Journal ofhttp://jad.sagepub.com/ Attention Disorders
Effects of Methylphenidate on Default-Mode Network/Task-Positive Network Synchronization in Children With ADHD Laurent Querne, Sidy Fall, Anne-Gaëlle Le Moing, Emilie Bourel-Ponchel, Aline Delignières, Anais Simonnot, Alain de Broca, Catherine Gondry-Jouet, Muriel Boucart and Patrick Berquin Journal of Attention Disorders published online 13 January 2014 DOI: 10.1177/1087054713517542 The online version of this article can be found at: http://jad.sagepub.com/content/early/2014/01/13/1087054713517542
Published by: http://www.sagepublications.com
Additional services and information for Journal of Attention Disorders can be found at: Email Alerts: http://jad.sagepub.com/cgi/alerts Subscriptions: http://jad.sagepub.com/subscriptions Reprints: http://www.sagepub.com/journalsReprints.nav Permissions: http://www.sagepub.com/journalsPermissions.nav
>> OnlineFirst Version of Record - Jan 13, 2014 What is This?
Downloaded from jad.sagepub.com at Umea University Library on October 11, 2014
517542
research-article2014
JADXXX10.1177/1087054713517542Journal of Attention DisordersQuerne et al.
Article
Effects of Methylphenidate on Default-Mode Network/Task-Positive Network Synchronization in Children With ADHD
Journal of Attention Disorders 1–13 © 2014 SAGE Publications Reprints and permissions: sagepub.com/journalsPermissions.nav DOI: 10.1177/1087054713517542 jad.sagepub.com
Laurent Querne1,2, Sidy Fall1,2, Anne-Gaëlle Le Moing1,2, Emilie Bourel-Ponchel1, Aline Delignières1,2, Anais Simonnot1, Alain de Broca1, Catherine Gondry-Jouet3, Muriel Boucart4, and Patrick Berquin1,2
Abstract Objective: A failure of the anti-phase synchronization between default-mode (DMN) and task-positive networks (TPN) may be involved in a main manifestation of ADHD: moment-to-moment variability. The study investigated whereby methylphenidate may improve TPN/DMN synchronization in ADHD. Method: Eleven drug-naive ADHD children and 11 typically developing (TD) children performed a flanker task during functional magnetic resonance imaging. The ADHD group was scanned without and 1 month later with methylphenidate. The signal was analyzed by independent component analysis. Results: The TD group showed anti-phase DMN/TPN synchronization. The unmedicated ADHD group showed synchronous activity in the posterior DMN only, which was positively correlated with response time variability for the flanker task. Methylphenidate initiated a partial anti-phase TPN/DMN synchronization, reduced variability, and abolished the variability/DMN correlation. Conclusion: Although results should be interpreted cautiously because the sample size is small, they suggest that a failure of the TPN/DMN synchronization could be involved in the moment-to-moment variability in ADHD. Methylphenidate initiated TPN/DMN synchronization, which in turn appeared to reduce variability. (J. of Att. Dis. 2014; XX(X) 1-XX) Keywords ADHD, methylphenidate, fMRI, variability ADHD is characterized by symptoms of inattention and hyperactivity/impulsivity that interfere with the social life and academic performances Diagnostic and Statistical Manual of Mental Disorders (4th ed.; DSM-IV; American Psychiatric Association [APA], 1994). Children with ADHD display moment-to-moment variability in daily life and in the classroom, making it difficult for them to maintain the continuity of mental activities over time (Brown et al., 2005; Castellanos & Tannock, 2002; Klein, Wendling, Huettner, Ruder, & Peper, 2006). Functional neuroimaging studies have revealed hypoactivity in the frontostriatal, parietal, and cerebellar regions suggesting that the latter underlie the networks involved in ADHD (Cortese et al., 2012; Durston, van Belle, & de Zeeuw, 2011). These regions contribute to several functional networks (referred to as task-positive networks [TPNs]) that are activated during the performance of externally oriented tasks. For example, both frontoparietal dorsal-attentional and ventral-attentional networks (DAN and VAN) are mainly solicited when normal individuals performed visuospatial tasks and constitute the
TPNs for these tasks (i.e., Corbetta, Patel, & Shulman, 2008). These networks and other functional networks also showed a spontaneous synchronic activity at rest as evidenced by intrinsic functional connectivity studies as demonstrated on a large sample size (Yeo et al., 2011). Children and adults with ADHD also show abnormal activation of a specific set of regions known as the default-mode network (DMN; Cortese et al., 2012; Hart, Radua, Nakao, MataixCols, & Rubia, 2013; Swanson, Baler, & Volkow, 2011). The DMN supported many internally focused cognitive processes, such as stimulus-independent thoughts or 1
GRAMFC INSERM U1105, Université de Picardie Jules-vernes, France Service de Neuropédiatrie, CHU Amiens-Picardie, France 3 Service de Radiologie, CHU Amiens-Picardie, France 4 LNFP EA4559, Université Lille-Nord, Lille, France 2
Corresponding Author: Laurent Querne, Service de Neuropédiatrie, GRAMFC INSERM U1105, CHU Amiens-Picardie, Place Victor Pauchet, Amiens 80054, France. Email: [email protected]
