* Correspondence to last author at Department of Neurology, University of Ulm, Oberer Eselsberg 45, D-89081 Ulm, Germany. E-mail: firstname.lastname@example.org
Recent studies have shown that changes in the basal ganglia circuitry and limbic loops may play an important role both in Tourette syndrome (TS) and attention-deficit–hyperactivity disorder (ADHD). This study aimed to investigate in vivo possible morphological alterations of the amygdala as a key component of the limbic system. Amygdalar and total brain volumes were measured in three-dimensional magnetic resonance imaging data sets of 17 male patients with TS (mean age 11y 8mo [SD 2y]; range 9–16y) and 17 age-matched comparison children (mean age 12y 6mo (SD 2y 1mo); range 9–17y) by volume-of-interest-based volumetry. Eight members of the TS group also fulfilled the diagnostic criteria for ADHD. A significant decrease in the left-hemispheric amygdalar volumes and in the proportions of amygdalar to total brain volumes was observed in members of the TS group compared with the comparison group. Amygdalar volumes did not correlate with tic severity, but with behavioural impairment and especially with symptoms of ADHD. The amygdalar volume reduction might be the pathoanatomical correlate of an impaired input of the amygdala to the striatum and frontal cortex. Future studies should investigate if the involvement of the amygdala is due to TS or rather caused by the genetically-linked most frequent comorbidity ADHD.
More than 120 years after its first description by Georges Gilles de la Tourette, the precise neurobiological basis of the complex neurobehavioural disorder labelled today as Tourette syndrome (TS) is still a topic of ongoing investigations. Tics are the clinical hallmark, i.e. sudden, brief, involuntary or semi-voluntary movements (motor tics) or sounds (phonic tics), that wax and wane in severity.1 Typically starting at the age of 3 to 8 years, the disorder is manifested by age 11 years in 96% of patients.1 Comorbidity with attention-deficit–hyperactivity disorder (ADHD), characterized by the three cardinal symptoms of impaired attention, hyperactive behaviour, and increased impulsivity, has been reported in 35 to 90% of children with TS.2 Furthermore, ADHD and TS seem to be genetically linked.3
The exact causes of the two disorders are still unknown; however, a dysfunction in the neurotransmitter systems in the circuitry involving the frontal cortex, the thalamus, and basal ganglia can at least be assumed for both entities. The concept of dopaminergic alterations in TS is supported by response to specific therapies, nuclear imaging, and postmortem studies. Using volumetric magnetic resonance imaging (MRI) techniques, Singer et al.4 found no statistically significant difference in the size of caudate, putamen, globus pallidus, or ventricles between 37 children with TS with or without comorbid ADHD and members of a comparison group. However, there were significant differences for measures of symmetry in the putamen and the lenticular region. These regional changes seemed to be related to both TS and the comorbidity ADHD, differing from those primarily associated with tics.4 In patients with ADHD without TS, a meta-analysis of 16 functional neuroimaging studies revealed significant patterns of frontal hypoactivity, affecting anterior cingulate, dorsolateral prefrontal, and inferior prefrontal cortices, as well as related regions including basal ganglia, thalamus, and portions of parietal cortex.5
In previous studies, a disinhibition of the cortico-striatal-thalamic-cortical (CSTC) pathway was postulated in TS and ADHD (see Plessen et al.6 for review). For instance, the ventral striatum (nucleus accumbens and rostroventral extensions of caudate and putamen) receives inputs from the amgydala as a key component of the limbic system. Jackson and Moghaddam7 showed in an animal study that the prefrontal cortex influences the behavioural impact of amygdala activation via a concomitant active suppression of accumbal dopamine release. Disturbance of the cortical influence in this triadic relationship – as it can be postulated in TS – appears to result in an aberrant pattern of behavioural expression in response to amygdala activation, including behavioural perseveration after stimulus termination.
