Get 20M+ Full-Text Papers For Less Than $1.50/day. Start a 14-Day Trial for You or Your Team.

Learn More →

Relationship between endophenotype and phenotype in ADHD

Relationship between endophenotype and phenotype in ADHD Background: It has been hypothesized that genetic and environmental factors relate to psychiatric disorders through the effect of intermediating, vulnerability traits called endophenotypes. The study had a threefold aim: to examine the predictive validity of an endophenotypic construct for the ADHD diagnosis, to test whether the magnitude of group differences at the endophenotypic and phenotypic level is comparable, and to investigate whether four factors (gender, age, IQ, rater bias) have an effect (moderation or mediation) on the relation between endophenotype and phenotype. Methods: Ten neurocognitive tasks were administered to 143 children with ADHD, 68 non- affected siblings, and 120 control children (first-borns) and 132 children with ADHD, 78 non- affected siblings, and 113 controls (second-borns) (5 – 19 years). The task measures have been investigated previously for their endophenotypic viability and were combined to one component which was labeled 'the endophenotypic construct': one measure representative of endophenotypic functioning across several domains of functioning. Results: The endophenotypic construct classified children with moderate accuracy (about 50% for each of the three groups). Non-affected children differed as much from controls at the endophenotypic as at the phenotypic level, but affected children displayed a more severe phenotype than endophenotype. Although a potentially moderating effect (age) and several mediating effects (gender, age, IQ) were found affecting the relation between endophenotypic construct and phenotype, none of the effects studied could account for the finding that affected children had a more severe phenotype than endophenotype. Conclusion: Endophenotypic functioning is moderately predictive of the ADHD diagnosis, though findings suggest substantial overlap exists between endophenotypic functioning in the groups of affected children, non-affected siblings, and controls. Results suggest other factors may be crucial and aggravate the ADHD symptoms in affected children. Page 1 of 14 (page number not for citation purposes) Behavioral and Brain Functions 2008, 4:4 http://www.behavioralandbrainfunctions.com/content/4/1/4 Studying endophenotypes may have certain advantages Background Psychiatric disorders as defined by the Diagnostic and Sta- over studying phenotypes. It has been proposed that tistical Manual of Mental Disorders (DSM-IV-TR) [1] have endophenotypes may be more suitable for detecting risk been hypothesized as reflecting the extreme end of under- genes, because endophenotypes are genetically less com- lying, continuously distributed traits [2-4]. In line with plex than phenotypes (i.e. related to fewer genes than phe- this, the behaviour of individuals suffering from psychiat- notypes) and, therefore, probably stronger linked to these ric disorders differs quantitatively but not necessarily disease genes than phenotypes [7-9]. Endophenotypes qualitatively from the behaviour of individuals without may also be useful in exploring different pathways leading psychiatric problems. The threshold of what is and what up to the disorder: Patients having the same diagnosis is not abnormal is to a certain extent arbitrarily deter- may differ strongly in the number and severity of symp- mined, but patients have in common that their behaviour toms they portray, suggesting heterogeneity in the causal interferes with their normal life and cause the patient (and pathways [10]. Creating more homogeneous subgroups his/her environment) to suffer. Why certain people pass of patients based on their endophenotypic functioning, this threshold and are diagnosed with a disorder and oth- may facilitate unravelling these differential causal path- ers do not is determined by additive and interacting ways. genetic and environmental risk factors [5,6]. Studies have shown that psychiatric disorders have genetic and envi- In the last two decades, substantial attention has been ronmental underpinnings which probably contribute to given to studying endophenotypes of psychiatric disor- certain neurocognitive abnormalities that, in turn, lead to ders. This has led to the discussion of what exactly consti- abnormal behaviour. It is theorized that these neurocog- tutes an endophenotype and what criteria must be met for nitive abnormalities form underlying, continuously dis- a neurocognitive function to be useful as candidate endo- tributed, vulnerability traits (endophenotypes) that phenotype [7-9,11-14]. Since an endophenotype forms a heighten the risk for developing a disorder (phenotype) link between susceptibility genes and the disorder, it fol- [7-9]. In this context, neurocognitive abnormalities refer lows that: (1) the neurocognitive dysfunction is heritable to mental functions that are mediated by brain processes; (and familial), in which at least partly the same genes these mental functions are not directly observable, but influence the endophenotype and phenotype; (2) the may be manipulated and measured using experimental neurocognitive dysfunction is associated with the disor- paradigms. Multiple endophenotypes interact to deter- der; (3) the neurocognitive dysfunction is observable in mine the finally observable behavior, the phenotype, non-affected first-degree relatives of an affected individ- which might be abnormal. In this context, the phenotype ual, because first-degree relatives are likely to carry some refers to directly observable symptoms of a disorder (Fig- of the susceptibility genes of the disorder. ure 1). Disease genes Environmental risk factors Neurobiological deviations Endophenotypes Neuropsychological deviations Phenotype Figure 1 The relationship between genetic and environmental risk factors, endophenotypes, and phenotype in ADHD The relationship between genetic and environmental risk factors, endophenotypes, and phenotype in ADHD. Page 2 of 14 (page number not for citation purposes) Behavioral and Brain Functions 2008, 4:4 http://www.behavioralandbrainfunctions.com/content/4/1/4 Thus far, several psychiatric conditions have been targeted motor endophenotypes, is predictive of the ADHD diag- for candidate endophenotypes and considerable knowl- nosis. edge has been gathered on the usefulness of different endophenotypes in for example, bipolar disorder, schizo- Practically all studies of ADHD endophenotypes con- phrenia, substance abuse, and depression [15-19]. This ducted thus far have reported the same type of results: The study targeted Attention-Deficit/Hyperactivity Disorder group of non-affected siblings performs in between the (ADHD), since it is one of the most prevalent child psy- affected siblings group and the normal control group. At chiatric disorders, yet the knowledge on the usefulness of a phenotypic level, non-affected siblings do not (or appar- ADHD endophenotypes is still limited. ently to a lesser extent) deviate from the controls. The reverse appears to be true for affected children: Their phe- Two studies have failed to find neurocognitive impair- notypic deficits are more pronounced than one would ments in parents of children with ADHD [20,21] and two expect based on their cognitive and motor dysfunction- studies found no conclusive evidence of cognitive dys- ing. Therefore, the second aim of the paper was to confirm functioning in non-affected siblings [22,23]. However, apparent observations that children with ADHD show one study did find evidence for cognitive functions as more severe ADHD symptoms than one would expect endophenotypes [24] and other studies, specifically tar- based on their endophenotypic dysfunctioning, whereas geting inhibition or interference control, found evidence non-affected siblings show less ADHD-like behaviour for these functions as endophenotypes [13,25-28]. In a than one would expect based on their endophenotypic previous study, we also found evidence for inhibition as vulnerabilities. well as visuo-spatial and verbal working memory as endo- phenotypes for ADHD [29]. Moreover, it appeared that The question that automatically arises if these observa- deficits in the various cognitive functions partly arose tions can be confirmed, is why endophenotypic vulnera- from the same genetic and/or environmental risk factors. bilities are not proportionally related to deviations at a Furthermore, evidence has been found for time reproduc- phenotypic level? It might be that certain factors moderate tion as ADHD endophenotype, a function related to the and/or mediate the relation between endophenotype and sense of time [30]. Less attention has been given to study- phenotype. Moderation would imply that the relation ing functions outside the cognitive domain, although between endophenotype and phenotype is not compara- ADHD is frequently associated with motor deficits in Cau- ble across different levels of the moderating factor [37]. casian and non-Caucasian subjects [31,32]. In previous Mediation would imply that (a part of) the relation studies, we and others have shown that non-affected sib- between endophenotype and phenotype can be explained lings have subtle problems, similar to their affected sib- through their correlation with a third factor. When this lings, in motor timing, motor control, motor speed and third factor is taken into account, the relation between variability, and speed of oculomotor control [33-36], sug- endophenotype and phenotype disappears or weakens gesting endophenotypes for ADHD may also lie inside the [37]. The third aim of this study was to explore whether motor domain. Moreover, it appeared that some of the four factors (gender, age, IQ, and rater bias) had a (mod- cognitive dysfunctioning in children with ADHD and erating and/or mediating) effect on the relationship their non-affected siblings may be related to problems between endophenotype and phenotype. already apparent on a simple reaction time task [31]. With respect to the first factor gender: ADHD is more fre- Taken together, the majority of studies have found sup- quently diagnosed in boys than girls [38], probably port for cognitive and motor endophenotypes for ADHD because boys are more vulnerable to the disorder. Since [24,26,28-31,33-35]. These studies have in common that affected children are more often boys than girls, whereas they administered one or a few measures tapping into a the gender ratio is more or less equal in the group of non- single domain. It is unlikely, however, that one such affected siblings, moderating and/or mediating effects of measure/domain mediates the relation between genotype gender may possibly be accountable for the apparent non- and phenotype and can predict the phenotype, because it comparable magnitude of group differences at the endo- is unlikely that all children with ADHD and their non- phenotypic and phenotypic level. With respect to the sec- affected siblings will show this endophenotype given the ond factor age: It is known from several studies that the causal heterogeneity of ADHD [13]. It is more likely that severity of ADHD symptoms appears to decline to some multiple endophenotypes mediate the relation between extent with age [39-41] and that the same might be true genotype and phenotype, and together are more powerful for underlying neurocognitive vulnerabilities [30]. Yet, in predicting the phenotype. Therefore, the first aim of our others have failed to find diminishing neurocognitive vul- study was to investigate if an endophenotypic construct, nerabilities with age [40,42,43]. It may thus be possible encompassing a broad battery of both cognitive and that age has a moderating and/or mediating influence on the relation between endophenotype and phenotype. Page 3 of 14 (page number not for citation purposes) Behavioral and Brain Functions 2008, 4:4 http://www.behavioralandbrainfunctions.com/content/4/1/4 With respect to the third factor IQ: A frequently reported relatives were required to have no formal or suspected finding is that children with ADHD have on average a ADHD diagnosis. For this study, we selected a subsample lower IQ than controls [44-46]. It has been suggested that of first- and second-borns to rule out dependency of the this lower IQ may underlie cognitive dysfunctioning, or data. In the first-borns, 143 affected children, 68 non- vice versa, that cognitive dysfunctioning is at the heart of affected siblings and 120 controls participated. In the sec- a lower intelligence, or that there is no hierarchical rela- ond-borns, 132 affected children, 78 non-affected sib- tion between both domains but both domains share com- lings, and 113 controls participated. All children were mon causes [47-49]. Either way, since IQ seems both between the ages of 5 and 19 years and were of European associated with the neurocognitive dysfunctioning as well Caucasian descent. Participants were excluded, if they had as with ADHD behaviour, a mediating effect of IQ may be an IQ < 70, a diagnosis of autism, epilepsy, general learn- expected on the relation between neurocognitive dysfunc- ing difficulties, brain disorders or known genetic disor- tioning and ADHD. ders, such as Down syndrome or Fragile-X-syndrome. An additional factor that may influence the relation The screening procedures and measures for phenotyping between endophenotype and phenotype may be rater have been described elsewhere [50]. Briefly, screening bias: Parents may underestimate the severity of ADHD questionnaires (parent and teacher Conners' long version symptoms in their undiagnosed/non-affected child and rating scales [51] parent and teacher Strengths and Diffi- overestimate the severity of ADHD symptoms in their culties Questionnaires [52] and parent Social Communi- diagnosed/affected child. Teachers, however, may be less cation Questionnaire [53]) were used to identify children likely to be affected by this contrast effect, since they often with ADHD symptoms and to screen for any autistic like do not have siblings of the same family in their class due behaviours. Scores were indicative for a diagnosis of to age differences between the siblings. Rater bias explain- ADHD if T-scores were ≥ 63 on the Conners' ADHD-sub- ing the non-comparable magnitude of group differences scales (DSM-IV Inattention, DSM-IV Hyperactive-Impul- th at the endophenotypic and phenotypic level may thus be sive, and DSM-IV ADHD Total) and > 90 percentile on investigated by comparing the relation between endophe- the SDQ-Hyperactivity scale. A score of ≥ 15 on the SCQ notype and phenotype as observed by parents and teach- was considered indicative of autistic like behaviours. Par- ers. ents and teachers were asked to rate the behavior of the child when off medication. Concerning all children rated The first aim was to investigate if an endophenotypic con- clinically on any of the questionnaires completed either struct, encompassing a broad battery of both cognitive by parents or teachers, the Parental Account of Children's and motor endophenotypes, is predictive of the ADHD Symptoms (PACS) was administered [54]. Data from the diagnosis. The second aim was to confirm apparent obser- questionnaires and the PACS were subjected to a stand- vations that children with ADHD show more severe ardised algorithm to derive each of the 18 DSM-IV ADHD ADHD symptoms than one would expect based on their symptoms, providing operational definitions for each endophenotypic dysfunctioning, whereas non-affected behavioural symptom [30]. With respect to control chil- siblings show less ADHD-like behaviour than one would dren, the Conners' long version for both parents and expect based on their endophenotypic vulnerabilities. The teachers was completed and all control children were third aim of this study was to explore the (moderating and required to obtain non-clinical scores. mediating) effects of four factors on the relationship between endophenotype and phenotype: gender, age, IQ, Full-scale IQ was estimated by four subtests of the and rater bias. Wechsler Intelligence Scale for Children III (WISC-III) or Wechsler Adult Intelligence Scale III (WAIS-III) (depend- Methods ing on the child's age): Vocabulary, Similarities, Block Sample design and Picture completion [55,56]. These subtests are Families with at least one child with the combined sub- known to correlate between .90–.95 with the Full-scale IQ type of ADHD (proband) and at least one additional sib- [57]. IQ testing took place while the children were off ling (regardless of possible ADHD-status) were recruited medication. in order to participate in the Dutch part of the Interna- tional Multicenter ADHD Genes study (IMAGE). The Procedure IMAGE project is an international collaborative study that Testing of children with ADHD and their siblings took aims to identify genes that increase the risk for ADHD place at the VU University Amsterdam or at the Radboud using QTL linkage and association strategies [50]. Addi- University Nijmegen Medical Centre and was conducted tional control families were recruited from primary and simultaneously for all children in a family. Psychostimu- high schools in the same geographical regions as the par- lants were discontinued for at least 48 hours before testing ticipating ADHD-families. Controls and their first degree took place [58]. Children were motivated with small Page 4 of 14 (page number not for citation purposes) Behavioral and Brain Functions 2008, 4:4 http://www.behavioralandbrainfunctions.com/content/4/1/4 breaks. At the end of the session, a gift worth approxi- inhibition) and third (cognitive flexibility) block could be mately € 4, – was given. Control children were tested in a compared. In all blocks, trials consisted of a horizontal similar way in a quiet room at their school. The study had bar with ten grey squares presented permanently at the medical-ethical approval by the local ethics committee centre of the screen. From trial to trial, a coloured square and was in accordance with the Helsinki Declaration. moved across