Executive functions as endophenotypes in ADHD: evidence from the Cambridge Neuropsychological Test Battery (CANTAB)
Conflict of interest statement: No conflicts declared.
Abstract
Background: Little is known about executive functions among unaffected siblings of children with attention deficit/hyperactivity disorder (ADHD), and there is lack of such information from non-Western countries. We examined verbal and nonverbal executive functions in adolescents with ADHD, unaffected siblings and controls to test whether executive functions could be potential endophenotypes for ADHD.
Methods: We assessed 279 adolescents (age range: 11–17 years) with a childhood diagnosis of DSM-IV ADHD, 136 biological siblings (108 unaffected, 79.4%), and 173 unaffected controls by using psychiatric interviews, the Wechsler Intelligence Scale for Children – 3rd edition (WISC-III), including digit spans, and the tasks involving executive functions of the Cambridge Neuropsychological Test Automated Battery (CANTAB): Intra-dimensional/Extra-dimensional Shifts (IED), Spatial Span (SSP), Spatial Working Memory (SWM), and Stockings of Cambridge (SOC).
Results: Compared with the controls, adolescents with ADHD and unaffected siblings had a significantly shorter backward digit span, more extra-dimensional shift errors in the IED, shorter spatial span length in the SSP, more total errors and poorer strategy use in the SWM, and fewer problems solved in the minimum number of moves and shorter initial thinking time in the SOC. The magnitudes of the differences in the SWM and SOC increased with increased task difficulties. In general, neither persistent ADHD nor comorbidity was associated with increased deficits in executive functions among adolescents with ADHD.
Conclusions: The lack of much difference in executive dysfunctions between unaffected siblings and ADHD adolescents suggests that executive dysfunctions may be useful cognitive endophenotypes for ADHD genetic studies.
Abbreviations:
-
- ADHD
-
- attention deficit/hyperactivity disorder
-
- CANTAB
-
- Cambridge Neuropsychological Test Automated Battery
-
- IED
-
- Intra-dimensional/Extra-dimensional Shifts
-
- SSP
-
- Spatial Span
-
- SWM
-
- Spatial Working Memory
-
- SOC
-
- Stockings of Cambridge
-
- K-SADS-E
-
- Kiddie epidemiologic version of the Schedule for Affective Disorders and Schizophrenia
-
- A
-
- ADHD
-
- S
-
- Sibling
-
- C
-
- Control
Attention deficit/hyperactivity disorder (ADHD) has been recognized as a common, early-onset, impairing, and heterogeneous neuropsychiatric disorder with high heritability (Faraone et al., 2005). Despite substantial evidence supporting the genetic etiology of ADHD (Faraone, 2000), molecular genetic studies, even using ADHD subgroup approaches such as comorbidity (Doyle & Faraone, 2002), persistency (Faraone, Biederman, & Monuteaux, 2000), and subtype (Chen et al., 2008), so far have provided few conclusive results. To address these challenges, there has been growing interest in using endophenotypes such as neuropsychological (Doyle et al., 2005; Slaats-Willemse, Swaab-Barneveld, De Sonneville, & Buitelaar, 2005), neuroimaging (Doyle et al., 2005), and electrophysiological paradigms (Doyle et al., 2005) for ADHD genetic studies. Among them, neuropsychological functioning is recognized as a valuable endophenotype for ADHD genetic studies.
An endophenotype is defined as a phenotype which can be measured at a cognitive or neurobiological level, which is more proximal to the biological etiology of a clinical disorder than the behavioral phenotype, and which is influenced by one or more of the same susceptibility genes as the condition. Proposed criteria for useful endophenotypes in psychiatry vary somewhat, but share several key elements. Specifically, researchers suggest that potential endophenotypes for ADHD should 1) be correlated with ADHD in the probands; 2) be measured by tools with good psychometric properties, including reliability; 3) show familial-genetic overlap with this disorder; 4) appear in unaffected relatives (Doyle et al., 2005; Nigg, Blaskey, Stawicki, & Sachek, 2004). Identification of potentially useful endophenotypes for ADHD may further augment the statistical power of molecular genetic studies of ADHD (Doyle et al., 2005). Furthermore, the studies of endophenotypes will result in the definition of genetically more homogeneous ADHD subtypes.
The literature documents lifelong executive function deficits in ADHD (Seidman, 2006), with the most consistent results found in planning, working memory, and inhibition, followed by set-shifting (Pennington & Ozonoff, 1996), and with greater reductions in visuo-spatial than verbal working memory (Martinussen, Hayden, Hogg-Johnson, & Tannock, 2005). Furthermore, impaired executive functions in ADHD are more apparent with increased task demands (Gau, Chiu, Shang, Cheng, & Soong, 2009; Rommelse, Oosterlaan, Buitelaar, Faraone, & Sergeant, 2007).
The findings of decreased performance in some domains of neuropsychological functions in unaffected siblings or other relatives of patients with ADHD (Rommelse, Van der Stigchel et al., 2008; Seidman, Biederman, Monuteaux, Weber, & Faraone, 2000), and deficits in attentional control (Slaats-Willemse, Swaab-Barneveld, De Sonneville, & Buitelaar, 2007) and response inhibition (Slaats-Willemse, Swaab-Barneveld, De Sonneville, van der Meulen, & Buitelaar, 2003) in unaffected relatives of patients with ADHD have provided evidence to support the familial overlap of ADHD and executive dysfunction (Seidman et al., 2000). However, a wide range of executive functions have not been comprehensively assessed among unaffected siblings of patients with ADHD in one study and no study has examined the neuropsychological functions in unaffected siblings of ADHD patients using the Cambridge Neuropsychological Test Automated Battery (CANTAB).
