Introduction
Attention deficit-hyperactivity disorder (ADHD) goes hand in hand with multiple dysfunctions in neuronal networks responsible for attention and response control (Pennington & Ozonoff, Reference Pennington and Ozonoff1996; Castellanos & Tannock, Reference Castellanos and Tannock2002). Heritability estimates exceeding 70% indicate a strong genetic impact on the expression of the disorder, despite weak associations between ADHD and each individual risk allele, and a limited understanding of developmental pathways from genetic and environmental factors (Castellanos & Tannock, Reference Castellanos and Tannock2002; Banaschewski et al. Reference Banaschewski, Hollis, Oosterlaan, Roeyers, Rubia, Willcutt and Taylor2005; Faraone et al. Reference Faraone, Perlis, Doyle, Smoller, Goralnick, Holmgren and Sklar2005). Endophenotypes, which are quantitative and heritable vulnerability traits characterized by more fundamental biological properties in between, on the one hand, genetic and environmental risk factors and on the other hand, the phenotype, may help to clarify these associations. It is expected that endophenotypes show larger and more specific genetic effects than diagnostic phenotypes (Doyle et al. Reference Doyle, Willcutt, Seidman, Biederman, Chouinard, Silva and Faraone2005) and may serve as useful intermediate constructs that explain the heterogeneity of the ADHD phenotype (Buitelaar, Reference Buitelaar2005; Banaschewski et al. Reference Banaschewski, Neale, Rothenberger and Roessner2007; Rommelse et al. Reference Rommelse, Altink, Oosterlaan, Buschgens, Buitelaar, De Sonneville and Sergeant2007a). This particularly holds for endophenotypes based on neural activity (Meyer-Lindenberg & Weinberger, Reference Meyer-Lindenberg and Weinberger2006).
In a recent study, we showed that action monitoring as reflected by several performance and electrophysiological parameters was impaired in boys with ADHD and intermediate in their non-affected siblings compared with controls without a family history of ADHD (Albrecht et al. Reference Albrecht, Brandeis, Uebel, Heinrich, Mueller, Hasselhorn, Steinhausen, Rothenberger and Banaschewski2008, Reference Albrecht, Brandeis, Uebel, Heinrich, Heise, Hasselhorn, Rothenberger and Banaschewski2009). Thus, deficits in these parameters may be a consequence of specific genes or environmental factors that are shared in the families. Similar familial deficits were also reported for performance in a Stroop task and an oral arithmetic test (Doyle et al. Reference Doyle, Willcutt, Seidman, Biederman, Chouinard, Silva and Faraone2005), with several parameters of motor inhibition (Slaats-Willemse et al. Reference Slaats-Willemse, Swaab-Barneveld, de Sonneville, van der Meulen and Buitelaar2003) and also for quantitative electroencephalography (EEG) parameters (Loo et al. Reference Loo, Hale, Hanada, Macion, Shrestha, McGough, McCracken, Nelson and Smalley2010).
While action-monitoring tasks require frequent responses to tap cognitive control mechanisms continuously, the assessment of preparation and response control requires a different approach that separates these processes in time to capture their temporal progression. In this study, the cued Continuous Performance Test (CPT) combining vigilance and cued Go-NoGo tasks (Rosvold et al. Reference Rosvold, Bransome, Mirsky, Rosvold and Sarason1956) was used to characterize these aspects of executive functions in ADHD (van Leeuwen et al. Reference van Leeuwen, Steinhausen, Overtoom, Pascual-Marqui, van't Klooster, Rothenberger, Sergeant and Brandeis1998; Banaschewski et al. Reference Banaschewski, Brandeis, Heinrich, Albrecht, Brunner and Rothenberger2003). The paradigm requires a simple response only if an infrequent cue stimulus is followed by a pre-specified target. Thus, the rare cues require preparation for a Go-NoGo task: if followed by a target a response needs to be executed, but if followed by a non-target, the prepared response must be withheld through inhibitory response control. CPT performance in children with ADHD is typically impaired, consistent with diminished sustained attention and control (Losier et al. Reference Losier, McGrath and Klein1996; Huang-Pollock et al. Reference Huang-Pollock, Nigg and Halperin2006). Additionally, event-related potentials (ERPs) can directly clarify deviance and timing of the covert brain processes leading to overt performance (Banaschewski & Brandeis, Reference Banaschewski and Brandeis2007).