Downloaded from jad.sagepub.com at Umea University Library on October 11, 2014
2
Journal of Attention Disorders
mind-wandering and shows spontaneous periodic activity at rest. In ADHD, default-mode activity may abnormally persist or intrude into periods of active task-specific processing, producing periodic fluctuations in attention that interfere with externally oriented activity (Sonuga-Barke & Castellanos, 2007). Convergent data have shown that the performance of an externally oriented task is supported by concomitant activation of the appropriate TPN and deactivation of the DMN. Time-domain analyses in normal individuals have revealed that DMN and TPN fluctuations are synchronized in anti-phase during externally oriented task (Kelly, Uddin, Biswal, Castellanos, & Milham, 2008; Sonuga-Barke & Castellanos, 2007). Moreover, DMN activation appears to be correlated with intra-individual response time (RT) variability during a task (Castellanos & Proal, 2012). This parameter is thought to result from moment-to-moment variability (Castellanos & Tannock, 2002) and is strongly impaired in ADHD whatever the task or the participant’s age or subtype (Bourel-Ponchel et al., 2011; Klein et al., 2006; Querne & Berquin, 2009). A positive correlation between RT variability and DMN fluctuations or DMN deactivation has been observed for various tasks in normal adults and in children with ADHD (Bonnelle et al., 2011; Fassbender et al., 2009; Kelly et al., 2008). Hence, moment-to-moment variability, which is so characteristic of functioning in ADHD, could be due (at least in part) to the maintenance of abnormal cyclic DMN activity that could interfere with task-related regions (Sonuga-Barke & Castellanos, 2007). In ADHD, stimulants relieve symptoms, improve academic achievement and the child’s quality of life, and reduce RT variability for externally oriented tasks (National Institutes of Health Consensus Development Conference Statement, 2000). Functional neuroimaging studies have shown that stimulants (and notably methylphenidate [MPH]) increased activation in frontostriatoparietal regions (Durston et al., 2011; Hart et al., 2013) and enhance deactivation of the DMN during inhibitory tasks (Liddle et al., 2011; Peterson et al., 2009). Nevertheless, synchronization of the TPN/DMN complex, the effect of stimulants on the latter and the relationship between synchronization and moment-to-moment variability have never been studied in children with ADHD. The present study sought to explore whether methylphenidate modifies synchronization of the TPN/DMN complex in relation to the moment-to-moment variability by applying an independent component analysis (ICA) to the time course of hemodynamic activity (Cole, Smith, & Beckmann, 2010). Our starting hypothesis was first that non-medicated ADHD children may present null or weak anti-phase synchronization of the TPN/DMN complex during an externally oriented task and second that methylphenidate may restore (at least partially) TPN/DMN anti-phase synchronization and thus reduce RT variability during the task.
Materials and Method Participants ADHD was diagnosed according to the DSM-IV criteria (APA, 1994) following a neurological evaluation of the child and an interview with his or her parents. The study’s main inclusion criteria were as follows: children aged 8 to 13, no previous drug treatment, a combined subtype (inattention criteria ≥ 6 and hyperactivity/impulsivity criteria ≥ 6), an intelligence quotient (IQ) > 85 and approval of methylphenidate treatment by the child’s parents. Children with a history of neurological, psychiatric, or developmental disorders (other than ADHD), depression or generalized anxiety were excluded from the study. A group of typical development children (TDC), presenting normal school achievement, and a normal IQ was constituted (IQ > 85, as evaluated with the Wechsler Intelligence Scale for Children 4th edition (WISC-IV) Similarities, Vocabulary, Block Design, and Picture Concepts). As in the ADHD group, the children in the TD group were medication-naive and did not have any history of neurological, psychiatric, or developmental disorders, depression or generalized anxiety. The children’s parents received comprehensive information about the study’s objectives and procedures, which were approved by the local ethics committee (Comité Pour la Protection des Personnes Nord-Ouest II). Children in the ADHD group underwent their first functional magnetic resonance imaging (fMRI) session prior to initiation of treatment. The methylphenidate treatment (20 or 30 mg, extended-release formulation) was initiated on the day after the first fMRI session. A second fMRI session was performed after 1 month of continuous treatment. The children took their medication 2 hr before the start of the second session. During the MRI session, the children’s parents filled out the French-language version of the Swanson, Nolan, and Pelham 4th edition (SNAP-IV) rating scale. Two ADHD children refused to go into the magnetic resonance imaging machine because of anxiety. Hence, our final analysis took account of 11 children in each group.
