Accepted Manuscript Attention-Deficit/Hyperactivity Disorder (ADHD) and Motor Timing in Adolescents and their Parents: Familial Characteristics of Reaction Time Variability Vary With Age Andrieke J.A.M. Thissen, PhD Marjolein Luman, PhD Catharina Hartman, PhD Pieter Hoekstra, MD/PhD Marloes van Lieshout, MSc Barbara Franke, PhD Jaap Oosterlaan, PhD Nanda N.J. Rommelse, PhD Jan K. Buitelaar, MD/PhD PII:
S0890-8567(14)00446-8
DOI:
10.1016/j.jaac.2014.05.015
Reference:
JAAC 1033
To appear in:
Journal of the American Academy of Child & Adolescent Psychiatry
Received Date: 8 February 2013 Revised Date:
14 March 2014
Accepted Date: 10 May 2014
Please cite this article as: Thissen AJAM, Luman M, Hartman C, Hoekstra P, van Lieshout M, Franke B, Oosterlaan J, Rommelse NNJ, Buitelaar JK, Attention-Deficit/Hyperactivity Disorder (ADHD) and Motor Timing in Adolescents and their Parents: Familial Characteristics of Reaction Time Variability Vary With Age, Journal of the American Academy of Child & Adolescent Psychiatry (2014), doi: 10.1016/ j.jaac.2014.05.015. This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
ACCEPTED MANUSCRIPT Attention-Deficit/Hyperactivity Disorder (ADHD) and Motor Timing in Adolescents and their Parents: Familial Characteristics of Reaction Time Variability Vary With Age
RH = ADHD and Motor Timing in Adolescents and Parents Andrieke J.A.M. Thissen, PhD; Marjolein Luman, PhD; Catharina Hartman, PhD; Pieter Hoekstra, MD/PhD; Marloes van
RI PT
Lieshout, MSc; Barbara Franke, PhD; Jaap Oosterlaan, PhD; Nanda N.J. Rommelse, PhD; Jan K. Buitelaar, MD/PhD Supplemental material cited in this article is available online. Accepted June 30, 2014
Drs. Thissen, Rommelse, and Buitelaar are with Donders Institute for Brain, Cognition and Behaviour and Karakter Child
SC
and Adolescent Psychiatry University Center, Radboud University Nijmegen Medical Center, the Netherlands. Drs. Luman and Oosterlaan and Ms. van Lieshout are with VU University Amsterdam. Drs. Hartman and Hoekstra are with University Medical Center Groningen, University of Groningen. Dr. Franke is with Donders Institute for Brain, Cognition and
M AN U
Behaviour, Radboud University Nijmegen Medical Center.
This work was supported by grants from the National Institutes of Health (NIH; R01MH62873, to Stephen V. Faraone, PhD, which helped build the NeuroIMAGE Database), Netherlands Organization for Scientific Research (NOW) Large Investment (1750102007010, JKB), and from Radboud University Nijmegen Medical Center, University Medical Center
TE D
Groningen, Accare, and Vrije Universiteit Amsterdam.
The authors thank all the families and teachers that took part in this study and all interns for their assistance in data collection. The authors also acknowledge Rogier Donders, PhD of Radboud University Medical Center for his advice with
AC C
Gaussian analyses.
EP
regard to statistical analyses, and Marcel Zwiers, PhD of Radboud University Medical Center for his assistance with ex-
Disclosure: Dr. Hoekstra has received advisory panel payments from Shire and Eli Lilly and Co. Dr. Buitelaar has been a consultant to / member of advisory board of and/or speaker for Janssen-Cilag BV, Eli Lilly and Co., Bristol-Myers Squibb, Organon/Shering-Plough, UCB, Shire, Medice, Roche, and Servier. Drs. Thissen, Luman, Hartman, Franke, Oosterlaan, Rommelse, and Ms. Van Lieshout report no biomedical financial interests or potential conflicts of interest.
Correspondence to: Andrieke J.A.M. Thissen, Reinier Postlaan 12 6525 GC Nijmegen, the Netherlands; email: [email protected].