The CANTAB, a computerized test battery targeting multiple neuropsychological functions, is suitable for cross-cultural comparison (De Luca et al., 2003) and measuring endophenotypes (Doyle et al., 2005) because of its standard procedures (De Luca et al., 2003), nonverbal nature (De Luca et al., 2003), and solid psychometric properties as established in Western (Luciana, 2003; Luciana & Nelson, 1998) and Taiwanese (Gau et al., 2009) populations. Using the CANTAB to assess executive functions, children with ADHD performed worse in spatial working memory (Rhodes, Coghill, & Matthews, 2004), spatial planning (Kempton et al., 1999), and set-shifting (Rhodes, Coghill, & Matthews, 2005). However, no study has used the CANTAB to assess the executive functions among unaffected relatives of ADHD.
In summary, the literature on neurocognitive impairments in unaffected relatives is inconsistent with small effect sizes. The CANTAB, which has not previously been used in studies of ADHD relatives, may have greater power than clinically administered measures to detect the subtle deficits we expect to see in unaffected siblings because it is a computerized battery that captures the reaction time of responses as well as errors in a wide range of executive functions. Second, the literature on ADHD and unaffected relatives is dominated by samples from the US/Western Europe (Seidman et al., 2000; Slaats-Willemse et al., 2007). Thus, the use of the CANTAB in Taiwanese families places this study in a position to advance the literature. Taken together, an investigation of the executive function as a potential endophenotype for ADHD in a large, ethnic Chinese population in Taiwan, by comprehensively assessing the verbal working memory and nonverbal executive functions using the CANTAB in ADHD adolescents, their behaviorally unaffected siblings, and unaffected controls, is warranted. We hypothesized that unaffected siblings, similar to adolescents with ADHD, are more likely to have deficits in verbal and visuo-spatial executive functions than unaffected healthy controls.
Methods
Participants
The sample consisted of 279 probands with ADHD, 108 unaffected siblings, and 173 school controls. The probands were 279 patients (male, 85.7%) who had had overt symptoms noted at ages 4.2 ± 1.6 years and were clinically diagnosed with DSM-IV ADHD at the mean age of 6.7 ± 2.9 years. They were recruited consecutively, mainly from the child psychiatric clinic of National Taiwan University Hospital (n = 240, 86%). Their current and lifetime ADHD and other psychiatric diagnoses were confirmed by the Chinese Kiddie epidemiologic version of the Schedule for Affective Disorders and Schizophrenia (K-SADS-E) interview at the mean age of 12.5 ± 1.6 years. We recruited 136 biological siblings (unaffected, 108), who were 8 years old or older and assessed by using the Chinese K-SADS-E at the mean age of 12.2 ± 3.3 years; and 173 controls (male, 72.8%) from the same school district as the ADHD probands, who were assessed to be without lifetime ADHD by inquiring into childhood and current symptoms using the Chinese K-SADS-E at the mean age of 12.6 ± 1.5 years. All participants who had a clinical diagnosis of psychosis or autism spectrum disorders or an intelligence quotient (IQ) score less than 80 were excluded.
Measures
Chinese K-SADS-E. The Chinese K-SADS-E was prepared by a two-stage translation and modification of several items with psycholinguistic equivalents relevant to the Taiwanese culture and further modification to meet the DSM-IV diagnostic criteria (Gau, Chong, Chen, & Cheng, 2005; Gau & Soong, 1999). It has been extensively used in a variety of clinical (Gau, Huang et al., 2007) and community studies (Gau et al., 2005; Gau, Chong et al., 2007) in Taiwan. The details of interview training have been described elsewhere (Gau & Chiang, 2009; Gau et al., in press) and are provided upon request.
Best estimate of diagnosis. The best estimate of each diagnostic category was made by S.S. Gau, who was blind to the diagnostic status and name of the participant, and who was not involved in direct K-SADS-E interviews of any of the participants or their parents at follow-up. The diagnosis was made based on the K-SADS-E interviews of the participants and their mothers, medical records, and teachers’ reports. The diagnostic coding was categorized into definite (reaching full DSM-IV diagnostic criteria), probable (either not reaching full, but more than half of the DSM-IV symptoms criteria, or no functional impairment), possible (some symptoms but no impairment), and no diagnosis. Those patients who received a rating as definite or probable by best estimate were categorized as having a particular mental disorder.
We categorized patients with a childhood diagnosis of ADHD into (1) patients with persistent ADHD, if their current symptoms reached the definite category (meeting all DSM-IV ADHD diagnostic criteria including impairment), based on best estimate, and (2) otherwise, patients without persistent ADHD.
The CANTAB involving executive abilities
Intra-dimensional/Extra-dimensional Shifts (IED). The IED assesses the ability to maintain attention to different stimuli within a relevant dimension (intra-dimensional shift, IDS) and then shift attention to a previously irrelevant dimension (extra-dimensional shift, EDS) (Luciana & Nelson, 1998). The task consists of nine stages. The participants progressed through the tests by satisfying a set criterion of learning at each stage (6 consecutive correct responses). If the participant failed to reach this criterion after 50 trials at any stage, the test was terminated. Six indices were presented: (1) the total numbers of stages successfully completed; (2) completed stage trials: the numbers of trials undertaken on all completed stages; (3) EDS errors: errors made in the EDS stage; (4) pre-EDS errors: errors made prior to the EDS stage; (5) total errors: the numbers of errors required to complete the stages; and (6) adjusted total errors: the adjusted score calculated by adding 25 for each stage not attempted due to failure (this value of 25 was used since the participants had to complete 50 trials to fail a stage and half of these could be correct by chance alone).