If the amygdala plays a core role in the functional network, including frontal cortex and ventral striatum, it seems to be an intriguing conclusion that functional or even structural alterations of the amygdala may be a correlate of symptoms in TS, possibly influenced by the comorbidity ADHD. Thus, we tested the main hypothesis in this volumetric study on high-resolution MRI that amygdalar volumes in TS patients would differ from those of age-matched comparison children and performed an additional correlation analysis with ADHD symptoms.
The protocol was approved by the Ethics Committee of the University of Ulm. Written informed consent was obtained from the parents of all patients and comparison children, and the children provided written assent. Part of the MRI data had been subjected to an analysis by an automated whole brain-based technique (voxel-based morphometry), i.e. 14 of the 17 participants with TS had also been included in this previous study.8
Patients were recruited in the specialized outpatient clinic for tic disorders and related movement disorders of the Department of Child and Adolescent Psychiatry at the University of Ulm, Germany, over the predefined period of 2 years. Seventeen male TS patients aged between 9 and 16 years (mean age 11y 7mo [SD 2y]) participated in the study. Only males were examined to rule out a sex bias. A clinical semi-structured diagnostic interview (modified after the Kiddie-SADS; http://www.wpic.pitt.edu/ksads/default.htm) of the patient alone, parents/custodians alone, and then parents/custodians and patient together was conducted by board-certified child and adolescent psychiatrists (AGL, GL) to assess comorbidities. ADHD comorbidity was quantified using the DSM-IV criteria checklist.9 By use of this checklist, ADHD was subdivided into the three different types (combined inattentive, hyperactive, and impulsive; predominantly inattentive; predominantly hyperactive and impulsive).
All patients fulfilled the diagnostic criteria of TS as defined by the TS Classification Study Group:10 (1) both multiple motor tics and one or more phonic tics must have been present and witnessed directly by a reliable examiner (AGL, GL) at some time during the illness; (2) tics must have occurred at least nearly every day or intermittently throughout a period of more than 1 year; (3) the type or severity of tics must have changed over time; and (4) involuntary movements and noises must not be explainable by other medical conditions. All patients were assessed with the Yale TS symptom checklist, a German version of the Yale Global Tic Severity Scale (YGTSS), modified by Steinhausen,11 which provides an evaluation of the number, frequency, intensity, complexity, and interference of motor and phonic symptoms and behavioural abnormalities. The Yale TS symptom checklist is divided into five subscales: (1) simple motor tics (11 items); (2) complex motor tics (11 items); (3) simple phonic tics (six items); (4) complex phonic tics (seven items); and (5) behaviour abnormalities (six items). Each item can be rated from 0 (not at all) to 5 (nearly constantly), resulting in a maximum rating score of 205. The behaviour subscale is composed of four definite items: quarrelsome/contentious, low tolerance of frustration, outbursts of fury, and provocative. The fifth and sixth items are classified ‘others’ and can be filled in by the rater, but this option was not used for the patients in this study.
The mean time between onset of tic symptoms and MRI was 3 years 5 months (range 2–5y). IQ was determined by use of the German version of the Wechsler Intelligence Scale for Children, 3rd edition – Revised (HAWIK-III R).12
Eight out of the 17 patients with TS fulfilled the DSM-IV criteria for the diagnosis of ADHD. Six out of these eight patients with ADHD comorbidity belonged to the predominantly inattentive type and two to the combined type. Four males took psychostimulants; no other psychotropic drugs was used. In particular, a history of intake of neuroleptics was an exclusion criterion, since neuroleptics are well known to influence subcortical volumes. Patients with any other additional psychiatric disorder or other medical problems, particularly neurological deficits or history of severe head trauma, were excluded. Table I shows the clinical data of the patients.
Table I. Characteristics of Tourette syndrome (TS) patients and comparison children
ADHD, attention-deficit–hyperactivity disorder; DSM-IV, Diagnostic and Statistical Manual of Mental Disorders, 4th edn.
