hyperactivity disorder and healthy controls

Author’s Accepted Manuscript Neural correlatesy of visuospatial working memory in attention-deficit/hyperactivity disorder and healthy controls Hanneke van Ewijk, Wouter D. Weeda, Dirk J. Heslenfeld, Marjolein Luman, Catharina A. Hartman, Pieter J. Hoekstra, Stephen V. Faraone, Barbara Franke, Jan K. Buitelaar, Jaap Oosterlaan

PII: DOI: Reference:

www.elsevier.com

S0925-4927(15)30027-5 http://dx.doi.org/10.1016/j.pscychresns.2015.07.003 PSYN10399

To appear in: Psychiatry Research: Neuroimaging Received date: 28 August 2014 Revised date: 21 May 2015 Accepted date: 3 July 2015 Cite this article as: Hanneke van Ewijk, Wouter D. Weeda, Dirk J. Heslenfeld, Marjolein Luman, Catharina A. Hartman, Pieter J. Hoekstra, Stephen V. Faraone, Barbara Franke, Jan K. Buitelaar and Jaap Oosterlaan, Neural correlatesy of visuospatial working memory in attention-deficit/hyperactivity disorder and healthy controls, Psychiatry Research: Neuroimaging, http://dx.doi.org/10.1016/j.pscychresns.2015.07.003 This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting galley proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

FMRI OF VSWM IN ADHD

VAN EWIJK ET AL.

Neural correlates of visuospatial working memory in attention-deficit/hyperactivity disorder and healthy controls Hanneke van Ewijka,*, Wouter D. Weedaa, Dirk J. Heslenfelda, Marjolein Lumana, Catharina A. Hartmanb, Pieter J. Hoekstrab, Stephen V. Faraonec, Barbara Franked, Jan K. Buitelaare,f, Jaap Oosterlaana a

Department of Clinical Neuropsychology, VU University Amsterdam, Amsterdam, The Netherlands b University of Groningen, University Medical Center Groningen, Department of Psychiatry, Groningen, The Netherlands c Departments of Psychiatry and of Neuroscience and Physiology, SUNY Upstate Medical University, Syracuse, NY, USA d Radboud University Medical Center, Donders Institute for Brain, Cognition and Behaviour, Centre for Neuroscience, Departments of Human Genetics and Psychiatry, Nijmegen, The Netherlands e Radboud University Medical Center, Donders Institute for Brain, Cognition and Behavior, Centre for Neuroscience, Department of Cognitive Neuroscience, Nijmegen, The Netherlands f Karakter, Child and Adolescent Psychiatry University Centre, Nijmegen, The Netherlands

Running head: fMRI of VSWM in ADHD Word count: Abstract: 194 Text: 5100

* Corresponding author: Hanneke van Ewijk, Department of Clinical Neuropsychology, Faculty of Psychology and Education, VU University Amsterdam, van der Boechorststraat 1, 1081 BT, Amsterdam, The Netherlands; Tel.: +31 (0)20 598 8770; fax +31 (0)20 598 8971; email: [email protected] Abstract Impaired visuospatial working memory (VSWM) is suggested to be a core neurocognitive deficit in attention-deficit/hyperactivity disorder (ADHD), yet the underlying neural activation patterns are poorly understood. Furthermore, it is unclear to what extent age and gender effects may play a role in VSWM-related brain abnormalities in ADHD. Functional magnetic resonance imaging (fMRI) data were collected from 109 individuals with ADHD (60% male) and 103 controls (53% male), aged 8-25 years, during a spatial span working

1

FMRI OF VSWM IN ADHD

VAN EWIJK ET AL.

memory task. VSWM-related brain activation was found in a widespread network, which was more widespread compared with N-back tasks used in the previous literature. Higher brain activation was associated with higher age and male gender. In comparison with controls, individuals with ADHD showed greater activation in the left inferior frontal gyrus (IFG) and the lateral frontal pole during memory load increase, effects explained by reduced activation on the low memory load in the IFG pars triangularis and increased activation during high load in the IFG pars opercularis. Age and gender effects did not differ between controls and individuals with ADHD. Results indicate that individuals with ADHD have difficulty in efficiently and sufficiently recruiting left inferior frontal brain regions with increasing task difficulty.

Keywords: Spatial span, Age, Gender, Development, fMRI, ADHD, Children

2

FMRI OF VSWM IN ADHD

VAN EWIJK ET AL.

1. Introduction Impaired working memory is suggested to be a core neurocognitive deficit in attentiondeficit/hyperactivity disorder (ADHD) (Martinussen et al., 2005; Willcutt et al., 2005), a deficit that is most pronounced on tasks that tap into the spatial domain (Martinussen et al., 2005). While several studies have used functional magnetic resonance imaging (fMRI) to investigate the neural correlates of working memory in ADHD (Cortese et al., 2012), these studies typically used non-spatial working memory tasks. In contrast, the literature on the neural correlates of visuospatial working memory (VSWM) in ADHD is surprisingly scarce. In healthy controls, VSWM is reported to be controlled by several (pre)frontal and parietal brain regions, including the dorsolateral and ventrolateral prefrontal cortex, premotor cortex, posterior parietal cortex, cingulate cortex, and intra-parietal sulcus (Nelson et al., 2000; Kwon et al., 2002; Wager and Smith, 2003; Klingberg, 2006). To our knowledge, only two fMRI studies of VSWM compared individuals with ADHD with controls. One study in ADHD showed higher activation for adults with ADHD in the left supplementary motor area in a low memory load condition (Ko et al., 2013), while the other reported higher activation in children with ADHD in the dorsolateral prefrontal cortex and posterior cingulate cortex during high, but not low, memory load (Bédard et al., 2014). Both studies reported no group differences in activation during memory load increase (high vs. low memory load) and no differences on any of the behavioural measures. Together, these findings suggest that individuals with ADHD may have difficulty recruiting prefrontal regions efficiently to obtain performance levels similar to those obtained by controls. Importantly, both fMRI VSWM studies in ADHD used a visuospatial N-back task, while most behavioural studies of VSWM use spatial span tasks. At the behavioural level, performance on N-back and performance on span tasks are only weakly correlated, and the construct validity of the N-back task as a measure of working memory has not been conclusively established, as opposed to span tasks. Researchers have been discouraged from

3

FMRI OF VSWM IN ADHD

VAN EWIJK ET AL.

treating the two tasks as measuring the same construct, or from generalizing results on underlying brain circuitries from N-back studies to the concept of working memory span (Kane et al., 2007; Redick and Lindsey, 2013). To our knowledge, only one study has described neural activation patterns associated with spatial span performance. The authors reported activation in the superior and middle frontal gyri, inferior and superior frontal sulci, cingulate cortex, and large parts of the parietal and occipital cortices in a small group of 13 typically developing children (Klingberg et al., 2002). Given that only two fMRI studies investigated VSWM in ADHD, and both studies used relatively small numbers of subjects (2030 subjects per group) and used N-back tasks (but not spatial span tasks), the current literature lacks a clear investigation of the neural correlates of VSWM in ADHD. Literature documenting the developmental trajectory of VSWM-related brain activation patterns in ADHD is rather limited, though this issue is highly important in light of the discussion of whether brain abnormalities in ADHD show normalization during adolescence or instead represent a more persistent deficit (Rubia, 2007; Shaw et al., 2007; Silk et al., 2009). At the behavioural level, some evidence exists for different developmental trajectories, with VSWM impairment in ADHD becoming more pronounced during adolescence (Westerberg et al., 2004), but there is also evidence of similar levels of impairment for children, adolescents and young adults with ADHD (van Ewijk et al., 2014). To our knowledge, no studies have investigated the neural correlates of the developmental trajectory of VSWM in ADHD. Evidence from studies in healthy controls suggests that children and adults recruit similar brain networks, including the bilateral prefrontal cortex and posterior parietal cortex, but that brain activation in these regions increases over age (Thomas et al., 1999; Kwon et al., 2002). Accordingly, it is possible that children with ADHD show a similar increase in brain activation over age, either with or without normalisation towards activation levels similar to those in controls. It is also possible, however, that differences in brain activation between children with ADHD and controls become even more