Cognitive Paradigm: The Flanker Task Stimuli were back-projected onto a screen viewed by children through a prism mirror incorporated into the fMRI antenna by using the SuperLab® software (Cedrus Corporation). Responses were collected via an MRI response box (Lumina Pad®, Cedrus Corporation) equipped with a two-push-button pad. The task stimuli consisted of five fish (considered to be more attractive for children than the usual arrows) arranged in a horizon array. The child was required to identify the orientation of the central fish (flanked by the four other fish) by pressing the right button with the right thumb (for a rightward orientation) or the left
Downloaded from jad.sagepub.com at Umea University Library on October 11, 2014
3
Querne et al.
also compared in terms of age and WISC-IV scores in Mann–Whitney U tests.
fMRI Acquisition
Figure 1. An example of three successive trials in the flanker task performed by children in the MRI.
Note. Top: A congruent trial with a rightward orientation of the central target. Middle: An incongruent trial with a leftward orientation of the central target. Bottom: A congruent trial with a leftward orientation of the central target. MRI = magnetic resonance imaging.
fMRI was performed with a Sigma 3T system (General Electric, Milwaukee, Wisconsin) using a T2*-weighted, single-shot, echo-planar imaging sequence (echo time [TE] = 55 ms, flip angle [FA] = 60°, in-plane matrix = 64 × 64, field of view [FOV] = 240 × 240 mm2, repetition time [TR] = 3,000 ms). In each of 107 functional volumes (including three dummy scans), 35 axial slices parallel to the AC–CP plane were acquired with an inter-slice interval of 4 mm (covering the whole brain). In addition to functional scans, high-resolution, axial, T1-weighted 3D-SPGR BRAVO images were collected from each participant (covering the whole brain) by using the following parameters: inversion time = 200 ms, TR/TE = 10.4/4.2 ms, FA = 20°, slice thickness = 1.4 mm, inter-slice interval = 0.7 mm, matrix = 256 × 256 pixels, and FOV = 256 × 256 mm2. This structural image was used to warp the functional images into Montreal Neurological Institute (MNI) standard space.
Independent Component Analysis
Behavioral Analysis
fMRI data pre-processing. The fMRI data from each group were fed into a multi-subject temporal concatenation analysis using MELODIC software (Multivariate Exploratory Linear Decomposition into Independent Components: FMRIB’s Software Library (FSL): Smith et al., 2004). Prestatistical processing of the fMRI data included brain extraction (to exclude non-brain structures) using the BET tool (Smith, 2002), removal of low-frequency drifts (using a high-pass filter cut-off of 156 s), within-subject motion correction (using the FMRIB’s linear registration tool, MCFLIRT; Jenkinson, Bannister, Brady, & Smith, 2002), mean-based intensity normalization of all volumes by the same factor, spatial smoothing (using a Gaussian kernel of 5 mm FWHM), co-registration of T1-weighted images, spatial normalization into the MNI 152 standard space (using a 12-mm affine/non-linear registration), and resampling the data into 2-mm isotropic voxels.
The analyzed flanker task parameters were the proportion of errors related to target orientations, the mean RT, RT variability (intra-individual standard error for the RTs), and the congruency-flanker effect (the mean RT for incongruent trials minus the mean RT for congruent trials). Wilcoxon’s tests were performed on all task parameters and the SNAP-IV scores in the ADHD group, with medication as a within-group factor (off/on-methylphenidate). Mann– Whitney U tests were performed on task parameters and SNAP-IV scores by comparing the TD group and ADHD group for the off-methylphenidate session and then onmethylphenidate session. The ADHD and TD groups were
Group level inference. Group fMRI data were reshaped into a three-dimensional tensor (Time points × Voxels × Subjects) and were then decomposed into a set of independent spatial maps together with associated time courses and subject-mode values, using the probabilistic ICA (Beckmann & Smith, 2005). For each estimated component, this approach is able to provide signal of interest in the spatial (group map of patterns of activation/deactivation), temporal (times-series for the cerebral group map), and subject domain (individual subject-mode values: For details, see the ICA and correlation between subject-mode values and
button with the left thumb (for a leftward orientation). Each trial started with presentation of the five fish for 1,000 ms, which were then replaced by a central fixation mark for 2,000 ms. The child was allowed to provide a response throughout the 3,000 ms trial (Figure 1). The trials were presented in a pseudorandom order. Thirty-second height blocks were preceded by a 9-s rest period, during which the child was instructed to look passively at five elliptical shapes that were very similar to the fish in size and color (Figure 1). The cognitive paradigm lasted 5.35 min in total and was preceded by a short training session.