Mean (SD) Range
Mean (SD) Range
Yale TS symptom checklist total
Yale checklist subscale behaviour
ADHD (DSM-IV criteria checklist)
For setting up an age-matched MRI comparison database, three-dimensional MRI data of 17 age-matched healthy males between 9 and 17 years (mean age 12y 6mo SD 2y 1mo) without any history of neurological or psychiatric symptoms or signs were acquired. Patients with TS and comparison children were well matched for age, handedness, and IQ (Table I).
Data acquisition and processing
Three-dimensional high-resolution MRI images of the whole brain were obtained on a 1 Tesla clinical scanner (Siemens, Erlangen, Germany) using a T1-weighted magnetization prepared rapid acquisition gradient echo sequence (MP-RAGE) which consisted of 160 to 180 sagittal partitions. The following acquisition parameters were applied: repetition time 9.7ms; echo time 3.93ms; flip angle 15°; matrix 256x256mm²; field of view 250mm. In order to avoid movement artifacts during recording, the patients were calmed by the presence of one of their parents in the scanning room. In addition, movement artifacts were minimized by using a vacuum mould adapted to the participants’ heads. If necessary, data acquisition was repeated. Post-acquisition, all data sets were thoroughly checked for motion artifacts and only those MRI data without any motion were used.
Four additional standard sequences were carried out in the patients to identify potential pathological findings which might be easier to assess in routine MRI (coronal fluid-attenuated inversion recovery, axial T2-weighted spin echo, sagittal T1-weighted spin echo, and axial T1-weighted spin echo).
Volumetric measurements were performed by use of the interactive software MRreg (Epilepsy Imaging Group, Department of Clinical and Experimental Epilepsy, Institute of Neurology, UCL, London, UK; http://www.erg.ion.ucl.ac.uk/MRreg.html).13 For the complex task of amygdala tracing, each of the two raters had received intensive training by an experienced rater (LTvE) in data of normal and pathologically altered brains different from the ones used in this study. Both raters, who were blinded to the patient characteristics, analyzed all MRI data sets. The manual delineation of the amygdala was performed according to the well-established protocol,14 separately for each hemisphere. The volume of the amygdala in each slice (in-slice volume) was calculated by multiplying the voxel number of each trace by the voxel volume and dividing this value by the magnification factor. The total amygdala volume was calculated as the sum of all in-slice volumes. To obtain the total brain volume, the cerebral structures were outlined in every single slice separately using the semi-automatic threshold and region-growing tool of MRreg. In order to provide a correction for individual brain size, the individual relative amygdala volume was calculated by dividing the amygdala volumes by the total brain volumes.
Reliability data were assessed by the two raters by calculating an intraclass correlation (ICC) coefficient from repeated measurements, based on the measurements of all MRI data sets. Intrarater reliability was assessed by three different methods: (1) the ratio of measured SDs to average amygdala volumes (coefficient of variation); (2) the coefficient of repeatability Cr; and (3) an ICC coefficient.
Patient and comparison groups were compared using the t-test with Bonferroni-Holm correction for multiple testing within SPSS software version 12.0). To measure the relationship between the volumetric data and the clinical tests (Yale TS symptom checklist total score, behavioural subscore, and DSM-IV criteria for ADHD) as associated research questions, the Pearson's correlation coefficient r was used within SPSS; 0.05 was defined as two-sided level of significance.
In none of the 34 participants was any pathological finding observed in the visual analysis of the routine MRI sequences.
In Table II, the volumetric findings are summarized. Both for left and for right amygdala, mean volumes were lower in the patient group. The results were statistically significant for the ratios total amygdala:whole brain volume and left amygdala:whole brain volume. The ratio right-sided amygdala: whole brain volume did not reach statistical significance. In the intragroup comparison, there was no significant asymmetry effect (left vs right amygdala volumes), neither for patients nor for comparison children. The mean total brain volume was not different between patients and comparison children (1279.3cm [SD 126cm³] in the patients vs 1365.63 [SD 168 cm³] in comparison group). Single amygdalar volumes plotted against total brain volumes are shown in Figure 1.