4

FMRI OF VSWM IN ADHD

VAN EWIJK ET AL.

pronounced as children grow older, in line with previous behavioural results (Westerberg et al., 2004). Exploring age effects on VSWM-related brain activation in ADHD may shed more light on this issue. Gender differences in VSWM-related brain activation are another relatively unexplored topic. In healthy controls, differences in cognition, as well as brain structure and function, between males and females are well documented (Speck et al., 2000; Goldstein et al., 2005; Bell et al., 2006; Hsu et al., 2008; Gong et al., 2011). These findings underline the importance of taking gender effects into account when studying the neural correlates of cognitive tasks (Goldstein et al., 2005). This is especially important in patient studies with imbalanced samples with regard to gender, such as ADHD studies that typically include more ADHD males and control females. Gender effects in ADHD (i.e., group × gender interactions) have been reported in relation to neurocognitive performance (Gaub and Carlson, 1997; Gershon and Gershon, 2002; Balint et al., 2009), though no studies specifically investigated VSWM. However, gender effects are rarely investigated in fMRI studies of ADHD. One fMRI study that specifically addressed this issue found underactivation for male, but not female, subjects with ADHD (Valera et al., 2010). This study, however, used a verbal N-back task, and a sample of adult subjects. Exploring gender effects in the neural correlates of VSWM with a spatial span task will enhance our insight into whether a similar interaction also exists in spatial span working memory, and in children and adolescents with ADHD as opposed to adults. The current study set out to map spatial span-related brain activation patterns in a large sample of children, adolescents and young adults with and without ADHD, to further unravel the neurophysiological underpinnings of VSWM impairment in ADHD. To the best of our knowledge, this is the first fMRI VSWM study in ADHD using a spatial span task, which is more comparable with the current literature on VSWM deficits in ADHD at the behavioural level. Moreover, considering our large gender-balanced sample with a broad age range, we

5

FMRI OF VSWM IN ADHD

VAN EWIJK ET AL.

were able to extend the existing literature by exploring the role of age and gender on neural correlates of VSWM, and to examine whether these effects differ between individuals with and without ADHD. Based on previous results by Klingberg and colleagues (Klingberg et al., 2002), we expected to find VSWM-related brain activation in inferior and superior frontal and parietal regions, the occipital cortex, and the cingulate. We hypothesized that individuals with ADHD would show underactivation in these regions compared with controls. Given the scarce and inconsistent literature on age and gender effects in fMRI studies of ADHD, we did not formulate explicit hypotheses for these exploratory analyses.

2. Methods 2.1. Participants Participants were part of the NeuroIMAGE cohort (von Rhein et al., 2014), including both ADHD and healthy control families. Inclusion criteria for the current study were as follows: age between 8 and 30 years, European Caucasian descent, IQ ≥ 70, and no known neurological or genetic disorder. Comorbid psychiatric disorders reported by parents were excluded, except oppositional defiant disorder (ODD), conduct disorder (CD), and pervasive developmental disorder not otherwise specified (PDD-NOS), given their high co-occurrence in ADHD. Furthermore, participants were excluded if they had any contraindication to MRI scanning (e.g., implanted metal or medical devices). Data were initially available of 281 participants who met inclusion criteria, aged between 8 and 25 years, including 156 ADHD subjects (all subtypes) and 125 healthy controls.

2.2. Diagnostic procedure To determine ADHD diagnoses, all participants were assessed with the Schedule for Affective Disorders and Schizophrenia for School-Age Children - Present and Lifetime Version (K-SADS; Kaufman et al., 1997) and Conners' ADHD questionnaires from multiple informants

6

FMRI OF VSWM IN ADHD

VAN EWIJK ET AL.

(Conners et al., 1998a, 1998b; Conners et al., 1999). A comprehensive diagnostic algorithm was used to determine combined symptom counts for inattentive and hyperactive/impulsive behaviour, derived from both measures. Participants with a combined symptom count of ≥ 6 symptoms on one or both of the dimensions were diagnosed with ADHD, provided they (a) met the DSM-IV and DSM-5 (American Psychiatric Association, 2000, 2013) criteria for pervasiveness and impact of the disorder (measures derived from the Kiddie Schedule for Affective Disorders and Schizophrenia for School Aged Children, K-SADS), (b) showed an age of onset before 12, in accordance with the DSM-5 (derived from the K-SADS; Polanczyk et al., 2010), and (c) the child received a T score ≥ 63 on at least one of the DSM ADHD scales on either one of the Conners' ADHD questionnaires. Controls were required to have a T score < 63 on the Conners' ADHD questionnaires, and to have ≤ 3 symptoms on the combined symptom count, to ensure they did not show signs of (subthreshold) ADHD. Criteria were slightly adapted for young adults (≥ 18 years), such that a combined symptom count of five symptoms was sufficient for a diagnosis (Kooij et al., 2005). Young adults were included as controls when they had ≤ two symptoms on both combined symptom counts. A more detailed description of diagnostic procedures has been published elsewhere (von Rhein et al., 2014). Additional sections of the K-SADS-PL (Present Lifetime version) were administered to assess the presence of comorbid ODD, CD, mood and anxiety disorders. A custom questionnaire filled in by parents assessed the presence of an official diagnosis of PDD-NOS.

2.3. Procedure The current study was part of a comprehensive assessment protocol encompassing phenotypic assessment, neurocognitive measures, and an MRI protocol including an fMRI scan (von Rhein et al., 2014). Previous studies have shown that ADHD medication may have a normalizing effect on working memory performance and brain activation patterns (e.g.,

7

FMRI OF VSWM IN ADHD

VAN EWIJK ET AL.

Cubillo et al., 2013). Participants were therefore asked to withhold the use of psychoactive medication for 48 h before measurement. Those who were not able to comply were excluded from the analyses (n=13 ADHD subjects). Before the scan, participants were familiarized with the scanning procedure and scanner sounds using a mock scanner to reduce anxiety and head movements. In addition, a practice version of the VSWM task was administered outside the scanner shortly before participants entered the scanner. During the testing day, participants were motivated with short breaks and, at the end of the day, participants received a reward of €50 and a copy of their MRI scan results. Full-scale IQ was estimated by the Vocabulary and Block Design subtests of the Wechsler Intelligence Scale for Children-III (WISC-III) or Wechsler Adult Intelligence Scale III (WAIS-III; for participants ≥ 17 years). Data acquisition was carried out in the Netherlands, either in Amsterdam at the VU University Amsterdam and VU University Medical Centre, or in Nijmegen at the Donders Centre for Cognitive Neuroimaging at the Radboud University Nijmegen. The study was approved by the Dutch local medical ethics committees. Informed consent was signed by all participants (parents signed informed consent for participants under 12 years of age).