Downloaded from jad.sagepub.com at Umea University Library on October 11, 2014
4
Journal of Attention Disorders
RT variability paragraph). The order of each group model was estimated by the Laplace approximation to the Bayesian evidence for a probabilistic ICA model (Beckmann & Smith, 2004). Group-specific spatial maps were then reduced to Z-statistics maps by dividing the estimated components’ weightings by the voxel-wise standard error of the residual noise. Clusters of voxels were thresholded by an alternative-hypothesis test (at a posterior probability threshold of p > .5), which was assessed by mixture-modeling of the Z-values’ probability density (Woolrich, Behrens, Beckmann, & Smith, 2005). The number of components generated was fixed by the MELODIC software according to an automatic dimensionality estimation of the model order.
Identification of DMN Components According to the procedure described in Kelly et al. (2010), the resulting spatial distributions of Z score maps and the associated power spectrum were examined in each group, to differentiate between the components corresponding to the neural signal of interest and those predominantly representing noise. We excluded components for which the high frequencies (>0.1 Hz) of the signal time-courses constituted 50% or more of the total power in the Fourier spectrum and brain activities, which were apparently scattered at random over a large section of the brain or mainly located within peripheral areas. Then, white matter, gray matter, and cerebral spinal fluid templates for each group were obtained from means of segmented individual structural images (within MNI standardized brain space) using FSL FMRIB Automated Segmentation Tool (Zhang, Brady, & Smith, 2001). Each segmented tissue was then compared with all the component maps corresponding to the neural signal of interest, to discard spatial maps with a high percentage (90%) of voxels in non-gray matter. We searched in each group the component that better capture the anti-correlation between the DMN and the frontoparietal TPN for visuospatial-conflicting tasks (Bonnelle et al., 2011; Kelly et al., 2008; Yeo et al., 2011). If DMN/TPN anti-correlation was not present, we searched for components included in the canonical DMN.
ICA and Correlation Between Subject-Mode Values and RT Variability The study of brain function can be carried out with the ICA approach, which can separate fMRI data into a set of uncorrelated spatio-temporal components describing the spatial and temporal characteristics of underlying signals. Each component consisted of voxel values (cerebral group map) and a unique associated time course of activation/ deactivation (McKeown et al., 1998). An advantage of this approach is the ability to detect unknown, yet structured
spatio-temporal processes (Beckmann & Smith, 2005; McKeown et al., 1998), which do not rely on a single seed region but can integrate the temporal information in the fMRI data across multiple distributed networks (Beckmann, Mackay, Filippini, & Smith, 2009). The ICA algorithm that we used in this study permits to decompose the group-specific fMRI data to spatially independent group maps with their associated time series and individual subject-mode values. This last parameter represents the processes of interest present in each spatial map in terms of the subject-dependent variations (Beckmann & Smith, 2005) and can be associated for each component as a weight indicating the contribution of individual temporal dynamics to each group-level spatial map identified. See the appendix for an illustration of the relationship linking the individual ICA subject-mode values and the correlations between individual time series in key regions of interest (ROIs) of the complex TPN/DMN and time series associated to ICA-selected group components. A potential relationship has been searched between the RT variability and the dynamic of the fluctuations of the DMN/TPN, which mediates the re-allocation of cognitive resources to the TPN when the participants were instructed to perform the task. To explore this question, we calculated separately for each group of children (TDC/ADHD off-MPH/ADHD on-MPH) the correlations between two sets of values (ri, bi), whereby r1, . . . r11 (ICA subject-mode values) denoting the contribution of individual temporal dynamics to each group-level spatial map in the TPN/DMN complex (or DMN if the anti-correlation was not present), and b1, . . . b11 denoting the individual RT variability during the flanker task for each participant (n1, . . . n11).