Cues elicit a broad positive deflection maximal over parietal areas associated with attentional orienting and resource allocation, which is reduced in children and adults with ADHD (van Leeuwen et al. Reference van Leeuwen, Steinhausen, Overtoom, Pascual-Marqui, van't Klooster, Rothenberger, Sergeant and Brandeis1998; Banaschewski et al. Reference Banaschewski, Brandeis, Heinrich, Albrecht, Brunner and Rothenberger2003; McLoughlin et al. Reference McLoughlin, Albrecht, Banaschewski, Rothenberger, Brandeis, Asherson and Kuntsi2010). This Cue-P3 is followed by a centrally negative slow cortical potential field that terminates with the presentation of the next stimulus (contingent negative variation; CNV) and reflects preparation, anticipation and time estimation, which is not present after irrelevant distractors or non-targets (Walter et al. Reference Walter, Cooper, Aldridge, McCallum and Winter1964). Neurophysiological studies have implicated dopaminergic control of the CNV (Linssen et al. Reference Linssen, Vuurman, Sambeth, Nave, Spooren, Vargas, Santarelli and Riedel2011; Kratz et al. Reference Kratz, Studer, Baack, Malcherek, Erbe, Moll and Heinrich2012) and an ensemble of thalamo-cortical structures including the dorsal anterior cingulate cortex, frontal cortex, thalamus and midbrain dopaminergic nuclei involved in CNV generation (Gomez et al. Reference Gomez, Marco and Grau2003; Fan et al. Reference Fan, Kolster, Ghajar, Suh, Knight, Sarkar and McCandliss2007; Lutcke et al. Reference Lutcke, Gevensleben, Albrecht and Frahm2008). Time processing and preparation problems in patients with ADHD are associated with reduced activation in brain regions implicated in CNV generation (Rubia et al. Reference Rubia, Overmeyer, Taylor, Brammer, Williams, Simmons and Bullmore1999; Smith et al. Reference Smith, Taylor, Brammer, Halari and Rubia2008) and are considered as a candidate endophenotype (Castellanos & Tannock, Reference Castellanos and Tannock2002; Rommelse et al. Reference Rommelse, Oosterlaan, Buitelaar, Faraone and Sergeant2007b; Plummer & Humphrey, Reference Plummer and Humphrey2008; Himpel et al. Reference Himpel, Banaschewski, Grüttner, Becker, Heise, Uebel, Albrecht, Rothenberger and Rammsayer2009). In line with these considerations, CNV is consistently reduced in ADHD (Hennighausen et al. Reference Hennighausen, Schulte-Korne, Warnke and Remschmidt2000; Perchet et al. Reference Perchet, Revol, Fourneret, Mauguiere and Garcia-Larrea2001; Banaschewski et al. Reference Banaschewski, Brandeis, Heinrich, Albrecht, Brunner and Rothenberger2003, Reference Banaschewski, Yordanova, Kolev, Heinrich, Albrecht and Rothenberger2008) and may represent a persistent deficit in patients with ADHD (Valko et al. Reference Valko, Doehnert, Müller, Schneider, Albrecht, Drechsler, Maechler, Steinhausen and Brandeis2009; Doehnert et al. Reference Doehnert, Brandeis, Imhof, Drechsler and Steinhausen2010).
In trials following such cues, response control is required, and the difference in task demand between Go and NoGo trials is paralleled in the ERP. Following enhanced N2 amplitude in NoGo trials indexing conflict monitoring (Banaschewski et al. Reference Banaschewski, Brandeis, Heinrich, Albrecht, Brunner and Rothenberger2004), also the subsequent P3 is modulated (Fallgatter & Strik, Reference Fallgatter and Strik1999). Cued non-targets that require suppressing a prepared response evoke a NoGo-P3 maximal at fronto-central sites. Several authors suggest that it may reflect response inhibition (Pfefferbaum et al. Reference Pfefferbaum, Ford, Weller and Kopell1985; Fallgatter et al. Reference Fallgatter, Mueller and Strik1999; Fallgatter & Strik, Reference Fallgatter and Strik1999) or termination of motor activation through cortical deactivation in motor areas (Kopp et al. Reference Kopp, Mattler, Goertz and Rist1996; Falkenstein et al. Reference Falkenstein, Hoormann and Hohnsbein1999; Verleger et al. Reference Verleger, Paehge, Kolev, Yordanova and Jaśkowski2006). Clearly, these aspects are interrelated, and thus the NoGo-P3 may be generated by several processes of terminal inhibitory response control in the medial or anterior cingulate cortex, premotor areas and frontal areas (Kiefer et al. Reference Kiefer, Marzinzik, Weisbrod, Scherg and Spitzer1998; Strik et al. Reference Strik, Fallgatter, Brandeis and Pascual-Marqui1998; Weisbrod et al. Reference Weisbrod, Kiefer, Marzinzik and Spitzer2000; Fallgatter et al. Reference Fallgatter, Bartsch and Herrmann2002; Verleger et al. Reference Verleger, Paehge, Kolev, Yordanova and Jaśkowski2006; Beste et al. Reference Beste, Saft, Andrich, Gold and Falkenstein2008). Clinical studies have shown that the NoGo-P3 is altered in amplitude or anteriorization in patients with ADHD, and represents a persistent neurophysiological deficit (Brandeis et al. Reference Brandeis, van Leeuwen, Steger, Imhof, Steinhausen, Hirata, Koga, Nagata and Yamazaki2002; Fallgatter et al. Reference Fallgatter, Ehlis, Seifert, Strik, Scheuerpflug, Zillessen, Herrmann and Warnke2004, Reference Fallgatter, Ehlis, Rosler, Strik, Blocher and Herrmann2005; Valko et al. Reference Valko, Doehnert, Müller, Schneider, Albrecht, Drechsler, Maechler, Steinhausen and Brandeis2009; Dhar et al. Reference Dhar, Been, Minderaa and Althaus2010; Doehnert et al. Reference Doehnert, Brandeis, Imhof, Drechsler and Steinhausen2010).
The more parietally distributed Go-P3 elicited after cued target stimuli has been associated with several attentional functions such as evaluation of stimuli, closure or resource allocation (Polich & Kok, Reference Polich and Kok1995; Kok, Reference Kok2000), and may thus reflect aspects of executive response control. Explicit analyses of the Go-P3 in ADHD are rare, but some instances of diminished amplitude for patients have been seen (Strandburg et al. Reference Strandburg, Marsh, Brown, Asarnow, Higa, Harper and Guthrie1996; Overtoom et al. Reference Overtoom, Verbaten, Kemner, Kenemans, van Engeland, Buitelaar, Camfferman and Koelega1998; Banaschewski et al. Reference Banaschewski, Brandeis, Heinrich, Albrecht, Brunner and Rothenberger2004; Lawrence et al. Reference Lawrence, Barry, Clarke, Johnstone, McCarthy, Selikowitz and Broyd2005).