Spatial Span (SSP). The SSP, similar to the Corsi blocks task (Milner, 1971), measures spatial short-term memory. This task requires the ability to remember the order in which visual stimuli are presented. At the beginning, nine white boxes were presented in fixed locations on the screen. In following, the boxes changed color, one after the other, in a predetermined sequence, and the end of the sequence was indicated by a tone. Then, the participants were asked to point to the boxes in the order in which they had changed color. The test began with 2-box problems, and then up to 9-box problems. Three indices were reported: (1) span length: the longest sequence successfully recalled; (2) total errors: the number of times an incorrect box was selected; and (3) total usage errors: the number of times a box was selected that was not in the sequence being recalled.
Spatial Working Memory (SWM). The SWM, based on a self-ordered search test (Petrides & Milner, 1982), assesses nonverbal working memory. Participants were asked to search through a number of colored boxes presented on the screen to find blue tokens hidden inside. Each box housed only one token per trial. Searching a box more than once during a sequence resulted in within errors, and returning to an emptied box resulted in between errors. A double error could be categorized as both a within and a between error. Two major indices were presented: (1) strategy utilization: the number of search sequences starting with a novel box in both 6- and 8-box problems, and (2) errors in total and three different levels of difficulty (4-, 6-, and 8-box problems): the total errors were calculated based on the between errors, within errors, and double errors of particular box problems (i.e., between errors + within errors – double errors).
Stockings of Cambridge (SOC). The SOC assesses spatial planning based on the Tower of London (Shallice, 1982), and requires participants to plan and execute a set of movements to replicate a goal arrangement of balls. At the beginning of each trial, three suspended vertical stockings and three colored balls are presented on the monitor screen. Participants were required to move the colored balls in a single move at a time between the stockings, to fulfill a goal position within a specified number of moves (2, 3, 4, and 5 moves) in the problem-solving condition. In the control condition, they were required to follow the ball movements as quickly as possible to provide baseline measures of reaction and movement times, with each trial being an exact replication of their earlier planning moves. Three major indices were presented: (1) number of problems solved in the specified minimum number of moves; (2) mean moves: the number of moves taken in excess of the specified minimum number, but within the maximum allowed; (3) initial thinking time: the difference in reaction time taken to select the first ball for the same problem under the two conditions, measuring cognitive impulsivity.
Temporal stability of the CANTAB. The test–retest reliability (intraclass correlation, ICC) for the CANTAB among 10 subjects revealed adequate temporal stability at an interval ranging from 14 days to 42 days (mean 22 ± 9.1) for the IED (ICC = .78∼1.00), SSP (ICC = .55∼1.00), SWM (ICC = .94∼.99), and SOC (ICC = .72∼1.00).
Digit span
Digit span is one of the subtests of the Wechsler Intelligence Scale for Children – 3rd edition (WISC-III); it first requires repetition of orally presented digit strings (digit forward) and then requires recall of the digits in a backward sequence (digit backward), which were used to represent the index of sustained attention and verbal working memory, respectively.
Procedure
The Research Ethics Committee of National Taiwan University Hospital approved this study prior to its implementation from January 2005 to December 2007. Written informed consent was obtained from the participants and parents. All participants received the same psychiatric and neuropsychological assessments. The participants and their mothers were interviewed independently by separate well-trained interviewers, and the participants’ DSM-IV psychiatric diagnoses at baseline and adolescence (past 6 months) were made using the Chinese K-SADS-E. The participants also underwent the CANTAB first, followed by the WISC-III. The participants, who took methylphenidate, were asked to halt medication for at least 24 hours before the tests.
Data analyses
We used SAS 9.1 software (SAS Institute Inc., Cary, NC, USA) to perform data analysis. The three comparison groups were probands with ADHD, unaffected siblings, and school controls. The descriptive results were displayed as frequency and percentage for categorical variables, and for continuous variables, mean and SD. We used a multi-level model with random and fixed effects to address the lack of independence of the probands and their siblings within the same family. The Proc Glimmix procedure with binomial distribution and logit link for the non-linear mixed model was used to compare the rate of psychiatric disorders. We used a linear multi-level model to compare the CANTAB performance and the Bonferroni method to adjust p values in post hoc analysis due to multiple comparisons, and we controlled for sex, age, IQ, comorbidity, and parental educational levels in the statistical model. Cohen’s d was used to compute the effect size (standardized difference between the two means) for the group comparisons, with the small, medium, and large effect sizes as Cohen’s d .2 to .5, .5 to .8, and ≥.8, respectively (Cohen, 1988).
For those tests with different levels of difficulty (SWM and SOC), we adjusted the repeated measures within the same subjects while we examined the interaction between the group and task difficulty. We used the Goodness of Fit test to compare the model treating the three groups in an order ranging from unaffected controls, unaffected siblings to ADHD probands, and as a categorical variable to the model treating the three groups as an ordinal variable. We found that the three groups can be treated as an ordinal variable. Then we tested the significance of the linear trend across the three groups in an order ranging from unaffected controls, unaffected siblings to ADHD probands on the CANTAB performance.
We further examined the effects of persistent ADHD, comorbidities, and current use and duration of methylphenidate treatment on executive functions among the ADHD probands (Supplementary Table 2). There was no interaction between the three groups and sex or age in the performance of the CANTAB and digit span, so the results were not stratified by sex and age in all our analyses. The alpha value was pre-selected at the level of p < .01.
