Reduced symptoms of inattention after dietary omega-3 fatty acid supplementation in boys with and without Attention Deficit/Hyperactivity Disorder

Dienke J. Bos MSc1, Bob Oranje PhD1, E. Sanne Veerhoek MSc1, Rosanne M. Van Diepen MSc1, Juliette M.H. Weusten MSc1, Hans Demmelmair PhD2, Berthold Koletzko MD PhD2, Monique G. M. de Sain-van der Velden PhD3, Ans Eilander PhD4, Marco Hoeksma PhD4, Sarah Durston PhD1

Affiliations

1NICHE Lab, Department of Psychiatry, Brain Center Rudolf Magnus, University Medical Center Utrecht, The Netherlands

2Division of Metabolic and Nutritional Medicine, University of Munich Medical Center, Dr. von Hauner Children's Hospital, Munich, Germany

3Department of Medical Genetics, University Medical Center Utrecht, The Netherlands Wilhelmina Children's Hospital, University Medical Center Utrecht, The Netherlands

4Unilever Research & Development, Vlaardingen, The Netherlands

Corresponding author

Dienke J. Bos, NICHE Lab, Department of Psychiatry, Brain Center Rudolf Magnus, University Medical Center Utrecht, HP A.01.126, Heidelberglaan 100, 3584 CX Utrecht, The Netherlands, T: +31 88 755 9840, E-mail:

Supplemental Materials 2: fMRI design, analysis and results

fMRI acquisition

All participants participated in a mock MRI-scanner practice session in our lab before participating in the fMRI-scan. The purpose of this practice session was to acclimate the children to the scanner environment, to reduce potential anxiety and to increase compliance by reducing the risk of motion artifacts (Durston et al, 2009). Children only participated in the actual MRI-scan after a successful practice session.

Data were acquired using a 3.0T Phillips Achieva MR scanner (Philips Medical Systems, Best, The Netherlands). Functional images were collected in five blocks of 119 dynamics with a 2D-EPI SENSE sequence (TR/TE=2000/35, flip angle = 70°, matrix 68x66, voxel size 3x3x3.5), with a total duration of 20 minutes. A high-resolution T1-weighted image was acquired to allow for spatial normalization and visualization (TR/TE=10/4.6, flip angle = 8°, matrix 304x299, voxel size 0.75x0.75x0.8).

During the fMRI-scan children performed a Go/No-Go task as described previously (Durston et al, 2003, 2006a, 2006b). In short, children were instructed to press a button as fast as possible in response to visually presented stimuli, and withhold their response following presentation of a rare non-target. The task consisted of five runs, each lasting 4 minutes, during which 57 trials were presented, containing 75% Go-trials. In order to make the task visually more attractive to young children, characters from the Pokemon series were used as stimuli. Stimuli were presented for 500ms and the interstimulus interval was 3500ms, adding up to a total trial length of 4000ms. One participant (RGActive) showed a chance level performance on the go-stimuli, and was excluded from the analysis.

Image Preprocessing

Standard preprocessing was performed using SPM8 (Wellcome Dept. of Cognitive Neurology, First, data were realigned, followed by unwarping to remove residual variance related to movement. Only movement not exceeding 3mm or the size of 1 voxel was accepted. Five participants in the ADHD group (three ADHDActive, two ADHDPlacebo), and two participants in the typically developing group (one RGActive, one RGPlacebo) had to be excluded from the fMRI analysis due to excessive movement. For eight participants in the ADHD group (four ADHDActive, four ADHDPlacebo), and four participants in the typically developing group (two RGActive, two RGPlacebo) only partial data could be included in the fMRI analyses (no more than two scans were excluded per subject). Next, the functional and anatomical images were co-registered and normalized into standard space (Montreal Neurological Institute (MNI) – template). Finally the functional images were smoothed with a 6mm3 FWHM Gaussian kernel.

For each subject, images were checked for any remaining artifacts caused by movement using the ArtRepair Toolbox (Mazaika et al. 2009). Slices marked as outliers in average global intensity (>1.5% of mean) were repaired by slice interpolation. No more than 20% of frames were repaired for any subject.

fMRI Task Analyses

At first level, four event types were defined: initial fixation, correct and incorrect NoGo-trials, and a parametric factor representing the number of Go-trials preceding a NoGo-event. These events included one event of interest (NoGo-trials) and four events of no interest (errors, fixation, Go-trials and the parametric factor). The event-types were time-locked to stimuli by a canonical synthetic hemodynamic response function (HRF) and its first-order temporal derivative (tHRF). Grand mean scaling and a high-pass filter were applied (cut-off = 128 sec) to account for differences in the global mean signal and low-frequency drifts.

For the group analysis, a random effects model was used in SPM8 to compute a voxel-wise T-statistic for the contrast NoGo trials > Go-trials. Differences in activation were tested at a threshold of p < .001, k > 10 voxels. ROIs included the regions activated by both the children with ADHD and the typically developing children in the NoGo > Go condition at a threshold of p < .001, k > 10 voxels. A Linear Mixed Effects analysis to account for missing data (i.e. MRI-scans lost to excessive motion at either baseline or follow-up) was also performed with 3dLME - recently implemented in the Analysis of Functional NeuroImages (AFNI) Toolbox - which uses functions of the R-based lme-package (

Covariates in the designs included age, and the combinations age and task-performance, as well as age and ADHD symptom severity as measured by the Child Behavior Checklist (CBCL).