Taken together, the CPT elicits several separable phases of attentional and executive processing. Attentional orienting and resource allocation after cue onset are associated with the Cue-P3, and followed by preparation for and anticipation of the next trial as reflected by the CNV. The strength of the CNV has been associated with response speed. If the subsequent trial is a Go trial, execution of the prepared response is required, and this target stimulus-evoked Go-P3 reflects perceptual resource allocation to the Go stimulus as well as context updating. NoGo trials instead require the termination of the prepared motor response which is associated with a P3 that has a more anterior topography than the Go-P3 (Banaschewski et al. Reference Banaschewski, Brandeis, Heinrich, Albrecht, Brunner and Rothenberger2004). The aim of this study was to test whether performance and electrophysiological parameters in the time course from preparation to response execution or response control show familiality and may thus reflect potential endophenotypes for ADHD. It is further tested whether preparation and response control are particularly impaired in ADHD if additional need for cognitive control is induced by the presentation of additional incompatible flanker stimuli in the CPT. Moreover, correlational analyses were performed to explore the relationship between performance parameters and brain activity measures, and electrical neuroimaging techniques were used to locate the electrical neuronal sources of the ERPs for clarifying the neuronal networks responsible.
Method
Subjects
Recruitment of subjects was carried out as part of the International Multi-Center ADHD Genetics Project (IMAGE). Ethical approval for this study was obtained from the local ethical review boards. Details concerning recruitment and the diagnostic procedure can be found elsewhere (Andreou et al. Reference Andreou, Neale, Chen, Christiansen, Gabriels, Heise, Meidad, Muller, Uebel, Banaschewski, Manor, Oades, Roeyers, Rothenberger, Sham, Steinhausen, Asherson and Kuntsi2007; Albrecht et al. Reference Albrecht, Brandeis, Uebel, Heinrich, Mueller, Hasselhorn, Steinhausen, Rothenberger and Banaschewski2008). Briefly, recruitment of ADHD sibling pairs was conducted as part of the IMAGE study (Asherson et al. Reference Asherson, Brookes, Franke, Chen, Gill, Ebstein, Buitelaar, Banaschewski, Sonuga-Barke, Eisenberg, Manor, Miranda, Oades, Roeyers, Rothenberger, Sergeant, Steinhausen and Faraone2007) from specialist clinics for ADHD or child and adolescent psychiatrists in private practice in Zurich, Switzerland and Göttingen, Germany. Controls were recruited from local schools or sports clubs. Following screening by behavioural rating questionnaires obtained from both parents and teacher [long versions of Conners' rating scales (Conners et al. Reference Conners, Sitarenios, Parker and Epstein1998a, b) and the Strengths and Difficulties Questionnaire (SDQ; Goodman, Reference Goodman1997, Woerner et al. Reference Woerner, Becker and Rothenberger2004)], a semi-structured clinical interview (Parental Account of Children's Symptoms; Taylor et al. Reference Taylor, Everitt, Thorley, Schachar, Rutter and Wieselberg1986, Reference Taylor, Schachar, Thorley, Wieselberg, Everitt and Rutter1987) was applied by trained investigators in order to verify ADHD diagnosis.
A total of 289 European Caucasian children and adolescents aged 8 to 16 years with an estimated Hamburg–Wechsler Intellegence Test for Children Kinder (HAWIK-IQ) above 80 (Sattler, Reference Sattler1992) and no known child psychiatric disorder that may mimic ADHD were initially included. They belong to one of three subgroups: (1) children and adolescents with a DSM-IV diagnosis of ADHD combined type; (2) non-affected siblings of those with a DSM-IV diagnosis of ADHD, without a clinical diagnosis of ADHD; and (3) unrelated healthy controls without a clinical diagnosis or a known family history of ADHD.
Since gender ratio differed across groups (χ2(2) = 40.8, p < 0.01) and girls with ADHD were outnumbered and furthermore considerably younger, only the subsample of boys (n = 193) was analysed here. Due to artefacts in the EEG, datasets from three controls, three non-affected siblings and 20 participants with ADHD needed to be excluded, but exclusion ratio did not differ significantly across groups (χ2(2) = 3.5, p = 0.17). Thus, groups of 97 boys with ADHD, 27 male non-affected siblings of ADHD patients and 43 controls were finally included. The groups were matched for age (F 2,164 = 0.3, p = 0.77, η2p <0.01), but showed a tendency towards differences in intelligence quotient (IQ) (F 2,164 = 2.8, p = 0.07, η2p <0.03; children with ADHD had lower IQ than controls). Non-affected sibling exhibited only subtle, subclinical increases in hyperactivity scores and parent ratings (for further description of the sample, see Table S1 of the Supplementary material).
Stimuli and task
Assessments were carried out on 2 days in video-controlled, noise-shielded and slightly dimmed rooms at both departments, with the order of neurophysiological and neuropsychological testing following a randomization scheme. To assess preparation and response control and the impact of attentional load thereon, the Standard-CPT as well as a CPT Flanker version (Flanker-CPT) incorporating additional incompatible stimuli were used (Doehnert et al. Reference Doehnert, Brandeis, Straub, Steinhausen and Drechsler2008, Reference Doehnert, Brandeis, Imhof, Drechsler and Steinhausen2010; Valko et al. Reference Valko, Doehnert, Müller, Schneider, Albrecht, Drechsler, Maechler, Steinhausen and Brandeis2009). On each trial of both CPT versions, a bold letter was presented for 150 ms subtending 0.6° viewing angle horizontally and 0.8° vertically at 120 cm viewing distance against a light grey background in the centre of a 17-inch (43 cm) cathode ray tube monitor. Since the Flanker-CPT incorporates two additional flanker letters that had to be ignored, horizontal viewing angle was approximately 2.0° (see Fig. 1).