Results
Sample description
The results showed that there were significant group differences in sex, parental educational levels, and mothers’ ages; and that 141 ADHD probands have been taking methylphenidate, with the treatment duration of 20 months (Table 1). The ADHD subtype distributions at the first clinical diagnosis of the probands with ADHD were 160 (57.3%) for combined type, 93 (33.3%) for predominantly inattentive type, and 26 (9.3%) for predominantly hyperactive-impulsive type. At adolescence, 121 probands (43.4%) met the criteria of inattentive type, followed by 102 (36.6%) combined type and 27 (9.7%) hyperactive-impulsive type; 29 probands (10.4%) did not reach the full criteria of DSM-IV ADHD (Table 2). The weighted kappa for the temporal agreement of ADHD subtypes was .50 (95% CI, .40, .60). The details of the changes of ADHD subtypes can be provided upon request (Supplementary Table 1).
| ADHD (N = 279) | Sibling (N = 108) | Control (N = 173) | F value or χ2 | |
|---|---|---|---|---|
| Mean (SD) or % | Mean (SD) or % | Mean (SD) or % | ||
| Age at assessment (in years) | 12.5 (1.6) | 12.5 (3.4) | 12.6 (1.5) | .06 |
| Gender, male, % | 85.7 | 40.7 | 72.8 | 79.79*** |
| Male–female ratio | 6.0 | .69 | 2.7 | |
| Current use of methylphenidate | 50.9 | |||
| Duration of methylphenidate treatment (in months) | 20.1 (22.1) | 4.31 | ||
| IQ | 103.0 (11.6) | 103.0 (10.6) | 110.7 (9.5) | 29.97*** |
| Mothers | ||||
| Current age (in years) | 42.7 (4.2) | 43.6 (3. 9) | 4.84* | |
| Educational level | ||||
| College or higher | 50.6 | 63.9 | 7.16* | |
| Senior high school | 40.0 | 29.7 | ||
| Junior high or lower | 9.4 | 6.4 | ||
| Employment status | ||||
| Professional | 4.1 | 6.2 | 4.09 | |
| Skilled work | 59.3 | 66.4 | ||
| Others | 36.7 | 27.4 | ||
| Fathers | ||||
| Current age (in years) | 45.7 (4.8) | 46.2 (4.5) | 1.07 | |
| Educational level | ||||
| College or higher | 59.9 | 76.0 | 11.41** | |
| Senior high school | 29.6 | 18.4 | ||
| Junior high or lower | 10.5 | 5.7 | ||
| Employment status | ||||
| Professional | 14.8 | 20.6 | 2.24 | |
| Skilled work | 79.9 | 74.0 | ||
| Others | 5.3 | 5.5 | ||
- Note: IQ = intelligence quotient; SD = standard deviation.
- *p < .05; **p < .01; ***p < .001.
| Psychiatric diagnoses | ADHD (N = 279) | Sibling (N = 108) | Control (N = 173) | Odds Ratio (95% CI) or p* | |||||
|---|---|---|---|---|---|---|---|---|---|
| N | (%) | N | |||||||
| (%) | N | (%) | ADHD vs. Control | Sibling vs. Control | ADHD vs. Sibling | ||||
| Attention-deficit/hyperactivity disorder | 250 | (89.6) | 0 | (0) | 0 | (0) | – | – | – |
| Combined type | 102 | (36.6) | 0 | (0) | 0 | (0) | – | – | – |
| Inattentive type | 121 | (43.4) | 0 | (0) | 0 | (0) | – | – | – |
| Hyperactive-impulsive type | 27 | (9.7) | 0 | (0) | 0 | (0) | – | – | – |
| Oppositional defiant disorder | 155 | (55.6) | 12 | (11.1) | 12 | (6.9) | 17.19(9.05–32.64) | 1.75(.75–4.11) | 9.80(5.09–18.87) |
| Conduct disorder | 54 | (19.4) | 4 | (3.7) | 2 | (1.2) | 20.80(4.92–88.00) | 3.42(.60–19.40) | 6.08(2.12–17.47) |
| Anxiety disorders | 77 | (27.6) | 32 | (29.6) | 33 | (19.1) | 1.63(1.02–2.62) | 1.80(1. 01–3.20) | .91(.55–1.50) |
| Mood disorders | 36 | (12.9) | 7 | (6.5) | 7 | (4.0) | 3.54(1.53–8.23) | 1.6 4(.55–4. 88) | 2.16(.92–5.05) |
| Depressive disorders | 28 | (10.0) | 5 | (4.6) | 7 | (4.0) | 2.67(1.13–6.31) | 1.15(.35–3.77) | 2.32(.86–6.24) |
| Bipolar disorders | 8 | (2.9) | 2 | (1.9) | 0 | (0) | .024* | .072* | .566* |
| Tic disorder | 18 | (6.5) | 1 | (.9) | 3 | (1.7) | 3.94(1.13–13.78) | .55(. 06–5.51) | 7.16(.92–55.65) |
- Note: CI = confidence interval.
- * Fisher exact p value.
Psychopathology
Twenty-eight siblings (20.6%) were diagnosed with DSM-IV ADHD and were excluded from the data analyses. ADHD probands were more likely than unaffected controls to have oppositional defiant disorder (ODD), conduct disorder (CD), tics, depressive disorders, and anxiety disorders, and were more likely than their unaffected siblings to have ODD and CD (Table 2). The unaffected siblings had higher rates of anxiety disorders than the controls (Table 2).
Executive functions ( Table 3)
Digit Span. ADHD probands had significantly fewer digits recalled forward (Cohen’s d, .25) and backward (Cohen’s d, .48) than unaffected controls. Unaffected siblings recalled fewer digits backward than unaffected controls (Cohen's d, .36, Table 3).
