × 1 min, 2 × 2 min, 2 × 3 min, 5 × 5 min and 2 × 10 min). Head movements have been minimized thanks to a head support band. DRD patients were allowed to continue using their levodopa treatment on the day of the examination.
Image processing and analysis
Reconstruction of dynamics [11C]DASB images were made using the 3D OSEM algorithm (3 iterations and 24 subsets), point spread function correction and time of flight, resulting in a matrix of 400 × 400 × 111 of 2 mm isotropic voxels, smoothed with 2- mm full-width half-height (FWHM) filter.
Image processing was performed with PMOD v3.8 software. We applied motion correction (frames 13 to 23, using 1 to 12 as a reference) before the individual PET image was matched to the individual MRI. A six-tissue probability map was used to estimate gray and white matter volumes and a normalization of the individual MRI to the Montreal Neurological Institute (MNI) standard space was calculated and applied to the corresponding PET image. Predefined volumes of interest (VOI) based on Hammers atlas were transformed into individual MRI space17. As no difference between the left and right hemispheres was identified, a weighted average, based on the volume of the region, was calculated from the corresponding regions in the two hemispheres. The VOIs for dorsal raphe nucleus (DRN) and medial raphe nucleus (MRN) were set manually by selecting the 80% of voxels with the highest absorption in a sphere according to visual inspection of the image PET, as previously described (subject-based: sDRN and sMRN)12. DRNs and MRNs were also defined based on their MNI coordinates (based on the atlas: aDRN and aMRN)18. A Gaussian kernel of 6 mm FWHM was applied before pharmacokinetic modeling was performed with the PXMOD tool of the PMOD V4.0 software. Non-displaceable binding potential (BPn/a) parametric maps were calculated using the Simplified Reference Tissue Model 2 (SRTM2) with the cerebellum (excluding the vermis) as the reference region and the striatum as the high-binding region for the estimation of k2′19.
For the VOI-based analysis, the average BPn/a values by region in the individual space were used. For voxel-based analysis and SSM/PCA, BPn/a parametric maps were used in MNI space.
An SSM/PCA analysis was performed to assess whether a BPn/a could discriminate DRD from CD and controls. SSM/PCA analysis was performed in MATLAB R2018b using in-house software that was validated against the original study where these mathematical principles are widely reported20.21. In short: BPn/a data were masked and centered twice, subtracting subject and region means to obtain a matrix of subject residuals. A covariance matrix was constructed from these residuals and a PCA was applied, yielding principal components (PC). PCs that cumulatively account for 50% of the variance were selected. A stepwise combination was used to generate the disease models and the model with the lowest Akaike information criteria was selected as the final disease model. A score per subject was calculated by taking the inner product of the disease pattern and the subject’s image. To robustly test the generated scores, leave-one-out cross-validation (LOOCV) was applied. A receiver operating characteristic (ROC) curve was plotted and the area under the curve (AUC) was calculated to assess the discriminant values of the pattern.
Statistical analysis was performed using IBM SPSS Statistics version 23. All baseline data were quantitatively described. Pearson’s χ2 test, one-way ANOVA, Mann Whitney U or Kruskall Wallis tests were used to compare demographic and clinical data between groups.
For VOI-based analysis, BPn/a the values were compared respectively between the DRD and CD and DRD and healthy control groups with a Mann Whitney U test. A correction for multiple comparisons (Bonferroni) was performed and the threshold for statistical significance was set at one p-value BPn/a values, was performed to correct for the presence of psychiatric comorbidity.
Voxel-based analysis was performed with SPM12 (Wellcome Trust Center for Neuroimaging, UK) and a two-sample assay you-test was used to assess the differences between DRD patients and CD patients, and DRD patients and healthy controls. T-map data was queried at p= 0.005 (uncorrected) and only clusters with p
Spearman’s rho test corrected for multiple tests (Benjamini–Hochberg) was used to perform a correlation analysis between the BPn/a in VOIs and clinical variables in all participants. To assess differences in BPn/a between participants with a lifetime psychiatric diagnosis and those without, a Mann Whitney U test was performed. To determine whether the significant correlations we found in the whole group were also present in the DRD group, the tests mentioned above were also performed in the DRD group.
Finally, to check for possible confounding factors, a Spearman rho test was performed between levodopa dose, age, La/La genotype and BPn/a.
This study was approved by the UMCG Medical Ethics Committee (METc 034/14).
Consent to participate
Written informed consent was obtained from all subjects in accordance with the Declaration of Helsinki.