ACCEPTED MANUSCRIPT Abstract Objective: There is consistent evidence that attention-deficit/hyperactivity disorder (ADHD) is strongly related to impaired motor timing as reflected in decreased accuracy and increased reaction time variability (RTV). Unknown is whether (a) motor timing impairments are present in
RI PT
adolescents and adults with ADHD and their unaffected relatives to the same extent as has been reported in children and (b) ADHD and motor timing share familial underpinnings, as reflected in parent-offspring co-segregation and sibling cross-correlations.
SC
Method: 589 parents and 808 children/adolescents from families with ADHD and control families (parent/offspring average age: 48.6/17.3 years) were included. All participants were thoroughly
M AN U
assessed for ADHD and performed a 40-trial motor timing task (1 s interval production). Dependent neurocognitive measures included RT median (RTM: representing accuracy), RTV and ex-Gaussian component τ (τ: representing infrequent long response times). Generalized Estimating Equations were used for analyses.
Results: Unaffected children from families with ADHD had RTV (but not RTM or τ) scores in
TE D
between those of affected and control children. However, during middle-to-late adolescence, unaffected offspring were not impaired compared to control offspring and differed from ADHD probands, whereas during late adolescence/early adulthood, all offspring groups performed
EP
equally. Affected and unaffected parents of families with ADHD showed increased RTV compared to controls, regardless of age (not significant after adjusting for IQ). There were
AC C
indications for shared familiality between RTV and ADHD as reflected by sibling crosscorrelations, and between RTM and ADHD, as reflected by sibling cross-correlations and a maternal parent-offspring relation (parent-of-origin effect). Conclusions: RTV and its familial characteristics are influenced by development during adolescence. Increased RTV in children with ADHD appears to reflect immaturities in their neurocognitive functioning. Maternal ADHD effects might be involved in transmission of RTM (not RTV), but overall RTM showed less compelling (familial) relationships with ADHD than RTV.
ACCEPTED MANUSCRIPT
AC C
EP
TE D
M AN U
SC
RI PT
Keywords: ADHD, motor timing, reaction time variability, familiality, development
ACCEPTED MANUSCRIPT Introduction
Attention deficit/hyperactivity disorder (ADHD) is a strongly heritable neurodevelopmental disorder1 defined by a persistent pattern of age-inappropriate impairing symptoms of inattention,
RI PT
hyperactivity, and impulsivity.2 Numerous studies have focused on the neurocognitive deficits underlying ADHD. Consistent evidence was found that ADHD is not only related to executive functioning (EF) deficits, such as deficits in working memory and motor inhibition3,4, but also
SC
strongly to deficits in a range of timing functions.5,6 These functions refer to a multidimensional construct of temporal information processing reflecting the ability to deal with timekeeping in
M AN U
behaviour and thoughts. This is often investigated by using motor timing paradigms6 including two components: central timekeeping and motor response variability.7 Both components have been proven to be related to ADHD, with patients showing premature responses when estimating short intervals (involving mainly subcortical brain regions), but showing overestimations when estimating long intervals (involving mainly cortical brain regions).8,9 Furthermore, patients with
TE D
ADHD often show a pattern of infrequent, abnormally slow responses, reflected in the long tail of the response time distribution as indicated by τ (tau)10,11, as well as a high variability in responding.12 Thus far, elevated response time variability (RTV) in general--including RTV in
EP
motor timing --emerges as the most universal finding in ADHD research,12-14 and it has even been proposed to be the true core deficit of ADHD.15,16
AC C
RTV is of particular interest to ADHD researchers, given its relation to a familial predisposition to the disorder, which has been suggested by impaired performance in unaffected (twin) relatives.17,18 Motor timing RTV, more than accuracy, even proved to be one of the strongest ADHD-intermediate phenotypes (i.e. a heritable vulnerability trait along the pathway between clinical phenotype and genotype, heightening the risk for developing ADHD19) of all neurocognitive intermediate phenotype candidates studied.20 This was indicated by two findings: unaffected siblings had RTV smaller than affected siblings but larger than healthy controls, and the presence of significant sibling (cross-)correlations. This represents a clear clue to the familial-