2.4. Spatial span task The spatial span task used to measure VSWM is an adapted version of a task developed by Klingberg and colleagues (Klingberg et al., 2002; McNab et al., 2008; van Ewijk et al., 2014). Two trial types (baseline and working memory) and two memory loads (low and high) were implemented in the task. Each trial consisted of a sequence of either three or six yellow circles (low and high memory load, respectively), displayed on a 4×4 grid for 500 ms each, with a 500-ms inter-stimulus interval in between (see Fig. 1). Subsequently, during a 2000-ms response window, a probe consisting of a number with a question mark was presented in one of the 16 locations. During working memory trials, participants were asked to remember the spatial location and temporal order of the presentation of cues, and

8

FMRI OF VSWM IN ADHD

VAN EWIJK ET AL.

indicate with a 'yes' or 'no' response (left or right button, respectively) whether the location of the probe had been stimulated before, at the indicated temporal position. During baseline trials, red circles followed by the probe (always the number 8) were presented sequentially in the four corners of the grid in a predictive manner, and participants were required to pay attention but not to try to remember the sequence, and always had to press the ‘no’ button. During both conditions, feedback was presented after the response in the form of a green or red coloured bar below the probe (for correct and incorrect responses, respectively), for the remainder of the response window. Accuracy on both conditions was determined by percentage of correct responses. The task was administered in four blocks of 24 trials each (presented in fixed random order), with a short break in between blocks to motivate participants and to avoid fatigue effects, with a total task duration of approximately 16 min. [ FIGURE 1 ]

2.5. Behavioural data analysis Possible group differences in sample characteristics and behavioural data of the spatial span task were analysed using SPSS (Statistical Package for the Social Sciences, version 21, IBM, Chicago, IL, USA). Behavioural data were analysed using Linear Mixed Models with a random intercept per family, to account for correlated data within families. A repeated measures model examined group differences in working memory performance, with memory load as a within-subject factor, and group (ADHD vs. controls), age and gender as predictors. Furthermore, scan site was included as a covariate to control for possible site differences, and performance on baseline trials was included to control for basic processing or motivational effects. Note that results for behavioural performance on the spatial span task in a largely overlapping sample have been described in detail elsewhere (van Ewijk et al., 2014).

9

FMRI OF VSWM IN ADHD

VAN EWIJK ET AL.

2.6. Imaging acquisition and (pre-)processing MRI scanning was carried out on either a 1.5 Tesla Sonata or a 1.5 Tesla Avanto MRI scanner (Siemens, Erlangen, Germany), using the same Siemens eight-channel head coil and closely matched pulse sequence parameters. Whole-brain, high-resolution T1-weighted anatomical images were acquired in the sagittal plane (magnetization-prepared rapid acquisition gradient echo (MP-RAGE), 176 slices, acquisition matrix 256×256, voxel size 1×1×1mm; echo time (TE)=2.95 ms, repetition time (TR)=2730 ms, inversion time (TI)=1000 ms, flip angle (FA)=7°, Generalized Autocalibrating Partially Parallel Acquisition (GRAPPA)-acceleration 2). Four functional runs were acquired (gradient-echo echo planar imaging (GE-EPI), TR=2340 ms, TE=40 ms, field of view (FOV)=224 mm, voxel size=3.5×3.5×3.0 mm, slice gap 0.5 mm, 38 slices). Functional images of each subject were realigned (using rigid body transformations), from which three translation and three rotation parameters were estimated, and subsequently images were slice-time-corrected using standard SPM8 routines (Statistical Parametric Mapping, Wellcome Trust Centre for Neuroimaging, London, UK). Subsequent fMRI data processing was carried out using FSL FEAT (FMRI Expert Analysis Tool; FMRIB Analysis group, Oxford, UK), including removal of the first three volumes, spatial smoothing (6-mm full width at half maximum (FWHM) Gaussian kernel) and highpass filtering (100 s). Subsequently, each participant's functional image was registered to his or her T1 image using linear registration in FLIRT (12 depth of field; DOF), followed by nonlinear registration of the images to the MNI152 template using FNIRT. Runs with >3mm motion in any direction were excluded from analysis (as determined by the maximum absolute displacement throughout the entire run). Data of 23 runs of ADHD and 8 runs of control subjects were discarded. Furthermore, the FSL Motion Outliers routine was run to detect time points with large motion and model these as a confound matrix, thereby regressing the effect of these time points out of the analysis.

10

FMRI OF VSWM IN ADHD

VAN EWIJK ET AL.

Fifty-six participants were excluded from analysis due to incidental findings after visual inspection (e.g., enlarged ventricles or unexpected hypo-intensities; 2 ADHD subjects, 4 controls), scan quality (e.g., artefacts, missing volumes, insufficient coverage of the entire brain; 12 ADHD subjects, 10 controls), insufficient data after exclusion of runs with >3 mm motion (19 ADHD, 8 controls), and one ADHD subject was excluded due to being an extreme outlier on blood oxygenation level-dependent (BOLD) activation parameters. Accordingly, the final dataset included 109 ADHD subjects and 103 controls.

2.7. fMRI analysis Four contrasts of interest were set up as follows: low memory load, high memory load, mean working memory (both loads averaged), and load difference (high minus low memory load). Each contrast consisted of activation during working memory trials corrected for activation during baseline trials of the same load(s), modelled from the start of the trial to the onset of the probe. Using FSL FEAT, each participant's activation map was estimated for each contrast for each run, while including confound regressors for motion (3 translation and 3 rotation parameters) and time points with extreme motion outliers (derived from FSL Motion Outliers). Second, for each participant, data from all runs were concatenated using a single-subject fixed effects model, resulting in four contrast maps (low load, high load, mean working memory, and load difference) per participant, reflecting contrast-specific activity across all available runs. Whole-brain analyses were conducted in FSL FEAT using a mixed effects model (FLAME1). Z statistical images were thresholded using a cluster forming threshold of Z>2.3 and a (corrected) cluster significance threshold of p<0.05 (Worsley, 2001). Automatic outlier de-weighting (Woolrich, 2008) was used to reduce the impact of remaining outlier data points on subsequent analyses. Whole-brain analyses examined brain activation on both the load difference and mean working memory contrasts. A General Linear Model (GLM) was set

11

FMRI OF VSWM IN ADHD

VAN EWIJK ET AL.

up to examine (a) spatial span-related brain activation (across all subjects), (b) group differences (ADHD vs. controls), (c) age and gender effects (across all subjects), and (d) group-by-age and group-by-gender interactions. In all analyses, scan site was included as a covariate to control for possible effects of site. In case of significant effects on the load difference contrast, follow-up analyses were conducted to further investigate the nature of this effect. To this end, the mean activation parameter (beta value) was extracted from the significant cluster for each participant for the low and high memory load contrasts separately, and analyses of variance were conducted in SPSS to examine group differences on both loads, while including the same predictors and covariates as the whole-brain analysis. To ensure that group differences were not driven by comorbid disorders in ADHD, all fMRI analyses were repeated while excluding participants with ODD, CD, affective or anxiety disorders, and PDD-NOS.