the bar in a random direction (either one square to the right or to the left). Responses were required Experimental tasks to be initiated between 150 to 5000 ms after a square The ten experimental tasks described in this study have moved one position, otherwise a trial was replaced. The been fully described elsewhere [29-31,33,34]. A short task was self-paced with post-response intervals of 250 description of each task will be given below. Based on pre- ms. In the first block, the moving square was coloured vious results [29-31,33,34], the variable per task that green, and compatible responses were required: children showed most optimal results in the endophenotypic anal- were instructed to press a response button as quickly and yses in the five previous studies was chosen for the current as accurately as possible that corresponded to the direc- analyses. tion in which the stimulus moved. In the second block, the moving square was coloured red, and incompatible Stop task responses were required. The suppression of the auto- The Stop task was used to measure speed of inhibition of matic compatible response, in order to generate a non- an ongoing response [59,60]. Subjects were presented two automatic incompatible response, was hypothesized as types of trials: go-trials and stop-trials. Go-trials consisted requiring inhibitory control. In the third block, the colour of the presentation of a go-stimulus (drawing of a plane) of the moving square alternated randomly between green that was either pointing to the right or to the left [61]. and red, and both compatible and incompatible Children were instructed to press a response button that responses were required. Thus, both the direction and the corresponded to the direction of the stimulus as quickly colour of the square were unpredictable. The mixture of and as accurately as possible. Stop-trials were identical to both compatible and incompatible trials was hypothe- the go-stimulus but in addition a stop-signal was pre- sized to require high levels of cognitive flexibility in addi- sented (drawing of a cross that was superimposed on the tion to inhibitory control [63]. The first and second block plane). Children were required to withhold their response consisted of 10 practice trials and 40 experimental trials. to the stop-signal. Go stimuli were displayed for 1000 ms, The third block consisted of 16 practice trials and 80 preceded by a 500 ms fixation point. Stop signals were dis- experimental trials. Administration took about 10 to 15 played for 1000 ms minus delay time, with a mean dis- min. The dependent measure was the percentage of errors play time of 709 ms (SD 125 ms). Inter-trial intervals were across blocks, which was the best indicator of endopheno- 3000 ms. The delay between the go- and stop-signal was typic vulnerabilities [31]. dynamically varied so that it could be estimated when the child successfully inhibited 50% of the stop-trials, and Time reproduction unsuccessfully inhibited the other 50%. At this point, the The Timetest application version 1.0 [64] was used to go-process and stop-process were of equal duration, measure time reproduction. Stimuli consisted of temporal which made it possible to estimate the latency of the stop- intervals with different durations (4, 8, 12, 16, 20 s) that process: the Stop signal reaction time (SSRT) [59]. A total had to be reproduced as accurately as possible. The task of 2 practice blocks and 4 experimental blocks were was administered first in the visual modality (light bulb) administered, each consisting of 60 trials. The first prac- and thereafter in the auditory modality (tone). Children tice block consisted of only go-trials. The second practice were not informed about the length of the intervals. In block and the 4 experimental blocks consisted of 75% go- both modalities, 3 practice and 20 experimental trials trials and 25% stop-trials. Go- and stop-trials were were administered. The five interval lengths were ran- pseudo-randomly presented. Task administration took domly presented four times. Task administration for both about 15 min. Based on previous results, the dependent modalities required 15 min. Based on previous results, the measure was the SSRT [29], which showed endopheno- main dependent measure was the precision of the repro- typic-like group differences and correlated between sib- duction (operationalized as the absolute discrepancy lings. between the response length and the stimulus length) averaged across trials and modalities, which was abnor- Shifting attentional set mal in children with ADHD and their non-affected sib- Shifting attentional set was designed to measure accuracy lings and correlated between siblings [30]. of motor inhibition and cognitive flexibility [62]. The task consisted of three blocks of which the first block was Visuo-spatial sequencing designed to acquire a baseline of the accuracy of respond- The Visuo-spatial sequencing task was used to measure ing with which the performance on the second (motor accuracy of visuo-spatial working memory [62]. Stimuli Page 5 of 14 (page number not for citation purposes) Behavioral and Brain Functions 2008, 4:4 http://www.behavioralandbrainfunctions.com/content/4/1/4 consisted of nine circles symmetrically organized in a invisible midline (radius 8 cm) between the inner and square (3 by 3). On each trial, a sequence of circles was outer circle as quickly and precisely as possible with a pointed at by a computer-driven hand. Subjects were mouse cursor. One practice and one experimental session instructed to replicate the exact same sequence of circles, were administered for both hands separately (clockwise by pointing to them with the small, self-driven hand. with the right hand and counter clockwise with the left There were no time constrictions. One practice trial and hand). Administration took about 3 min. The dependent 24 experimental trials were presented. Every succeeding measure was the precision (mean distance to midline in trial increased in difficulty level: an increase in the mm averaged across 60 equal parts of the circle) of the left number of circles required to be remembered and/or an hand. Previous results have shown that precision of the increase in the complexity of the spatial pattern (i.e. the non-dominant hand showed endophenotypic-like char- trial consisted of circles that were spatially further acteristics [33]. removed from one another instead of being close to one another), hence manipulating working memory Tapping demands. Task administration took about 7 min. Based This task measured variability of self-generated motor out- on previous results, the total number of correct targets in put [62]. This task required the child to tap as frequently the correct order was used as dependent measure reflect- as possible within a certain time period. During tapping, ing endophenotypic-like group differences and correlat- the number of taps was continuously counted and dis- ing between siblings [29]. played on the screen. One practice session (5 s) and one experimental session (18 s) were administered for both Digit span hands separately. The task was first practised and executed The Digit span backwards of the WISC-III and WAIS-III with the index finger of the non-preferred hand, thereafter [55,56] was used to obtain an indication of verbal work- practised and executed with the index finger of the pre- ing memory. The backward part consisted of repeating a ferred hand. Administration took about 3 min. The sequence of numbers in the opposite order. Children were dependent measure was the variability (SD of intertap instructed to reproduce sequences as accurately as possi- intervals in ms) averaged across hands. Previous results ble. One digit was added to the sequence if a child repro- have shown that this measure correlates between siblings duced the sequence successfully. Two practice trials with a [34]. 2 digit sequence and (dependent on the child's perform- ance) a maximum of 8 experimental sequences were Baseline speed administered. Dependent measure was the maximum This task was designed to measure variability on a simple Digit span backwards, which proved useful as endophe- reaction time task [62]. Stimuli consisted of a fixation notypic candidate [29]. cross in the centre of a computer screen that changed unpredictably into a white square. Immediately following Pursuit the response, the white square changed back into the fix- This task was designed to measure precision of motor con- ation cross. The time interval between a response and the trol under continuous adaptation [62]. The stimulus con- emergence of the next white square varied randomly sisted of a randomly moving target (asterisk) that was between 500 to 2500 ms in order to prevent anticipation required to be 'caught' by moving a mouse cursor on top strategies. Subjects were required to press a key as quickly of the asterisk. The target moved at a constant speed of 10 as possible when the white square appeared. A practice mm/s. Children were instructed to 'catch' the randomly session (10 trials) and an experimental session (32 trials) moving target as precisely as possible. One practice (13 s) were administered for both hands separately. The task was and one experimental session (60 s) were administered first practised and executed with the index finger of the for both hands separately. Administration took about 5 non-preferred hand, thereafter practised and executed min. The dependent measure was the precision (mean with the index finger of the preferred hand. Administra- distance in mm between target and cursor calculated per tion took about 5 min. Dependent measure was the vari- second and averaged across the 60 s experimental session) ability (SD of reaction times in ms) of responses averaged of the left hand. Previous results have shown that mainly across hands. Previous results have shown that this meas- the performance of the non-dominant hand was most ure was associated with ADHD and correlated between strongly associated with ADHD [33]. siblings [34]. Tracking Motor timing This task aimed to measure precision of motor control This task was designed to measure variability of motor without continuous adaptation required [62]. The stimu- timing [65]. In this task a 1 s interval had to be produced. lus consisted of an inner and outer circle (radius 7.5 and The start of the interval was announced by a tone (80 db, 8.5 cm, respectively). Children were instructed to trace an 50 ms). After the subject's response, visual feedback was Page 6 of 14 (page number not for citation purposes) Behavioral and Brain Functions 2008, 4:4 http://www.behavioralandbrainfunctions.com/content/4/1/4 given, indicating whether the response was correct, too total raw score was averaged across parents and teachers short or too long. A response was regarded as correct, if it and labelled 'the phenotype'. fell between the lower and upper boundary set by a dynamic tracking algorithm. Boundaries were set at 500 to The first aim of this paper was to investigate whether the 1500 ms at the beginning of the task. If the response fell endophenotypic construct was predictive of the ADHD within these boundaries, the boundaries for the subse- diagnosis (affected, non-affected, control) using a multi- quent trial were narrowed by 100 ms. Likewise, the ple discriminant analysis. Age was also entered in the boundaries of the subsequent trial were widened with 100 model, because age differences had arisen between the ms, if the response on the previous trial fell outside those first- and second-borns by splitting up the sample. There boundaries. Subjects were instructed to produce as accu- was assessed what percentage of children was correctly rately as possible the 1 s interval. Twenty practice trials classified by the endophenotypic construct. and 80 experimental trials were administered. Both ses- sions were preceded by presenting 10 times a cartoon fig- The second aim of this paper was to test whether children ure for exactly 1 s on the screen to demonstrate the with ADHD showed a more severe phenotype than one duration of 1 s [65]. Administration took about 8 min. would expect based on their endophenotypical dysfunc- The dependent measure was the variability (SD of produc- tioning and whether the reverse was true for the non- tions in ms). Previous results have shown that this meas- affected siblings. Therefore, the interaction of group ure to be a viable endophenotypic candidate [34]. (affected, non-affected, control) by functioning (2 within subject levels: endophenotypic construct and phenotype) Data analyses was tested in a repeated measures ANOVA. Age was again Missing data for all variables for the sample described in implemented as covariate. this study was less than 5% and were replaced by means of expectation maximization [66]. The task measures were The third aim of this paper was to investigate whether the successfully normalized and standardized to z-scores by factors gender, age, and IQ may moderate and/or mediate applying a Van der Waerden transformation (SPSS version the relation between the endophenotypic construct and 14). Some of the z-scores were multiplied by -1, so that the the phenotype. Using regression analyses, significant z-scores of all task variables would have the same mean- moderation for a factor would be demonstrated if the pre- ing: A higher z-score was indicative of poor performance dictive effect of the endophenotypic construct was not or underestimation. Correction for multiple comparisons constant for that factor and was tested by examining the according to the false discovery rate (FDR) controlling interaction between the endophenotypic construct and procedure was applied to the analyses with a q-value set- the factor in predicting the phenotype [37]. Mediation ting of 0.05 [67]. was tested by calculating the path coefficients between (a) the endophenotypic construct and the factor, (b) the fac- In order to rule out dependency of the data (more than tor and the phenotype, and (c) the endophenotypic con- one child per family participated in the study), analyses struct and phenotype (see Figure 2). The effect of were performed with the children split by birth order. mediation (in which endophenotypical dysfunctioning First, analyses were run on the data from the first-borns (N lead to ADHD through gender, age, and/or IQ) is termed = 331), thereafter repeated on the second-born children the indirect effect [36] and can be tested using the formula 2 2 2 2 s + a s ), in which a represents (N = 323). In this way, not only was the dependency of [68]: z-value = a × b/√(b a b data handled, but this also gave the opportunity to study the unstandardized path coefficient of the endopheno- replicability of the results. Since most families consisted typic construct on the mediating factor, b represents the of two children, the samples of third- and fourth-borns unstandardized path coefficient of the mediating factor 2 2 were substantially smaller (N = 76 and N = 11, respec- on the phenotype, s and s represent the square of the a b tively) than the samples of first- and second-borns. These standard error of the path coefficients a and b. It was sample ratio differences were considered too large (> 4) examined whether the direct effect between the endophe- and therefore, the third- and fourth-borns were excluded notypic construct and phenotype was still significant after from the analyses. controlling for the indirect effect. To combine the task variables to a component that would Rater bias could not be examined for moderation and simplify the analyses and reduce error variance, a princi- mediation, since the phenotype was confounded with the pal component analysis was performed on the ten task moderating or mediating factor. Therefore, a different variables, separately for the first- and second-borns. The approach was taken. Paired sample t-tests were used to component explaining the largest amount of variance was compare parental and teacher ratings of ADHD to assess used in the further analyses and is labelled 'the endophe- whether the phenotype differed when rated by parents or notypic construct'. The standardized Conners' ADHD teachers. Additionally, in order to investigate whether the Page 7 of 14 (page number not for citation purposes) direct effect Behavioral and Brain Functions 2008, 4:4 http://www.behavioralandbrainfunctions.com/content/4/1/4 Results en dophenotype The percentage of affected, non-affected and controls was equally distributed across the first- and second-borns (χ = 1.24, df = 2, p = .54). The characteristics of both samples are presented in Table 1. Principal component analysis A principal component analysis was performed on the ten task measures, separately for the first- and second-borns. mediating fact or Similar results were obtained in both samples: All ten task measures related to one major component, explaining respectively 40.5% and 47.0% of the variance in the task gender, age, IQ measures. Additional components did not have an eigen- value above 1 and explained only 10.3% or less additional variance. Therefore, the following results report only the main component labelled 'the endophenotypic con- struct'. This endophenotypic construct was representative of endophenotypic functioning across ten tasks, because all tasks loaded on this component score (Table 2). Results were similar, when the analysis was repeated using the raw (not standardized and not normalized) task vari- ables, with a one-component solution explaining 34.9% and 41.5% of the variance in the first- and second-borns respectively, and with all variables correlated with this phen otype component (Table 2). Dis b Figure 2 etween play of en po dophenot ssible medi ype ati and phenotype ng effects of gender, age, and IQ First aim: Predictive validity of the endophenotypic Display of possible mediating effects of gender, age, and IQ construct for the ADHD diagnosis between endophenotype and phenotype. A multiple discriminant analysis was performed with age and the endophenotypic construct as predictors and diag- nosis as the grouping variable. To correct