Table II. Synopsis of volumetric findings in patients with Tourette syndrome (TS) and comparison children, mean values (SD)
Proportion of right amygdala to total brain volume [×10−3]
Proportion of left amygdala to total brain volume [×10−3]
Proportion of amygdala to total brain volume [×10−3]
95% confidence interval
The findings of the ratio of bilateral amygdalar volumes to whole brain volume did not correlate with severity of tic symptoms (Yale TS symptom checklist total score; Fig. 2a), but correlated with the behavioural subscale of the Yale TS symptom checklist (r=−0.553, p=0.039; Fig. 2b) and with the DSM-IV ADHD criteria checklist (r=−0.68, p=0.025; Fig. 2c). The separate correlation analyses for the right and left amygdalar volumes did not yield significant results.
The coefficient of variation was 7.8%. The coefficient of repeatability Cr was 22% of the overall mean. The ICC coefficient was 0.92 for intrarater reliability and 0.64 for interrater reliability. All values corresponded well to what has been reported in the literature.14
The amygdala measurements in males with TS showed a significantly decreased proportion of amygdalar to total brain volume in TS patients compared with members of the comparison group. Amygdalar alterations did not correlate with tic severity but with ADHD symptoms.
Whereas there are several volumetric studies on changes in cerebral structures in TS,15 amygdala volumes were specifically investigated by use of modern MRI techniques in the present study. A key role of the amygdala in TS had been suggested very early.16 In addition, Coffey et al.17 found a significant association between anxiety disorder and tic severity in young patients with TS. These clinical observations fit the hypothesis that not only the CSTC pathway is involved in TS, but that the triadic interaction between prefrontal cortex, amygdala, and ventral striatum7 might also influence the aberrant motor behaviour in TS. Approximately half of the males in our sample also fulfilled the diagnostic criteria for ADHD, the most common comorbidity in TS, while none was classified for any other psychiatric axis I diagnosis besides TS or ADHD (neither obsessive–compulsive disorder nor anxiety disorder nor depression).
The present result of a reduction of the overall amygdalar volume in TS was significantly correlated with ADHD symptoms and behavioural abnormalities as reflected in the Yale TS symptom checklist. In a previous MRI study, Plessen et al.18 found an altered amygdalar complex bilaterally in the ADHD group in a large cohort of patients aged 7 to 18 years: the overall size of the amygdala was not reduced, but surface analyses detected an altered size of the basolateral amygdala bilaterally. Thus, it is most likely that the volumetric findings in the present study reflect less the tic symptoms (since it did not correlate with Yale TS symptom checklist total score), but the behavioural deviations as a part of the ADHD comorbidity. Already Castellanos et al.19 contrasted ADHD males with and without TS to a comparison group and detected a significant loss of the normal globus pallidus asymmetry in ADHD patients, but did not find an influence of TS symptoms. The authors concluded that accounting for ADHD comorbidity would be important in future TS morphometric studies.
In a previous voxel-based morphometric MRI analysis of nearly the same patient group as in the present study, abnormalities in amygdalar volumes were not detected8 although the left hippocampal gyrus, as a core component of the limbic system, had been observed to be significantly decreased in volume, confirming the association of alterations within the limbic loops and TS. This discrepancy between the volume-of-interest-based results and the voxel-based morphometric results is most probably due to the different methodological approach, as described in previous studies.20 Voxel-based morphometric analyses, as a fully automated whole-brain measurement technique, test for residual brain tissue concentration differences that remain after all investigated patients’ MRIs have been spatially normalized into the same standardized stereotaxic space, and such analyses are thus less sensitive to shape differences. In contrast, landmark-based volume-of-interest measurements have the limitation that manually delineated regions are subject to inaccurancies due to variations of local individual neuroanatomy, but have the strength of substantial anatomical validity and output absolute accountings of the numbers of voxels in the areas being investigated.20 Consequently, the hypothesis-based nature of the volume-of-interest measurement makes it possible to search only for differences in volume in prespecified areas, such as the amygdala in the present study. Voxel-based morphometric and volume-of-interest methods provide different types of information and can be used in tandem.