3. Results 3.1. Behavioural results Sample characteristics are described in Table 1. The Linear Mixed Model for VSWM performance showed a main effect of memory load, with better performance on the low compared with the high memory load across all subjects (Table 2). A significant group difference was found with lower accuracy for ADHD compared with controls, which did not interact with memory load (Fig. 2). Age was significantly associated with VSWM performance (better performance with increasing age), and did not interact with group. Furthermore, better performance on baseline trials was associated with better VSWM performance. Neither a significant effect of gender nor a group-by-gender interaction was found.

[ FIGURE 2 ]

12

FMRI OF VSWM IN ADHD

VAN EWIJK ET AL. [ TABLE 1, 2 ]

3.2. fMRI results Anatomical labels of all fMRI results are summarized in Supplementary Table S1. Whole-brain analyses were conducted on both the mean working memory and load difference contrasts. First, to estimate spatial span-related brain activation, mean activation patterns for each contrast were estimated across all participants. For the mean working memory contrast, a widespread pattern of brain activation was found. To aid interpretation, activation patterns were evaluated at a more conservative threshold (Z>8) than the one used in the whole-brain analysis to obtain regions showing the strongest effect. Significant clusters were bilaterally located in the middle frontal gyrus, precentral gyrus, superior parietal cortex, insula, thalamus, bilateral occipital cortex, and cerebellum (Fig. 3, left panel). Activation associated with load difference was found bilaterally in the lateral frontal pole, paracingulate gyrus, the inferior, middle and superior frontal gyri, middle temporal gyrus, and occipital and posterior parietal regions (Fig. 3, right panel).

[ FIGURE 3 ]

Second, group differences between the ADHD group and the control group were investigated. No significant group differences were found on the mean working memory contrast. On the load difference contrast, ADHD subjects showed higher activation compared with controls in the left inferior frontal cortex, more specifically in the left inferior frontal gyrus (IFG; Brodmann areas (BA) 44/45), extending to the lateral frontal pole (BA 10) (see Fig. 4). For follow-up analyses, the significant cluster was evaluated at a more conservative threshold (Z>2.5, minimum cluster size = 100 voxels) compared with the one

13

FMRI OF VSWM IN ADHD

VAN EWIJK ET AL.

used in the whole-brain analysis to subdivide it into more meaningful regions. This threshold yielded the following three separate clusters: the IFG - pars opercularis, IFG - pars triangularis, and the lateral frontal pole (Table 3). Follow-up analyses examined group differences on the low and the high memory load conditions separately in each of the three clusters. Results revealed significant group differences in the IFG pars opercularis on the high memory load (F1,207=11.08, p=0.001) but not the low load (F1,207=0.503, p=0.479), and the IFG pars triangularis on the low memory load (F1,207=24.77, p<0.001) but not the high load (F1,207=0.373, p=0.542) (see Fig. 5). No group differences were found in the lateral frontal pole (all p>0.05). Activation in none of the three clusters showed a significant correlation with behavioural performance on the task (represented by mean accuracy across both memory loads, corrected for baseline; all p>0.05).

[ TABLE 3 ] [ FIGURE 4, 5 ]

Third, age and gender effects were examined. Age-related increases were found on mean working memory in the bilateral precentral gyrus, middle and superior frontal gyrus, frontal operculum, cingulate and paracingulate gyrus, superior parietal cortex, pallidum, occipital cortex and cerebellum, and in the right supramarginal gyrus, putamen, and orbitofrontal cortex. Activation associated with load difference significantly decreased with age in the posterior division of the cingulate gyrus and the bilateral supplementary motor cortex. To further illustrate the nature of this age-by-memory load interaction, follow-up analyses were conducted while splitting the sample at the mean age (17.0 years), in order to compare activation in the younger (n=106) compared with the older (n=106) subjects. Analyses showed that the age-by-memory load interaction was due to higher activation on the low load for the older compared with the younger age group (F1,207=7.541, p=0.007),

14

FMRI OF VSWM IN ADHD

VAN EWIJK ET AL.

while activation on the high memory load was similar for both age groups (F1,207=0.070, p=0.791) (Fig. 6, upper panel). Gender effects were found for mean working memory, with higher activation for males than females in the bilateral frontal pole, paracingulate gyrus, superior frontal gyrus, posterior parietal regions (precuneus), and the right angular gyrus. Group differences for load difference were located in partly the same regions, including the bilateral frontal pole, paracingulate gyrus and superior frontal gyrus, with higher activation for females compared with males. Follow-up analyses showed that this was due to higher activation for males compared with females on the low memory load (F1,207=25.275, p<0.001), but similar activation on the high memory load (F1,207=0.012, p=0.911) (Fig. 6, lower panel). Fourth, whole-brain analyses were conducted to determine whether age and gender effects were similar for subjects with and without ADHD. No group-by-age or group-bygender interactions were found on either the mean working memory or the load difference contrast, indicating that group differences were similar for males and females, and across the whole age range investigated. Last, all fMRI analyses were re-run while excluding subjects with comorbid disorders to ensure that results were not driven by the high percentage of subjects with comorbid diagnoses. Results of all analyses were qualitatively the same compared with the original analyses.

[FIGURE 6 ]

4. Discussion The current study set out to investigate the neural correlates of VSWM impairment in ADHD. Using a spatial span task, we examined VSWM-related brain activation patterns in a large sample of individuals with and without ADHD, and additionally investigated the effects of

15

FMRI OF VSWM IN ADHD

VAN EWIJK ET AL.

age and gender as well as group-by-age and group-by-gender interactions. Behavioural results showed impaired performance for individuals with ADHD on both the high and the low memory load, with similar decreases in performance during memory load increase. fMRI results showed higher brain activation for ADHD subjects compared with controls during memory load increase in left inferior frontal regions, which was similar for males and females, and similar across development from childhood to young adulthood. VSWMrelated age and gender effects were found across all participants and both memory loads in widespread regions, with higher activation with increasing age, and higher activation in males compared with females. Across all participants (ADHD subjects and controls alike), a widespread pattern of spatial span-related brain activation was found in all four lobes in cortical as well as subcortical regions (Fig. 3). Regions of spatial span-related activation are consistent with previous findings in a small sample of typically developing children, using a similar task (Klingberg et al., 2002), and largely coincide with activation patterns associated with executive control (Gilbert et al., 2006; Dosenbach et al., 2007; Matthews et al., 2014), an important component of working memory (Baddeley, 2003). Considering our large sample size, broad age range and even distribution of gender, we here provide a robust estimate of spatial span-related brain activation patterns in children, adolescents and young adults, with and without ADHD. Compared with previously obtained results with spatial N-back tasks, which elicit activation in frontal and parietal regions (Nelson et al., 2000; Kwon et al., 2002; Wager and Smith, 2003), our spatial span task yielded a more widespread network of activation that also extends to occipital, subcortical and cerebellar regions. This discrepancy indicates that, as previously suggested (Kane et al., 2007; Redick and Lindsey, 2013), spatial span and N-back tasks may indeed be controlled by (partly) different brain networks, and results from one working memory task should not automatically be generalized to the other. Given that at the behavioural level, VSWM impairments in ADHD have mainly been

16

FMRI OF VSWM IN ADHD

VAN EWIJK ET AL.