Results Age, Psychometric Data, and SNAP-IV Scores The two groups did not differ significantly in terms of age and IQ (age = 9.8 ± 1.7 years, verbal IQ = 112.5 ± 14.4, performance IQ = 103.1 ± 14.4 in the ADHD group; age = 10.8 ± 1.7 years, verbal IQ = 122.4 ± 11.7, performance IQ = 111.9 ± 8.8 in the TD group). The parent-reported SNAP-IV scores were significantly higher for the ADHD group (inattention = 7.9 ± 1.4; hyperactivity/impulsivity = 7.1 ± 2.1) than for the TD group (inattention = 0.5 ± 1.8; hyperactivity/impulsivity = 0.2 ± 0.6; Z = 3.9, p < .0001; Z = 4.0, p
Effects of Methylphenidate on Default-Mode Network/Task-Positive Network Synchronization in Children With ADHD Laurent Querne, Sidy Fall, Anne-Gaëlle Le Moing, Emilie Bourel-Ponchel, Aline Delignières, Anais Simonnot, Alain de Broca, Catherine Gondry-Jouet, Muriel Boucart and Patrick Berquin Journal of Attention Disorders published online 13 January 2014 DOI: 10.1177/1087054713517542 The online version of this article can be found at: http://jad.sagepub.com/content/early/2014/01/13/1087054713517542
Published by: http://www.sagepublications.com
Additional services and information for Journal of Attention Disorders can be found at: Email Alerts: http://jad.sagepub.com/cgi/alerts Subscriptions: http://jad.sagepub.com/subscriptions Reprints: http://www.sagepub.com/journalsReprints.nav Permissions: http://www.sagepub.com/journalsPermissions.nav
>> OnlineFirst Version of Record - Jan 13, 2014 What is This?
Downloaded from jad.sagepub.com at Umea University Library on October 11, 2014
517542
research-article2014
JADXXX10.1177/1087054713517542Journal of Attention DisordersQuerne et al.
Article
Effects of Methylphenidate on Default-Mode Network/Task-Positive Network Synchronization in Children With ADHD
Journal of Attention Disorders 1–13 © 2014 SAGE Publications Reprints and permissions: sagepub.com/journalsPermissions.nav DOI: 10.1177/1087054713517542 jad.sagepub.com
Laurent Querne1,2, Sidy Fall1,2, Anne-Gaëlle Le Moing1,2, Emilie Bourel-Ponchel1, Aline Delignières1,2, Anais Simonnot1, Alain de Broca1, Catherine Gondry-Jouet3, Muriel Boucart4, and Patrick Berquin1,2
Abstract Objective: A failure of the anti-phase synchronization between default-mode (DMN) and task-positive networks (TPN) may be involved in a main manifestation of ADHD: moment-to-moment variability. The study investigated whereby methylphenidate may improve TPN/DMN synchronization in ADHD. Method: Eleven drug-naive ADHD children and 11 typically developing (TD) children performed a flanker task during functional magnetic resonance imaging. The ADHD group was scanned without and 1 month later with methylphenidate. The signal was analyzed by independent component analysis. Results: The TD group showed anti-phase DMN/TPN synchronization. The unmedicated ADHD group showed synchronous activity in the posterior DMN only, which was positively correlated with response time variability for the flanker task. Methylphenidate initiated a partial anti-phase TPN/DMN synchronization, reduced variability, and abolished the variability/DMN correlation. Conclusion: Although results should be interpreted cautiously because the sample size is small, they suggest that a failure of the TPN/DMN synchronization could be involved in the moment-to-moment variability in ADHD. Methylphenidate initiated TPN/DMN synchronization, which in turn appeared to reduce variability. (J. of Att. Dis. 2014; XX(X) 1-XX) Keywords ADHD, methylphenidate, fMRI, variability ADHD is characterized by symptoms of inattention and hyperactivity/impulsivity that interfere with the social life and academic performances Diagnostic and Statistical Manual of Mental Disorders (4th ed.; DSM-IV; American Psychiatric Association [APA], 1994). Children with ADHD display moment-to-moment variability in daily life and in the classroom, making it difficult for them to maintain the continuity of mental activities over time (Brown et al., 2005; Castellanos & Tannock, 2002; Klein, Wendling, Huettner, Ruder, & Peper, 2006). Functional neuroimaging studies have revealed hypoactivity in the frontostriatal, parietal, and cerebellar regions suggesting that the latter underlie the networks involved in ADHD (Cortese et al., 2012; Durston, van Belle, & de Zeeuw, 2011). These regions contribute to several functional networks (referred to as task-positive networks [TPNs]) that are activated during the performance of externally oriented tasks. For example, both frontoparietal dorsal-attentional and ventral-attentional networks (DAN and VAN) are mainly solicited when normal individuals performed visuospatial tasks and constitute the
TPNs for these tasks (i.e., Corbetta, Patel, & Shulman, 2008). These networks and other functional networks also showed a spontaneous synchronic activity at rest as evidenced by intrinsic functional connectivity studies as demonstrated on a large sample size (Yeo et al., 2011). Children and adults with ADHD also show abnormal activation of a specific set of regions known as the default-mode network (DMN; Cortese et al., 2012; Hart, Radua, Nakao, MataixCols, & Rubia, 2013; Swanson, Baler, & Volkow, 2011). The DMN supported many internally focused cognitive processes, such as stimulus-independent thoughts or 1
GRAMFC INSERM U1105, Université de Picardie Jules-vernes, France Service de Neuropédiatrie, CHU Amiens-Picardie, France 3 Service de Radiologie, CHU Amiens-Picardie, France 4 LNFP EA4559, Université Lille-Nord, Lille, France 2