Task description. Standard Continuous Performance Test (Standard-CPT) and Flanker-CPT in comparison. In both tasks, a sequence of relevant letters was presented in the centre of a screen between two vertical fixation marks. Participants were instructed to respond if a cue A was followed by X, but to withhold responding in all other cases. In the Flanker-CPT, additional distracting letters X or O were presented alongside.
In both CPTs, the children had to press the response button with the index finger of their dominant hand only if a central ‘O’ (cues, presented 80 times, see Fig. 1) was followed by a central ‘X’ (Go-trials, n = 40), but to withhold responding if the cue was followed by a non-target (NoGo trials, n = 40), or if the ‘X’ was uncued (40 trials). Most trials were neutral distractors (letters B, C, D, E, F, G, J or L, n = 20 each, or the letter H, n = 80) which also did not require a response, making up a total of 400 trials presented at a rate of 1/1650 ms. The sequences and neutral distractors were pseudo-randomly distributed.
The flankers consisted either of O's or X's to induce conflict. Targets and distractors H were flanked by O's (‘OXO’ and ‘OHO’), while cues as well as the remaining distractor stimuli were flanked by X's. As a consequence, cues and cued distractors now required additional response control in terms of inhibition (as they are flanked by target stimuli which can require a response in the context of the CPT paradigm). Cued targets require additional response control in terms of execution (since the flanking cue stimuli would require no response).
Both CPTs lasted 11 min following 25 practice trials. Their order was counter-balanced after 5 min of resting EEG with a 13 min action-monitoring task in between, and short breaks between tasks as desired.
Electrophysiological recording and processing
For the subjects assessed in Göttingen, the EEG was recorded with Ag/AgCl electrodes from 24 sites according to an extended 10–20 system using a BrainAmp amplifier (BrainProducts, Germany) including the FCz recording reference, a 500 Hz sampling-rate, 0.016–100 Hz filters, a 50 Hz notch filter, and a ground electrode placed on the forehead. The electrooculogram (EOG) was recorded from additional electrodes placed above and below the right eye and at the outer canthi. In Zürich, the EEG was recorded in the same way except for including additional channels, and using a SynAmps amplifier (Neuroscan, USA) and Fpz as reference at 500 Hz with 0.1–70 Hz filters. The EOG was recorded from electrodes below the eyes. Impedances were kept below 10 kΩ. Post-processing (see below) eliminated all differences between sites to ensure full compatibility.
Altogether 24 common electrodes were analysed here. After down-sampling to 256 Hz, the EEG was re-referenced to the average reference and filtered offline with 0.1–30 Hz, 24 db/oct Butterworth filters. Occular artefacts were corrected using a linear regression method without raw average subtraction (Gratton et al. Reference Gratton, Coles and Donchin1983). If the amplitude at any EEG electrode exceeded ±100 μV, a section −100 to +800 ms around was excluded from further analyses. Segments −125 to 1875 ms around stimulus onset of Cue, Go and NoGo trials with correct responses [i.e. no responses on cues and NoGo trials, responses (‘hits’) on Go trials] were subsequently visually checked and averaged. All averages contained at least 15 sweeps. To avoid distortion of ERP topography, no baseline subtraction was applied.
Analyses
The performance parameters mean reaction time (RT) of correct target responses and inter-individual RT variability (RT-SD) were analysed in an analysis of variance (ANOVA) with the between-subject factor Group (controls v. non-affected siblings v. children with ADHD) and the within-subject factor Task (Standard-CPT v. Flanker-CPT). Since commission errors in Go trials and false alarms in NoGo trials were rare and distribution was not normal, accuracy was analysed non-parametrically with Wilcoxon tests on the factor Task for the total sample, and overall Kruskal–Wallis tests on the factor Group followed by post hoc Mann–Whitney U tests separately for each task.
The CNVs following Cues over centro-parietal electrodes (Fig. 2) were analysed as mean amplitudes from 1200 to 1650 ms at electrodes Cz and Pz. The Cue- and Go-P3 had a parietal maximum and were scored as the most positive peak in the time-frame 250 to 700 ms at electrode Pz (see Figs 2 and 3), while the NoGo-P3 was maximal at fronto-central sites and thus scored 300 to 550 ms at FCz (Fig. 4). The electrophysiological parameters were analysed using ANOVAs with the between-subject factor ‘Group’ and within-subject factors ‘Task’ and ‘Site’ (centro-parietal electrodes Cz and Pz for CNV, fronto-central electrodes Fz, FCz and Cz for NoGo-P3 and electrodes Cz, P3, Pz, P4 and Oz for Cue- and Go-P3 which are maximal over the mid-parietal cortex).
Topography and sources of the Cue-P3 and the contingent negative variation (CNV). Stimulus-locked grand average waves (above) of controls, non-affected siblings and attention deficit-hyperactivity disorder (ADHD) patients with spline-interpolated maps and standardized low-resolution electromagnetic tomography (sLORETA) source localizations of the Cue-P3 (middle) and mean CNV (below, mean area 1200 to 1650 ms post-cue onset) evoked by correct responded cues. CPT, Standard Continuous Performance Test.