demonstrated using spatial span tasks, current results underline the importance of adopting similar paradigms in future fMRI research in ADHD, to be able to compare and integrate behavioural and fMRI findings more reliably. During memory load increase, individuals with ADHD showed higher brain activation compared with controls in regions of the left inferior frontal cortex, more specifically in regions of the left IFG, pars opercularis and triangularis, extending to the lateral frontal pole (Figs. 4 and 5). Abnormal brain activation in the left IFG in ADHD concurs with previous literature on working memory-related brain activation in ADHD (Cortese et al., 2012), though most of these studies used non-spatial tasks. Notably, brain activation associated with the spatial domain of working memory is typically reported to be more specific to the right hemisphere, while left hemispheric activity is more strongly associated with verbal working memory and more general phonological processes (e.g., Smith et al., 1996; d'Esposito et al., 1998). More specifically, the left IFG corresponds to Broca's area, strongly implicated in speech production as well as internal speech (Hinke et al., 1993; Bookheimer, 2002). The spatial span task used in the current study requires active rehearsal of cues and is correlated with complex reasoning skills (Klingberg, 2006), and as with most spatial span tasks, participants are likely to rehearse the items verbally. Consequently, it is possible that our finding of higher IFG activation in ADHD during memory load increase represents greater difficulty or effort within this group to internally verbalise the location of the cues, when the task becomes more complex. Alternatively, higher activation in the ADHD group in left inferior frontal regions could signify reduced inhibition of these regions by other functionally connected - brain regions (such as the right inferior frontal cortex, or more posterior regions). For example, it has been shown that functional connectivity between medial frontal regions and the posterior cingulate cortex plays an important in working memory, and that the strength of these connections is correlated with working memory performance (Hampson et al., 2006). Consequently, we could speculate that functional

17

FMRI OF VSWM IN ADHD

VAN EWIJK ET AL.

connectivity within or between working memory-related brain networks is weaker in ADHD, leading to decreased inhibition within these networks, ultimately resulting in relatively higher activation of specific regions in ADHD compared to controls. Future functional connectivity studies could provide us with more insight into these mechanisms. Follow-up analyses determining the nature of the group-by-memory load interaction (see Fig. 5) indicated underactivation in ADHD during low memory load in the IFG pars triangularis. It is possible that during low memory load, a relatively simple condition, ADHD subjects failed to recruit this region due to a lack of arousal or attention, which improved when the task became more complex. Conversely, in the IFG pars opercularis, individuals with ADHD showed higher activation compared with controls on the high memory load. However, higher activation did not transfer to better behavioural performance. This discrepancy points towards reduced efficiency of prefrontal brain regions in ADHD during VSWM, in line with previous findings (Bédard et al., 2014). For the lateral frontal pole cluster, no group differences were found on either the low or the high memory load. Group differences did not interact with age, suggesting parallel delayed developmental trajectories of VSWM-related brain activation for individuals with ADHD compared with their typically developing peers, between childhood and young adulthood. This finding is consistent with our previous report on behavioural VSWM performance in ADHD in a partially overlapping sample (van Ewijk et al., 2014), and implies that abnormalities in VSWM-related brain activation do not catch up, but also do not worsen when children with ADHD grow into adulthood (in contrast to one other study; Westerberg et al., 2004). The absence of a gender interaction indicates that abnormal brain activation is present in males as well as females with ADHD. This finding is consistent with the absence of a group-by-gender interaction at the behavioural level, but is in contrast to our hypothesis based on a previous report, which suggested that only males with ADHD show abnormal VSWM-related brain activation patterns (Valera et al., 2010). It is possible that our large

18

FMRI OF VSWM IN ADHD

VAN EWIJK ET AL.

sample size allowed us to detect group differences in females, which were not captured by previous, smaller, studies. Another explanation may lie in the age range investigated. While our sample consisted of mostly adolescents and young adults, the study by Valera and colleagues included adults with mean ages between 32 and 36 years. It is possible that observed brain abnormalities in females in our sample reflect a developmental delay in girls with ADHD, which may normalize in their late twenties or early thirties. Longitudinal studies, with large numbers of both male and female participants, could provide more insight. Across all participants, age-related increases in VSWM-related brain activation were found in regions that largely overlap with our general VSWM-related activation patterns (Figs. 3 and 6). This indicates that children recruit similar brain regions as young adults, but start to activate these regions more intensively throughout their development, consistent with a previous report using a spatial N-back task (Kwon et al., 2002). Additionally, on the low memory load, we found higher activation in the supplementary motor cortex and posterior cingulate gyrus for older compared with younger participants. Given the role of the posterior cingulate gyrus in memory functioning (Sutherland and Hoesing, 1993; Kim, 2011), this finding could signify a developmental effect, either representing an increase in recruitment of memory-related brain regions over age, or a decrease in inhibition of these regions by functionally connected regions within the same working memory-related brain network. Gender effects were mainly found in frontal brain regions, with higher activation for males compared with females (Fig. 6), consistent with previous research using N-back or spatial attention tasks (Gur et al., 2000; Goldstein et al., 2005). In the lateral frontal pole and paracingulate and superior frontal gyri, the gender effect was only present at the low, but not the high memory load. Importantly, behavioural performance was similar between male and female subjects at both memory loads, indicating a discrepancy between brain activation and behavioural performance. These results may indicate that males recruit more

19

FMRI OF VSWM IN ADHD

VAN EWIJK ET AL.

brain resources than females to reach similar levels of VSWM performance. In sum, these findings underline the importance of taking age and gender into account when studying VSWM-related brain activation, especially when dealing with psychiatric disorders that are subject to uneven gender distributions. The current results should be considered in light of some limitations. First, it should be noted that the developmental trajectory of VSWM should ideally be investigated in longitudinal samples, following children with ADHD into young adulthood with assessment at multiple time points. However, our large sample size with a broad age range did allow us to provide exploratory insights into the developmental trajectory of VSWM. Second, the majority of our ADHD participants had a history of medication treatment. While we ruled out acute pharmacological effects by using a 48-h washout period before scanning, possible long-term effects cannot be ruled out. Scarce literature on long-term effects suggests that stimulants may have an ameliorating effect on deviant levels of brain activation levels in ADHD during attention and interference, but not inhibition tasks (Rubia et al., 2014). It is unknown whether similar normalizing effects may exist on working memory-related brain activation; hence it is possible that a medication-naive sample would show more extensive or more pronounced group differences. Third, generalization of results is restricted to clinically referred samples of European-Caucasian descent. In conclusion, we here show spatial span-related brain activation patterns in more widespread regions than have been reported with N-back tasks, indicating that these tasks may not measure exactly the same construct and should not be used interchangeably. Individuals with ADHD showed underactivation during low memory load in the IFG pars triangularis, and greater activation compared with controls during high memory load in the IFG pars opercularis. Abnormal brain functioning in ADHD was present in males as well as females, and developmental trajectories were similar for ADHD subjects and controls. Furthermore, we report effects of age and gender on the location and intensity of these

20

FMRI OF VSWM IN ADHD

VAN EWIJK ET AL.

activation patterns across all participants, underlining the importance of taking these factors into account in future studies.

21

FMRI OF VSWM IN ADHD

VAN EWIJK ET AL.

Acknowledgements The authors thank all families that participated in this study and all researchers who collected the data. Furthermore, we acknowledge the Department of Pediatrics of the VU University Medical Center for giving us the opportunity to use the mock scanner for preparation of our participants.