| Variables, Mean (SD) | ADHD (N = 279) | Unaffected sibling (N = 108) | Control (N = 173) | Univariate analysis | Multivariate analysis† | β | Cohen’s d | ||||
|---|---|---|---|---|---|---|---|---|---|---|---|
| F | Comparison | F | Comparison | A vs. S | A vs. C | S vs. C | |||||
| Digit Span (Verbal Executive Functioning) | |||||||||||
| Digit span, forward | 8.28 (.96) | 8.37 (.89) | 8.50 (.79) | 3.48* | A<C | 7.01** | A<S,C | −.11** | .10 | .25 | .15 |
| Digit span, backward | 5.29 (1.69) | 5.50 (1.61) | 6.08 (1.58) | 11.93*** | A,S<C | 9.09*** | A<C | −.39*** | .13 | .48 | .36 |
| Intradimension/Extradimension Shift | |||||||||||
| Trials of completed stages | 76.49 (19.98) | 73.15 (17.06) | 73.84 (15.23) | 1.87 | – | 1.27 | – | 1.42 | −.18 | −.15 | .04 |
| Extra-dimensional shift errors | 11.20 (10.34) | 12.20 (10.65) | 8.78 (9.32) | 4.46* | A,S>C | 1.94 | – | 1.08* | .10 | −.25 | −.34 |
| Pre-extra-dimensional shift errors | 7.84 (4.70) | 7.35 (3.68) | 6.92 (3.89) | 3.00 | – | .48 | – | .48* | −.12 | −.21 | −.11 |
| Completed stages | 8.40 (1.02) | 8.34 (1.00) | 8.63 (.87) | 3.86* | A<C | 1.62 | – | −.11* | −.06 | .24 | .31 |
| Total errors | 23.45 (13.29) | 22.04 (12.39) | 18.38 (11.40) | 8.72*** | A>C | 4.52* | A>C | 2.47*** | −.11 | −.41 | −.31 |
| Total errors(adjusted) | 30.26 (24.89) | 30.14 (23.47) | 22.71 (21.97) | 5.62** | A>C | 2.21 | – | 3.48** | −.00 | −.32 | −.33 |
| Spatial Span | |||||||||||
| Span length | 6.62 (1.49) | 6.87 (1.66) | 7.55 (1.30) | 21.70*** | A,S<C | 13.78*** | A<C | −.45*** | .16 | .67 | .46 |
| Total errors | 14.28 (6.85) | 12.58 (6.56) | 12.91 (6.65) | 3.54* | – | 1.29 | – | .74* | −.25 | −.20 | .05 |
| Total usage errors | 2.11 (1.59) | 1.55 (1.50) | 1.43 (1.37) | 12.72*** | A>S,C | 4.95** | A>S,C | .35*** | −.36 | −.46 | −.08 |
| Spatial Working Memory | |||||||||||
| Total errors | 29.56 (17.53) | 26.58 (17.66) | 17.31 (13.33) | 30.17*** | A,S>C | 18.24*** | A,S>C | 5.94*** | −.17 | −.79 | −.59 |
| 4 box problems | .89 (1.60) | .99 (1.85) | .36 (.95) | 8.37*** | A,S>C | 2.70 | – | .24*** | .06 | −.40 | −.43 |
| 6 box problems | 7.84 (6.30) | 6.84 (6.32) | 4.24 (5.15) | 19.60*** | A,S>C | 10.69*** | A>C | 1.76*** | −.16 | −.63 | −.45 |
| 8 box problems | 20.84 (12.03) | 18.93 (12.28) | 12.72 (9.56) | 27.56*** | A,S>C | 17.63*** | A,S>C | 3.94*** | −.16 | −.75 | −.56 |
| Strategy utilization | 34.09 (4.59) | 33.66 (4.86) | 31.95 (4.66) | 11.46*** | A,S>C | 5.43** | A>C | 1.04*** | −.09 | −.46 | −.36 |
| Stockings of Cambridge | |||||||||||
| Problems solved in minimum moves | 7.78 (2.05) | 8.06 (1.93) | 8.80 (1.93) | 13.88*** | A,S<C | 5.67** | A<C | −.49*** | .14 | .51 | .38 |
| Total moves | 18.08 (2.29) | 17.57 (2.07) | 16.88 (2.07) | 16.26*** | A,S>C | 7.15* | A>C | .59*** | −.23 | −.55 | −.33 |
| 2 move problem | 2.04 (.20) | 2.05 (.24) | 2.03 (.20) | .32 | – | .30 | – | .00 | .05 | −.05 | −.09 |
| 3 move problem | 3.30 (.57) | 3.29 (.54) | 3.21 (.44) | 1.76 | – | .08 | – | .05 | −.02 | −.18 | −.16 |
| 4 move problem | 5.67 (1.09) | 5.49 (1.03) | 5.28 (1.04) | 7.06** | A>C | 3.75* | A>C | .19*** | −.17 | −.37 | −.20 |
| 5 move problem | 7.14 (1.52) | 6.75 (1.38) | 6.36 (1.28) | 16.59*** | A>S,C | 8.57*** | A>C | .39*** | −.27 | −.56 | −.29 |
| Mean initial thinking time(ms) | 3669.92 (2254.19) | 3779.31 (2223.85) | 4904.03 (2816.78) | 13.47*** | A,S<C | 8.28*** | A,S<C | −568.8*** | .05 | .48 | .44 |
| 2 move problem | 1305.24 (950.36) | 1221.85 (1076.20) | 1379.26 (1000.46) | .72 | – | .86 | – | −30.0 | −.08 | .08 | .15 |
| 3 move problem | 3413.27 (2453.87) | 3129.58 (2420.21) | 3578.50 (2308.79) | 1.26 | – | 1.75 | – | −49.3 | −.12 | .07 | .19 |
| 4 move problem | 4890.07 (3737.23) | 4358.06 (3193.88) | 5990.34 (4303.43) | 5.82** | A,S<C | 3.66* | S<C | −414.8* | −.15 | .27 | .43 |
| 5 move problem | 5540.52 (5210.22) | 6492.62 (5386.51) | 8961.20 (7211.99) | 17.34*** | A,S<C | 9.77*** | A,S<C | −1652.1*** | .18 | .54 | .39 |
- Note: SD = standard deviation; A = ADHD; S = Sibling; C = Control; β = regression coefficient estimates for linear trend from the controls, unaffected siblings, to patients with ADHD.