for the unequal predictive validity of the endophenotypic construct was group sizes, prior probabilities for all groups were set to 1/ different for the phenotype as rated by the parents or by 3. The endophenotypic construct significantly predicted the teachers, the correlation coefficients between the diagnostic status (first-borns F (2, 328) = 33.24, p < .001; endophenotypic construct – phenotype rated by parents second-borns F (2, 320) = 29.73, p < .001). In all groups, on the one hand and the endophenotypic construct – phe- correct classification percentages were roughly around notype rated by teachers on the other hand were com- 50%. In the first-borns, respectively 55%, 52%, and 48% pared using two-sided dependent samples t-tests [69]. of the affected children, non-affected siblings, and con- trols were correctly classified. In the second-borns, respec- Table 1: Sample characteristics first-borns (N = 331) second-borns (N = 323) affected non-affected control affected non-affected control N = 143 (43%) N = 68 (21%) N = 120 (36%) N = 132 (41%) N = 78 (24%) N = 113 (35%) 1 1 MSD M SD M SD F contrasts MSD M SD M SD F contrasts 2,328 2,320 age in years 13.2 2.6 14.9 2.7 13.3 2.7 10.9 2 > 1 = 3 11.6 2.7 9.7 2.4 10.6 2.9 11.3 1 > 2 = 3 2 2 % male 76.9 45.6 44.2 34.8 1 > 2 = 3 75.8 48.7 38.1 37.4 1 > 2 = 3 estimated full scale IQ 99.8 11.8 101.8 10.7 106.2 9.9 11.3 1 = 2 < 3 99.2 11.8 105.5 10.9 105.6 10.1 13.2 1 < 2 = 3 Conners' parent DSM-IV inattentive 68.6 10.0 48.6 7.2 46.3 4.5 306.6 1 > 2 = 3 69.7 10.4 46.8 6.5 46.7 4.9 329.9 1 > 2 = 3 hyperactive-impulsive 75.4 12.9 48.4 6.9 47.3 4.9 343.1 1 > 2 = 3 76.0 11.3 49.6 6.8 47.6 5.2 410.0 1 > 2 = 3 Conners' teacher DSM-IV inattentive 64.1 9.5 48.8 6.4 46.2 4.7 211.4 1 > 2 = 3 65.7 9.3 48.0 5.6 46.3 4.2 279.0 1 > 2 = 3 hyperactive-impulsive 68.2 11.7 48.5 7.7 46.3 3.9 236.2 1 > 2 = 3 69.0 11.5 47.7 4.7 47.2 5.5 260.5 1 > 2 = 3 1 2 2 Note. 1 = affected; 2 = non-affected; 3 = controls. = contrasts based on p-values of ≤ .05. = χ . th ADHD = Attention-Deficit/Hyperactivity Disorder; DSM-IV = Diagnostic and Statistical Manual for Mental Disorders (4 edition). Page 8 of 14 (page number not for citation purposes) indirect effect Behavioral and Brain Functions 2008, 4:4 http://www.behavioralandbrainfunctions.com/content/4/1/4 Table 2: Correlations between the individual task variables and the endophenotypic construct first-borns second-borns Task Measure z-scores raw data z-scores raw data Stop task Stop signal reaction time .66 .63 .67 .64 Shifting attentional set % errors .63 .67 .67 .71 Time reproduction Absolute deviation .70 .59 .74 .67 Visuo-spatial sequencing N identified targets correct order .73 .72 .79 .80 Digit span N backwards .55 .53 .69 .69 Pursuit Precision .74 .64 .77 .65 Tracking Precision .60 .55 .54 .48 Tapping Variability .46 .50 .49 .47 Baseline speed Variability .52 .48 .70 .62 Motor timing Variability .71 .56 .73 .52 tively 67%, 41%, and 45% of the affected children, non- Age affected siblings, and controls were correctly classified. Age had a moderating effect on the relation between the endophenotypic construct and phenotype in the second- Second aim: Group differences at an endophenotypic and borns (t = 2.92, β = .58, p = .004), though this effect was phenotypic level completely non-significant in the first-borns (t = -.02, β = The group by level interaction was analyzed to assess -.004, p = .99). The significant interaction between the whether group differences were comparable at the endo- endophenotypic construct and age in predicting ADHD in phenotypic and phenotypic level. This interaction was sig- second-borns appeared to be related to the non-affected nificant (first-borns F (2, 327) = 45.46, p < .001, η = .22; siblings: Younger non-affected siblings performed in second-borns F (2, 319) = 56.89, p < .001, η = .26), sug- between their affected siblings and controls (non-affected gesting group contrasts to be different at the endopheno- versus affected p < .001 and non-affected versus control p typical and phenotypical level. When the analysis was = .002), but older non-affected siblings performed more repeated with affected children and controls, the interac- like controls (non-affected versus affected p < .001 and tion remained significant (first-borns F (1, 260) = 67.68, non-affected versus control p = .54). Strong evidence was p < .001, η = .21; second-borns F (1, 242) = 92.13, p < found for a partially mediating effect of age: The endophe- .001, η = .28). As is visible in Figure 3, affected children notypic construct related to age, and age related to ADHD deviated more from controls at the phenotypic level than severity (Table 3). Correcting for this partial mediating at the endophenotypic level. No such interaction was present for non-affected siblings compared to controls (first-borns F (1, 185) = 0.24, p = .63, η < .01; second- borns F (1, 188) = 0.02, p = .89, η < .01). Third aim: Moderating and mediating effects of gender, age, and IQ and the influence of rater bias on the relation between endophenotypic construct and phenotype Gender No evidence was found for a moderating effect of gender (first-borns t = -.34, β = -.05, p = .73; second-borns t = -.41, β = -.06, p = .69), which suggested that the relation between the endophenotypic construct and phenotype was comparable for boys and girls. There was evidence, however, for a partially mediating effect of gender in the first-borns (Table 3). Part of the relationship between the endophenotypic construct and ADHD was related to gen- der: Boys performed slightly worse than girls on the endo- Group di iz phenot and teacher A Figure 3 ed as a ypic level (operat com fferences at the endo pD os H ite D s qc uor estionna ie o onalized f te ph ire) n enotypic level a task variables) and at the s a composite of pa (operational- rental phenotypic construct and boys had more severe ADHD Group differences at the endophenotypic level (operational- phenotypes than girls. ized as a composite score of ten task variables) and at the phenotypic level (operationalized as a composite of parental and teacher ADHD questionnaire). Page 9 of 14 (page number not for citation purposes) Behavioral and Brain Functions 2008, 4:4 http://www.behavioralandbrainfunctions.com/content/4/1/4 Table 3: Mediating effects of gender, age, and IQ on the relation between the endophenotypic construct and phenotype mediator endophenotype → mediator mediator → phenotype endophenotype → phenotype test of mediation: endophenotype → phenotype 2 2 2 2 z = a × b/√(b s + a s ) corrected for mediation a b as pb s pc s pz p c s a b c c gender first-borns -0.129 .026 <.001 -0.319 .108 .003 7.215 .657 <.001 2.538 .01 5.963 0.629 <.001 second-borns -0.020 .028 .47 -0.285 .113 .01 2.390 .771 .002 0.687 .49 2.148 0.697 age first-borns -1.724 .117 <.001 -1.383 .271 <.001 7.215 .657 <.001 4.822 <.001 8.041 0.846 <.001 second-borns -2.130 .102 <.001 0.889 .275 .001 2.390 .771 .002 -3.195 .001 10.150 1.044 <.001 IQ first-borns -2.264 .608 <.001 -0.255 .067 <.001 7.215 .657 <.001 2.662 .008 6.916 0.667 <.001 second-borns -1.222 .633 .05 -0.328 .066 <.001 2.390 .771 .002 1.799 .07 2.012 0.752 effect of age did not, however, result in a non-significant dictive validity of the endophenotypic construct was direct effect between the endophenotypic construct and comparable for the phenotype as rated by the parents or phenotype in ADHD. by the teachers. IQ Discussion IQ did not have a moderating influence (first-borns t = - This study aimed at examining the predictive validity of 1.11, β = -.49, p = .27; second-borns t = -.81, β = -.37, p = endophenotypic functioning for the ADHD diagnosis, the .42), indicating that the relation between the endopheno- magnitude of group differences at an endophenotypic and typic construct and phenotype to be constant across the phenotypic level, and the mediating and moderating IQ range studied here. However, there was a partial medi- effects of gender, age, and IQ as well as the influence of ating effect of IQ (Table 3), suggesting the endopheno- rater bias on the relation between endophenotype and typic performance related to IQ, and IQ related to ADHD. phenotype. Analyses were separately conducted on first- Correcting for this partial mediating effect of IQ did not and second-borns to accommodate for the non-independ- result in a non-significant direct route between the endo- ency of data. First- and second-borns did not differ with phenotypic construct and phenotype. respect to proportion of affected, non-affected and control children, and results were largely consistent across first- Rater bias and second-borns, suggesting ADHD was not related to Paired samples t-tests indicated that parents rated the birth-order as has been previously observed [70]. ADHD symptoms of their affected child as more severe than did teachers (first-borns t = 4.92, p < .001; second- Concerning the first aim of our study: The endopheno- borns t = 5.32, p < .001), which was in line with our expec- typic construct significantly predicted the diagnostic sta- tations. However, in contrast to our expectations, parents tus (affected, non-affected, and control), which is in line and teachers ratings agreed on the degree of ADHD-like with the aetiology of psychiatric disorders, in which endo- behaviour in non-affected children (first-borns t = -.51, p phenotypic vulnerabilities lead to phenotypic symptoms. = .61; second-borns t = .21, p = .84), which is similar to the The status of all groups was predicted with roughly similar findings in control children (first-borns t = .92, p = .36; percentages (around 50%). However, these percentages second-borns t = .86, p = .39). It appeared that parents were only moderate and not high, indicating a substantial were indeed possibly 'overestimating' the degree of overlap in endophenotypic functioning of children with ADHD in their affected child, but they did not seem to different diagnostic status. Thus, even an aggregated com- 'underestimate' the degree of ADHD in their non-affected ponent encompassing multiple endophenotypic meas- child. ures is not a firm predictor of diagnostic status and illustrates the causal heterogeneity of ADHD [13]. None of the correlation coefficients between the endo- phenotypic construct and ADHD rated by parents, on the In the second aim of our study, it was confirmed that one hand, and the endophenotypic construct and ADHD affected children indeed portrayed a more severe ADHD rated by teachers, on the other hand, differed significantly phenotype than one would expect based on their endo- from each other (t-values between 0.47 – 2.47 and p-val- phenotypic dysfunctioning. This is in line with previous ues between .36 – .98), when calculated separately for studies on cognitive dysfunctioning in patients with affected, non-affected, and control children or when cal- ADHD, in which not all patients portray neurocognitive culated across groups. These findings suggest that the pre- dysfunctioning and effect sizes are generally modest [71]. Page 10 of 14 (page number not for citation purposes) Behavioral and Brain Functions 2008, 4:4 http://www.behavioralandbrainfunctions.com/content/4/1/4 Based on previous endophenotypic studies, we expected recting for this partially mediating effect of age did not to find the reverse pattern for non-affected siblings, por- result in a non-significant direct effect between the endo- traying a less severe phenotype compared to their endo- phenotypic construct and phenotype. The findings of phenotypic construct. This appeared not to be the case: non-moderation of age in affected children are in line The difference between non-affected siblings and controls with studies showing symptoms of ADHD persist into was comparable for the endophenotypic construct and adolescence and adulthood for the majority of patients phenotype. This suggests that endophenotypic dysfunc- [40,74] and suggest that the effect of age does not appear tioning related proportionally to (subtle) phenotypical to contribute to the group differences at the endopheno- deviations in non-affected children, but other factors pos- typic and phenotypic levels. sibly come into play that aggravate the eventually observ- able phenotypical problems in affected children. These Like gender and age, IQ also did not appear to account for factors may be environmental in nature, such as differ- these non-comparable group differences, since IQ did not ences in upbringing and schooling, which apparently had have a moderating influence on the relationship between a more positive outcome for the non-affected sibling than the endophenotypic construct and phenotype. IQ did for the affected sibling. It may also be that certain pheno- have a mediating effect, but correcting for this mediating typic symptoms may aggravate other phenotypic symp- effect did not result a non-significant direct route between toms, resulting in a disproportionate severe phenotype in the endophenotypic construct and phenotype, suggesting affected children in relation to their endophenotype. For the influence of IQ on this relation is not substantial. example, a child being hyperactive, may become even These findings give support to the idea that endopheno- more hyperactive because his/her inattention during a typic dysfunctioning and a lower IQ are both related to school-task results in the child leaving his/her seat. each other and to ADHD [47-49], though a lower IQ can not account for the endophenotypic dysfunctioning asso- In line with the idea that some factors may influence the ciated with ADHD. relation between endophenotype and phenotype, we ana- lyzed the (moderating and mediating) effects of gender, Another factor we studied was the effect of rater bias on age, IQ, and rater bias. No evidence was found for a mod- the relation between the endophenotypic construct and erating effect of gender, which suggests that the relation phenotype. We hypothesized that parents, compared to between the endophenotypic construct and phenotype in teachers, may possibly overrate the ADHD severity in their ADHD is comparable for boys and girls. There was evi- affected children, whereas they may underrate symptoms dence, though, for a partially mediating effect of gender in in their non-affected children. This, in turn, might explain the first-borns. However, since this partial mediation was the disproportionate severe phenotype of the affected completely non-significant in the second-borns, and the group compared to their endophenotypic construct. Par- direct effect between the endophenotypic construct and ents indeed rated the degree of ADHD in their affected phenotype was still present after correcting for this par- children more severely than teachers. This may be because tially mediating effect of gender, it appears that gender did parents wanted to secure their participation in the study not have a large impact on the relation between the endo- [75] and/or because the problem behaviour was genu- phenotypic construct and phenotype in ADHD. It could inely (experienced as) more severe at home than at not account for the apparent disproportionately severe school, possibly because the use of symptom reducing phenotype in affected children. Similar non-influential medication is more consistent and/or effective at school effects of gender in relation to ADHD have been docu- [75,76]. Either way, although parents did rate the severity mented previously [72,73]. of their child's ADHD as more serious than teachers, a possible effect of rater bias could not explain the larger Another factor possibly influencing the endophenotypic phenotypic dysfunctions compared to endophenotypic construct – phenotype relation is age. Age moderated the dysfunctions in affected children, because the endophe- relation between the endophenotypic construct and phe- notypic construct related comparably to the phenotype notype in the second-borns, though not in the first-borns. when rated by parents and teachers. Furthermore, no evi- This significant interaction between the endophenotypic dence was found for an underestimation of ADHD by par- construct and age in predicting the ADHD phenotype ents in their non-affected siblings. Therefore, possible appeared to be related to the non-affected siblings: differences in parental and teacher ratings of ADHD do Younger non-affected siblings performed in between their not underlie the disproportionate severe phenotype com- affected siblings and controls, but older non-affected sib- pared to endophenotype in children with ADHD. lings performed more like controls, possibly suggesting endophenotypic vulnerabilities ease somewhat with Limitations increasing age in non-affected siblings. Strong evidence Several possible limitations of this study warrant consid- was found for a partial mediating relation of age, but cor- eration. Dividing the sample in first- and second-borns Page 11 of 14 (page number not for citation purposes) Behavioral and Brain Functions 2008, 4:4 http://www.behavioralandbrainfunctions.com/content/4/1/4 doubled the number of statistical tests that were per- ant for Eli Lilly, McNeil Consumer & Specialty Pharma- formed. However, we corrected for multiple testing and ceuticals, Shire, Noven Pharmaceuticals, and Cephalon; this approach gave the opportunity to investigate replica- has been on the speakers' bureaus of Eli Lilly, McNeil bility of the results and showed that almost all findings Consumer & Specialty Pharmaceuticals, Shire, and were comparable in first- and second-borns. Another pos- Cephalon. JB has been a consultant to/member of advi- sible limitation was the interpretability of the endopheno- sory board of/and/or speaker for Janssen Cilag BV, Eli typic component measure. Since it was a composition of Lilly, Bristol-Myer Squibb, UBC, Shire, Medice. JS has various cognitive and motor task variables, its exact repre- been a member of advisory board of Eli Lilly, Shire, Jans- sentation remains unclear. It may be hypothesized that sen Cilag. This study was partly funded by a grant assigned the endophenotypic construct taps into an underlying fac- to SF by the National Institute of Mental Health (NIH tor, which may represent general cognitive functioning grant # R01 MH62873-01A1). ('g') possibly in combination with variability of reacting. We feel though, that combining individual task measures Authors' contributions to one more robust component will facilitate heritability NR contributed to the data ascertainment, study design, research in ADHD, since a component probably entails literature searches, analyses, and writing. MA and CB con- less error variance and may be a more reliable measure tributed to the data ascertainment and writing. NM con- than individual task measures. tributed to the analyses and writing. SF, JB, JS and JO have written the study protocol and have contributed to the Conclusion writing. All authors have read and approved the final An endophenotypic construct encompassing multiple manuscript. endophenotypic measures is moderately predictive of diagnostic status, but substantial overlap exists between Acknowledgements The authors thank all of the parents, teachers and children who partici- endophenotypic functioning in the groups of affected pated. children, non-affected siblings and controls. Group differ- ences at an endophenotypic and phenotypic level are not References comparable for affected children, displaying a more severe 1. American Psychiatric Association: Diagnostic and Statistical phenotype than one would expect based on their endo- Manual for Mental Disorders. 