Possible alterations of cerebral asymmetry and lateralization in TS patients, i.e. reduced normal asymmetry especially in subcortical (basal ganglia) structures, have been discussed previously. In the very first volumetric imaging study of the basal ganglia, the volume of the left but not the right lenticular nucleus was found to be significantly smaller and the normal basal ganglia asymmetry was reduced in the group of 14 adult patients with TS,21 confirmed later in a sample of 37 children with TS.4 In a recent large-scale study of TS patients, however, the hemisphere × diagnosis × region interaction was not significant for the basal ganglia.22
In our study, only the left-sided amygdala was significantly reduced. Remarkably, volumetric changes of the left-sided amygdala were also observed in further psychiatric disorders known to be dopamine-related, i.e. obsessive compulsive disorder and schizophrenia. Our results are also in agreement with a recent voxel-based morphometric MRI study of 12 children with conduct disorder (related to ADHD).23 Here, significantly smaller grey matter volumes were observed, among others, in the medial part of the left amygdala of conduct disorder patients, and the left amygdala grey matter volume correlated most strongly with attention problems. In our sample, six out of eight TS patients with ADHD belonged to the predominantly inattentive type, whereas this type is reported to be diagnosed in only 10 to 15% of children with ADHD alone. However, these findings cannot yet be integrated into one single underlying pathoanatomical concept. Phelps et al.24 observed an activation only in the left and not in the right amygdala in a functional MRI study in response to threat versus safe conditions. It seems reasonable to assume from these imaging studies that the left and right amygdala fulfil different functions. The left-sided volume reduction in TS supports the notion that the asymmetric changes in amygdala size might reflect altered amygdalar function and might thus be a correlate of clinical symptomatology. In this context, it has to be noted again that our results suggest a correlation with the behaviour subscale of the YGTSS and with ADHD symptoms. Thus, the comorbidity of TS and ADHD has to be considered an important influence on the results, but is difficult to assess in a quantitative way. Interestingly, Lee et al.25 reported enlarged thalamic volume also only on the left side in males with TS without any ADHD comorbidity.
In the present study of males with TS, we observed regional volume alterations of the amygdala. The exact nature of the volume changes remains to be clarified: one possibility might be neuronal hypotrophy, i.e. loss of neurites, as a correlate of disturbed connectivity between the amygdaloid complex and prefrontal cortex and/or ventral striatum. Since we did not find a correlation with the tic symptoms, but with the behaviour subscale of the Yale TS symptom checklist and especially with ADHD symptoms, we interpret our data as a correlate of ADHD comorbidity. The well-known interindividual variability of TS and ADHD cautions against too generalized conclusions, but the aetiological relationship between TS and ADHD needs to be further clarified. These considerations might be also relevant for treatment options in a clinical setting, since psychostimulants, which increase intrasynaptic dopamine level, do not seem to worsen tic symptoms26 and atomoxetine, a norepinephrine transporter inhibitor and the first approved non-stimulant medication for ADHD, even seems to ameliorate tic symptoms as well.27
The results of the present study provide evidence of the possible involvement of limbic structures in the pathophysiology of TS. Future studies should focus on correlation analyses with different clinical parameters in samples of TS patients, with ADHD comorbidity as a possible confounder.
The first two authors contributed to this article equally. Assistance in MRI acquisition was provided by S Fuchs. We thank L Lemieux PhD (Epilepsy Research Group, Institute of Neurology, Queen Square, London, UK) for the software package MRreg.