ACCEPTED MANUSCRIPT genetic underpinnings of ADHD.19 Motor timing RTV has also been linked to molecular genetics by demonstrating strong genetic linkage to chromosome 2q21.1 (logarithm of the odds [LOD] score 3.944).21 Furthermore, it was found to tap into unique genetic aspects of ADHD, possibly related to loci conferring risk for ADHD (12q24.3 and 17p13.3).14 Therefore, motor timing RTV could be of
RI PT
help in unravelling the genetic factors involved in ADHD. Previous support for the role of motor timing RTV as an intermediate phenotype mediating the link between genes and ADHD was based mostly on young sibling and twin designs. So far, two
SC
key aspects in the relation between ADHD and motor timing are still underexposed: the cosegregation of ADHD and motor timing deficits from parents to offspring, and the role of
M AN U
developmental factors. Previous studies have found that significant parent-offspring relations exist for both ADHD and motor timing separately. These studies, however, did not examine the cosegregation of these traits, i.e. did not study whether ADHD and motor timing deficits are transmitted together.18,22,23 Earlier findings suggested this may not necessarily be the case, since no association was found between parental ADHD symptoms and offspring RTV (on a 1 s interval
TE D
motor timing task) in a study including parents and childhood offspring from the IMAGE-sample that preceded the current follow-up study five to six years ago.24 Further, selective parent-of-origin effects were detected for related functions like motor control and long time interval reproduction.
EP
This suggests that parental gender may be a relevant factor in understanding the causal pathways underlying neurocognitive deficits in ADHD and should be included in data analyses.25
AC C
The second issue is the role of developmental factors. The vast majority of ADHD neurocognitive literature focuses on children rather than adults, and the role of adolescence in the developmental trajectory of ADHD from childhood to adulthood has been studied much less. This is particularly problematic given the remission of ADHD that is mainly driven by prefrontal cortex maturation in a proportion of cases during this exact period of development.26,27 The brain structures that appear to be mostly involved in timing perception networks are the cerebellum and basal ganglia.28,29 These brain structures undergo fewer developmental changes during adolescence and young adulthood than the prefrontal regions.30 However, subjects with ADHD show reductions
ACCEPTED MANUSCRIPT in cerebellar volume throughout adolescence.31 On a behavioural level, it is found that (motor) timing deficits and motor timing variability in ADHD are present throughout adolescence 32,33 and adulthood 34, but it is unclear to what extent the co-segregation of ADHD and motor timing deficits is influenced by age. Therefore, the two main aims of the current study were to clarify whether (a)
RI PT
motor timing impairments are present in adolescents and (young) adults with ADHD and their unaffected relatives to the same extent as has been reported in children, and (b) ADHD and motor timing share familial underpinnings as reflected by co-segregation from parent to offspring and
AC C
EP
TE D
M AN U
SC
cross-correlations between siblings.
ACCEPTED MANUSCRIPT Method Participants Participants were selected from the Dutch follow-up (NeuroIMAGE [2009-2012]) of the International Multicenter ADHD Genetics (IMAGE) study performed between 2003-2006 (as
RI PT
described previously).35 Inclusion criteria for all participants were then: age between 5-19 years, European Caucasian descent, IQ ≥ 70, and no diagnosis of autism, epilepsy, general learning
difficulties, brain disorders, or known genetic disorders (such as Fragile X syndrome or Down
SC
syndrome). All family members were invited for follow-up measurement, with a mean follow-up period of 5.9 years (SD=.72). No additional inclusion criteria were applied, except a minimum age
M AN U
of 8 years for neurocognitive testing (not interviews/questionnaires). In the current study, 259 families with members having ADHD and 98 control families with at least one family member completing neurocognitive assessment participated. Mean age was 17.3 years (SD=3.7) for offspring, and 48.6 years (SD=5.0) for parents. Table 1 provides offspring and parents’
TE D
characteristics.