Topography and sources of the Go-P3. Stimulus-locked grand average waves of controls, non-affected siblings and attention deficit-hyperactivity disorder (ADHD) patients with spline-interpolated Go-P3 maps evoked by cued targets with correct responses (hits). CPT, Standard Continuous Performance Test.
Topography and sources of the NoGo-P3. Stimulus-locked grand average waves of controls, non-affected siblings and attention deficit-hyperactivity disorder (ADHD) patients with spline-interpolated NoGo-P3 maps evoked by cued non-targets with correctly inhibited responses. CPT, Standard Continuous Performance Test.
Familiality is indicated by a significant Group main effect or interaction with the factor Group, and by significant differences of the means between controls and ADHD as well as between controls and non-affected siblings, which was tested in post hoc analyses of confidence intervals with p < 0.05. The functional significance of the ERP components and their relation to general cognitive ability was explored in the control group through partial correlations between ERP amplitudes, and CPT performance parameters (controlled for age as an indicator of developmental trends in the assessed parameters that may lead to artificial correlations).
The scalp topography of ERP group differences was explored using t-maps. ERP sources were calculated using standardized low-resolution electromagnetic tomography (sLORETA) (Fuchs et al. Reference Fuchs, Kastner, Wagner, Hawes and Ebersole2002; Pascual-Marqui, Reference Pascual-Marqui2002) for each group's average ERP component.
Results
Performance data
RTs of correct responses
Correct reactions to cued targets were slower in the Flanker-CPT than in the Standard-CPT (Task: F 1,164 = 50.9, p < 0.01, η2p = 0.24). There was also a trend towards group differences (Group: F 2,164 = 2.8, p = 0.06, η2p = 0.03; non-affected siblings, and as a tendency also ADHD patients responded slower than controls), but no Group × Task interaction (F 2,164 = 2.2, p = 0.11, η2p <0.03). For means and standard deviations and further details of the performance data statistics, see Table S2 of the Supplementary material.
RT-SDs of correct responses
Intra-individual standard deviations of RTs (RT-SDs) were higher in the Flanker-CPT than in the Standard-CPT (Task: F 1,164 = 5.5, p = 0.02, η2p = 0.03). ADHD patients had higher RT-SDs than both non-affected siblings and controls (Group: F 2,164 = 4.4, p = 0.01, η2p = 0.05).
Hit rate
Since performance accuracy was generally high and close to ceiling, it was analysed non-parametrically. In an overall comparison, the tasks did not differ regarding hit rate (Wilcoxon test, Z = 1.0, p = 0.31). Patients with ADHD performed less accurately than controls in both tasks, and less accurately than non-affected siblings for the Standard-CPT only (overall Kruskal–Wallis χ2(2) > 11.3, p < 0.01 and post-hoc Mann–Whitney U tests p < 0.03; for all other post-hoc comparisons p > 0.14).
False alarms
The Flanker-CPT provoked more false alarms than the Standard-CPT (Wilcoxon test, Z = 6.9, p < 0.01), but neither task revealed group differences (χ2(2) < 2.2, p > 0.35).
ERP data
Cue-P3
The Cue-P3 with its parietal maximum had sources localized by sLORETA in the superior parietal cortex [Brodmann Area (BA) 5, 7, 31 and 40; see Fig. 2]. The latency detected at Pz was prolonged in the Flanker-CPT [marginal means 534 (s.d. = 122) ms v. 558 (s.d. = 118) ms, Task: F 1,164 = 8.3, p < 0.01, η2p = 0.05), but similar in the three groups (Group: F 2,164 = 0.2, p > 0.50, η2p <0.01, and Task × Group: F 2,164 = 1.5, p = 0.22, η2p = 0.02).
Cue-P3 amplitude was largest at the middle-parietal site Pz (compared with the surrounding electrodes; Site: F 4,656 = 101.4, p < 0.01, η2p = 0.38), and showed a more anterior topography in the Flanker-CPT (positivity larger at Cz and diminished at Oz, Task × Site: F 4,656 = 8.1, p < 0.01, η2p = 0.05). Moreover, the Cue-P3 showed familiality effects (Group: F 2,164 = 4.0, p = 0.02, η2p = 0.05), with significantly larger amplitudes in controls than in ADHD, and as a tendency (p = 0.08) the intermediate values of the non-affected siblings also differed from both groups. T-maps revealed significantly diminished Cue-P3 amplitude in case–control and non-affected siblings–control comparisons at parietal and occipital sites. For further details, see Table S3 and Fig. S1a of the Supplementary material.
Cue-CNV
The CNV following cues shows the characteristic transient negativity with centro-parietal maximum that terminates with the onset of the following stimulus. Source localizations using sLORETA detected activity in the medial frontal cortex and rostral and dorsolateral cingulate cortex (BA 6, 11 and 24; see Fig. 2). Particularly for the Flanker-CPT, the CNV mean amplitude following cues was largest at Cz (Site: F 1,164 = 31.4, p < 0.01, η2p = 0.16, and Task × Site: F 1,164 = 4.5, p = 0.04, η2p = 0.03). It was generally diminished for boys with ADHD as compared with both their non-affected siblings and controls, and non-affected siblings showed smaller CNVs than controls (Group: F 2,164 = 9.7, p < 0.01, η2p = 0.11). Moreover, this pattern was more prominent for the Flanker-CPT (Task × Group: F 2,164 = 3.3, p = 0.04, η2p = 0.04).
The t-map comparisons between controls and patients with ADHD or non-affected siblings revealed that the detected effects coincide with the CNV topography. For further details, see Table S3 and Fig. S1b of the Supplementary material.