- † controlling for sex, age, IQ, comorbidity, and parental educational levels
- *p < .05; **p < .01; ***p < .001.
Intra-dimensional/Extra-dimensional Shifts. ADHD probands (as well as unaffected siblings) had more EDS errors, and more total raw and adjusted errors than unaffected controls, with small effect sizes (Table 3). The significant differences, except total raw errors, disappeared in multivariate analyses.
Spatial Span. Both univariate and multivariate ana-lyses revealed that ADHD probands (Cohen’s d, .67) and unaffected siblings (Cohen’s d, .46) had significantly shorter span sequences successfully recalled than unaffected controls; and ADHD probands had more total usage errors than the other two groups, with small effect sizes (Cohen's d, .36, Table 3).
Spatial Working Memory. ADHD probands and unaffected siblings showed poorer use of strategy, with small effect sizes, and had more total errors in searching the box than the controls, with medium effect sizes; the significant differences were the same in the 4-, 6- and 8-box searches (Table 3). The majority of patterns of significant differences remained in multivariate analyses, except for the 4-box search.
Further interaction analyses revealed significant main effects from Group and Task Difficulty, and from a Group × Task Difficulty interaction (Figure 1a). Table 4 presents the significance of each level of interaction by backward model selection, revealing interactions between ADHD (vs. control) and Task Difficulty (6-box vs. 4-box, 8-box vs. 4-box), and between unaffected siblings (vs. control) and Task Difficulty (8-box vs. 4-box) controlling for all the confounding variables.
a) Total errors in the 4-box, 6-box, and 8-box problems of the Spatial Working Memory task; b) mean moves, and c) mean initial thinking time in the 2-move, 3-move, 4-move, and 5-move problems of the Stockings of Cambridge task for probands with ADHD, unaffected siblings, and the controls
| β | 95% CI | F statistics | p† | |
|---|---|---|---|---|
| Spatial Working Memory | ||||
| Total Errors | ||||
| ADHD vs. Control | .90 | (−.45, 2.25) | 1.71 | .192 |
| Sibling vs. Control | 1.60 | (.15, 3.05) | 4.70 | .030 |
| 6- vs. 4-box problem | 4.63 | (3.61, 5.65) | 77.90 | <.001 |
| 8- vs. 4-box problem | 12.73 | (11.48, 13.98) | 401.37 | <.001 |
| ADHD*(6- vs. 4-box problem) | 2.32 | (.87, 3.77) | 9.74 | .002 |
| ADHD*(8- vs. 4-box problem) | 7.22 | (5.59, 8.85) | 76.59 | <.001 |
| Sibling*(8- vs. 4-box problem) | 4.47 | (2.65, 6.29) | 22.91 | <.001 |
| Stockings of Cambridge | ||||
| Mean moves | ||||
| ADHD vs. Control | .07 | (−.07, .21) | 1.17 | .280 |
| Sibling vs. Control | .10 | (−.04, .24) | 1.85 | .174 |
| 3- vs. 2-move problem | 1.24 | (1.14, 1.34) | 579.90 | <.001 |
| 4- vs. 2-move problem | 3.34 | (3.20, 3.48) | 2424.01 | <.001 |
| 5- vs. 2-move problem | 4.38 | (4.22, 4.54) | 2812.79 | <.001 |
| ADHD*(4- vs. 2-move problem) | .27 | (.09, .45) | 9.46 | .002 |
| ADHD*(5- vs. 2-move problem) | .72 | (.52, .92) | 51.57 | <.001 |
| Sibling*(5- vs. 2-move problem) | .29 | (.05, .53) | 5.69 | .017 |
| Initial thinking time | ||||
| ADHD vs. Control | −118.38 | (−684.78, 448.02) | .17 | .682 |
| Sibling vs. Control | −303.08 | (−1020.73, 414.57) | .69 | .408 |
| 3- vs. 2-move problem | 2094.03 | (1714.05, 2474.02) | 116.66 | <.001 |
| 4- vs. 2-move problem | 4584.51 | (3962.97, 5206.05) | 209.01 | <.001 |
| 5- vs. 2-move problem | 7546.75 | (6923.92, 8169.58) | 564.03 | <.001 |
| ADHD*(4- vs. 2-move problem) | −973.42 | (−1727.73, −219.11) | 6.40 | .012 |
| ADHD*(5- vs. 2-move problem) | −3284.02 | (−4041.05, −2526.99) | 72.29 | <.001 |
| Sibling*(4- vs. 2-move problem) | −1338.79 | (−2296.64, −380.94) | 7.50 | .006 |
| Sibling*(5- vs. 2-move problem) | −2168.25 | (−3125.08, −1211.42) | 19.73 | <.001 |
- Note: CI = confidence interval; β = regression coefficient estimates
- † controlling for sex, age, IQ, comorbidity, treatment with methylphenidate, and parental educational levels.
Stocking of Cambridge. ADHD probands and unaffected siblings solved fewer problems in the minimum number of moves, mean moves, and shorter initial thinking time, with small to medium effect sizes (absolute Cohen’s d, .33 ∼ .56; Table 3), particularly in the 4-move and 5-move tasks. The majority of patterns of significant differences remained in multivariate analyses.
Further testing for the interactions on number of moves (Figure 1b) and initial thinking times (Figure 1c) showed significant effects from Group, Task Difficulty, and Group × Task Difficulty.
Table 4 presents the significant interactions between ADHD and Task Difficulty on the number of moves (4-move vs. 2-move, 5-move vs. 2-move), and between unaffected siblings and Task Difficulty (5-move vs. 2-move). There were significant interactions between ADHD and Task Difficulty on initial thinking time (4-move vs. 2-move, 5-move vs. 2-move), and between unaffected siblings and Task Difficulty (4-move vs. 2-move, 5-move vs. 2-move).