[TABLE_1]
EP
ADHD Diagnoses
All participants were reassessed using a semi-structured diagnostic interview (Dutch translation
AC C
of the Kaufman Schedule for Affective Disorders and Schizophrenia for School-Age Children Present and Lifetime Version: K-SADS-PL) and Conners' ADHD questionnaires to determine current ADHD diagnoses. A diagnostic algorithm was applied to combine symptom counts on the K-SADS-PL and Conners’ questionnaire (Supplement 1, available online, provides a detailed description) for calculation of the diagnostic status of the three groups included in this study: affected, unaffected, and control.
Measures
ACCEPTED MANUSCRIPT Parent and Offspring ADHD Symptoms Current ADHD information was obtained using the Dutch version of the Conners’ Adult ADHD Rating Scale - Observer: Screen Version (CAARS-O:SV) for parents and the CPRS-R:L for offspring.36 For parents and offspring, the DSM Total Score was used as a measure of current
RI PT
ADHD severity.
Motor Timing Paradigm
SC
A self-paced time production paradigm was employed to measure motor timing in parents and offspring.37 Details on the task are provided elsewhere.20 Briefly, participants were instructed to
M AN U
produce an interval of 1,000 ms in a 40 trial computer task by pressing a button as soon as they thought a 1-s time interval had elapsed after hearing a brief tone.
Motor Timing Performance Measures
Central timekeeping was measured by calculating accuracy in terms of the median of production
TE D
times (RTM) in ms, thereby limiting the influence of outliers, in line with earlier studies using a similar paradigm.37 Furthermore, response time variability (trial-by-trial variability in ms) of response timing was calculated: √ (Σ (RTi - RTi-1)2 / (n -1)), where i = trial number, n = number of
EP
trials and RT = response time.38 This measure of response time variability (RTV) correlated highly with RT standard deviation (RT SD): r=.89 in offspring and r=.93 in parents. Because the short
AC C
interval duration (1 s) yielded a positive skew in RT distribution, RTV was included as it provides for a more pure measure of variability compared to the mean dependent RT SD. In addition, exGaussian analyses were performed using MATLAB in order to acquire τ, the exponential part of the distribution with higher τ representing more infrequent long response times.10,11
Intelligence Full-scale IQ was estimated for parents and offspring by combining scores on the two subtests of the Wechsler Intelligence Scale for Children (WISC)/ Wechsler Adult Intelligence Scale (WAIS-
ACCEPTED MANUSCRIPT III) that show the highest correlation with full-scale IQ score: Vocabulary and Block Design (.88 for WISC and .90 for WAIS).39,40 Individual correlations between IQ, motor timing variables, and ADHD are provided in Table S1, available online.
RI PT
Procedure
The current study was part of a comprehensive assessment protocol (see
http://www.neuroimage.nl for an overview of the complete protocol and previous publications on
SC
one or more of the other cognitive tasks41,42). Participants were asked to withhold use of
psychostimulant drugs and other psychotropic medications for 48 hours and at least 3 days before
M AN U
assessment, respectively. After the study procedures had been explained, informed consent was signed by all participants, whereby parents signed for offspring .05), followed by a step down to the quadratic model, and so on (e.g. 44). If an interaction with age was present, parents or offspring were divided into three age groups for post
M AN U
hoc analyses (specified in the results section). Since IQ differed significantly between the three groups (for parents as well as offspring), analyses on all motor timing measures were conducted with and without IQ as a covariate. Effect sizes in the models were defined as standardized b (parent-offspring analyses), or Cohen’s d (main group comparisons).
To test whether parental ADHD and motor timing predicted offspring ADHD and motor timing,
TE D
the aforementioned model in GEE was used. In all parent-offspring analyses---executed for fathers and mothers separately--age and sex of the child, parental age, and their interactions with the main predictor were implemented in the model. If an interaction with age was present, a median split for
EP
age groups was applied for post hoc analyses (specified in the results section). Individual correlations between motor timing and ADHD symptomatology, and sibling (between first- and
AC C
second-born) and spouse (cross-)correlations for motor timing and ADHD symptomatology were calculated with correction for age difference (spouse and sibling) and sex difference (sibling).
