Lower serotonin transporter binding in patients with cervical dystonia is associated with psychiatric symptoms

Included subjects also participated in another study recently published [5]. In short, we included patients with idiopathic CD with stable disease severity for at least 1 year on the Tsui scale, age between 35 and 80 years, and ongoing treatment with botulinum neurotoxin (BoNT) injections, which is the current standard treatment for focal dystonia. BoNT injections were administered on the day of SPECT scanning or a maximum of 7 days prior to or after scanning. Exclusion criteria were other relevant neurological conditions at inclusion or in the past, treatment with deep brain stimulation (DBS), use of medication with a known dopaminergic or serotonergic effect in the past 6 months, including SSRIs [18], and pregnancy or lactation. Patients were allowed to use other medication including low-dose benzodiazepines (for example clonazepam 1 mg twice daily). Healthy age- and sex-matched subjects with a normal neurological examination and no self-reported (family) history of dystonia, myoclonus, or psychiatric illness recruited through flyers and databases served as the control group. Written informed consent was obtained in all subjects, and the study was approved by the local medical ethics committee.

The neurological examination of patients was videotaped and blindly scored by two independent clinicians. The Toronto Western Spasmodic Torticollis Rating Scale (TWSTRS) [19] and the Tsui scale were used to score dystonic symptoms [20]. The Unified Myoclonic Rating Scale (UMRS) was used to score jerks [21]. In our previous study [5], we showed good interobserver agreement for the TWSRTS and Tsui (> 0.80 intraclass correlation coefficients) and reasonable agreement for the UMRS (0.73 intraclass correlation coefficients). The average score of the two experts on the Tsui, TWSTRS, and UMRS was used in the statistical analysis. The psychiatric examination consisted of an interview, performed by a trained investigator (EZ; YD). The Mini International Neuropsychiatric Interview (MINI)-Plus and Montgomery-Åsberg Depression Rating Scale (MADRS) were included, as well as several questionnaires concerning symptoms of depression (Beck Depression Inventory (BDI)) and anxiety (Liebowitz Social Anxiety Scale (LSAS) and the Beck Anxiety Inventory (BAI)for anxiety ). The psychiatric interview was performed on the day of the SPECT scan, and questionnaires were answered in the week preceding the SPECT scan. Subjects were judged to have a depressive disorder when they fulfilled the according criteria on the MINI (current depression) and/or had a BDI ≥ 14 points or MADRS ≥ 20 points, consistent with moderate-severe depression. Subjects were judged to have an anxiety disorder when they fulfilled the according criteria on the MINI (current anxiety disorder) and/or had a BAI ≥ 16 points or LSAS ≥ 30 points, consistent with moderate-severe anxiety.

All subjects received 300 mg potassium iodide (three capsules of 100 mg each) to block thyroid uptake of free radioactive iodide before administration of the tracer. Then, all subjects received a mean dose of 100 MBq (2.7 mCi) [¹²³I]FP-CIT intravenously (produced according to GMP criteria by GE Healthcare) as a single bolus [22]. Scans were acquired 2 h after bolus injection to optimally visualize extrastriatal SERT binding, a technique that has been validated before [23]. SPECT studies were performed using a 12-detector single-slice brain-dedicated scanner (Neurofocus 810, which is an upgrade of the Strichman Medical Equipment); this system was described extensively by Stoddart and Stoddart [24], with a full-width at half-maximum in-slice resolution of approximately 5–6 mm, throughout the 20-cm field-of-view, and a system volume sensitivity of 0.22 (counts/s)/(Bq/ml) [25]. After positioning of the subjects with the head parallel to the orbitomeatal line, axial slices parallel, and upward from the orbitomeatal line to the vertex were acquired in 5-mm steps. An average of 15 slices with 3.5-min scanning time per slice was acquired in a 64 × 64 matrix. The energy window was set at 140–178 keV. Images were corrected for attenuation, as earlier described, and reconstructed in a 3-D mode using Neurofocus proprietary software (Neurofocus Inc., USA) and the manufacturer’s recommended iterative reconstruction algorithm, which was based on maximum a-posteriori (MAP) reconstruction methods [26]. The 3-D reconstructed images were randomly numbered and analyzed blinded for the subject group (CD patient with/without jerks or control) by one observer (EZ). Fixed regions of interest (ROIs) for the diencephalon and midbrain combined were positioned as earlier described [23, 27]. The four consecutive slices with highest activity in the ROI were pooled together, and average activity was calculated. This was also done for the cerebellum as reference region using the two to three consecutive slices with highest activity. Specific to non-specific binding ratio was calculated as [(activity in ROI − activity in reference region)/activity in reference region], representing the binding potential (BP_ND) [28]. Test-retest variability of this analysis was calculated in a random subset of subjects (n = 10) by using the formula (test-retest)/((test + retest)/2) × 100%. Average variability was 10%, which is comparable to previously published variability in extrastriatal [¹²³I]FP-CIT binding in the midbrain [29] and only slightly higher than variability in striatal [¹²³I]FP-CIT binding, which is ~ 7.5% [22].