Linear trend of three groups in the executive task performance
There were significant linear trends in the CANTAB performances across unaffected controls, unaffected siblings, and ADHD probands in terms of digit spans, total errors of the IED, spatial length and total usage errors in the SSP, total errors and spatial strategy utilization in the SWM, and problems solved in minimum moves, total moves, and mean initial thinking time (all p < .01, Table 3).
Effect of persistent ADHD and comorbidity
We did not find any significant difference in executive function between probands with (n = 251) and without persistent ADHD (n = 25, p, .079∼.989), except that probands with persistent ADHD had fewer problems solved in minimum moves in the SOC (β = −.81, p = .037) than their counterparts. There was no difference between probands with (n = 212) and without (n = 68) psychiatric comorbidity (p, .144∼.905), except that comorbidity was associated with increased errors prior to the EDS (p = .038).
Probands treated with methylphenidate in the past 6 months (n = 141) had more total usage errors (β = .39, p = .044) in the SSP, and more errors in the 6-box problems (β = 1.70, p = .024) and poorer usage of strategies (β = 1.14, p = .037) in the SWM than probands who had not received methylphenidate treatment in the past 6 months (n = 136). Increased duration of treatment with methylphenidate was correlated to decreased spatial length (β = −.01, p = .003) and increased total usage errors (β = .01, p = .027) in the SSP; increased total errors (β = .11, p = .046), particularly in 8-box problem tasks (β = .07, p = .040), and poorer strategy utilization (β = .03, p = .030) in the SWM; and a decreased number of problems solved in minimum moves (β = −.01, p = .020), an increased number of moves in 2-move problems (β = .00, p = .016) and 3-move problems (β = .00, p = .027), and decreased initial thinking time in 3-move problems (β = −16.25, p = .030) in the SOC.
The detailed results regarding the effects of persistent ADHD, comorbidity, and treatment with methylphenidate (use and duration) on the executive function among ADHD probands are provided in Supplementary Table 2.
Discussion
The current study is the first to comprehensively examine executive functions, which consisted of low (spatial span and digit span forward) and high (verbal and spatial working memory) executive tasks, by using the CANTAB and digit spans in a large sample of ADHD probands and their unaffected siblings. Our findings consistently demonstrated that despite no obvious ADHD symptoms, unaffected siblings (Rommelse, Altink et al., 2008), like ADHD probands, performed significantly worse than unaffected controls in verbal working memory measured by digit spans backward, and in the majority of nonverbal executive functions, such as spatial short-term memory, spatial working memory, spatial planning, and response inhibition, measured by the CANTAB. These findings indicate the significant familiarity of executive dysfunction in ADHD, consistent with the idea that subtle cognitive traits may be more closely linked to the underlying genetic factors than the behavioral phenotype (Slaats-Willemse et al., 2007). And, this strongly implies that executive functions measured by the digit span backward and the CANTAB fulfill some of the important criteria of an endophenotype (Doyle et al., 2005): executive dysfunctions co-occur with ADHD and are manifested in unaffected relatives. The CANTAB is also shown to be a suitable instrument with good psychometric properties. Our results suggest that studies on executive dysfunction in ADHD probands and their unaffected siblings can shed light on the effort to explore the genetic etiology of this disorder (Nigg et al., 2004; Rommelse, Altink, Oosterlaan, Buschgens, Buitelaar, and Sergeant, 2008a; Seidman et al., 2000; Slaats-Willemse et al., 2005, 2007).
Although some studies failed to find significant deficits in some neuropsychological tests with unaffected siblings of ADHD (Asarnow et al., 2002; Seidman et al., 2000), our findings support the sensitivity and usefulness of the CANTAB tasks in assessing the executive dysfunction associated with ADHD in unaffected siblings.
The current study not only confirmed the findings of Western studies (Martinussen et al., 2005) but also those of our previous studies on 69 children with ADHD aged 7–10 (Chiang & Gau, 2008) and 53 adolescents with ADHD aged 11–16 (Gau et al., 2009), as compared to school children and adolescents without ADHD in Taiwan, that children and adolescents with ADHD performed worse in nonverbal executive functions, yet less significantly in set-shifting tasks (Pennington & Ozonoff, 1996), than unaffected controls, but also provided some evidence to support a worse performance in verbal sustained attention and working memory in ADHD (Chiang & Gau, 2008). The large sample size makes our findings regarding impaired verbal working memory in ADHD more convincing relative to the negative findings in our previous study (Gau et al., 2009), and others (Martinussen et al., 2005). In line with our previous work (Gau et al., 2009), the vulnerability to deficits in executive functions in ADHD probands emerged more remarkably with systematically increased task difficulties. Such a relationship was also clearly demonstrated in unaffected siblings, a novel finding of this study leading to important clinical implications for unaffected siblings of ADHD.
Given that ADHD probands had a worse performance in executive tasks than unaffected controls, the findings that there were no differential performances for unaffected siblings of ADHD probands from ADHD probands, or from unaffected controls, or that unaffected siblings performed worse than unaffected controls in digit spans and all the CANTAB tasks except for total usages errors in the SSP, may reflect the fact that the unaffected siblings occupy an intermediate position. The significant linear trends in the CANTAB performances across the three groups regarding sustained attention and verbal working memory measured by digit spans, errors in set-shifting tasks, spatial short-term memory length and errors, total errors and strategy utilization in spatial working memory, and in spatial planning, problem solving and cognitive impulsivity measured by the SOC (Table 3) imply that ADHD may be considered as the extreme of a cognitive-behavioral phenotype with genetic susceptibility throughout the population (Slaats-Willemse et al., 2003).