ACCEPTED MANUSCRIPT Results
Offspring and Parent Motor Timing Group Differences All group results are provided in Table 2. Significant group x age interactions were found for
RI PT
RTV in offspring (group x age b=-6.6, p=.009 [affected vs. control]; group x age b=-7.4, p=.008 [unaffected vs. control]; group x age3 b=0.1, p=.009 [affected vs. control]). Post hoc analyses with RTV as dependent variable in three groups (children [aged ≤14 y, n=229], adolescents [aged 15-21
SC
y, n=494] and young adults [aged ≥22 y, n=85]) indicated that at childhood age, unaffected siblings performed in between affected siblings and controls (affected [m=298 SD=97] > unaffected [m=295
M AN U
SD=85] > control [m=244 SD=75]), but older unaffected siblings and affected siblings performed equally to controls in the adolescence and young adult time-window, respectively (adolescents: affected [m=269 SD=80] > unaffected [m=233 SD=73] = control [m=222 SD=76]; young adults: affected [m=226 SD=80] = unaffected [m=217 SD=79] = control [m=216 SD=70]). With regard to RTM and τ, no group x age or sex interactions appeared. On RTM, affected
TE D
siblings scored significantly lower (m=978 SD=71) than unaffected siblings (m=998 SD=75), but affected and unaffected siblings did not differ from controls (m=994 SD=65), indicating most pronounced time underestimations in affected siblings. Group comparisons with τ revealed affected
EP
siblings to show higher τ (m=0.10 SD=0.9) than controls (m=-0.21 SD=0.9). Main effects for IQ were found on all measures (p-values ranging from 3
Hyperactive-Impulsive
54.9 (11.1)
45.1 (7.1)
43.7 (5.1)
93.2**
1>2=3
Inattentive
26 (15)
-
-
Hyperactive-Impulsive
15 (8)
-
Combined
65 (37)
-
-
Subthreshold
68 (39)
-
-
ADOLESCENTS
n=406
n=209
Mean (SD) years of age
17.4 (3.4)
17.4 (4.0)
16.8 (3.8)
NS
9.6
13.4
17.1
7.1*
12=3
96.6 (16.3)
101.2 (15.4)
107.3 (14.5)
31.2**
12=3
67.2 (48.8)
48.8 (7.9)
46.6 (5.0)
283.5**
1>2=3
d
AC C
% Male
EP
%< 12 years of age
-
TE D
ADHD Diagnosis n(%)
M AN U
Conners’ Observer DSM-IV M(SD)
Estimated IQ M (SD)
e
n=193
Conners’ Parent DSM-IVd M(SD)
Inattention
Hyperactive-Impulsive
ADHD diagnosis DSM-IV n(%)
ACCEPTED MANUSCRIPT Inattentive
155 (37)
-
-
Hyperactive-Impulsive
37 (9)
-
-
Combined
153 (37)
-
-
Subthreshold
61 (17)
Note: NS = not significant. number of families with offspring=1, 2, 3 or 4=35, 199, 93, and 24, respectively.
b
F-statistic; group differences for % male were tested with χ² tests
c
1=Affected 2=Unaffected 3=Control
d
T-scores based on American norm group; Dutch T scores are found to be lower (Dutch [Adult] Twin Register)
e
11% (n=23) remitted cases
SC
EP
TE D
M AN U
* pU C .003
.33
U>C .005
.32
-
-
-
-
-
-
-
-
-
-
-
-
A = U = C†
A = U = C₪
A>C C@
d
p post hoc
A>U .008
.40
A=U >.45
-
.61a
A>C .45
-
U>C .04
.26
U=C >.45
-
U=C .25a
-
AC .002
.31
A=U .19
-
U=C .10
-
Note: RTV is reflected trial by trial. A = ADHD-affected group; C = healthy control group; U = ADHD-unaffected group (related to
EP
affected siblings or offspring); τ = infrequent long response times.
a
Post hoc split up in three age groups for RTV group results. For RTV, these post hoc p-values and effects sizes in adolescents
column represent age group 15-21 years.