Go-P3
The Go-P3 had a parietal maximum with sources detected in the cingulate gyrus, the superior paracentral and parietal lobulae and the praecuneus (BA 5, 7 and 31; see Fig. 4). Its latency was higher for the Flanker-CPT [453 (s.d. = 107) ms v. 399 (s.d. = 120) ms, Task: F 1,164 = 29.0, p < 0.01, η2p = 0.15], and showed no group differences or Task × Group interactions (both F 2,164 <1, p > 0.48, η2p <0.01).
Go-P3 amplitude was highest at Pz (Site: F 4,656 = 154.0, p < 0.01, η2p = 0.48) and showed a tendency towards higher amplitudes in the Standard-CPT at the left-parietal P3 electrode (Task × Site: F 4,656 = 2.2, p = 0.09, η2p = 0.02), without any overall differences between groups or any further significant interactions (all p > 0.21). The t-maps showed reduced activity for non-affected siblings at centro-parietal sites (see Table S4 and Fig. S1c of the Supplementary material).
An additional analysis subdividing the Go-P3 into P3a (250–400 ms) and P3b (400–700 ms) components also revealed no familiality effects or interactions.
NoGo-P3
The NoGo-P3 with its fronto-central maximum showed sLORETA sources in the superior and medial frontal cortex and cingulate gyrus (BA 6, 8, 9, 24 and 32; see Fig. 3). Its latency was longer for the Flanker-CPT than the Standard-CPT [414 (s.d. = 60) ms v. 388 (s.d. = 48) ms, respectively, Task: F 1,164 = 33.7, p < 0.01, η2p = 0.17] without a main effect of Group or an interaction (both F 2,164 <1, p > 0.85, η2p <0.01).
The amplitude of the NoGo-P3 was higher in the Standard-CPT (Task: F 1,164 =7.8, p < 0.01, η2p = 0.05), and highest at FCz (Site: F 2,328 = 88.5, p < 0.01, η2p = 0.35). It was generally diminished for boys with ADHD as compared with controls, while non-affected siblings did not differ significantly from either of the other groups (Group: F 2,164 = 3.8, p = 0.03, η2p = 0.04, post hoc controls >ADHD, p < 0.05, differences involving non-affected siblings both p > 0.20). The t-map comparisons showed lower amplitudes for ADHD as compared with controls over central sites for both tasks, and non-affected siblings showed higher amplitudes than ADHD in the Standard-CPT only, but a tendency towards reduction compared with controls in the Flanker-CPT (see Fig. S1d of the Supplementary material).
Fig. 5 summarizes the CNV, Cue-, Go- and NoGo-P3 amplitude findings of the Standard-CPT and Flanker-CPT for all groups.
Synopsis of electrophysiological parameters. Means and 95% confidence intervals with p < 0.05 for the mean [of Flanker Standard Continuous Performance Test (Flanker-CPT) and Standard-CPT] and difference (Flanker-CPT minus Standard-CPT) event-related potential amplitudes. Impaired cue processing was familially driven in attention deficit-hyperactivity disorder (ADHD), with impairments in (a) Cue-P3 and (b) contingent negative variation (CNV) mean amplitude present in children with ADHD and their non-affected siblings when compared with controls. Patients with ADHD showed (d) further impaired NoGo-P3 while non-affected siblings did not differ significantly from either of the other two groups. (c) No clear group differences were evident for the Go-P3.
Correlations between performance and ERP parameters
In the control group, mean RT and RT-SD were positively correlated with mean CNV, while all P3 component amplitudes showed negative correlations (all p < 0.05). The RT and RT-SD differences between the two CPTs correlated negatively with the respective change in the NoGo-P3 amplitude. General cognitive ability was not related to performance (all r < 0.12, p > 0.43), and only showed by trend a correlation with mean Go-P3 amplitude (see Fig. S2 of the Supplementary material).
Discussion
In this study, the time course of familial ADHD effects upon sequential neural markers from preparation to response control was assessed. Behavioural ratings obtained from parents and teachers with the SDQ confirmed the problems in children with ADHD, while strengths and difficulties were within the normal range for healthy siblings and controls. The mean hyperactivity scores of non-affected siblings also fell within the normal range but were slightly higher than for controls. The aim of this study was to assess brain processes during the time course from preparation to response execution or control along with the impact of stimulus incongruency.