4th edition. Washington, DC: phenotype when compared to controls. Group differences American Psychiatric Press; 1994. 2. Clark LA: Toward a consensual set of symptom clusters for were comparable for non-affected siblings compared to assessment of personality disorder. In Advances in personality controls, suggesting subtle endophenotypic vulnerabili- assessment Volume 8. Edited by: Butcher JN, Spielberger CD. Hillsdale, NJ: Lawrence Erlbaum; 1990:243-266. ties translate proportionally into phenotypic deviations. 3. Faraone SV, Tsuang MT: Measuring diagnostic accuracy in the Even though a potentially moderating effect (age) and absence of a "gold standard". Am J Psychiatry 1994, 151:650-657. several mediating effects (gender, age, IQ) have been 4. Hudziak JJ, Heath AC, Madden PF, Reich W, Bucholz KK, Slutske W, Bierut LJ, Neuman RJ, Todd RD: Latent class and factor analysis found affecting the relation between the endophenotypic of DSM-IV ADHD: a twin study of female adolescents. J Am construct and phenotype, none of the effects studied (gen- Acad Child Adolesc Psychiatry 1998, 37:848-857. der, age, IQ, and rater bias) could account for the finding 5. Kendler KS, Eaves LJ: Models for the joint effect of genotype and environment on liability to psychiatric illness. Am J Psychi- that affected children deviated more from controls at the atry 1986, 143:279-289. phenotypic than endophenotypic level. These findings 6. Kendler KS, Prescott CA, Myers J, Neale MC: The structure of genetic and environmental risk factors for common psychi- suggest other factors come into play and aggravate the atric and substance abuse disorders in men and woman. Arch phenotype in affected children. Gen Psychiatry 2003, 60:929-937. 7. Gottesman II, Gould TD: The endophenotype concept in psy- chiatry: etymology and strategic intentions. Am J Psychiatry Abbreviations 2003, 160:636-645. ADHD = Attention-Deficit/Hyperactivity Disorder; DSM- 8. Waldman ID: Statistical approaches to complex phenotypes: evaluating neuropsychological endophenotypes for atten- IV = Diagnostic and Statistical Manual for Mental Disor- tion-deficit/hyperactivity disorder. Biol Psychiatry 2005, th ders 4 edition; IMAGE = International Multicenter 57:1347-1356. ADHD Genes study; IQ = Intelligence Quotient; PACS = 9. Kuntsi J, Neale BM, Chen W, Faraone SV, Asherson A: The IMAGE project: methodological issues for the molecular genetic Parental Account of Children's Symptoms; SCQ = Social analysis of ADHD. Behav Brain Funct 2006, 2:27. Communication Questionnaire; SDQ = Strengths and 10. Faraone SV, Biederman J: Neurobiology of attention-deficit Difficulties Questionnaires; SSRT = Stop Signal Reaction hyperactivity disorder. Biol Psychiatry 1998, 44:951-958. 11. Almasy L, Blangero J: Endophenotypes as quantitative risk fac- Time. tors for psychiatric disease: rationale and study design. Am J Med Genet 2001, 105:42-44. 12. Castellanos FX, Tannock R: Neuroscience of attention-deficit/ Competing interests hyperactivity disorder: the search for endophenotypes. Nat NR, MA, NM, CB, and JO have no competing interests. SF Rev Neurosci 2002, 3:617-628. has received grants from Eli Lilly, McNeil Consumer & 13. Doyle AE, Faraone SV, Seidman LJ, Willcutt EG, Nigg JT, Waldman ID, Pennington BF, Peart J, Biederman J: Are endophenotypes based Specialty Pharmaceuticals, and Shire; has been a consult- on measures of executive functions useful for molecular Page 12 of 14 (page number not for citation purposes) Behavioral and Brain Functions 2008, 4:4 http://www.behavioralandbrainfunctions.com/content/4/1/4 genetic studies of ADHD? J Child Psychol Psychiatry 2005, 34. Rommelse NN, Altink ME, Oosterlaan J, Beem L, Buschgens CJ, Buite- 46:774-803. laar J, Sergeant JA: Speed, variability, and timing of motor out- 14. Skuse DH: Endophenotypes and child psychiatry. Br J Psychiatry put in ADHD: which measures are useful for endophenotypic 2001, 178:395-396. research? Behav Genet in press. 15. Faraone S, Kremen W, Lyons M, Pepple J, Seidman L, Tsuang M: 35. Slaats-Willemse D, De Sonneville L, Swaab-Barneveld H, Buitelaar J: Diagnostic accuracy and linkage analysis: how useful are Motor flexibility problems as a marker for genetic suscepti- schizophrenia spectrum phenotypes. Am J Psychiatry 1995, bility to attention-deficit/hyperactivity disorder. Biol Psychiatry 152:1286-1290. 2005, 58:233-238. 16. Faraone SV, Adamson JJ, Wilens T, Monuteaux MC, Biederman J: 36. Van der Stigchel S, Rommelse NN, Deijen JB, Geldof CJ, Witlox J, Deriving phenotypes for molecular genetic studies of sub- Oosterlaan J, Sergeant JA, Theeuwes J: Oculomotor capture in stance use disorders: a family approach. Drug Alcohol Depend ADHD. Cogn Neuropsychol 2007, 24:535-549. 2007, 88:244-250. 37. Baron RM, Kenny DA: The moderator-mediator variable dis- 17. Hasler G, Drevets WC, Manji HK, Charney DS: Discovering endo- tinction in social psychological research: conceptual, strate- phenotypes for major depression. Neuropsychopharmacology gic, and statistical considerations. J Pers Soc Psychol 1986, 2004, 29:1765-1781. 51:1173-1182. 18. Lenox RH, Gould TD, Manji HK: Endophenotypes in bipolar dis- 38. Gaub M, Carlson CL: Gender differences in ADHD: a meta- order. Am J Med Genet 2002, 114:391-406. analysis and critical review. J Am Acad Child Adolesc Psychiatry 19. Weiser M, Van Os J, Davidson M: Time for a shift in focus in 1997, 36:1036-1045. schizophrenia: from narrow phenotypes to broad endophe- 39. Faraone S, Biederman J, Mick E: The age dependent decline of notypes. Br J Psychiatry 2005, 187:203-205. attention-deficit/hyperactivity disorder: a meta-analysis of 20. Asarnow RF, Nuechterlein KH, Subotnik KL, Fogelson DL, Torquato follow-up studies. Psychol Med 2006, 36:159-165. RD, Payne DL, Asamen J, Mintz J, Guthrie D: Neurocognitive 40. Drechsler R, Brandeis D, Földényi M, Imhof K, Steinhausen H-C: The impairments in nonpsychotic parents of children with schiz- course of neuropsychological functions in children with ophrenia and Attention-Deficit/Hyperactivity Disorder. Arch attention deficit hyperactivity disorder from late childhood Gen Psychiatry 2002, 59:1053-1060. to early adolescence. J Child Psychol Psychiatry 2005, 46:824-836. 21. Murphy KR, Barkley RA: Parents of children with Attention- 41. Willoughby MT: Developmental course of ADHD symptoma- Deficit/Hyperactivity Disorder: psychological and atten- tology during the transition from childhood to adolescents: tional impairment. Am J Orthopsychiatry 1996, 66:93-102. a review with recommendations. J Child Psychol Psychiatry 2003, 22. Nigg JT, Blaskey LG, Stawicki JA, Sachek J: Evaluating the endo- 44:88-106. phenotype model of ADHD neuropsychological deficit: 42. Fischer M, Barkley RA, Edelbrock CS, Smallish L: The adolescent results for parents and siblings of children with ADHD com- outcome of hyperactive children diagnosed by research cri- bined and inattentive subtypes. J Abnorm Psychol 2004, teria: academic, attentional, and neuropsychological status. 113:614-625. J Consult Clin Psychol 1990, 58:580-588. 23. Seidman LJ, Biederman J, Monuteaux M, Weber W, Faraone SV: Neu- 43. Seidman L, Biederman J, Faraone SV, Weber W, Ouellette C: ropsychological functioning in nonreferred siblings of chil- Toward defining a neuropsychology of Attention Deficit- dren with attention deficit hyperactivity disorder. J Abnorm Hyperactivity Disorder: performance of children and adoles- Psychol 2000, 109:252-265. cents from a large clinically referred sample. J Consult Clin Psy- 24. Bidwell LC, Willcutt EG, DeFries JC, Pennington BF: Testing for chol 1997, 65:150-160. neuropsychological endophenotypes in siblings discordant 44. Antshel KM, Phillips MH, Gordon M, Barkley R, Faraone SV: Is for attention-deficit/hyperactivity disorder. Biol Psychiatry 2007 ADHD a valid disorder in children with intellectual delays? in press. Clin Psychol Rev 2006, 26:555-572. 25. Crosbie J, Schachar R: Deficient inhibition as a marker for famil- 45. Frazier TW, Demaree HA, Youngstrom EA: Meta-analysis of intel- ial ADHD. Am J Psychiatry 2001, 158:1884-1890. lectual and neuropsychological test performance in atten- 26. Schachar RJ, Crosbie J, Barr CL, Ornstein TJ, Kennedy J, Malone M, tion-deficit/hyperactivity disorder. Neuropsychology 2004, Roberts W, Ickowicz A, Tannock R, Chen S, Pathare T: Inhibition of 18:543-555. motor responses in siblings concordant and discordant for 46. Kuntsi J, Eley TC, Taylor A, Hughes C, Asherson P, Caspi A, Moffitt Attention Deficit Hyperactivity Disorder. Am J Psychiatry 2005, TE: Co-occurrence of ADHD and low IQ has genetic origins. 162:1076-1082. Am J Med Genet B Neuropsychiatr Genet 2004, 124B:41-47. 27. Slaats-Willemse D, Swaab-Barneveld H, De Sonneville L, Buitelaar J: 47. Conway AR, Kane MJ, Engle RW: Working memory capacity and Familial clustering of executive functioning in affected sibling its relation to general intelligence. Trends Cogn Sci 2003, pair families with ADHD. J Am Acad Child Adolesc Psychiatry 2005, 7:547-552. 44:385-391. 48. Engle RW: Working memory capacity as executive attention. 28. Slaats-Willemse D, Swaab-Barneveld H, De Sonneville L, Van der Curr Dir Psychol Sci 2002, 11:19-23. Meulen E, Buitelaar J: Deficient response inhibition as a cogni- 49. Schretlen D, Pearlson GD, Anthony JC, Aylward EH, Augustine AM, tive endophenotype of ADHD. J Am Acad Child Adolesc Psychiatry Davis A, Barta P: Elucidating the contributions of processing 2003, 42:1242-1248. speed, executive ability, and frontal lobe volume to normal 29. Rommelse NN, Altink ME, Oosterlaan J, Buschgens CJ, Buitelaar J, age-related differences in fluid intelligence. J Int Neuropsychol Sergeant JA: Support for an independent familial segregation Soc 2000, 6:52-61. of executive and intelligence endophenotypes in ADHD- 50. Brookes K, Xu X, Chen W, Zhou K, Neale B, Lowe N, Anney R, families. Psychol Med 2008 in press. Franke B, Gill M, Ebstein R, Buitelaar J, Sham P, Campbell D, Knight J, 30. Rommelse NN, Oosterlaan J, Buitelaar J, Faraone SV, Sergeant JA: Andreou P, Altink M, Arnold R, Boer F, Buschgens C, Butler L, Chris- Time reproduction in children with ADHD and their non- tiansen H, Feldman L, Fleischman K, Fliers E, Howe-Forbes R, Gold- affected siblings. J Am Acad Child Adolesc Psychiatry 2007, farb A, Heise A, Gabriels I, Korn-Lubetzki I, Marco R, et al.: The 46:582-590. analysis of 51 genes in DSM-IV combined type attention def- 31. Rommelse NN, Altink ME, De Sonneville LM, Buschgens CJ, Buitelaar icit hyperactivity disorder: association signals in DRD4, J, Oosterlaan J, Sergeant JA: Are motor inhibition and cognitive DAT1 and 16 other genes. Mol Psychiatry 2006, 11:934-953. flexibility dead ends in ADHD? J Abnorm Child Psychol 2007, 51. Conners K: Rating scales in ADHD Duke University Medical Center; 35:957-967. 1996. 32. Meyer A, Sagvolden T: Fine motor skills in South African chil- 52. Goodman R: The strengths and difficulties questionnaire: a dren with symptoms of ADHD: influence of subtype, gender, research note. J Child Psychol Psychiatry 1997, 38:581-586. age, and hand dominance. Behav Brain Funct 2006, 2:33. 53. Berument SK, Rutter M, Lord C, Pickles A, Bailey A: Autism screen- 33. Rommelse NN, Altink ME, Oosterlaan J, Buschgens CJ, Buitelaar J, De ing questionnaire: diagnostic validity. Br J Psychiatry 1999, Sonneville LM, Sergeant JA: Motor control in children with 175:444-451. ADHD and non-affected siblings: deficits most pronounced 54. Taylor EA: Childhood hyperactivity. Br J Psychiatry 1986, using left hand. J Child Psychol Psychiatry 2007, 48:1071-1079. 149:562-573. Page 13 of 14 (page number not for citation purposes) Behavioral and Brain Functions 2008, 4:4 http://www.behavioralandbrainfunctions.com/content/4/1/4 55. Wechsler D: WAIS-III Nederlandstalige bewerking. Technische handlei- ding London: The Psychological Corporation; 2000. 56. Wechsler D: WISC-III Handleiding London: The Psychological Corpo- ration; 2002. 57. Groth-Marnat G: Handbook of psychological assessment 3rd edition. New York: Wiley; 1997. 58. Pelham WE, Aronoff HR, Midlam JK, Shapiro CJ, Gnagy EM, Chronis AM, Onyango AN, Forehand G, Nguyen A, Waxmonsky J: A com- parison of Ritalin and Adderall: efficacy and time-course in children with attention-deficit/hyperactivity disorder. Pediat- rics 1999, 103:e43. 59. Logan GD: On the ability to inhibit thought and action: a users' guide to the stop signal paradigm. In Inhibitory processes in attention, memory, and language Edited by: Dagenbach D, Carr TH. San Diego: Academic Press; 1994:189-239. 60. Logan GD, Cowan WB: On the ability to inhibit thought and action: a theory of an act of control. Psychol Rev 1984, 91:295-327. 61. Scheres A, Oosterlaan J, Sergeant JA: Speed of inhibition predicts teacher-rated medication response in boys with attention deficit hyperactivity disorder. Journal of Disability, Development and Education 2006, 1:93-109. 62. De Sonneville LMJ: Amsterdam Neuropsychological Task: a computer-aided assessment program. In Cognitive ergonomics, clinical assessment and computer-assisted learning: Computers in psychol- ogy Volume 6. Edited by: Den Brinker BPLM, Beek PJ, Brand AN, Maarse SJ, Mulder LJM. Lisse, The Netherlands: Swets & Zeitlinger; 1999:204-217. 63. Los SA: On the origin of mixing costs: exploring information processing in pure and mixed blocks of trials. Acta Psychol 1996, 94:145-188. 64. Barkley RA: Time perception application, (version 1.0) (Com- puter software). University of Massachussetts Medical Center: Chesapeake Technology; 1998. 65. Van Meel CS, Oosterlaan J, Heslenfeld DJ, Sergeant JA: Motivational effects on motor timing in ADHD. J Am Acad Child Adolesc Psy- chiatry 2005, 44:451-460. 66. Tabachnick BG, Fidell LS: Using multivariate statistics 4th edition. Need- ham Heights: Allyn and Bacon; 2001. 67. Benjamini Y, Hochberg Y: Controlling the false discovery rate: a practical and powerful approach to multiple testing. J R Stat Soc B 1995, 57:289-300. 68. Sobel ME: Asymptotic confidence intervals for indirect effects in structural equation models. In Sociological Methodology Edited by: Leinhardt S. Washington, DC: American Sociological Association; 1992:290-312. 69. Chen PY, Popovich PM: Correlation: Parametric and nonpara- metric measures. Thousand Oaks, CA: Sage Publications; 2002. 70. Gimpel GA, Kuhn BR: Maternal report of attention deficit hyperactivity disorder symptoms in preschool children. Child Care Health Dev 1998, 26:163-179. 71. Nigg JT, Willcutt EG, Doyle AE, Sonuga-Barke EJS: Causal hetero- geneity in Attention-Deficit/Hyperactivity Disorder: do we need neuropsychologically impaired subtypes? Biol Psychiatry 2005, 57:1224-1230. 72. Biederman J, Faraone SV, Spencer T, Wilens T, Mick E, Lapey KA: Gender differences in a sample of adults with attention defi- cit hyperactivity disorder. Psychiatry Res 1994, 53:13-29. 73. Seidman LJ, Biederman J, Monuteaux MC, Valera E, Doyle AE, Faraone SV: Impact of gender and age on executive functioning: do girls and boys with and without Attention Deficit Hyperac- Publish with Bio Med Central and every tivity Disorder differ neuropsychologically in preteen and scientist can read your work free of charge teenage years? Dev Neuropsychol 2005, 27:79-105. 74. Biederman J, Faraone S, Milberger S, Curtis S, Chen L, Marrs A, Ouel- "BioMed Central will be the most significant development for lette C, Moore P, Spencer T: Predictors of persistence and disseminating the results of biomedical researc h in our lifetime." remission of ADHD into adolescence: results from a four- Sir Paul Nurse, Cancer Research UK year prospective follow-up study. J Am Acad Child Adolesc Psychi- atry 1996, 35:343-351. Your research papers will be: 75. Klassen AF, Miller A, Fine S: Health-related quality of life in chil- available free of charge to the entire biomedical community dren and adolescents who have a diagnosis of Attention-Def- icit/Hyperactivity Disorder. Pediatrics 2004, 114:e541-e547. peer reviewed and published immediately upon acceptance 76. Schachar RJ, Tannock R, Cunningham C, Corkum P: Behavioral, sit- cited in PubMed and archived on PubMed Central uational, and temporal effects of treatment of ADHD with methylphenidate. J Am Acad Child Adolesc Psychiatry 1997, yours — you keep the copyright 36:754-763. BioMedcentral Submit your manuscript here: http://www.biomedcentral.com/info/publishing_adv.asp Page 14 of 14 (page number not for citation purposes) http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png Behavioral and Brain Functions Springer Journals

Loading next page...
 
/lp/springer-journals/relationship-between-endophenotype-and-phenotype-in-adhd-TNGUgfWqk3
Publisher
Springer Journals
Copyright
Copyright © 2008 by Rommelse et al; licensee BioMed Central Ltd.