Declaration of interest Dr. van Ewijk, Dr.Weeda, Dr. Heslenfeld, Dr. Luman, Dr. Hartman, and Dr. Franke report no conflicts of interest. Dr. Hoekstra is a paid advisor of Shire. In the past year, Dr. Faraone received income, travel expenses and/or research support from and/or has been on an Advisory Board for Pfizer, Ironshore, Shire, Akili Interactive Labs, CogCubed, Alcobra, VAYA Pharma, Neurovance, Impax, NeuroLifeSciences and research support from the National Institutes of Health (NIH). His institution is seeking a patent for the use of sodium-hydrogen exchange inhibitors in the treatment of ADHD. In previous years, he received consulting fees or was on Advisory Boards or participated in continuing medical education programs sponsored by: Shire, Alcobra, Otsuka, McNeil, Janssen, Novartis, Pfizer and Eli Lilly. Dr. Faraone receives royalties from books published by Guilford Press: Straight Talk about Your Child’s Mental Health and Oxford University Press: Schizophrenia: The Facts. Dr. Buitelaar reports grants from Netherlands Organization for Scientific Research NWO, grants from Radboudumc; VU University Amsterdam; UMC Groningen/Accare, grants from EU FP7 programmes TACTICS, MATRICS, and AGGRESSOTYPE, during the conduct of the study; other from Janssen Cilag BV, other from Lilly, other from Servier, outside the submitted work. Dr. Oosterlaan has received an unrestricted investigator initiated grant from Shire.

Funding

22

FMRI OF VSWM IN ADHD

VAN EWIJK ET AL.

This work was supported by NIH Grant R01MH62873 (to Stephen V. Faraone), NWO Large Investment Grant 1750102007010 (to Jan Buitelaar), and grants from Radboud University Nijmegen Medical Center, University Medical Center Groningen and Accare, and VU University Amsterdam.

Author contributions Dr. van Ewijk was responsible for the study concept and design, contributed to the acquisition of data, performed the analyses, and drafted the manuscript. Dr. Weeda assisted in the study design and data analysis, contributed to the interpretation of findings, and critically reviewed the manuscript. Dr. Heslenfeld, Dr. Luman, Dr. Hartman, Dr. Hoekstra, Dr. Faraone, Dr. Franke, Dr. Buitelaar and Dr. Oosterlaan contributed to the interpretation of findings and provided critical reviews of the manuscript. All authors approved the final version for publication.

23

FMRI OF VSWM IN ADHD

VAN EWIJK ET AL.

References American Psychiatric Association, 2000. Diagnostic and Statistical Manual of Mental Disorders: DSMIV-TR. American Psychiatric Publishing, Inc., Arlington, VA. American Psychiatric Association, 2013. The Diagnostic and Statistical Manual of Mental Disorders: DSM 5. bookpointUS. Baddeley, A., 2003. Working memory: looking back and looking forward. Nature Reviews. Neuroscience 4, 829-839. Balint, S., Czobor, P., Komlosi, S., Meszaros, A., Simon, V., Bitter, I., 2009. Attention deficit hyperactivity disorder (ADHD): gender- and age-related differences in neurocognition. Psychological Medicine 39, 1337-1345. Bédard, A.C.V., Newcorn, J.H., Clerkin, S.M., Krone, B., Fan, J., Halperin, J.M., Schulz, K.P., 2014. Reduced prefrontal efficiency for visuospatial working memory in attention-deficit/hyperactivity disorder. Journal of the American Academy of Child and Adolescent Psychiatry 53 (9), 1020-1030. Bell, E.C., Willson, M.C., Wilman, A.H., Dave, S., Silverstone, P.H., 2006. Males and females differ in brain activation during cognitive tasks. Neuroimage 30, 529-538. Bookheimer, S., 2002. Functional MRI of language: new approaches to understanding the cortical organization of semantic processing. Annual Review of Neuroscience 25, 151-188. Conners, C.K., Erhardt, D., Sparrow, E.P., 1999. Conner's Adult ADHD Rating Scales: CAARS. MultiHealth Systems, North Tonawanda, NY. Conners, C.K., Sitarenios, G., Parker, J.D.A., Epstein, J.N., 1998a. The revised Conners' Parent Rating Scale (CPRS-R): factor structure, reliability, and criterion validity. Journal of Abnormal Child Psychology 26, 257-268. Conners, C.K., Sitarenios, G., Parker, J.D.A., Epstein, J.N., 1998b. Revision and restandardization of the Conners Teacher Rating Scale (CTRS-R): factor structure, reliability, and criterion validity. Journal of Abnormal Child Psychology 26, 279-291. Cortese, S., Kelly, C., Chabernaud, C., Proal, E., Di Martino, A., Milham, M.P., Castellanos, F.X., 2012. Toward systems neuroscience of ADHD: a meta-analysis of 55 fMRI studies. American Journal of Psychiatry 169, 1038-1055. Cubillo, A., Smith, A.B., Barrett, N., Giampietro, V., Brammer, M., Simmons, A., Rubia, K., 2013. Drugspecific laterality effects on frontal lobe activation of atomoxetine and methylphenidate in attention deficit hyperactivity disorder boys during working memory. Psychological Medicine 44, 633-646. d'Esposito, M., Aguirre, G., Zarahn, E., Ballard, D., Shin, R., Lease, J., 1998. Functional MRI studies of spatial and nonspatial working memory. Cognitive Brain Research 7, 1-13. Dosenbach, N.U., Fair, D.A., Miezin, F.M., Cohen, A.L., Wenger, K.K., Dosenbach, R.A., Fox, M.D., Snyder, A.Z., Vincent, J.L., Raichle, M.E., 2007. Distinct brain networks for adaptive and stable task control in humans. Proceedings of the National Academy of Sciences of the United States of America 104, 11073-11078. Gaub, M., Carlson, C.L., 1997. Gender differences in ADHD: a meta-analysis and critical review. Journal of the American Academy of Child and Adolescent Psychiatry 36, 1036-1045. Gershon, J., Gershon, J., 2002. A meta-analytic review of gender differences in ADHD. Journal of Attention Disorders 5, 143-154. Gilbert, S., Spengler, S., Simons, J., Steele, J., Lawrie, S., Frith, C., Burgess, P., 2006. Functional specialization within rostral prefrontal cortex (area 10): a meta-analysis. Journal of Cognitive Neuroscience 18, 932-948. Goldstein, J.M., Jerram, M., Poldrack, R., Anagnoson, R., Breiter, H.C., Makris, N., Goodman, J.M., Tsuang, M.T., Seidman, L.J., 2005. Sex differences in prefrontal cortical brain activity during fMRI of auditory verbal working memory. Neuropsychology 19, 509. Gong, G., He, Y., Evans, A.C., 2011. Brain connectivity gender makes a difference. The Neuroscientist 17, 575-591. Gur, R.C., Alsop, D., Glahn, D., Petty, R., Swanson, C.L., Maldjian, J.A., Turetsky, B.I., Detre, J.A., Gee, J., Gur, R.E., 2000. An fMRI study of sex differences in regional activation to a verbal and a spatial task. Brain and Language 74, 157-170. Hampson, M., Driesen, N.R., Skudlarski, P., Gore, J.C., Constable, R.T., 2006. Brain connectivity related to working memory performance. J. Neurosci. 26, 13338-13343.

24

FMRI OF VSWM IN ADHD

VAN EWIJK ET AL.