Performance
Means and variability of correct response RT as well as false alarm rate confirm that the Flanker-CPT is generally more demanding than the Standard version. The main RT analyses indicate that robust CPT performance impairments in ADHD are limited to the intra-individual RT variability, which is regarded as a particularly sensitive (van Leeuwen et al. Reference van Leeuwen, Steinhausen, Overtoom, Pascual-Marqui, van't Klooster, Rothenberger, Sergeant and Brandeis1998; Banaschewski et al. Reference Banaschewski, Brandeis, Heinrich, Albrecht, Brunner and Rothenberger2003; Klein et al. Reference Klein, Wendling, Huettner, Ruder and Peper2006), albeit probably not specific, marker of ADHD. Responses to cued targets were slower in the Flanker-CPT than in the Standard-CPT, which probably reflects the additional attentional load imposed by the flanking targets by cue stimuli that (if presented centrally) would not require a response. Post hoc tests provided some evidence that the Flanker-CPT is also capable of uncovering slower responses in patients with ADHD, as also reported for more demanding tasks tapping cognitive control (Albrecht et al. Reference Albrecht, Brandeis, Uebel, Heinrich, Mueller, Hasselhorn, Steinhausen, Rothenberger and Banaschewski2008) and for the Flanker-CPT in adults with ADHD (McLoughlin et al. Reference McLoughlin, Albrecht, Banaschewski, Rothenberger, Brandeis, Asherson and Kuntsi2010, Reference McLoughlin, Asherson, Albrecht, Banaschewski, Rothenberger, Brandeis and Kuntsi2011). Accuracy was lower in the ADHD patients than in the other two groups irrespective of task version. Thus, these results indicate that children with ADHD display several problems in CPT performance, consistent with most previous studies (van Leeuwen et al. Reference van Leeuwen, Steinhausen, Overtoom, Pascual-Marqui, van't Klooster, Rothenberger, Sergeant and Brandeis1998; Banaschewski et al. Reference Banaschewski, Brandeis, Heinrich, Albrecht, Brunner and Rothenberger2003). However, as non-affected siblings did not consistently display intermediate results, overt behavioural CPT performance in ADHD may be less subject to shared familial factors. This is in contrast to our previous findings regarding action monitoring and may reflect specific qualitative differences in task demands (Albrecht et al. Reference Albrecht, Brandeis, Uebel, Heinrich, Mueller, Hasselhorn, Steinhausen, Rothenberger and Banaschewski2008, Reference Albrecht, Brandeis, Uebel, Heinrich, Heise, Hasselhorn, Rothenberger and Banaschewski2009).
ERPs
Cue processing appears to be the most consistent familially driven electrophysiological deficit in childhood ADHD in this test. Following the presentation of a cue which indicates that the next stimulus may be relevant for task performance, children with ADHD and also to a smaller degree their non-affected siblings showed diminished brain activity related to attentional orienting to the cue as reflected in the Cue-P3 with sources in the parietal lobe. The subsequent preparation for or anticipation of the next trial following cues was also familially driven as reflected by a markedly reduced CNV in children with ADHD as compared with controls, while non-affected siblings had intermediate activity. CNV was diminished by increasing conflict through additional flankers in non-affected siblings and in ADHD boys as compared with controls whose CNV remained stable. This indicates that incompatible stimuli implicating response conflict and accordingly increasing attentional load efficiently hampered preparatory processes in ADHD (Valko et al. Reference Valko, Doehnert, Müller, Schneider, Albrecht, Drechsler, Maechler, Steinhausen and Brandeis2009; Doehnert et al. Reference Doehnert, Brandeis, Imhof, Drechsler and Steinhausen2010; McLoughlin et al. Reference McLoughlin, Albrecht, Banaschewski, Rothenberger, Brandeis, Asherson and Kuntsi2010), which is in agreement with previously reported difficulties in action monitoring (Albrecht et al. Reference Albrecht, Brandeis, Uebel, Heinrich, Mueller, Hasselhorn, Steinhausen, Rothenberger and Banaschewski2008). Consequently, group differences were particularly pronounced in the more challenging Flanker-CPT, which suggests that impaired processing of conflicting signals shows familiality in ADHD already for preparatory activity, and may thus indicate that shared environmental or genetic factors have an impact on cognitive control even during response preparation. Sources of CNV were located in the dorsal and rostral cingulate cortex, consistent with the hypothesis that CNV reductions may be a consequence of impaired dopaminergic modulations. Since several studies have shown that the CNV can be increased by successful neurofeedback training (Heinrich et al. Reference Heinrich, Gevensleben, Freisleder, Moll and Rothenberger2004; Doehnert et al. Reference Doehnert, Brandeis, Straub, Steinhausen and Drechsler2008; Wangler et al. Reference Wangler, Gevensleben, Albrecht, Studer, Rothenberger, Moll and Heinrich2010), these findings demonstrate that even trait-like endophenotype markers can change with effective behavioural treatment, and may reflect the plasticity of dopaminergic functions which can also be altered by working memory training (McNab et al. Reference McNab, Varrone, Farde, Jucaite, Bystritsky, Forssberg and Klingberg2009). Conversely, one may speculate which shared familial factors influence CNV: it may be that potential genetic factors related to the dopaminergic system along with environmental factors within the family like psychosocial interactions, participation in intellectual and physical activities (Nigg et al. Reference Nigg, Nikolas and Burt2010; Halperin & Healey, Reference Halperin and Healey2011) or stress may play important roles.
Both Go- and NoGo-P3s had longer latencies in the Flanker-CPT. Since Go-P3 latency was positively correlated with Go-RT, this probably reflects additional time needed for conflict resolution due to processing of incongruent stimuli (Kok, Reference Kok2001). The amplitude of the NoGo-P3 was diminished after incongruent stimuli but reflected less clearly familiality of ADHD. It was significantly diminished in boys with ADHD as compared with controls and intermediate in non-affected siblings, but the siblings' scores did not differ significantly from either of the other two groups. Impairments detected for NoGo-P3 amplitude did not extend towards the Go-P3. Executive response control in Go trials was associated with a large positive deflection maximal at posterior electrodes that peaked (like the NoGo-P3) earlier in the Standard-CPT than in the Flanker-CPT. We found no Go-P3 differences in latency or amplitude between groups, consistent with previous findings using this type of CPT (van Leeuwen et al. Reference van Leeuwen, Steinhausen, Overtoom, Pascual-Marqui, van't Klooster, Rothenberger, Sergeant and Brandeis1998; Banaschewski et al. Reference Banaschewski, Brandeis, Heinrich, Albrecht, Brunner and Rothenberger2003), although such reductions have also been reported in highly co-morbid ADHD samples (Overtoom et al. Reference Overtoom, Verbaten, Kemner, Kenemans, van Engeland, Buitelaar, Camfferman and Koelega1998; Banaschewski et al. Reference Banaschewski, Brandeis, Heinrich, Albrecht, Brunner and Rothenberger2003).