Corresponding Author: Laurent Querne, Service de Neuropédiatrie, GRAMFC INSERM U1105, CHU Amiens-Picardie, Place Victor Pauchet, Amiens 80054, France. Email: [email protected]
Downloaded from jad.sagepub.com at Umea University Library on October 11, 2014
2
Journal of Attention Disorders
mind-wandering and shows spontaneous periodic activity at rest. In ADHD, default-mode activity may abnormally persist or intrude into periods of active task-specific processing, producing periodic fluctuations in attention that interfere with externally oriented activity (Sonuga-Barke & Castellanos, 2007). Convergent data have shown that the performance of an externally oriented task is supported by concomitant activation of the appropriate TPN and deactivation of the DMN. Time-domain analyses in normal individuals have revealed that DMN and TPN fluctuations are synchronized in anti-phase during externally oriented task (Kelly, Uddin, Biswal, Castellanos, & Milham, 2008; Sonuga-Barke & Castellanos, 2007). Moreover, DMN activation appears to be correlated with intra-individual response time (RT) variability during a task (Castellanos & Proal, 2012). This parameter is thought to result from moment-to-moment variability (Castellanos & Tannock, 2002) and is strongly impaired in ADHD whatever the task or the participant’s age or subtype (Bourel-Ponchel et al., 2011; Klein et al., 2006; Querne & Berquin, 2009). A positive correlation between RT variability and DMN fluctuations or DMN deactivation has been observed for various tasks in normal adults and in children with ADHD (Bonnelle et al., 2011; Fassbender et al., 2009; Kelly et al., 2008). Hence, moment-to-moment variability, which is so characteristic of functioning in ADHD, could be due (at least in part) to the maintenance of abnormal cyclic DMN activity that could interfere with task-related regions (Sonuga-Barke & Castellanos, 2007). In ADHD, stimulants relieve symptoms, improve academic achievement and the child’s quality of life, and reduce RT variability for externally oriented tasks (National Institutes of Health Consensus Development Conference Statement, 2000). Functional neuroimaging studies have shown that stimulants (and notably methylphenidate [MPH]) increased activation in frontostriatoparietal regions (Durston et al., 2011; Hart et al., 2013) and enhance deactivation of the DMN during inhibitory tasks (Liddle et al., 2011; Peterson et al., 2009). Nevertheless, synchronization of the TPN/DMN complex, the effect of stimulants on the latter and the relationship between synchronization and moment-to-moment variability have never been studied in children with ADHD. The present study sought to explore whether methylphenidate modifies synchronization of the TPN/DMN complex in relation to the moment-to-moment variability by applying an independent component analysis (ICA) to the time course of hemodynamic activity (Cole, Smith, & Beckmann, 2010). Our starting hypothesis was first that non-medicated ADHD children may present null or weak anti-phase synchronization of the TPN/DMN complex during an externally oriented task and second that methylphenidate may restore (at least partially) TPN/DMN anti-phase synchronization and thus reduce RT variability during the task.
Materials and Method Participants ADHD was diagnosed according to the DSM-IV criteria (APA, 1994) following a neurological evaluation of the child and an interview with his or her parents. The study’s main inclusion criteria were as follows: children aged 8 to 13, no previous drug treatment, a combined subtype (inattention criteria ≥ 6 and hyperactivity/impulsivity criteria ≥ 6), an intelligence quotient (IQ) > 85 and approval of methylphenidate treatment by the child’s parents. Children with a history of neurological, psychiatric, or developmental disorders (other than ADHD), depression or generalized anxiety were excluded from the study. A group of typical development children (TDC), presenting normal school achievement, and a normal IQ was constituted (IQ > 85, as evaluated with the Wechsler Intelligence Scale for Children 4th edition (WISC-IV) Similarities, Vocabulary, Block Design, and Picture Concepts). As in the ADHD group, the children in the TD group were medication-naive and did not have any history of neurological, psychiatric, or developmental disorders, depression or generalized anxiety. The children’s parents received comprehensive information about the study’s objectives and procedures, which were approved by the local ethics committee (Comité Pour la Protection des Personnes Nord-Ouest II). Children in the ADHD group underwent their first functional magnetic resonance imaging (fMRI) session prior to initiation of treatment. The methylphenidate treatment (20 or 30 mg, extended-release formulation) was initiated on the day after the first fMRI session. A second fMRI session was performed after 1 month of continuous treatment. The children took their medication 2 hr before the start of the second session. During the MRI session, the children’s parents filled out the French-language version of the Swanson, Nolan, and Pelham 4th edition (SNAP-IV) rating scale. Two ADHD children refused to go into the magnetic resonance imaging machine because of anxiety. Hence, our final analysis took account of 11 children in each group.