Task Performance Analyses

Behavioral data from the cognitive control task included; mean reaction time (RT) on correct responses to Go-trials, mean standard deviation (SD) for correct responses on Go-trials, accuracy on Go-trials and accuracy on NoGo-trials. Standardized residuals of a linear regression with NoGo-accuracy corrected for RT were used to reflect the speed-accuracy trade-off. Furthermore, the intra-individual coefficient of variation (ICV) was calculated (mean SD Go-trials/mean RT Go-trials). Developmental changes in performance were investigated using Pearson’s correlations (r) with age.

Independent samples T-tests or Mann-Whitney U-tests were used to analyze group differences at baseline between the diagnostic groups, and between the intervention groups. To investigate treatment effects, all variables of interest were fed into an ANCOVA (analysis of covariance) model with the baseline measurement entered as a covariate (Vickers, 2005).

Results

Task Performance

All subjects (N = 56) in the fMRI-analysis performed the task adequately. Mean accuracy on Go-trials was 98% for typically developing children and 95% for children with ADHD, both at baseline and follow-up, indicating that subjects were performing the task correctly (see SupplementalMaterials 3).

Mean reaction time (RT) on Go-trials decreased with age (r= -.551, p= .000), while accuracy on NoGo-trials increased with age (r= .245, p= .047). The standard deviation of RT on Go-trials showed a decrease with age (r= -.369, p= .002) and correlated significantly with performance on NoGo-trials (r= -.453, p= .000). Therefore, mean and SD of RT were combined to calculate the intra-individual coefficient of variation (ICV = SD RTgo/mean Rtgo). ICV correlated negatively with age (r = -.285, p = .020). As some children may adopt a strategy in which accuracy is favored over speed, standardized residuals of NoGo-performance corrected for RT were calculated as a measure reflecting the speed-accuracy trade-off. Age did not correlate with the speed-accuracy trade-off (r= .134, p= .284), suggesting that there was no age-related shift in strategy that could influence the results.

Children with ADHD and typically developing children differed in baseline performance on Go-trials (F(1,66)= 7.00, p = .010). In addition, mean RT on Go-trials was slower for children with ADHD (F(1,66)= 14.47, p= .000), and mean SD (F(1,66)=9.06, p= .004) and ICV (F(1,66)=6.71, p= .012) were higher in children with ADHD, reflecting higher variability in responses, again with age as covariate. There was neither a difference between the diagnostic groups in NoGo-performance nor in speed-accuracy trade-off at baseline (age as covariate).No effects of treatment with omega-3 PUFAs were found on any of the measures of task performance.

fMRI results

Typically developing children showed increased activation (p< .001, k= 10 voxels) in the bilateral anterior ACC, bilateral middle frontal gyrus, bilateral inferior parietal lobule and right precuneus and middle temporal gyrus during NoGo-trials (Figure S1).Children with ADHD showed less extensive increases in activation during NoGo-trials in the right frontal pole, the right paracingulate gyrus, right insula and orbitofrontal cortex, bilateral PCC and left middle temporal gyrus compared to the controls.

An ROI-analysis of regions activated in both control- and ADHD subjects during NoGo-trials (see Figure S1) indeed showed significantly greater activation for controls than for children with ADHD in the bilateral inferior parietal lobule, and the right middle frontal gyrus (Figure S1). Both a traditional GLM performed in SPM8, and a Linear Mixed Effects analysis performed in AFNI yielded no significant results for treatment effects, nor when age, task-performance, or CBCL-scores were added as covariates.

Motion

Average head motion in the group carried forward for analysis was 1.05 mm and even though movement was slightly higher in the children with ADHD, a repeated measures analysisof the motion parameters of the four treatment conditionsshowed no significant group by motioninteractions (F(6,102)= 1.33 ,p = .253).

Figure S1.Group differences in brain activation during cognitive control at the baseline measurement. Displayed in yellow is the t-map of the NoGo > Gocondition in the cognitive control fMRI-task in typically developing controls at a threshold of t 3.38, p< 0.001. In blue, regions are shown with group differences in activation during Nogo-trials,where children with ADHD show less activation than controls.

References

Durston S, Davidson MC, Tottenham N, Galvan A, Spicer J, Fossella J a, et al (2006a). A shift from diffuse to focal cortical activity with development. Dev Sci9: 1–8.

Durston S, Mulder M, Casey BJ, Ziermans T, Engeland H van (2006b). Activation in ventral prefrontal cortex is sensitive to genetic vulnerability for attention-deficit hyperactivity disorder. Biol Psychiatry60: 1062–70.

Durston S, Nederveen H, Dijk S van, Belle J van, Zeeuw P de, Langen M, et al (2009). Magnetic resonance simulation is effective in reducing anxiety related to magnetic resonance scanning in children. J Am Acad Child Adolesc Psychiatry48: 206–7.

Durston S, Tottenham NT, Thomas KM, Davidson MC, Eigsti I-M, Yang Y, et al (2003). Differential patterns of striatal activation in young children with and without ADHD. Biol Psychiatry53: 871–878.

Mazaika P., Hoeft, F. Glover, G.H., Allan L. Reiss (2009): Methods and Software for fMRI Analysis for Clinical Subjects, Human Brain Mapping Conference 2009

Vickers AJ (2005). Parametric versus non-parametric statistics in the analysis of randomized trials with non-normally distributed data. BMC Med Res Methodol5: 35.