Subject
Biomedicine; Neurosciences; Neurology; Behavioral Therapy; Psychiatry
eISSN
1744-9081
DOI
10.1186/1744-9081-4-4
pmid
18234079
Publisher site
See Article on Publisher Site

Abstract

Background: It has been hypothesized that genetic and environmental factors relate to psychiatric disorders through the effect of intermediating, vulnerability traits called endophenotypes. The study had a threefold aim: to examine the predictive validity of an endophenotypic construct for the ADHD diagnosis, to test whether the magnitude of group differences at the endophenotypic and phenotypic level is comparable, and to investigate whether four factors (gender, age, IQ, rater bias) have an effect (moderation or mediation) on the relation between endophenotype and phenotype. Methods: Ten neurocognitive tasks were administered to 143 children with ADHD, 68 non- affected siblings, and 120 control children (first-borns) and 132 children with ADHD, 78 non- affected siblings, and 113 controls (second-borns) (5 – 19 years). The task measures have been investigated previously for their endophenotypic viability and were combined to one component which was labeled 'the endophenotypic construct': one measure representative of endophenotypic functioning across several domains of functioning. Results: The endophenotypic construct classified children with moderate accuracy (about 50% for each of the three groups). Non-affected children differed as much from controls at the endophenotypic as at the phenotypic level, but affected children displayed a more severe phenotype than endophenotype. Although a potentially moderating effect (age) and several mediating effects (gender, age, IQ) were found affecting the relation between endophenotypic construct and phenotype, none of the effects studied could account for the finding that affected children had a more severe phenotype than endophenotype. Conclusion: Endophenotypic functioning is moderately predictive of the ADHD diagnosis, though findings suggest substantial overlap exists between endophenotypic functioning in the groups of affected children, non-affected siblings, and controls. Results suggest other factors may be crucial and aggravate the ADHD symptoms in affected children. Page 1 of 14 (page number not for citation purposes) Behavioral and Brain Functions 2008, 4:4 http://www.behavioralandbrainfunctions.com/content/4/1/4 Studying endophenotypes may have certain advantages Background Psychiatric disorders as defined by the Diagnostic and Sta- over studying phenotypes. It has been proposed that tistical Manual of Mental Disorders (DSM-IV-TR) [1] have endophenotypes may be more suitable for detecting risk been hypothesized as reflecting the extreme end of under- genes, because endophenotypes are genetically less com- lying, continuously distributed traits [2-4]. In line with plex than phenotypes (i.e. related to fewer genes than phe- this, the behaviour of individuals suffering from psychiat- notypes) and, therefore, probably stronger linked to these ric disorders differs quantitatively but not necessarily disease genes than phenotypes [7-9]. Endophenotypes qualitatively from the behaviour of individuals without may also be useful in exploring different pathways leading psychiatric problems. The threshold of what is and what up to the disorder: Patients having the same diagnosis is not abnormal is to a certain extent arbitrarily deter- may differ strongly in the number and severity of symp- mined, but patients have in common that their behaviour toms they portray, suggesting heterogeneity in the causal interferes with their normal life and cause the patient (and pathways [10]. Creating more homogeneous subgroups his/her environment) to suffer. Why certain people pass of patients based on their endophenotypic functioning, this threshold and are diagnosed with a disorder and oth- may facilitate unravelling these differential causal path- ers do not is determined by additive and interacting ways. genetic and environmental risk factors [5,6]. Studies have shown that psychiatric disorders have genetic and envi- In the last two decades, substantial attention has been ronmental underpinnings which probably contribute to given to studying endophenotypes of psychiatric disor- certain neurocognitive abnormalities that, in turn, lead to ders. This has led to the discussion of what exactly consti- abnormal behaviour. It is theorized that these neurocog- tutes an endophenotype and what criteria must be met for nitive abnormalities form underlying, continuously dis- a neurocognitive function to be useful as candidate endo- tributed, vulnerability traits (endophenotypes) that phenotype [7-9,11-14]. Since an endophenotype forms a heighten the risk for developing a disorder (phenotype) link between susceptibility genes and the disorder, it fol- [7-9]. In this context, neurocognitive abnormalities refer lows that: (1) the neurocognitive dysfunction is heritable to mental functions that are mediated by brain processes; (and familial), in which at least partly the same genes these mental functions are not directly observable, but influence the endophenotype and phenotype; (2) the may be manipulated and measured using experimental neurocognitive dysfunction is associated with the disor- paradigms. Multiple endophenotypes interact to deter- der; (3) the neurocognitive dysfunction is observable in mine the finally observable behavior, the phenotype, non-affected first-degree relatives of an affected individ- which might be abnormal. In this context, the phenotype ual, because first-degree relatives are likely to carry some refers to directly observable symptoms of a disorder (Fig- of the susceptibility genes of the disorder. ure 1). Disease genes Environmental risk factors Neurobiological deviations Endophenotypes Neuropsychological deviations Phenotype Figure 1 The relationship between genetic and environmental risk factors, endophenotypes, and phenotype in ADHD The relationship between genetic and environmental risk factors, endophenotypes, and phenotype in ADHD. Page 2 of 14 (page number not for citation purposes) Behavioral and Brain Functions 2008, 4:4 http://www.behavioralandbrainfunctions.com/content/4/1/4 Thus far, several psychiatric conditions have been targeted motor endophenotypes, is predictive of the ADHD diag- for candidate endophenotypes and considerable knowl- nosis. edge has been gathered on the usefulness of different endophenotypes in for example, bipolar disorder, schizo- Practically all studies of ADHD endophenotypes con- phrenia, substance abuse, and depression [15-19]. This ducted thus far have reported the same type of results: The study targeted Attention-Deficit/Hyperactivity Disorder group of non-affected siblings performs in between the (ADHD), since it is one of the most prevalent child psy- affected siblings group and the normal control group. At chiatric disorders, yet the knowledge on the usefulness of a phenotypic level, non-affected siblings do not (or appar- ADHD endophenotypes is still limited. ently to a lesser extent) deviate from the controls. The reverse appears to be true for affected children: Their phe- Two studies have failed to find neurocognitive impair- notypic deficits are more pronounced than one would ments in parents of children with ADHD [20,21] and two expect based on their cognitive and motor dysfunction- studies found no conclusive evidence of cognitive dys- ing. Therefore, the second aim of the paper was to confirm functioning in non-affected siblings [22,23]. However, apparent observations that children with ADHD show one study did find evidence for cognitive functions as more severe ADHD symptoms than one would expect endophenotypes [24] and other studies, specifically tar- based on their endophenotypic dysfunctioning, whereas geting inhibition or interference control, found evidence non-affected siblings show less ADHD-like behaviour for these functions as endophenotypes [13,25-28]. In a than one would expect based on their endophenotypic previous study, we also found evidence for inhibition as vulnerabilities. well as visuo-spatial and verbal working memory as endo- phenotypes for ADHD [29]. Moreover, it appeared that The question that automatically arises if these observa- deficits in the various cognitive functions partly arose tions can be confirmed, is why endophenotypic vulnera- from the same genetic and/or environmental risk factors. bilities are not proportionally related to deviations at a Furthermore, evidence has been found for time reproduc- phenotypic level? It might be that certain factors moderate tion as ADHD endophenotype, a function related to the and/or mediate the relation between endophenotype and sense of time [30]. Less attention has been given to study- phenotype. Moderation would imply that the relation ing functions outside the cognitive domain, although between endophenotype and phenotype is not compara- ADHD is frequently associated with motor deficits in Cau- ble across different levels of the moderating factor [37]. casian and non-Caucasian subjects [31,32]. In previous Mediation would imply that (a part of) the relation studies, we and others have shown that non-affected sib- between endophenotype and phenotype can be explained lings have subtle problems, similar to their affected sib- through their correlation with a third factor. When this lings, in motor timing, motor control, motor speed and third factor is taken into account, the relation between variability, and speed of oculomotor control [33-36], sug- endophenotype and phenotype disappears or weakens gesting endophenotypes for ADHD may also lie inside the [37]. The third aim of this study was to explore whether motor domain. Moreover, it appeared that some of the four factors (gender, age, IQ, and rater bias) had a (mod- cognitive dysfunctioning in children with ADHD and erating and/or mediating) effect on the relationship their non-affected siblings may be related to problems between endophenotype and phenotype. already apparent on a simple reaction time task [31]. With respect to the first factor gender: ADHD is more fre- Taken together, the majority of studies have found sup- quently diagnosed in boys than girls [38], probably port for cognitive and motor endophenotypes for ADHD because boys are more vulnerable to the disorder. Since [24,26,28-31,33-35]. These studies have in common that affected children are more often boys than girls, whereas they administered one or a few measures tapping into a the gender ratio is more or less equal in the group of non- single domain. It is unlikely, however, that one such affected siblings, moderating and/or mediating effects of measure/domain mediates the relation between genotype gender may possibly be accountable for the apparent non- and phenotype and can predict the phenotype, because it comparable magnitude of group differences at the endo- is unlikely that all children with ADHD and their non- phenotypic and phenotypic level. With respect to the sec- affected siblings will show this endophenotype given the ond factor age: It is known from several studies that the causal heterogeneity of ADHD [13]. It is more likely that severity of ADHD symptoms appears to decline to some multiple endophenotypes mediate the relation between extent with age [39-41] and that the same might be true genotype and phenotype, and together are more powerful for underlying neurocognitive vulnerabilities [30]. Yet, in predicting the phenotype. Therefore, the first aim of our others have failed to find diminishing neurocognitive vul- study was to investigate if an endophenotypic construct, nerabilities with age [40,42,43]. It may thus be possible encompassing a broad battery of both cognitive and that age has a moderating and/or mediating influence on the relation between endophenotype and phenotype. Page 3 of 14 (page number not for citation purposes) Behavioral and Brain Functions 2008, 4:4 http://www.behavioralandbrainfunctions.com/content/4/1/4 With respect to the third factor IQ: A frequently reported relatives were required to have no formal or suspected finding is that children with ADHD have on average a ADHD diagnosis. For this study, we selected a subsample lower IQ than controls [44-46]. It has been suggested that of first- and second-borns to rule out dependency of the this lower IQ may underlie cognitive dysfunctioning, or data. In the first-borns, 143 affected children, 68 non- vice versa, that cognitive dysfunctioning is at the heart of affected siblings and 120 controls participated. In the sec- a lower intelligence, or that there is no hierarchical rela- ond-borns, 132 affected children, 78 non-affected sib- tion between both domains but both domains share com- lings, and 113 controls participated. All children were mon causes [47-49]. Either way, since IQ seems both between the ages of 5 and 19 years and were of European associated with the neurocognitive dysfunctioning as well Caucasian descent. Participants were excluded, if they had as with ADHD behaviour, a mediating effect of IQ may be an IQ < 70, a diagnosis of autism, epilepsy, general learn- expected on the relation between neurocognitive dysfunc- ing difficulties, brain disorders or known genetic disor- tioning and ADHD. ders, such as Down syndrome or Fragile-X-syndrome. An additional factor that may influence the relation The screening procedures and measures for phenotyping between endophenotype and phenotype may be rater have been described elsewhere [50]. Briefly, screening bias: Parents may underestimate the severity of ADHD questionnaires (parent and teacher Conners' long version symptoms in their undiagnosed/non-affected child and rating scales [51] parent and teacher Strengths and Diffi- overestimate the severity of ADHD symptoms in their culties Questionnaires [52] and parent Social Communi- diagnosed/affected child. Teachers, however, may be less cation Questionnaire [53]) were used to identify children likely to be affected by this contrast effect, since they often with ADHD symptoms and to screen for any autistic like do not have siblings of the same family in their class due behaviours. Scores were indicative for a diagnosis of to age differences between the siblings. Rater bias explain- ADHD if T-scores were ≥ 63 on the Conners' ADHD-sub- ing the non-comparable magnitude of group differences scales (DSM-IV Inattention, DSM-IV Hyperactive-Impul- th at the endophenotypic and phenotypic level may thus be sive, and DSM-IV ADHD Total) and > 90 percentile on investigated by comparing the relation between endophe- the SDQ-Hyperactivity scale. A score of ≥ 15 on the SCQ notype and phenotype as observed by parents and teach- was considered indicative of autistic like behaviours. Par- ers. ents and teachers were asked to rate the behavior of the child when off medication. Concerning all children rated The first aim was to investigate if an endophenotypic con- clinically on any of the questionnaires completed either struct, encompassing a broad battery of both cognitive by parents or teachers, the Parental Account of Children's and motor endophenotypes, is predictive of the ADHD Symptoms (PACS) was administered [54]. Data from the diagnosis. The second aim was to confirm apparent obser- questionnaires and the PACS were subjected to a stand- vations that children with ADHD show more severe ardised algorithm to derive each of the 18 DSM-IV ADHD ADHD symptoms than one would expect based on their symptoms, providing operational definitions for each endophenotypic dysfunctioning, whereas non-affected behavioural symptom [30]. With respect to control chil- siblings show less ADHD-like behaviour than one would dren, the Conners' long version for both parents and expect based on their endophenotypic vulnerabilities. The teachers was completed and all control children were third aim of this study was to explore the (moderating and required to obtain non-clinical scores. mediating) effects of four factors on the relationship between endophenotype and phenotype: gender, age, IQ, Full-scale IQ was estimated by four subtests of the and rater bias. Wechsler Intelligence Scale for Children III (WISC-III) or Wechsler Adult Intelligence Scale III (WAIS-III) (depend- Methods ing on the child's age): Vocabulary, Similarities, Block Sample design and Picture completion [55,56]. These subtests are Families with at least one child with the combined sub- known to correlate between .90–.95 with the Full-scale IQ type of ADHD (proband) and at least one additional sib- [57]. IQ testing took place while the children were off ling (regardless of possible ADHD-status) were recruited medication. in order to participate in the Dutch part of the Interna- tional Multicenter ADHD Genes study (IMAGE). The Procedure IMAGE project is an international collaborative study that Testing of children with ADHD and their siblings took aims to identify genes that increase the risk for ADHD place at the VU University Amsterdam or at the Radboud using QTL linkage and association strategies [50]. Addi- University Nijmegen Medical Centre and was conducted tional control families were recruited from primary and simultaneously for all children in a family. Psychostimu- high schools in the same geographical regions as the par- lants were discontinued for at least 48 hours before testing ticipating ADHD-families. Controls and their first degree took place [58]. Children were motivated with small Page 4 of 14 (page number not for citation purposes) Behavioral and Brain Functions 2008, 4:4 http://www.behavioralandbrainfunctions.com/content/4/1/4 breaks. At the end of the session, a gift worth approxi- inhibition) and third (cognitive flexibility) block could be mately € 4, – was given. Control children were tested in a compared. In all blocks, trials consisted of a horizontal similar way in a quiet room at their school. The study had bar with ten grey squares presented permanently at the medical-ethical approval by the local ethics committee centre of the screen. From trial to trial, a coloured square and was in accordance with the