Cognitive Paradigm: The Flanker Task Stimuli were back-projected onto a screen viewed by children through a prism mirror incorporated into the fMRI antenna by using the SuperLab® software (Cedrus Corporation). Responses were collected via an MRI response box (Lumina Pad®, Cedrus Corporation) equipped with a two-push-button pad. The task stimuli consisted of five fish (considered to be more attractive for children than the usual arrows) arranged in a horizon array. The child was required to identify the orientation of the central fish (flanked by the four other fish) by pressing the right button with the right thumb (for a rightward orientation) or the left
Downloaded from jad.sagepub.com at Umea University Library on October 11, 2014
3
Querne et al.
also compared in terms of age and WISC-IV scores in Mann–Whitney U tests.
fMRI Acquisition
Figure 1. An example of three successive trials in the flanker task performed by children in the MRI.
Note. Top: A congruent trial with a rightward orientation of the central target. Middle: An incongruent trial with a leftward orientation of the central target. Bottom: A congruent trial with a leftward orientation of the central target. MRI = magnetic resonance imaging.
fMRI was performed with a Sigma 3T system (General Electric, Milwaukee, Wisconsin) using a T2*-weighted, single-shot, echo-planar imaging sequence (echo time [TE] = 55 ms, flip angle [FA] = 60°, in-plane matrix = 64 × 64, field of view [FOV] = 240 × 240 mm2, repetition time [TR] = 3,000 ms). In each of 107 functional volumes (including three dummy scans), 35 axial slices parallel to the AC–CP plane were acquired with an inter-slice interval of 4 mm (covering the whole brain). In addition to functional scans, high-resolution, axial, T1-weighted 3D-SPGR BRAVO images were collected from each participant (covering the whole brain) by using the following parameters: inversion time = 200 ms, TR/TE = 10.4/4.2 ms, FA = 20°, slice thickness = 1.4 mm, inter-slice interval = 0.7 mm, matrix = 256 × 256 pixels, and FOV = 256 × 256 mm2. This structural image was used to warp the functional images into Montreal Neurological Institute (MNI) standard space.
Independent Component Analysis
Behavioral Analysis
fMRI data pre-processing. The fMRI data from each group were fed into a multi-subject temporal concatenation analysis using MELODIC software (Multivariate Exploratory Linear Decomposition into Independent Components: FMRIB’s Software Library (FSL): Smith et al., 2004). Prestatistical processing of the fMRI data included brain extraction (to exclude non-brain structures) using the BET tool (Smith, 2002), removal of low-frequency drifts (using a high-pass filter cut-off of 156 s), within-subject motion correction (using the FMRIB’s linear registration tool, MCFLIRT; Jenkinson, Bannister, Brady, & Smith, 2002), mean-based intensity normalization of all volumes by the same factor, spatial smoothing (using a Gaussian kernel of 5 mm FWHM), co-registration of T1-weighted images, spatial normalization into the MNI 152 standard space (using a 12-mm affine/non-linear registration), and resampling the data into 2-mm isotropic voxels.
The analyzed flanker task parameters were the proportion of errors related to target orientations, the mean RT, RT variability (intra-individual standard error for the RTs), and the congruency-flanker effect (the mean RT for incongruent trials minus the mean RT for congruent trials). Wilcoxon’s tests were performed on all task parameters and the SNAP-IV scores in the ADHD group, with medication as a within-group factor (off/on-methylphenidate). Mann– Whitney U tests were performed on task parameters and SNAP-IV scores by comparing the TD group and ADHD group for the off-methylphenidate session and then onmethylphenidate session. The ADHD and TD groups were
Group level inference. Group fMRI data were reshaped into a three-dimensional tensor (Time points × Voxels × Subjects) and were then decomposed into a set of independent spatial maps together with associated time courses and subject-mode values, using the probabilistic ICA (Beckmann & Smith, 2005). For each estimated component, this approach is able to provide signal of interest in the spatial (group map of patterns of activation/deactivation), temporal (times-series for the cerebral group map), and subject domain (individual subject-mode values: For details, see the ICA and correlation between subject-mode values and
button with the left thumb (for a leftward orientation). Each trial started with presentation of the five fish for 1,000 ms, which were then replaced by a central fixation mark for 2,000 ms. The child was allowed to provide a response throughout the 3,000 ms trial (Figure 1). The trials were presented in a pseudorandom order. Thirty-second height blocks were preceded by a 9-s rest period, during which the child was instructed to look passively at five elliptical shapes that were very similar to the fish in size and color (Figure 1). The cognitive paradigm lasted 5.35 min in total and was preceded by a short training session.