AC C
d
.49a
M AN U
RTM
Agea
Group
TE D
RTV
Adolescents
RI PT
Parents
* p = .01; ‡ p = .004; † p = .25; ₪ p = .29; # p = .012 (group x age); $.02 (group x age₪); @p = .002
ACCEPTED MANUSCRIPT
Table 3 Significant Familial Effects of Attention-Deficit/Hyperactivity Disorder (ADHD) Symptoms and Motor Timing Measures and Significant Shared Familial Relations Between ADHD Symptoms and Motor Timing Measures; Expressed in Correlations or Standardized b’s (Parent-Offspring Relations). ADHD Totala
RTV
Tau
Parent-offspring relation (father)
ns
.18
ns
.27
Parent-offspring relation (mother)
ns
.27
ns
.25 c
Sibling correlation
ns
.24
ns
.13
Spouse correlation
-.11
.20
.13
.20
RI PT
RTM
Shared Familiality
M AN U
Mother-offspring cross-relationb
SC
Familiality
maternal ADHDoffspring motor timing
Sibling cross-correlation motor timing sibling 1ADHD sibling 2
.
.24 d
ns
ns
-
.18
.12
ns
-
TE D
Note: significant correlations are indicated in bold, and reported effects did not differ between ADHD and healthy control families, and only between fathers and mothers in case of an underscored effect size. RT=reaction time; RTM=reaction time median; RTV=consecutive response time variability. ADHD scores obtained from observer reports.
b
Shared familial effects apply to mother-offspring relations only and were not significant for fathers.
c
b is main effect; b in interaction with maternal RT median=0.09 (p=.001).
d
b is main effect for young mothers (median split at age 47.6); b in interaction with maternal age = -0.02 (p=.007).
AC C
EP
a
ACCEPTED MANUSCRIPT Figures Figure 1: Adolescent group x age interactions on motor timing reaction time consecutive variability. Note: A = participant with attention-deficit/hyperactivity disorder (ADHD); C = control participant; U = unaffected sibling. a p values between .001 and .036 b p (A>U/C) .45
AC C
EP
TE D
M AN U
SC
RI PT
Figure 2: Schematic and simplified representation of cross-sectional group patterns for trial by trial variability, reaction time median, and infrequent long response times on a 1s motor timing task. The three measurements during adolescence represent findings from the current study. Childhood measurements are based on previous results from the IMAGE-sample that studied the current NeuroIMAGE cohort five to six years ago (τ not included there;20). Note: a. shows trial by trial reaction time variability; b. shows reaction time median; c. shows infrequent long motor response times (t). ADHD = attention-deficit/hyperactivity disorder.
ACCEPTED MANUSCRIPT Supplement 1 Diagnostic Algorithm for Attention-Deficit/Hyperactivity Disorder (ADHD) in the NeuroIMAGE Sample Participants were administered the Dutch translation of the Schedule for Affective Disorders and Schizophrenia for School-Age Children - Present and Lifetime Version (K-SADS-PL),1 which is compatible with the DSM-IV-TR.2 Both parents and children, if ≥ 12 years old, were interviewed separately and were initially only administered the ADHD screening interview. Participants with elevated screen scores were administered the full
RI PT
ADHD section. For participants using medication, ratings concerned functioning of medication. Furthermore, each child was assessed with a parent-rated questionnaire (Conners’ Parenting Rating Scale – Revised: Long version [CPRS-R:L]) combined with either a teacher-rating (Conners’ Teacher Rating Scale – Revised: Long version [CTRS-R:L]) applied for children
S0890-8567(14)00446-8
DOI:
10.1016/j.jaac.2014.05.015
Reference:
JAAC 1033
To appear in:
Journal of the American Academy of Child & Adolescent Psychiatry
Received Date: 8 February 2013 Revised Date:
14 March 2014
Accepted Date: 10 May 2014