Helsinki Declaration. moved across the bar in a random direction (either one square to the right or to the left). Responses were required Experimental tasks to be initiated between 150 to 5000 ms after a square The ten experimental tasks described in this study have moved one position, otherwise a trial was replaced. The been fully described elsewhere [29-31,33,34]. A short task was self-paced with post-response intervals of 250 description of each task will be given below. Based on pre- ms. In the first block, the moving square was coloured vious results [29-31,33,34], the variable per task that green, and compatible responses were required: children showed most optimal results in the endophenotypic anal- were instructed to press a response button as quickly and yses in the five previous studies was chosen for the current as accurately as possible that corresponded to the direc- analyses. tion in which the stimulus moved. In the second block, the moving square was coloured red, and incompatible Stop task responses were required. The suppression of the auto- The Stop task was used to measure speed of inhibition of matic compatible response, in order to generate a non- an ongoing response [59,60]. Subjects were presented two automatic incompatible response, was hypothesized as types of trials: go-trials and stop-trials. Go-trials consisted requiring inhibitory control. In the third block, the colour of the presentation of a go-stimulus (drawing of a plane) of the moving square alternated randomly between green that was either pointing to the right or to the left [61]. and red, and both compatible and incompatible Children were instructed to press a response button that responses were required. Thus, both the direction and the corresponded to the direction of the stimulus as quickly colour of the square were unpredictable. The mixture of and as accurately as possible. Stop-trials were identical to both compatible and incompatible trials was hypothe- the go-stimulus but in addition a stop-signal was pre- sized to require high levels of cognitive flexibility in addi- sented (drawing of a cross that was superimposed on the tion to inhibitory control [63]. The first and second block plane). Children were required to withhold their response consisted of 10 practice trials and 40 experimental trials. to the stop-signal. Go stimuli were displayed for 1000 ms, The third block consisted of 16 practice trials and 80 preceded by a 500 ms fixation point. Stop signals were dis- experimental trials. Administration took about 10 to 15 played for 1000 ms minus delay time, with a mean dis- min. The dependent measure was the percentage of errors play time of 709 ms (SD 125 ms). Inter-trial intervals were across blocks, which was the best indicator of endopheno- 3000 ms. The delay between the go- and stop-signal was typic vulnerabilities [31]. dynamically varied so that it could be estimated when the child successfully inhibited 50% of the stop-trials, and Time reproduction unsuccessfully inhibited the other 50%. At this point, the The Timetest application version 1.0 [64] was used to go-process and stop-process were of equal duration, measure time reproduction. Stimuli consisted of temporal which made it possible to estimate the latency of the stop- intervals with different durations (4, 8, 12, 16, 20 s) that process: the Stop signal reaction time (SSRT) [59]. A total had to be reproduced as accurately as possible. The task of 2 practice blocks and 4 experimental blocks were was administered first in the visual modality (light bulb) administered, each consisting of 60 trials. The first prac- and thereafter in the auditory modality (tone). Children tice block consisted of only go-trials. The second practice were not informed about the length of the intervals. In block and the 4 experimental blocks consisted of 75% go- both modalities, 3 practice and 20 experimental trials trials and 25% stop-trials. Go- and stop-trials were were administered. The five interval lengths were ran- pseudo-randomly presented. Task administration took domly presented four times. Task administration for both about 15 min. Based on previous results, the dependent modalities required 15 min. Based on previous results, the measure was the SSRT [29], which showed endopheno- main dependent measure was the precision of the repro- typic-like group differences and correlated between sib- duction (operationalized as the absolute discrepancy lings. between the response length and the stimulus length) averaged across trials and modalities, which was abnor- Shifting attentional set mal in children with ADHD and their non-affected sib- Shifting attentional set was designed to measure accuracy lings and correlated between siblings [30]. of motor inhibition and cognitive flexibility [62]. The task consisted of three blocks of which the first block was Visuo-spatial sequencing designed to acquire a baseline of the accuracy of respond- The Visuo-spatial sequencing task was used to measure ing with which the performance on the second (motor accuracy of visuo-spatial working memory [62]. Stimuli Page 5 of 14 (page number not for citation purposes) Behavioral and Brain Functions 2008, 4:4 http://www.behavioralandbrainfunctions.com/content/4/1/4 consisted of nine circles symmetrically organized in a invisible midline (radius 8 cm) between the inner and square (3 by 3). On each trial, a sequence of circles was outer circle as quickly and precisely as possible with a pointed at by a computer-driven hand. Subjects were mouse cursor. One practice and one experimental session instructed to replicate the exact same sequence of circles, were administered for both hands separately (clockwise by pointing to them with the small, self-driven hand. with the right hand and counter clockwise with the left There were no time constrictions. One practice trial and hand). Administration took about 3 min. The dependent 24 experimental trials were presented. Every succeeding measure was the precision (mean distance to midline in trial increased in difficulty level: an increase in the mm averaged across 60 equal parts of the circle) of the left number of circles required to be remembered and/or an hand. Previous results have shown that precision of the increase in the complexity of the spatial pattern (i.e. the non-dominant hand showed endophenotypic-like char- trial consisted of circles that were spatially further acteristics [33]. removed from one another instead of being close to one another), hence manipulating working memory Tapping demands. Task administration took about 7 min. Based This task measured variability of self-generated motor out- on previous results, the total number of correct targets in put [62]. This task required the child to tap as frequently the correct order was used as dependent measure reflect- as possible within a certain time period. During tapping, ing endophenotypic-like group differences and correlat- the number of taps was continuously counted and dis- ing between siblings [29]. played on the screen. One practice session (5 s) and one experimental session (18 s) were administered for both Digit span hands separately. The task was first practised and executed The Digit span backwards of the WISC-III and WAIS-III with the index finger of the non-preferred hand, thereafter [55,56] was used to obtain an indication of verbal work- practised and executed with the index finger of the pre- ing memory. The backward part consisted of repeating a ferred hand. Administration took about 3 min. The sequence of numbers in the opposite order. Children were dependent measure was the variability (SD of intertap instructed to reproduce sequences as accurately as possi- intervals in ms) averaged across hands. Previous results ble. One digit was added to the sequence if a child repro- have shown that this measure correlates between siblings duced the sequence successfully. Two practice trials with a [34]. 2 digit sequence and (dependent on the child's perform- ance) a maximum of 8 experimental sequences were Baseline speed administered. Dependent measure was the maximum This task was designed to measure variability on a simple Digit span backwards, which proved useful as endophe- reaction time task [62]. Stimuli consisted of a fixation notypic candidate [29]. cross in the centre of a computer screen that changed unpredictably into a white square. Immediately following Pursuit the response, the white square changed back into the fix- This task was designed to measure precision of motor con- ation cross. The time interval between a response and the trol under continuous adaptation [62]. The stimulus con- emergence of the next white square varied randomly sisted of a randomly moving target (asterisk) that was between 500 to 2500 ms in order to prevent anticipation required to be 'caught' by moving a mouse cursor on top strategies. Subjects were required to press a key as quickly of the asterisk. The target moved at a constant speed of 10 as possible when the white square appeared. A practice mm/s. Children were instructed to 'catch' the randomly session (10 trials) and an experimental session (32 trials) moving target as precisely as possible. One practice (13 s) were administered for both hands separately. The task was and one experimental session (60 s) were administered first practised and executed with the index finger of the for both hands separately. Administration took about 5 non-preferred hand, thereafter practised and executed min. The dependent measure was the precision (mean with the index finger of the preferred hand. Administra- distance in mm between target and cursor calculated per tion took about 5 min. Dependent measure was the vari- second and averaged across the 60 s experimental session) ability (SD of reaction times in ms) of responses averaged of the left hand. Previous results have shown that mainly across hands. Previous results have shown that this meas- the performance of the non-dominant hand was most ure was associated with ADHD and correlated between strongly associated with ADHD [33]. siblings [34]. Tracking Motor timing This task aimed to measure precision of motor control This task was designed to measure variability of motor without continuous adaptation required [62]. The stimu- timing [65]. In this task a 1 s interval had to be produced. lus consisted of an inner and outer circle (radius 7.5 and The start of the interval was announced by a tone (80 db, 8.5 cm, respectively). Children were instructed to trace an 50 ms). After the subject's response, visual feedback was Page 6 of 14 (page number not for citation purposes) Behavioral and Brain Functions 2008, 4:4 http://www.behavioralandbrainfunctions.com/content/4/1/4 given, indicating whether the response was correct, too total raw score was averaged across parents and teachers short or too long. A response was regarded as correct, if it and labelled 'the phenotype'. fell between the lower and upper boundary set by a dynamic tracking algorithm. Boundaries were set at 500 to The first aim of this paper was to investigate whether the 1500 ms at the beginning of the task. If the response fell endophenotypic construct was predictive of the ADHD within these boundaries, the boundaries for the subse- diagnosis (affected, non-affected, control) using a multi- quent trial were narrowed by 100 ms. Likewise, the ple discriminant analysis. Age was also entered in the boundaries of the subsequent trial were widened with 100 model, because age differences had arisen between the ms, if the response on the previous trial fell outside those first- and second-borns by splitting up the sample. There boundaries. Subjects were instructed to produce as accu- was assessed what percentage of children was correctly rately as possible the 1 s interval. Twenty practice trials classified by the endophenotypic construct. and 80 experimental trials were administered. Both ses- sions were preceded by presenting 10 times a cartoon fig- The second aim of this paper was to test whether children ure for exactly 1 s on the screen to demonstrate the with ADHD showed a more severe phenotype than one duration of 1 s [65]. Administration took about 8 min. would expect based on their endophenotypical dysfunc- The dependent measure was the variability (SD of produc- tioning and whether the reverse was true for the non- tions in ms). Previous results have shown that this meas- affected siblings. Therefore, the interaction of group ure to be a viable endophenotypic candidate [34]. (affected, non-affected, control) by functioning (2 within subject levels: endophenotypic construct and phenotype) Data analyses was tested in a repeated measures ANOVA. Age was again Missing data for all variables for the sample described in implemented as covariate. this study was less than 5% and were replaced by means of expectation maximization [66]. The task measures were The third aim of this paper was to investigate whether the successfully normalized and standardized to z-scores by factors gender, age, and IQ may moderate and/or mediate applying a Van der Waerden transformation (SPSS version the relation between the endophenotypic construct and 14). Some of the z-scores were multiplied by -1, so that the the phenotype. Using regression analyses, significant z-scores of all task variables would have the same mean- moderation for a factor would be demonstrated if the pre- ing: A higher z-score was indicative of poor performance dictive effect of the endophenotypic construct was not or underestimation. Correction for multiple comparisons constant for that factor and was tested by examining the according to the false discovery rate (FDR) controlling interaction between the endophenotypic construct and procedure was applied to the analyses with a q-value set- the factor in predicting the phenotype [37]. Mediation ting of 0.05 [67]. was tested by calculating the path coefficients between (a) the endophenotypic construct and the factor, (b) the fac- In order to rule out dependency of the data (more than tor and the phenotype, and (c) the endophenotypic con- one child per family participated in the study), analyses struct and phenotype (see Figure 2). The effect of were performed with the children split by birth order. mediation (in which endophenotypical dysfunctioning First, analyses were run on the data from the first-borns (N lead to ADHD through gender, age, and/or IQ) is termed = 331), thereafter repeated on the second-born children the indirect effect [36] and can be tested using the formula 2 2 2 2 s + a s ), in which a represents (N = 323). In this way, not only was the dependency of [68]: z-value = a × b/√(b a b data handled, but this also gave the opportunity to study the unstandardized path coefficient of the endopheno- replicability of the results. Since most families consisted typic construct on the mediating factor, b represents the of two children, the samples of third- and fourth-borns unstandardized path coefficient of the mediating factor 2 2 were substantially smaller (N = 76 and N = 11, respec- on the phenotype, s and s represent the square of the a b tively) than the samples of first- and second-borns. These standard error of the path coefficients a and b. It was sample ratio differences were considered too large (> 4) examined whether the direct effect between the endophe- and therefore, the third- and fourth-borns were excluded notypic construct and phenotype was still significant after from the analyses. controlling for the indirect effect. To combine the task variables to a component that would Rater bias could not be examined for moderation and simplify the analyses and reduce error variance, a princi- mediation, since the phenotype was confounded with the pal component analysis was performed on the ten task moderating or mediating factor. Therefore, a different variables, separately for the first- and second-borns. The approach was taken. Paired sample t-tests were used to component explaining the largest amount of variance was compare parental and teacher ratings of ADHD to assess used in the further analyses and is labelled 'the endophe- whether the phenotype differed when rated by parents or notypic construct'. The standardized Conners' ADHD teachers. Additionally, in order to investigate whether the Page 7 of 14 (page number not for citation purposes) direct effect Behavioral and Brain Functions 2008, 4:4 http://www.behavioralandbrainfunctions.com/content/4/1/4 Results en dophenotype The percentage of affected, non-affected and controls was equally distributed across the first- and second-borns (χ = 1.24, df = 2, p = .54). The characteristics of both samples are presented in Table 1. Principal component analysis A principal component analysis was performed on the ten task measures, separately for the first- and second-borns. mediating fact or Similar results were obtained in both samples: All ten task measures related to one major component, explaining respectively 40.5% and 47.0% of the variance in the task gender, age, IQ measures. Additional components did not have an eigen- value above 1 and explained only 10.3% or less additional variance. Therefore, the following results report only the main component labelled 'the endophenotypic con- struct'. This endophenotypic construct was representative of endophenotypic functioning across ten tasks, because all tasks loaded on this component score (Table 2). Results were similar, when the analysis was repeated using the raw (not standardized and not normalized) task vari- ables, with a one-component solution explaining 34.9% and 41.5% of the variance in the first- and second-borns respectively, and with all variables correlated with this phen otype component (Table 2). Dis b Figure 2 etween play of en po dophenot ssible medi ype ati and phenotype ng effects of gender, age, and IQ First aim: Predictive validity of the endophenotypic Display of possible mediating effects of gender, age, and IQ construct for the ADHD diagnosis between endophenotype and phenotype. A multiple discriminant analysis was performed with age