Downloaded from jad.sagepub.com at Umea University Library on October 11, 2014
4
Journal of Attention Disorders
RT variability paragraph). The order of each group model was estimated by the Laplace approximation to the Bayesian evidence for a probabilistic ICA model (Beckmann & Smith, 2004). Group-specific spatial maps were then reduced to Z-statistics maps by dividing the estimated components’ weightings by the voxel-wise standard error of the residual noise. Clusters of voxels were thresholded by an alternative-hypothesis test (at a posterior probability threshold of p > .5), which was assessed by mixture-modeling of the Z-values’ probability density (Woolrich, Behrens, Beckmann, & Smith, 2005). The number of components generated was fixed by the MELODIC software according to an automatic dimensionality estimation of the model order.
Identification of DMN Components According to the procedure described in Kelly et al. (2010), the resulting spatial distributions of Z score maps and the associated power spectrum were examined in each group, to differentiate between the components corresponding to the neural signal of interest and those predominantly representing noise. We excluded components for which the high frequencies (>0.1 Hz) of the signal time-courses constituted 50% or more of the total power in the Fourier spectrum and brain activities, which were apparently scattered at random over a large section of the brain or mainly located within peripheral areas. Then, white matter, gray matter, and cerebral spinal fluid templates for each group were obtained from means of segmented individual structural images (within MNI standardized brain space) using FSL FMRIB Automated Segmentation Tool (Zhang, Brady, & Smith, 2001). Each segmented tissue was then compared with all the component maps corresponding to the neural signal of interest, to discard spatial maps with a high percentage (90%) of voxels in non-gray matter. We searched in each group the component that better capture the anti-correlation between the DMN and the frontoparietal TPN for visuospatial-conflicting tasks (Bonnelle et al., 2011; Kelly et al., 2008; Yeo et al., 2011). If DMN/TPN anti-correlation was not present, we searched for components included in the canonical DMN.
ICA and Correlation Between Subject-Mode Values and RT Variability The study of brain function can be carried out with the ICA approach, which can separate fMRI data into a set of uncorrelated spatio-temporal components describing the spatial and temporal characteristics of underlying signals. Each component consisted of voxel values (cerebral group map) and a unique associated time course of activation/ deactivation (McKeown et al., 1998). An advantage of this approach is the ability to detect unknown, yet structured
spatio-temporal processes (Beckmann & Smith, 2005; McKeown et al., 1998), which do not rely on a single seed region but can integrate the temporal information in the fMRI data across multiple distributed networks (Beckmann, Mackay, Filippini, & Smith, 2009). The ICA algorithm that we used in this study permits to decompose the group-specific fMRI data to spatially independent group maps with their associated time series and individual subject-mode values. This last parameter represents the processes of interest present in each spatial map in terms of the subject-dependent variations (Beckmann & Smith, 2005) and can be associated for each component as a weight indicating the contribution of individual temporal dynamics to each group-level spatial map identified. See the appendix for an illustration of the relationship linking the individual ICA subject-mode values and the correlations between individual time series in key regions of interest (ROIs) of the complex TPN/DMN and time series associated to ICA-selected group components. A potential relationship has been searched between the RT variability and the dynamic of the fluctuations of the DMN/TPN, which mediates the re-allocation of cognitive resources to the TPN when the participants were instructed to perform the task. To explore this question, we calculated separately for each group of children (TDC/ADHD off-MPH/ADHD on-MPH) the correlations between two sets of values (ri, bi), whereby r1, . . . r11 (ICA subject-mode values) denoting the contribution of individual temporal dynamics to each group-level spatial map in the TPN/DMN complex (or DMN if the anti-correlation was not present), and b1, . . . b11 denoting the individual RT variability during the flanker task for each participant (n1, . . . n11).
Results Age, Psychometric Data, and SNAP-IV Scores The two groups did not differ significantly in terms of age and IQ (age = 9.8 ± 1.7 years, verbal IQ = 112.5 ± 14.4, performance IQ = 103.1 ± 14.4 in the ADHD group; age = 10.8 ± 1.7 years, verbal IQ = 122.4 ± 11.7, performance IQ = 111.9 ± 8.8 in the TD group). The parent-reported SNAP-IV scores were significantly higher for the ADHD group (inattention = 7.9 ± 1.4; hyperactivity/impulsivity = 7.1 ± 2.1) than for the TD group (inattention = 0.5 ± 1.8; hyperactivity/impulsivity = 0.2 ± 0.6; Z = 3.9, p < .0001; Z = 4.0, p