Although individual measures of executive function, such as response inhibition (Slaats-Willemse et al., 2003), have been identified as potential endophenotypes for ADHD, the use of isolated measures may be problematic and the results may be inconsistent across settings. In contrast, our findings have shown that a range of executive function components may be relevant to the familial risk for this disorder. Because children with ADHD show a range of deficits on measures of executive function, a comprehensive battery approach is assumed to be maximally informative.
In contrast, ADHD probands performed worse than unaffected siblings and controls in total usage errors in the SSP, and problems solved in 5-move tasks in the SOC. These cognitive deficits may not relate to a familial predisposition for ADHD and may not be suitable for an ADHD endophenotype. These deficits may be caused by the presence of ADHD itself or may be the cause of ADHD, or relate to some risk factors not shared between the probands and their unaffected siblings (Durston et al., 2004).
The lack of an effect of persistent ADHD on verbal and the majority of nonverbal executive functions (Gau et al., 2009; Seidman et al., 2000) suggests that the underlying executive dysfunction associated with ADHD may be enduring into adolescence, regardless of reduction in ADHD symptoms. Accordingly, these findings suggest that executive dysfunction may be independent of ADHD symptoms as it is seen in unaffected probands. Therefore, executive dysfunction can probably be treated as a trait marker for ADHD.
As suggested by others (Chen et al., 2008), an increased risk of ADHD in siblings suggests that ADHD runs in the family. Significantly higher comorbidity with other psychiatric disorders lends evidence to support the findings from numerous studies (Faraone, Biederman, Mennin, Gershon, & Tsuang, 1996). Although ADHD probands were more likely than their unaffected siblings to have ODD and CD, previous research has not revealed more severe executive dysfunctions in ADHD combined with ODD/CD than for ADHD alone (Oosterlaan, Scheres, & Sergeant, 2005). Lack of an effect of comorbidity on most executive functions in ADHD and significant executive dysfunction in ADHD probands and unaffected siblings after adjusting for comorbidity suggest that psychiatric comorbidity is unlikely to account for the executive dysfunction associated with ADHD (Seidman, Biederman, Faraone, Weber, & Ouellette, 1997; Seidman et al., 2000; Willcutt, Doyle, Nigg, Faraone, & Pennington, 2005).
Methodological consideration
This study is the first to investigate, with satisfactory internal validity, executive function in unaffected siblings in an ethnic Chinese population, and the first to use the CANTAB in unaffected siblings of ADHD. The direct interviews with the participants and their mothers may have minimized the misclassification of the psychiatric diagnosis, e.g., ADHD in unaffected siblings. Moreover, the large sample size and comprehensive assessments of a wide range of verbal and nonverbal executive functions using standardized neuropsychological tests (CANTAB and WISC-III) (Sonuga-Barke, 2005) and satisfactory psychometric properties (Gau et al., 2009; Luciana, 2003; Luciana & Nelson, 1998) make our findings much more convincing.
The major methodological limitation is the questionable generalization of our findings to community-based samples. However, a meta-analysis has reported a similar effect size on the executive function between community-based and clinic-based studies (Willcutt et al., 2005). Moreover, as a naturalistic design, although the currently medicated participants held medication for at least 24 hours before receiving neuropsychological tests, the concerns about the medication effect on the executive function remain. However, given that many ADHD children receive pharmacotherapy, the executive function of treated samples is relevant to clinical practice (Seidman et al., 1997). Moreover, the use of only one verbal task (digit span forward and backward) is another limitation of this study. Lastly, this study is limited by the fact that the majority of patients (86%) were male. Hence, further studies of relatives of females with ADHD are warranted.
Implications
Taken together, the findings suggest that verbal and nonverbal executive dysfunctions measured by digit spans and the CANTAB may be independent of ADHD symptoms as it is seen in unaffected probands and unaffected siblings; therefore, it may constitute a cognitive endophenotype for ADHD. The executive function as an endophenotype approach may increase the power to detect susceptibility loci and candidate genes, and can also be used as a paradigm for functional brain imaging studies in ADHD. In addition to providing educational assistance to adolescents with childhood diagnosis of ADHD, regardless of declining ADHD symptoms at adolescence, particularly when complex tasks are assigned to them, the clinical implications of this study are that screening and preventive interventions for executive dysfunction may be also needed for their unaffected siblings as well.
Key points
- •
ADHD is associated with deficits in some executive functions, with greater deficits in visuo-spatial than verbal executive functions.
- •
Little is known about whether unaffected siblings also have increased risks for executive dysfunctions. No study has examined the executive functions among unaffected siblings using the CANTAB, or in Asian populations.
- •
Deficits in several executive functions, such as short-term spatial memory, verbal and spatial working memory, spatial planning, and sustained attention noted in unaffected siblings without much difference from those in ADHD adolescents suggest that executive dysfunctions may be useful cognitive endophenotypes for ADHD.
- •
The clinical implications are that educational assistance should be provided to unaffected siblings of individuals with ADHD, particularly, when complex tasks are assigned to them.
Acknowledgements
This work was supported by grants from the National Health Research Institute (NHRI-EX94-9407PC, NHRI-EX95-9407PC, NHRI-EX96-9407PC, NHRI-EX97-9407PC), Taiwan. The authors would like to thank Ming-Fang Chen, Yi-Chun Lai, Chi-Mei Lee, Hsin-Yi Luo, and Wang-Ling Tseng for conducting psychiatric interviews and administering the CANTAB, and Ming-Fang Chen for her assistance in manuscript preparation and data analysis. We also thank Drs. Yu-Yu Wu and Liang-Yin Lin for referring patients with ADHD.
Correspondence to
Susan Shur-Fen Gau, Department of Psychiatry, National Taiwan University Hospital & College of Medicine, No. 7, Chung-Shan South Road, Taipei 10002, Taiwan; Email: gaushufe@ntu.edu.tw