and the endophenotypic construct as predictors and diag- nosis as the grouping variable. To correct for the unequal predictive validity of the endophenotypic construct was group sizes, prior probabilities for all groups were set to 1/ different for the phenotype as rated by the parents or by 3. The endophenotypic construct significantly predicted the teachers, the correlation coefficients between the diagnostic status (first-borns F (2, 328) = 33.24, p < .001; endophenotypic construct – phenotype rated by parents second-borns F (2, 320) = 29.73, p < .001). In all groups, on the one hand and the endophenotypic construct – phe- correct classification percentages were roughly around notype rated by teachers on the other hand were com- 50%. In the first-borns, respectively 55%, 52%, and 48% pared using two-sided dependent samples t-tests [69]. of the affected children, non-affected siblings, and con- trols were correctly classified. In the second-borns, respec- Table 1: Sample characteristics first-borns (N = 331) second-borns (N = 323) affected non-affected control affected non-affected control N = 143 (43%) N = 68 (21%) N = 120 (36%) N = 132 (41%) N = 78 (24%) N = 113 (35%) 1 1 MSD M SD M SD F contrasts MSD M SD M SD F contrasts 2,328 2,320 age in years 13.2 2.6 14.9 2.7 13.3 2.7 10.9 2 > 1 = 3 11.6 2.7 9.7 2.4 10.6 2.9 11.3 1 > 2 = 3 2 2 % male 76.9 45.6 44.2 34.8 1 > 2 = 3 75.8 48.7 38.1 37.4 1 > 2 = 3 estimated full scale IQ 99.8 11.8 101.8 10.7 106.2 9.9 11.3 1 = 2 < 3 99.2 11.8 105.5 10.9 105.6 10.1 13.2 1 < 2 = 3 Conners' parent DSM-IV inattentive 68.6 10.0 48.6 7.2 46.3 4.5 306.6 1 > 2 = 3 69.7 10.4 46.8 6.5 46.7 4.9 329.9 1 > 2 = 3 hyperactive-impulsive 75.4 12.9 48.4 6.9 47.3 4.9 343.1 1 > 2 = 3 76.0 11.3 49.6 6.8 47.6 5.2 410.0 1 > 2 = 3 Conners' teacher DSM-IV inattentive 64.1 9.5 48.8 6.4 46.2 4.7 211.4 1 > 2 = 3 65.7 9.3 48.0 5.6 46.3 4.2 279.0 1 > 2 = 3 hyperactive-impulsive 68.2 11.7 48.5 7.7 46.3 3.9 236.2 1 > 2 = 3 69.0 11.5 47.7 4.7 47.2 5.5 260.5 1 > 2 = 3 1 2 2 Note. 1 = affected; 2 = non-affected; 3 = controls. = contrasts based on p-values of ≤ .05. = χ . th ADHD = Attention-Deficit/Hyperactivity Disorder; DSM-IV = Diagnostic and Statistical Manual for Mental Disorders (4 edition). Page 8 of 14 (page number not for citation purposes) indirect effect Behavioral and Brain Functions 2008, 4:4 http://www.behavioralandbrainfunctions.com/content/4/1/4 Table 2: Correlations between the individual task variables and the endophenotypic construct first-borns second-borns Task Measure z-scores raw data z-scores raw data Stop task Stop signal reaction time .66 .63 .67 .64 Shifting attentional set % errors .63 .67 .67 .71 Time reproduction Absolute deviation .70 .59 .74 .67 Visuo-spatial sequencing N identified targets correct order .73 .72 .79 .80 Digit span N backwards .55 .53 .69 .69 Pursuit Precision .74 .64 .77 .65 Tracking Precision .60 .55 .54 .48 Tapping Variability .46 .50 .49 .47 Baseline speed Variability .52 .48 .70 .62 Motor timing Variability .71 .56 .73 .52 tively 67%, 41%, and 45% of the affected children, non- Age affected siblings, and controls were correctly classified. Age had a moderating effect on the relation between the endophenotypic construct and phenotype in the second- Second aim: Group differences at an endophenotypic and borns (t = 2.92, β = .58, p = .004), though this effect was phenotypic level completely non-significant in the first-borns (t = -.02, β = The group by level interaction was analyzed to assess -.004, p = .99). The significant interaction between the whether group differences were comparable at the endo- endophenotypic construct and age in predicting ADHD in phenotypic and phenotypic level. This interaction was sig- second-borns appeared to be related to the non-affected nificant (first-borns F (2, 327) = 45.46, p < .001, η = .22; siblings: Younger non-affected siblings performed in second-borns F (2, 319) = 56.89, p < .001, η = .26), sug- between their affected siblings and controls (non-affected gesting group contrasts to be different at the endopheno- versus affected p < .001 and non-affected versus control p typical and phenotypical level. When the analysis was = .002), but older non-affected siblings performed more repeated with affected children and controls, the interac- like controls (non-affected versus affected p < .001 and tion remained significant (first-borns F (1, 260) = 67.68, non-affected versus control p = .54). Strong evidence was p < .001, η = .21; second-borns F (1, 242) = 92.13, p < found for a partially mediating effect of age: The endophe- .001, η = .28). As is visible in Figure 3, affected children notypic construct related to age, and age related to ADHD deviated more from controls at the phenotypic level than severity (Table 3). Correcting for this partial mediating at the endophenotypic level. No such interaction was present for non-affected siblings compared to controls (first-borns F (1, 185) = 0.24, p = .63, η < .01; second- borns F (1, 188) = 0.02, p = .89, η < .01). Third aim: Moderating and mediating effects of gender, age, and IQ and the influence of rater bias on the relation between endophenotypic construct and phenotype Gender No evidence was found for a moderating effect of gender (first-borns t = -.34, β = -.05, p = .73; second-borns t = -.41, β = -.06, p = .69), which suggested that the relation between the endophenotypic construct and phenotype was comparable for boys and girls. There was evidence, however, for a partially mediating effect of gender in the first-borns (Table 3). Part of the relationship between the endophenotypic construct and ADHD was related to gen- der: Boys performed slightly worse than girls on the endo- Group di iz phenot and teacher A Figure 3 ed as a ypic level (operat com fferences at the endo pD os H ite D s qc uor estionna ie o onalized f te ph ire) n enotypic level a task variables) and at the s a composite of pa (operational- rental phenotypic construct and boys had more severe ADHD Group differences at the endophenotypic level (operational- phenotypes than girls. ized as a composite score of ten task variables) and at the phenotypic level (operationalized as a composite of parental and teacher ADHD questionnaire). Page 9 of 14 (page number not for citation purposes) Behavioral and Brain Functions 2008, 4:4 http://www.behavioralandbrainfunctions.com/content/4/1/4 Table 3: Mediating effects of gender, age, and IQ on the relation between the endophenotypic construct and phenotype mediator endophenotype → mediator mediator → phenotype endophenotype → phenotype test of mediation: endophenotype → phenotype 2 2 2 2 z = a × b/√(b s + a s ) corrected for mediation a b as pb s pc s pz p c s a b c c gender first-borns -0.129 .026 <.001 -0.319 .108 .003 7.215 .657 <.001 2.538 .01 5.963 0.629 <.001 second-borns -0.020 .028 .47 -0.285 .113 .01 2.390 .771 .002 0.687 .49 2.148 0.697 age first-borns -1.724 .117 <.001 -1.383 .271 <.001 7.215 .657 <.001 4.822 <.001 8.041 0.846 <.001 second-borns -2.130 .102 <.001 0.889 .275 .001 2.390 .771 .002 -3.195 .001 10.150 1.044 <.001 IQ first-borns -2.264 .608 <.001 -0.255 .067 <.001 7.215 .657 <.001 2.662 .008 6.916 0.667 <.001 second-borns -1.222 .633 .05 -0.328 .066 <.001 2.390 .771 .002 1.799 .07 2.012 0.752 effect of age did not, however, result in a non-significant dictive validity of the endophenotypic construct was direct effect between the endophenotypic construct and comparable for the phenotype as rated by the parents or phenotype in ADHD. by the teachers. IQ Discussion IQ did not have a moderating influence (first-borns t = - This study aimed at examining the predictive validity of 1.11, β = -.49, p = .27; second-borns t = -.81, β = -.37, p = endophenotypic functioning for the ADHD diagnosis, the .42), indicating that the relation between the endopheno- magnitude of group differences at an endophenotypic and typic construct and phenotype to be constant across the phenotypic level, and the mediating and moderating IQ range studied here. However, there was a partial medi- effects of gender, age, and IQ as well as the influence of ating effect of IQ (Table 3), suggesting the endopheno- rater bias on the relation between endophenotype and typic performance related to IQ, and IQ related to ADHD. phenotype. Analyses were separately conducted on first- Correcting for this partial mediating effect of IQ did not and second-borns to accommodate for the non-independ- result in a non-significant direct route between the endo- ency of data. First- and second-borns did not differ with phenotypic construct and phenotype. respect to proportion of affected, non-affected and control children, and results were largely consistent across first- Rater bias and second-borns, suggesting ADHD was not related to Paired samples t-tests indicated that parents rated the birth-order as has been previously observed [70]. ADHD symptoms of their affected child as more severe than did teachers (first-borns t = 4.92, p < .001; second- Concerning the first aim of our study: The endopheno- borns t = 5.32, p < .001), which was in line with our expec- typic construct significantly predicted the diagnostic sta- tations. However, in contrast to our expectations, parents tus (affected, non-affected, and control), which is in line and teachers ratings agreed on the degree of ADHD-like with the aetiology of psychiatric disorders, in which endo- behaviour in non-affected children (first-borns t = -.51, p phenotypic vulnerabilities lead to phenotypic symptoms. = .61; second-borns t = .21, p = .84), which is similar to the The status of all groups was predicted with roughly similar findings in control children (first-borns t = .92, p = .36; percentages (around 50%). However, these percentages second-borns t = .86, p = .39). It appeared that parents were only moderate and not high, indicating a substantial were indeed possibly 'overestimating' the degree of overlap in endophenotypic functioning of children with ADHD in their affected child, but they did not seem to different diagnostic status. Thus, even an aggregated com- 'underestimate' the degree of ADHD in their non-affected ponent encompassing multiple endophenotypic meas- child. ures is not a firm predictor of diagnostic status and illustrates the causal heterogeneity of ADHD [13]. None of the correlation coefficients between the endo- phenotypic construct and ADHD rated by parents, on the In the second aim of our study, it was confirmed that one hand, and the endophenotypic construct and ADHD affected children indeed portrayed a more severe ADHD rated by teachers, on the other hand, differed significantly phenotype than one would expect based on their endo- from each other (t-values between 0.47 – 2.47 and p-val- phenotypic dysfunctioning. This is in line with previous ues between .36 – .98), when calculated separately for studies on cognitive dysfunctioning in patients with affected, non-affected, and control children or when cal- ADHD, in which not all patients portray neurocognitive culated across groups. These findings suggest that the pre- dysfunctioning and effect sizes are generally modest [71]. Page 10 of 14 (page number not for citation purposes) Behavioral and Brain Functions 2008, 4:4 http://www.behavioralandbrainfunctions.com/content/4/1/4 Based on previous endophenotypic studies, we expected recting for this partially mediating effect of age did not to find the reverse pattern for non-affected siblings, por- result in a non-significant direct effect between the endo- traying a less severe phenotype compared to their endo- phenotypic construct and phenotype. The findings of phenotypic construct. This appeared not to be the case: non-moderation of age in affected children are in line The difference between non-affected siblings and controls with studies showing symptoms of ADHD persist into was comparable for the endophenotypic construct and adolescence and adulthood for the majority of patients phenotype. This suggests that endophenotypic dysfunc- [40,74] and suggest that the effect of age does not appear tioning related proportionally to (subtle) phenotypical to contribute to the group differences at the endopheno- deviations in non-affected children, but other factors pos- typic and phenotypic levels. sibly come into play that aggravate the eventually observ- able phenotypical problems in affected children. These Like gender and age, IQ also did not appear to account for factors may be environmental in nature, such as differ- these non-comparable group differences, since IQ did not ences in upbringing and schooling, which apparently had have a moderating influence on the relationship between a more positive outcome for the non-affected sibling than the endophenotypic construct and phenotype. IQ did for the affected sibling. It may also be that certain pheno- have a mediating effect, but correcting for this mediating typic symptoms may aggravate other phenotypic symp- effect did not result a non-significant direct route between toms, resulting in a disproportionate severe phenotype in the endophenotypic construct and phenotype, suggesting affected children in relation to their endophenotype. For the influence of IQ on this relation is not substantial. example, a child being hyperactive, may become even These findings give support to the idea that endopheno- more hyperactive because his/her inattention during a typic dysfunctioning and a lower IQ are both related to school-task results in the child leaving his/her seat. each other and to ADHD [47-49], though a lower IQ can not account for the endophenotypic dysfunctioning asso- In line with the idea that some factors may influence the ciated with ADHD. relation between endophenotype and phenotype, we ana- lyzed the (moderating and mediating) effects of gender, Another factor we studied was the effect of rater bias on age, IQ, and rater bias. No evidence was found for a mod- the relation between the endophenotypic construct and erating effect of gender, which suggests that the relation phenotype. We hypothesized that parents, compared to between the endophenotypic construct and phenotype in teachers, may possibly overrate the ADHD severity in their ADHD is comparable for boys and girls. There was evi- affected children, whereas they may underrate symptoms dence, though, for a partially mediating effect of gender in in their non-affected children. This, in turn, might explain the first-borns. However, since this partial mediation was the disproportionate severe phenotype of the affected completely non-significant in the second-borns, and the group compared to their endophenotypic construct. Par- direct effect between the endophenotypic construct and ents indeed rated the degree of ADHD in their affected phenotype was still present after correcting for this par- children more severely than teachers. This may be because tially mediating effect of gender, it appears that gender did parents wanted to secure their participation in the study not have a large impact on the relation between the endo- [75] and/or because the problem behaviour was genu- phenotypic construct and phenotype in ADHD. It could inely (experienced as) more severe at home than at not account for the apparent disproportionately severe school, possibly because the use of symptom reducing phenotype in affected children. Similar non-influential medication is more consistent and/or effective at school effects of gender in relation to ADHD have been docu- [75,76]. Either way, although parents did rate the severity mented previously [72,73]. of their child's ADHD as more serious than teachers, a possible effect of rater bias could not explain the larger Another factor possibly influencing the endophenotypic phenotypic dysfunctions compared to endophenotypic construct – phenotype relation is age. Age moderated the dysfunctions in affected children, because the endophe- relation between the endophenotypic construct and phe- notypic construct related comparably to the phenotype notype in the second-borns, though not in the first-borns. when rated by parents and teachers. Furthermore, no evi- This significant interaction between the endophenotypic dence was found for an underestimation of ADHD by par- construct and age in predicting the ADHD phenotype ents in their non-affected siblings. Therefore, possible appeared to be related to the non-affected siblings: differences in parental and teacher ratings of ADHD do Younger non-affected siblings performed in between their not underlie the disproportionate severe phenotype com- affected siblings and controls, but older non-affected sib- pared to endophenotype in children with ADHD. lings performed more like controls, possibly suggesting endophenotypic vulnerabilities ease somewhat with Limitations increasing age in non-affected siblings. Strong evidence Several possible limitations of this study warrant consid- was found for a partial mediating relation of age, but cor- eration. Dividing the sample in first- and second-borns Page 11 of 14 (page number not for citation purposes) Behavioral and Brain Functions 2008, 4:4 http://www.behavioralandbrainfunctions.com/content/4/1/4 doubled the number of statistical tests that were per- ant for Eli Lilly, McNeil Consumer & Specialty Pharma- formed. However, we corrected for multiple testing and ceuticals, Shire, Noven Pharmaceuticals, and Cephalon; this approach gave the opportunity to investigate replica- has been on the speakers' bureaus of Eli Lilly, McNeil bility of the results and showed that almost all findings Consumer & Specialty Pharmaceuticals, Shire, and were comparable in first- and second-borns. Another pos- Cephalon. JB has been a consultant to/member of advi- sible limitation was the interpretability of the endopheno- sory board of/and/or speaker for Janssen Cilag BV, Eli typic component measure. Since it was a composition of Lilly, Bristol-Myer Squibb, UBC, Shire, Medice. JS has various cognitive and motor task variables, its exact repre- been a member of advisory board of Eli Lilly, Shire, Jans- sentation remains unclear. It may be hypothesized that sen Cilag. This study was partly funded by a grant assigned the endophenotypic construct taps into an underlying fac- to SF by the National Institute of Mental Health (NIH tor, which may represent general cognitive functioning grant # R01 MH62873-01A1). ('g') possibly in combination with variability of reacting. We feel though, that combining individual task measures Authors' contributions to one more robust component will facilitate heritability NR contributed to the data ascertainment, study design, research in ADHD, since a component probably entails literature searches, analyses, and writing. MA and CB con- less error variance and may be a more reliable measure tributed to the data ascertainment and writing. NM con- than individual task measures. tributed to the analyses and writing. SF, JB, JS and JO have written the study protocol and have contributed to the Conclusion writing. All authors have read and approved the final An endophenotypic construct encompassing multiple manuscript. endophenotypic measures is moderately predictive of diagnostic status, but substantial overlap exists between Acknowledgements The authors thank all of the parents, teachers and children who partici- endophenotypic functioning in the groups of affected pated. children, non-affected siblings and controls. Group differ- ences at an endophenotypic and phenotypic level are not References comparable for affected children, displaying a more severe 1. American Psychiatric Association: Diagnostic and Statistical phenotype than one would expect based on their endo- Manual for Mental Disorders. 