The functional meaning of these ERP parameters was assessed in the control sample, in order to avoid artificial relationships due to group differences or potential confounds in ADHD symptom severity. In general, CPT performance and the assessed ERP components were not related to differences in cognitive ability in the control group. This supports our approach to conduct analysis which does not control for IQ differences between the groups and suggests that IQ cannot explain the reported findings. Moreover, a diminished IQ as often reported in ADHD research may be a consequence of the symptomatology, which would question the need for a statistical adjustment. Larger (negative) CNV as well as (positive) Go- and NoGo-P3 amplitudes were associated with faster and more homogeneous RTs in both task versions; the additional cognitive control needed in the Flanker-CPT compared with the Standard-CPT led to slower and more variable performance, which was significantly correlated with the respective reduction in NoGo-P3 amplitude. This highlights that brain activity reflected in these ERP parameters may be crucial for proper performance.
These ERP parameters and the underlying constructs also relate to several components of the cognitive energetic model of ADHD (Sergeant, Reference Sergeant2005): particularly the diminished Cue-P3 and CNV amplitudes correlate with performance and may index attentional orienting and subsequent activation problems in ADHD and also non-affected siblings. Higher cognitive load due to additional incompatible flankers may affect effort available for processing the target stimuli, and consequently increase familially driven problems in ADHD.
Limitations
The current study has a number of limitations that need to be considered in the evaluation of the results.
Although the sample size of the current study (n 167) is relatively large for a clinical ERP investigation, the study may still be limited by statistical power. With α set to 5%, and power to 80%, it is possible to detect a medium effect size (η2p >5.5% explained variance) between the three groups. This statistical power is sufficient to detect the medium to large differences between ADHD cases and controls commonly reported for performance and several electrophysiological parameters in post hoc pairwise comparisons, but differences between the ADHD cases and the (intermediate) non-affected siblings may be of considerably lower effect size. The sample size required to detect small-to-medium effect sizes would be 260, and accepting also a trend with α set to 10% as a control–sibling difference would require at least 204.
As a consequence, this study may have failed to detect possible small sibling–control effects, but our differences between non-affected siblings and controls were even below Cohen's suggestion for small effect sizes (d = 0.12) for the mean NoGo-P3 amplitude, and for the Go-P3 where no familiality at all was detected, the differences between the three groups were even smaller. Taken together, our conclusion that particularly cue processing is familially driven in ADHD remains supported despite the limited statistical power of the study. ADHD case–control differences in the CPTs NoGo-P3 related to terminal response control did not show clear familiality: although the non-affected sibling group mean was intermediate between controls and ADHD children, they did not statistically differ from either one, and the small effect size of the sibling–control differences suggests a very limited practical importance. Other tasks with more demands on response inhibition like Stop-NoGo or Go-NoGo tasks with more Go-Go than NoGo trials may be better capable of showing familiality in these ‘inhibitory’ parameters. This is further supported by the current data, since the non-affected siblings also showed a strong reduction in NoGo-P3 amplitude with additional incongruent flankers.
A second limitation of the current study is the source localization precision of the assessed ERPs. For a clinical study with children suffering from attention deficits, hyperactivity and impulsivity, it is necessary to limit the preparation and recording time. As a consequence, the spatial sampling of the electrical field on the scalp was limited to a relatively low number of electrodes that could be attached with good electrical impedance (ensuring quality and stability) within a reasonable time in a multicentre study. The 24 scalp electrodes available for the current study are sufficient to capture the basic scalp topography of the evoked potentials, but the precision of the brain electrical neuroimaging is limited. There is considerable evidence that sLORETA gives qualitatively valid source localizations in children and adults with lower numbers of electrodes as long as the potential field is adequately covered, although optimal spatial resolution would require more (at least 64 to 100) electrodes (Ryynanen et al. Reference Ryynanen, Hyttinen and Malmivuo2006; Michel & Brandeis, Reference Michel, Brandeis, Michel, Koenig, Brandeis, Gianotti and Wackermann2009). Therefore, the sLORETA neuroimaging provided here is clearly limited in resolution.
Conclusion
Neural ADHD markers of covert preparation following cues, and partly markers of response inhibition in NoGo trials were familially driven in ADHD. These familial ADHD markers reflect specific attentional and inhibitory processing stages correlated with performance, and thus appear crucial for the functional understanding of genetic and environmental pathways leading to ADHD. Particularly preparatory processes as reflected by the Cue-P3 and CNV associated with attentional orienting, resource allocation and preparation or anticipation were significantly impaired in non-affected siblings, which indicates that genetic or environmental factors shared in families with ADHD have an impact on these functions. Moreover, as both preparation and response inhibition depend on attentional load, ADHD familiality effects seemed to become larger if the going gets tough.
Supplementary material
For supplementary material accompanying this paper visit http://dx.doi.org/10.1017/S003329171200270X.
Acknowledgements
Recruitment of ADHD sibling pairs was supported by National Institute of Mental Health grant no. R01MH062873 to Stephen V. Faraone. D.B. received support from the Swiss National Science Foundation (grant no. 32–109591).
The authors thank all children and their families for participation. Christa Dahlmann, Renate Kolle and Antonia Seitz conducted ERP recordings; Anke Fillmer-Otte, Anne Reiners, Gudrun Schneider and Nicola Wöstmann performed IQ and further neuropsychological testing.
Declaration of Interest
None.