Hinke, R.M., Hu, X., Stillman, A.E., Kim, S.-G., Merkle, H., Salmi, R., Ugurbil, K., 1993. Functional magnetic resonance imaging of Broca's area during internal speech. Neuroreport 4, 675-678. Hsu, J.-L., Leemans, A., Bai, C.-H., Lee, C.-H., Tsai, Y.-F., Chiu, H.-C., Chen, W.-H., 2008. Gender differences and age-related white matter changes of the human brain: a diffusion tensor imaging study. Neuroimage 39, 566-577. Kane, M.J., Conway, A.R., Miura, T.K., Colflesh, G.J., 2007. Working memory, attention control, and the n-back task: a question of construct validity. Journal of Experimental Psychology: Learning, Memory, and Cognition 33, 615. Kaufman, J., Birmaher, B., Brent, D., Rao, U., Flynn, C., Moreci, P., Williamson, D., Ryan, N., 1997. Schedule for Affective Disorders and Schizophrenia for School-Age Children-Present and Lifetime Version (K-SADS-PL): initial reliability and validity data. Journal of the American Academy of Child and Adolescent Psychiatry 36, 980-988. Kim, H., 2011. Neural activity that predicts subsequent memory and forgetting: a meta-analysis of 74 fMRI studies. Neuroimage 54, 2446-2461. Klingberg, T., 2006. Development of a superior frontal-intraparietal network for visuo-spatial working memory. Neuropsychologia 44, 2171-2177. Klingberg, T., Forssberg, H., Westerberg, H., 2002. Increased brain activity in frontal and parietal cortex underlies the development of visuospatial working memory capacity during childhood. Journal of Cognitive Neuroscience 14, 1-10. Ko, C.H., Yen, J.Y., Yen, C.F., Chen, C.S., Lin, W.C., Wang, P.W., Liu, G.C., 2013. Brain activation deficit in increased-load working memory tasks among adults with ADHD using fMRI. European Archives of Psychiatry and Clinical Neuroscience 263, 561-573. Kooij, J.J., Buitelaar, J.K., van den Oord, E.J., Furer, J.W., Rijnders, C.A., Hodiamont, P.P., 2005. Internal and external validity of attention-deficit hyperactivity disorder in a population-based sample of adults. Psychological Medicine 35, 817-827. Kwon, H., Reiss, A.L., Menon, V., 2002. Neural basis of protracted developmental changes in visuospatial working memory. Proceedings of the National Academy of Sciences of the United States of America 99, 13336-13341. Martinussen, R., Hayden, J., Hogg-Johnson, S., Tannock, R., 2005. A meta-analysis of working memory impairments in children with attention-deficit/hyperactivity disorder. Journal of the American Academy of Child and Adolescent Psychiatry 44, 377-384. Matthews, M., Nigg, J.T., Fair, D.A., 2014. Attention deficit hyperactivity disorder. Current Topics in Behavioral Neurosciences 16, 235-266. McNab, F., Leroux, G., Strand, F., Thorell, L., Bergman, S., Klingberg, T., 2008. Common and unique components of inhibition and working memory: an fMRI, within-subjects investigation. Neuropsychologia 46, 2668-2682. Nelson, C.A., Monk, C.S., Lin, J., Carver, L.J., Thomas, K.M., Truwit, C.L., 2000. Functional neuroanatomy of spatial working memory in children. Developmental Psychology 36, 109. Polanczyk, G., Caspi, A., Houts, R., Kollins, S.H., Rohde, L.A., Moffitt, T.E., 2010. Implications of extending the ADHD age-of-onset criterion to age 12: results from a prospectively studied birth cohort. Journal of the American Academy of Child and Adolescent Psychiatry 49, 210-216. Redick, T.S., Lindsey, D.R., 2013. Complex span and n-back measures of working memory: A metaanalysis. Psychonomic Bulletin and Review 20, 1102-1113. Rubia, K., 2007. Neuro-anatomic evidence for the maturational delay hypothesis of ADHD. Proceedings of the National Academy of Sciences of the United States of America 104, 19663-19664. Rubia, K., Alegria, A.A., Cubillo, A.I., Smith, A.B., Brammer, M.J., Radua, J., 2014. Effects of stimulants on brain function in attention-deficit/hyperactivity disorder: a systematic review and meta-analysis. Biological Psychiatry 76, 616-628. Shaw, P., Eckstrand, K., Sharp, W., Blumenthal, J., Lerch, J., Greenstein, D., Clasen, L., Evans, A., Giedd, J., Rapoport, J., 2007. Attention-deficit/hyperactivity disorder is characterized by a delay in cortical maturation. Proceedings of the National Academy of Sciences of the United States of America 104, 19649-19654. Silk, T.J., Vance, A., Rinehart, N., Bradshaw, J.L., Cunnington, R., 2009. Structural development of the basal ganglia in attention deficit hyperactivity disorder: a diffusion tensor imaging study. Psychiatry Research: Neuroimaging 172, 220-225. Smith, E.E., Jonides, J., Koeppe, R.A., 1996. Dissociating verbal and spatial working memory using PET. Cerebral Cortex 6, 11-20.

25

FMRI OF VSWM IN ADHD

VAN EWIJK ET AL.

Speck, O., Ernst, T., Braun, J., Koch, C., Miller, E., Chang, L., 2000. Gender differences in the functional organization of the brain for working memory. Neuroreport 11, 2581-2585. Sutherland, R.J., Hoesing, J.M., 1993. Posterior Cingulate Cortex and Spatial Memory: A Microlimnology Analysis, Neurobiology of Cingulate Cortex and Limbic Thalamus. Birkhäuser Boston, Boston, MA, pp. 461-477. Thomas, K.M., King, S.W., Franzen, P.L., Welsh, T.F., Berkowitz, A.L., Noll, D.C., Birmaher, V., Casey, B.J., 1999. A developmental functional MRI study of spatial working memory. Neuroimage 10, 327338. Valera, E.M., Brown, A., Biederman, J., Faraone, S.V., Makris, N., Monuteaux, M.C., Whitfield-Gabrieli, S., Vitulano, M., Schiller, M., Seidman, L.J., 2010. Sex differences in the functional neuroanatomy of working memory in adults with ADHD. American Journal of Psychiatry 167, 86-94. van Ewijk, H., Heslenfeld, D.J., Luman, M., Rommelse, N.N.J., Hartman, C.A., Hoekstra, P.J., Franke, B., Buitelaar, J.K., Oosterlaan, J., 2014. Visuospatial working memory in ADHD patients, unaffected siblings, and healthy controls. Journal of Attention Disorders 18, 369-378. von Rhein, D., Mennes, M., van Ewijk, H., Groenman, A.P., Zwiers, M.P., Oosterlaan, J., Heslenfeld, D., Franke, B., Hoekstra, P.J., Faraone, S.V., 2014. The NeuroIMAGE study: a prospective phenotypic, cognitive, genetic and MRI study in children with attention-deficit/hyperactivity disorder. Design and descriptives. European Child and Adolescent Psychiatry DOI: 10.1007/s00787-014-0573-4. Wager, T.D., Smith, E.E., 2003. Neuroimaging studies of working memory. Cognitive, Affective, & Behavioral Neuroscience 3, 255-274. Westerberg, H., Hirvikoski, T., Forssberg, H., Klingberg, T., 2004. Visuo-spatial working memory span: a sensitive measure of cognitive deficits in children with ADHD. Child Neuropsychology 10, 155-161. Willcutt, E.G., Doyle, A.E., Nigg, J.T., Faraone, S.V., Pennington, B.F., 2005. Validity of the executive function theory of attention-deficit/hyperactivity disorder: a meta-analytic review. Biological Psychiatry 57, 1336-1346. Woolrich, M., 2008. Robust group analysis using outlier inference. Neuroimage 41, 286-301. Worsley, K., 2001. Statistical analysis of activation images. Functional MRI: an introduction to methods 14, 251-270.