4th edition. Washington, DC: phenotype when compared to controls. Group differences American Psychiatric Press; 1994. 2. Clark LA: Toward a consensual set of symptom clusters for were comparable for non-affected siblings compared to assessment of personality disorder. In Advances in personality controls, suggesting subtle endophenotypic vulnerabili- assessment Volume 8. Edited by: Butcher JN, Spielberger CD. Hillsdale, NJ: Lawrence Erlbaum; 1990:243-266. ties translate proportionally into phenotypic deviations. 3. Faraone SV, Tsuang MT: Measuring diagnostic accuracy in the Even though a potentially moderating effect (age) and absence of a "gold standard". Am J Psychiatry 1994, 151:650-657. several mediating effects (gender, age, IQ) have been 4. Hudziak JJ, Heath AC, Madden PF, Reich W, Bucholz KK, Slutske W, Bierut LJ, Neuman RJ, Todd RD: Latent class and factor analysis found affecting the relation between the endophenotypic of DSM-IV ADHD: a twin study of female adolescents. J Am construct and phenotype, none of the effects studied (gen- Acad Child Adolesc Psychiatry 1998, 37:848-857. der, age, IQ, and rater bias) could account for the finding 5. Kendler KS, Eaves LJ: Models for the joint effect of genotype and environment on liability to psychiatric illness. Am J Psychi- that affected children deviated more from controls at the atry 1986, 143:279-289. phenotypic than endophenotypic level. These findings 6. Kendler KS, Prescott CA, Myers J, Neale MC: The structure of genetic and environmental risk factors for common psychi- suggest other factors come into play and aggravate the atric and substance abuse disorders in men and woman. Arch phenotype in affected children. Gen Psychiatry 2003, 60:929-937. 7. Gottesman II, Gould TD: The endophenotype concept in psy- chiatry: etymology and strategic intentions. Am J Psychiatry Abbreviations 2003, 160:636-645. ADHD = Attention-Deficit/Hyperactivity Disorder; DSM- 8. Waldman ID: Statistical approaches to complex phenotypes: evaluating neuropsychological endophenotypes for atten- IV = Diagnostic and Statistical Manual for Mental Disor- tion-deficit/hyperactivity disorder. Biol Psychiatry 2005, th ders 4 edition; IMAGE = International Multicenter 57:1347-1356. ADHD Genes study; IQ = Intelligence Quotient; PACS = 9. Kuntsi J, Neale BM, Chen W, Faraone SV, Asherson A: The IMAGE project: methodological issues for the molecular genetic Parental Account of Children's Symptoms; SCQ = Social analysis of ADHD. Behav Brain Funct 2006, 2:27. Communication Questionnaire; SDQ = Strengths and 10. Faraone SV, Biederman J: Neurobiology of attention-deficit Difficulties Questionnaires; SSRT = Stop Signal Reaction hyperactivity disorder. Biol Psychiatry 1998, 44:951-958. 11. Almasy L, Blangero J: Endophenotypes as quantitative risk fac- Time. tors for psychiatric disease: rationale and study design. Am J Med Genet 2001, 105:42-44. 12. Castellanos FX, Tannock R: Neuroscience of attention-deficit/ Competing interests hyperactivity disorder: the search for endophenotypes. Nat NR, MA, NM, CB, and JO have no competing interests. SF Rev Neurosci 2002, 3:617-628. has received grants from Eli Lilly, McNeil Consumer & 13. Doyle AE, Faraone SV, Seidman LJ, Willcutt EG, Nigg JT, Waldman ID, Pennington BF, Peart J, Biederman J: Are endophenotypes based Specialty Pharmaceuticals, and Shire; has been a consult- on measures of executive functions useful for molecular Page 12 of 14 (page number not for citation purposes) Behavioral and Brain Functions 2008, 4:4 http://www.behavioralandbrainfunctions.com/content/4/1/4 genetic studies of ADHD? J Child Psychol Psychiatry 2005, 34. Rommelse NN, Altink ME, Oosterlaan J, Beem L, Buschgens CJ, Buite- 46:774-803. laar J, Sergeant JA: Speed, variability, and timing of motor out- 14. Skuse DH: Endophenotypes and child psychiatry. Br J Psychiatry put in ADHD: which measures are useful for endophenotypic 2001, 178:395-396. research? Behav Genet in press. 15. Faraone S, Kremen W, Lyons M, Pepple J, Seidman L, Tsuang M: 35. Slaats-Willemse D, De Sonneville L, Swaab-Barneveld H, Buitelaar J: Diagnostic accuracy and linkage analysis: how useful are Motor flexibility problems as a marker for genetic suscepti- schizophrenia spectrum phenotypes. Am J Psychiatry 1995, bility to attention-deficit/hyperactivity disorder. Biol Psychiatry 152:1286-1290. 2005, 58:233-238. 16. Faraone SV, Adamson JJ, Wilens T, Monuteaux MC, Biederman J: 36. Van der Stigchel S, Rommelse NN, Deijen JB, Geldof CJ, Witlox J, Deriving phenotypes for molecular genetic studies of sub- Oosterlaan J, Sergeant JA, Theeuwes J: Oculomotor capture in stance use disorders: a family approach. Drug Alcohol Depend ADHD. Cogn Neuropsychol 2007, 24:535-549. 2007, 88:244-250. 37. Baron RM, Kenny DA: The moderator-mediator variable dis- 17. Hasler G, Drevets WC, Manji HK, Charney DS: Discovering endo- tinction in social psychological research: conceptual, strate- phenotypes for major depression. Neuropsychopharmacology gic, and statistical considerations. J Pers Soc Psychol 1986, 2004, 29:1765-1781. 51:1173-1182. 18. Lenox RH, Gould TD, Manji HK: Endophenotypes in bipolar dis- 38. Gaub M, Carlson CL: Gender differences in ADHD: a meta- order. Am J Med Genet 2002, 114:391-406. analysis and critical review. J Am Acad Child Adolesc Psychiatry 19. Weiser M, Van Os J, Davidson M: Time for a shift in focus in 1997, 36:1036-1045. schizophrenia: from narrow phenotypes to broad endophe- 39. Faraone S, Biederman J, Mick E: The age dependent decline of notypes. Br J Psychiatry 2005, 187:203-205. attention-deficit/hyperactivity disorder: a meta-analysis of 20. Asarnow RF, Nuechterlein KH, Subotnik KL, Fogelson DL, Torquato follow-up studies. Psychol Med 2006, 36:159-165. RD, Payne DL, Asamen J, Mintz J, Guthrie D: Neurocognitive 40. Drechsler R, Brandeis D, Földényi M, Imhof K, Steinhausen H-C: The impairments in nonpsychotic parents of children with schiz- course of neuropsychological functions in children with ophrenia and Attention-Deficit/Hyperactivity Disorder. Arch attention deficit hyperactivity disorder from late childhood Gen Psychiatry 2002, 59:1053-1060. to early adolescence. J Child Psychol Psychiatry 2005, 46:824-836. 21. Murphy KR, Barkley RA: Parents of children with Attention- 41. Willoughby MT: Developmental course of ADHD symptoma- Deficit/Hyperactivity Disorder: psychological and atten- tology during the transition from childhood to adolescents: tional impairment. Am J Orthopsychiatry 1996, 66:93-102. a review with recommendations. J Child Psychol Psychiatry 2003, 22. Nigg JT, Blaskey LG, Stawicki JA, Sachek J: Evaluating the endo- 44:88-106. phenotype model of ADHD neuropsychological deficit: 42. Fischer M, Barkley RA, Edelbrock CS, Smallish L: The adolescent results for parents and siblings of children with ADHD com- outcome of hyperactive children diagnosed by research cri- bined and inattentive subtypes. J Abnorm Psychol 2004, teria: academic, attentional, and neuropsychological status. 113:614-625. J Consult Clin Psychol 1990, 58:580-588. 23. Seidman LJ, Biederman J, Monuteaux M, Weber W, Faraone SV: Neu- 43. Seidman L, Biederman J, Faraone SV, Weber W, Ouellette C: ropsychological functioning in nonreferred siblings of chil- Toward defining a neuropsychology of Attention Deficit- dren with attention deficit hyperactivity disorder. J Abnorm Hyperactivity Disorder: performance of children and adoles- Psychol 2000, 109:252-265. cents from a large clinically referred sample. J Consult Clin Psy- 24. Bidwell LC, Willcutt EG, DeFries JC, Pennington BF: Testing for chol 1997, 65:150-160. neuropsychological endophenotypes in siblings discordant 44. Antshel KM, Phillips MH, Gordon M, Barkley R, Faraone SV: Is for attention-deficit/hyperactivity disorder. Biol Psychiatry 2007 ADHD a valid disorder in children with intellectual delays? in press. Clin Psychol Rev 2006, 26:555-572. 25. Crosbie J, Schachar R: Deficient inhibition as a marker for famil- 45. Frazier TW, Demaree HA, Youngstrom EA: Meta-analysis of intel- ial ADHD. Am J Psychiatry 2001, 158:1884-1890. lectual and neuropsychological test performance in atten- 26. Schachar RJ, Crosbie J, Barr CL, Ornstein TJ, Kennedy J, Malone M, tion-deficit/hyperactivity disorder. Neuropsychology 2004, Roberts W, Ickowicz A, Tannock R, Chen S, Pathare T: Inhibition of 18:543-555. motor responses in siblings concordant and discordant for 46. Kuntsi J, Eley TC, Taylor A, Hughes C, Asherson P, Caspi A, Moffitt Attention Deficit Hyperactivity Disorder. Am J Psychiatry 2005, TE: Co-occurrence of ADHD and low IQ has genetic origins. 162:1076-1082. Am J Med Genet B Neuropsychiatr Genet 2004, 124B:41-47. 27. Slaats-Willemse D, Swaab-Barneveld H, De Sonneville L, Buitelaar J: 47. Conway AR, Kane MJ, Engle RW: Working memory capacity and Familial clustering of executive functioning in affected sibling its relation to general intelligence. Trends Cogn Sci 2003, pair families with ADHD. J Am Acad Child Adolesc Psychiatry 2005, 7:547-552. 44:385-391. 48. Engle RW: Working memory capacity as executive attention. 28. Slaats-Willemse D, Swaab-Barneveld H, De Sonneville L, Van der Curr Dir Psychol Sci 2002, 11:19-23. Meulen E, Buitelaar J: Deficient response inhibition as a cogni- 49. Schretlen D, Pearlson GD, Anthony JC, Aylward EH, Augustine AM, tive endophenotype of ADHD. J Am Acad Child Adolesc Psychiatry Davis A, Barta P: Elucidating the contributions of processing 2003, 42:1242-1248. speed, executive ability, and frontal lobe volume to normal 29. Rommelse NN, Altink ME, Oosterlaan J, Buschgens CJ, Buitelaar J, age-related differences in fluid intelligence. J Int Neuropsychol Sergeant JA: Support for an independent familial segregation Soc 2000, 6:52-61. of executive and intelligence endophenotypes in ADHD- 50. Brookes K, Xu X, Chen W, Zhou K, Neale B, Lowe N, Anney R, families. Psychol Med 2008 in press. Franke B, Gill M, Ebstein R, Buitelaar J, Sham P, Campbell D, Knight J, 30. Rommelse NN, Oosterlaan J, Buitelaar J, Faraone SV, Sergeant JA: Andreou P, Altink M, Arnold R, Boer F, Buschgens C, Butler L, Chris- Time reproduction in children with ADHD and their non- tiansen H, Feldman L, Fleischman K, Fliers E, Howe-Forbes R, Gold- affected siblings. J Am Acad Child Adolesc Psychiatry 2007, farb A, Heise A, Gabriels I, Korn-Lubetzki I, Marco R, et al.: The 46:582-590. analysis of 51 genes in DSM-IV combined type attention def- 31. Rommelse NN, Altink ME, De Sonneville LM, Buschgens CJ, Buitelaar icit hyperactivity disorder: association signals in DRD4, J, Oosterlaan J, Sergeant JA: Are motor inhibition and cognitive DAT1 and 16 other genes. Mol Psychiatry 2006, 11:934-953. flexibility dead ends in ADHD? J Abnorm Child Psychol 2007, 51. Conners K: Rating scales in ADHD Duke University Medical Center; 35:957-967. 1996. 32. Meyer A, Sagvolden T: Fine motor skills in South African chil- 52. Goodman R: The strengths and difficulties questionnaire: a dren with symptoms of ADHD: influence of subtype, gender, research note. J Child Psychol Psychiatry 1997, 38:581-586. age, and hand dominance. Behav Brain Funct 2006, 2:33. 53. Berument SK, Rutter M, Lord C, Pickles A, Bailey A: Autism screen- 33. Rommelse NN, Altink ME, Oosterlaan J, Buschgens CJ, Buitelaar J, De ing questionnaire: diagnostic validity. Br J Psychiatry 1999, Sonneville LM, Sergeant JA: Motor control in children with 175:444-451. ADHD and non-affected siblings: deficits most pronounced 54. Taylor EA: Childhood hyperactivity. Br J Psychiatry 1986, using left hand. J Child Psychol Psychiatry 2007, 48:1071-1079. 149:562-573. Page 13 of 14 (page number not for citation purposes) Behavioral and Brain Functions 2008, 4:4 http://www.behavioralandbrainfunctions.com/content/4/1/4 55. Wechsler D: WAIS-III Nederlandstalige bewerking. Technische handlei- ding London: The Psychological Corporation; 2000. 56. Wechsler D: WISC-III Handleiding London: The Psychological Corpo- ration; 2002. 57. Groth-Marnat G: Handbook of psychological assessment 3rd edition. New York: Wiley; 1997. 58. Pelham WE, Aronoff HR, Midlam JK, Shapiro CJ, Gnagy EM, Chronis AM, Onyango AN, Forehand G, Nguyen A, Waxmonsky J: A com- parison of Ritalin and Adderall: efficacy and time-course in children with attention-deficit/hyperactivity disorder. Pediat- rics 1999, 103:e43. 59. Logan GD: On the ability to inhibit thought and action: a users' guide to the stop signal paradigm. In Inhibitory processes in attention, memory, and language Edited by: Dagenbach D, Carr TH. San Diego: Academic Press; 1994:189-239. 60. Logan GD, Cowan WB: On the ability to inhibit thought and action: a theory of an act of control. Psychol Rev 1984, 91:295-327. 61. Scheres A, Oosterlaan J, Sergeant JA: Speed of inhibition predicts teacher-rated medication response in boys with attention deficit hyperactivity disorder. Journal of Disability, Development and Education 2006, 1:93-109. 62. De Sonneville LMJ: Amsterdam Neuropsychological Task: a computer-aided assessment program. In Cognitive ergonomics, clinical assessment and computer-assisted learning: Computers in psychol- ogy Volume 6. Edited by: Den Brinker BPLM, Beek PJ, Brand AN, Maarse SJ, Mulder LJM. Lisse, The Netherlands: Swets & Zeitlinger; 1999:204-217. 63. Los SA: On the origin of mixing costs: exploring information processing in pure and mixed blocks of trials. Acta Psychol 1996, 94:145-188. 64. Barkley RA: Time perception application, (version 1.0) (Com- puter software). University of Massachussetts Medical Center: Chesapeake Technology; 1998. 65. Van Meel CS, Oosterlaan J, Heslenfeld DJ, Sergeant JA: Motivational effects on motor timing in ADHD. J Am Acad Child Adolesc Psy- chiatry 2005, 44:451-460. 66. Tabachnick BG, Fidell LS: Using multivariate statistics 4th edition. Need- ham Heights: Allyn and Bacon; 2001. 67. Benjamini Y, Hochberg Y: Controlling the false discovery rate: a practical and powerful approach to multiple testing. J R Stat Soc B 1995, 57:289-300. 68. Sobel ME: Asymptotic confidence intervals for indirect effects in structural equation models. In Sociological Methodology Edited by: Leinhardt S. Washington, DC: American Sociological Association; 1992:290-312. 69. Chen PY, Popovich PM: Correlation: Parametric and nonpara- metric measures. Thousand Oaks, CA: Sage Publications; 2002. 70. Gimpel GA, Kuhn BR: Maternal report of attention deficit hyperactivity disorder symptoms in preschool children. Child Care Health Dev 1998, 26:163-179. 71. Nigg JT, Willcutt EG, Doyle AE, Sonuga-Barke EJS: Causal hetero- geneity in Attention-Deficit/Hyperactivity Disorder: do we need neuropsychologically impaired subtypes? Biol Psychiatry 2005, 57:1224-1230. 72. Biederman J, Faraone SV, Spencer T, Wilens T, Mick E, Lapey KA: Gender differences in a sample of adults with attention defi- cit hyperactivity disorder. Psychiatry Res 1994, 53:13-29. 73. Seidman LJ, Biederman J, Monuteaux MC, Valera E, Doyle AE, Faraone SV: Impact of gender and age on executive functioning: do girls and boys with and without Attention Deficit Hyperac- Publish with Bio Med Central and every tivity Disorder differ neuropsychologically in preteen and scientist can read your work free of charge teenage years? Dev Neuropsychol 2005, 27:79-105. 74. Biederman J, Faraone S, Milberger S, Curtis S, Chen L, Marrs A, Ouel- "BioMed Central will be the most significant development for lette C, Moore P, Spencer T: Predictors of persistence and disseminating the results of biomedical researc h in our lifetime." remission of ADHD into adolescence: results from a four- Sir Paul Nurse, Cancer Research UK year prospective follow-up study. J Am Acad Child Adolesc Psychi- atry 1996, 35:343-351. Your research papers will be: 75. Klassen AF, Miller A, Fine S: Health-related quality of life in chil- available free of charge to the entire biomedical community dren and adolescents who have a diagnosis of Attention-Def- icit/Hyperactivity Disorder. Pediatrics 2004, 114:e541-e547. peer reviewed and published immediately upon acceptance 76. Schachar RJ, Tannock R, Cunningham C, Corkum P: Behavioral, sit- cited in PubMed and archived on PubMed Central uational, and temporal effects of treatment of ADHD with methylphenidate. J Am Acad Child Adolesc Psychiatry 1997, yours — you keep the copyright 36:754-763. BioMedcentral Submit your manuscript here: http://www.biomedcentral.com/info/publishing_adv.asp Page 14 of 14 (page number not for citation purposes)

Journal

Behavioral and Brain FunctionsSpringer Journals

Published: Jan 30, 2008

References