As published before [5], we also measured DAT BP_ND in the striatum 3 h after the same [¹²³I]FP-CIT injection in the same subjects. We used these values to test the correlation between SERT and DAT BP_ND.

Mann-Whitney U test and Kruskal-Wallis test were used to assess differences in baseline characteristics, SERT BP_ND, and SERT to DAT ratio between different groups of subjects. Chi-square and Fisher’s exact test were used to assess dichotomous variables. Linear regression was used to assess if differences in baseline characteristics explained differences in SERT BP_ND, both between patients with and without jerks/tremor, and between CD patients and healthy controls. Linear regression was also used for assessing relationships between SERT BP_ND and motor and psychiatric scores. Since age and sex are known to have an effect on SERT BP_ND as measured with [¹²³I]FP-CIT SPECT [30], we corrected for these factors. Analyses were carried out using SPSS version 23, and differences were considered significant at p < 0.05.

We planned to include 29 CD patients and compare them to 15 matched healthy controls. Due to technical difficulties, one [¹²³I]FP-CIT scan of a control and six [¹²³I]FP-CIT scans of patients had to be removed from the analysis. More specific, not in all scans the whole cerebellum was scanned because of a skewed position of the head (the brain-dedicated system is a tomographic system), and consequently, this region could not be used as a reliable reference region [5]. We could use data from 23 CD patients (12 with jerks/tremor and 11 without) and 14 controls. Baseline characteristics are depicted in Table 1, no significant differences between groups were detected. Low-dose benzodiazepines were used by 6/23 patients: oxazepam or diazepam on as needed base (three patients), oxazepam 5 mg once daily (one patient), and clonazepam 0.5 mg once or twice daily (two patients). In total, 8/11 (73%) of the patients without comorbid jerks/tremor fulfilled the criteria for a psychiatric diagnosis (any diagnosis on the MINI and/or a score above the before mentioned cutoff values on the BDI, MADRS, LSAS, or BAI) compared to 6/12 (50%) of the patients with comorbid jerks/tremor. The most common condition was social anxiety disorder (7/23 patients; 30%). In one patient, both a depressive disorder as well as social anxiety disorder were present. As mentioned in our previous study, one control fulfilled the criteria for alcohol abuse in the past and one control scored 34 on the LSAS meeting the criteria of social anxiety disorder [5].

Median SERT BP_ND was 0.26 in controls (IQR 0.10–0.40) and 0.25 in patients (IQR 0.13–0.40; p = 0.89, Mann-Whitney U test). In addition, median SERT BP_ND was not significantly different in patients with jerks/tremor (0.28; IQR 0.15–0.42) compared to patients without jerks/tremor (0.17; IQR 0.11–0.34; p = 0.49, Mann-Whitney U test). There was no significant correlation between the diencephalon/midbrain SERT BP_ND and scores on Tsui (p = 0.79), TWSTRS (p = 0.20), or UMRS (p = 0.32; linear regression) in CD patients.

Median SERT BP_ND was 0.20 in patients with psychiatric co-morbidity (n = 14; IQR 0.11–0.28) and 0.42 in patients without a psychiatric diagnosis (n = 9; IQR 0.16–0.47; p = 0.05, Mann-Whitney U test). This trend was not present between patients with any anxiety disorder (n = 10; median SERT BP_ND 0.24 (IQR 0.15–0.28)) and patients without any anxiety disorder (n = 13; median SERT BP_ND 0.31 (IQR 0.11–0.44); p = 0.45, Mann-Whitney U test). There was a trend towards a difference between patients with a depression (n = 5; median SERT BP_ND 0.14 (IQR 0.10–0.22)) and patients without (n = 18; median SERT BP_ND 0.28 (IQR 0.15–0.42); p = 0.09, Mann-Whitney U test). There was no significant correlation between the diencephalon/midbrain SERT BP_ND and scores on BDI (p = 0.43), MADRS (p = 0.43), LSAS (p = 0.29), or BAI (p = 0.34, linear regression) in CD patients.

Correcting for age and sex by means of linear regression did not change the results in any of our analyses. Regression coefficients and p values before and after correction are depicted in Table 2.

There was a statistically significant correlation between the diencephalon/midbrain SERT BP_ND and striatal DAT BP_ND in patients with jerks (R² 0.51, regression coefficient 0.27 (0.06–0.49; p = 0.02; see Fig. 1b). Correcting for any psychiatric diagnosis or depression did not change these results significantly. This correlation was not present in patients without jerks (R² 0.04, regression coefficient − 0.11–0.20; p = 0.56, see Fig. 1a) or in controls (p = 0.14, see Fig. 1c).