26

FMRI OF VSWM IN ADHD

VAN EWIJK ET AL.

Figure captions Note: All images are to be reproduced in colour on the Web (free of charge) and in blackand-white in print. Fig. 1. Schematic overview of a working memory trial with low memory load on the visuospatial working memory (VSWM) task. Adapted from (van Ewijk et al., 2014). Each trial consisted of a sequence of either three or six circles, displayed on a 4 × 4 grid for 500 ms each, with a 500-ms interstimulus interval in between (a). Subsequently, during a 2,000-ms response window, the probe was presented and the participant was required to press a response button (b), after which feedback was presented for the remainder of the response window (c). Each trial was followed by a 3.000-ms intertrial interval, consisting of a black screen (d). Fig. 2. Behavioural performance on the visuospatial working memory (VSWM) task. High memory load was associated with worse performance compared to low memory load for all subjects (p<0.001), and individuals with ADHD showed lower accuracy compared to controls across both memory loads (p=0.032). No significant memory load by group interaction was found (p>0.05). Error bars represent the standard error (SE) of the mean accuracy. Fig 3. Mean activation patterns across all participants for the mean working memory contrast (average of both loads, corrected for baseline trials) and the load difference contrast (high minus low memory load, corrected for baseline trials). Fig 4. Higher activation for ADHD compared with controls on the load difference contrast (high minus low memory load, corrected for baseline trials). The significant cluster was located in the left inferior frontal gyrus (pars opercularis and pars triangularis; Brodmann areas 44/45), and extended to the lateral frontal pole (Brodmann area 10). Fig 5. Regions of interest (ROIs) used for follow-up analyses. ROIs were derived from the significant cluster from the group analysis in which ADHD showed higher activation compared with controls on the load difference contrast (see Fig. 4). This cluster was thresholded at a more conservative threshold (Z>2.5) than in the whole-brain analysis to obtain separate clusters with unique anatomical labels. Resulting clusters were located in the inferior frontal gyrus, pars triangularis (a) and pars opercularis (b), and the lateral frontal pole (c). The right panel shows extracted activation parameters in each cluster on both memory loads, showing a significant difference (marked with an asterisk) in the IFG pars opercularis on the high memory load (ADHD > controls), and in the IFG pars triangularis on the low memory load (ADHD < controls). Boxes represent the median value (horizontal line) and the first and third quartile (bottom and top of the box, respectively) of the data, with vertical lines extending to 1.5 interquartile range (IQR) of the lower and upper quartile. Fig 6. Effects of age and gender on spatial span-related brain activation, across all participants. Upper panel: age was positively associated with activation on the mean working memory contrast (green) and negatively with activation on the load difference contrast (yellow). Post-hoc analyses of the load difference contrast revealed that this was due to significantly higher activation for older subjects on the low load (marked with an asterisk), but not the high load (graph shown on the right). Lower panel: Gender effects

27

FMRI OF VSWM IN ADHD

VAN EWIJK ET AL.

represented higher activation for males compared with females on the mean working memory contrast (red), and higher activation for females compared with males associated with the load difference contrast (orange). Post-hoc analyses of the load difference contrast showed that this was due to higher activation for males compared with females on the low load (marked with an asterisk), but not the high load (graph shown on the right). Boxes represent the median value (horizontal line) and the first and third quartile (bottom and top of the box, respectively) of the data, with vertical lines extending to 1.5 interquartile range (IQR) of the lower and upper quartile.

28

FMRI OF VSWM IN ADHD

VAN EWIJK ET AL.

Table 1. Sample characteristics.

Age (M, SD) Gender (% male) IQ (M, SD)

Control (n=103) 16.8 (2.9) 53.4 104.0 (13.5) 73.8

ADHD (n=109) 17.1 (3.3) 59.6 99.1 (15.6) 46.8

Test statistic

p-value

F1,210 = 0.392 χ21,N=212 = 0.838 F1,210 = 6.096

0.532 0.406 0.014

χ21,N=212 = <0.001 16.070 0.7 (1.3) 13.1 (3.0) F1,210 = <0.001 1512.704 86.1 88.0 χ22,N=212 = 0.832 0.660 0.0 82.9 χ22,N=196 = <0.001 135.582 13.2 (2.6) 11.7 (2.2) F1,203 = 21.505 <0.001 2 0 24.8 χ 1,N=212 = <0.001 29.237 Internalizing disorder (%) 1.0 7.7 χ21,N=212 = 5.622 0.035 PDD-NOS (%) 0 15.6 χ21,N=212=22.026 <0.001 a Total number of symptoms, derived from the diagnostic algorithm (see Methods) Scan site (% scanned in Amsterdam) Number of ADHD symptomsa (M, SD) Hand preference (% right-handed) ADHD medication (% with a history of medication use) Socio-economic status (M, SD) ODD/CD (%)

Table 2. Multilevel model for working memory performance (behavioural performance). Coefficient SE Test statistics p-value Fixed effects 1 6. Intercept 2 -14.15 0 t(394)=-0.87 0.383 1. Group (ADHD versus controls) 5 -3.44 8 t(148)=-2.17 0.032 0. Baseline accuracy 1 0.75 6 t(385)=4.64 <0.001 0. Memory load (high versus low) 8 -4.91 2 t(219)=-6.00 <0.001 0. Age 2 1.15 3 t(207)=5.06 <0.001 1. Gender (male versus female) 5 2.19 0 t(207)=1.45 0.148 1. Scan site (Nijmegen versus 6 Amsterdam) -0.76 8 t(162)=-0.45 0.653 Random effects 1 Wald Z = 2.14 0.033 3. Family 29.00 5 7

29

FMRI OF VSWM IN ADHD

VAN EWIJK ET AL.

Table 3. Significant clusters of higher activation for ADHD versus controls on the load difference contrast. Cluster Anatomical label HemisN voxels MNI coordinates # phere x y z 1 Inferior frontal gyrus, pars Left 413 -36 36 6 triangularis 2 Lateral frontal pole Left 159 -34 56 10 3 Inferior frontal gyrus, pars Left 125 -52 20 4 opercularis Note: Results were evaluated at a more conservative threshold (Z>2.5) than in the wholebrain analysis to obtain meaningful clusters. MNI coordinates represent the location of the maximum intensity voxel (maximum z-statistic). Highlights * Neural activation patterns underlying visuospatial working memory (VSWM) deficits in ADHD were investigated using fMRI in a large sample * VSWM-related activation as well as age- and gender effects were found in widespread regions * ADHD showed higher activation than controls in left inferior frontal regions associated with memory load increase, indicating reduced efficiency in recruitment of these regions * Abnormal activation in ADHD was present in males as well as females * Abnormal activation in ADHD was present across the whole age range investigated

30

FMRI OF VSWM IN ADHD

VAN EWIJK ET AL.

31