Please cite this article as: Thissen AJAM, Luman M, Hartman C, Hoekstra P, van Lieshout M, Franke B, Oosterlaan J, Rommelse NNJ, Buitelaar JK, Attention-Deficit/Hyperactivity Disorder (ADHD) and Motor Timing in Adolescents and their Parents: Familial Characteristics of Reaction Time Variability Vary With Age, Journal of the American Academy of Child & Adolescent Psychiatry (2014), doi: 10.1016/ j.jaac.2014.05.015. This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
ACCEPTED MANUSCRIPT Attention-Deficit/Hyperactivity Disorder (ADHD) and Motor Timing in Adolescents and their Parents: Familial Characteristics of Reaction Time Variability Vary With Age
RH = ADHD and Motor Timing in Adolescents and Parents Andrieke J.A.M. Thissen, PhD; Marjolein Luman, PhD; Catharina Hartman, PhD; Pieter Hoekstra, MD/PhD; Marloes van
RI PT
Lieshout, MSc; Barbara Franke, PhD; Jaap Oosterlaan, PhD; Nanda N.J. Rommelse, PhD; Jan K. Buitelaar, MD/PhD Supplemental material cited in this article is available online. Accepted June 30, 2014
Drs. Thissen, Rommelse, and Buitelaar are with Donders Institute for Brain, Cognition and Behaviour and Karakter Child
SC
and Adolescent Psychiatry University Center, Radboud University Nijmegen Medical Center, the Netherlands. Drs. Luman and Oosterlaan and Ms. van Lieshout are with VU University Amsterdam. Drs. Hartman and Hoekstra are with University Medical Center Groningen, University of Groningen. Dr. Franke is with Donders Institute for Brain, Cognition and
M AN U
Behaviour, Radboud University Nijmegen Medical Center.
This work was supported by grants from the National Institutes of Health (NIH; R01MH62873, to Stephen V. Faraone, PhD, which helped build the NeuroIMAGE Database), Netherlands Organization for Scientific Research (NOW) Large Investment (1750102007010, JKB), and from Radboud University Nijmegen Medical Center, University Medical Center
TE D
Groningen, Accare, and Vrije Universiteit Amsterdam.
The authors thank all the families and teachers that took part in this study and all interns for their assistance in data collection. The authors also acknowledge Rogier Donders, PhD of Radboud University Medical Center for his advice with
AC C
Gaussian analyses.
EP
regard to statistical analyses, and Marcel Zwiers, PhD of Radboud University Medical Center for his assistance with ex-
Disclosure: Dr. Hoekstra has received advisory panel payments from Shire and Eli Lilly and Co. Dr. Buitelaar has been a consultant to / member of advisory board of and/or speaker for Janssen-Cilag BV, Eli Lilly and Co., Bristol-Myers Squibb, Organon/Shering-Plough, UCB, Shire, Medice, Roche, and Servier. Drs. Thissen, Luman, Hartman, Franke, Oosterlaan, Rommelse, and Ms. Van Lieshout report no biomedical financial interests or potential conflicts of interest.
Correspondence to: Andrieke J.A.M. Thissen, Reinier Postlaan 12 6525 GC Nijmegen, the Netherlands; email: [email protected].
ACCEPTED MANUSCRIPT Abstract Objective: There is consistent evidence that attention-deficit/hyperactivity disorder (ADHD) is strongly related to impaired motor timing as reflected in decreased accuracy and increased reaction time variability (RTV). Unknown is whether (a) motor timing impairments are present in
RI PT
adolescents and adults with ADHD and their unaffected relatives to the same extent as has been reported in children and (b) ADHD and motor timing share familial underpinnings, as reflected in parent-offspring co-segregation and sibling cross-correlations.
SC
Method: 589 parents and 808 children/adolescents from families with ADHD and control families (parent/offspring average age: 48.6/17.3 years) were included. All participants were thoroughly
M AN U
assessed for ADHD and performed a 40-trial motor timing task (1 s interval production). Dependent neurocognitive measures included RT median (RTM: representing accuracy), RTV and ex-Gaussian component τ (τ: representing infrequent long response times). Generalized Estimating Equations were used for analyses.
Results: Unaffected children from families with ADHD had RTV (but not RTM or τ) scores in
TE D
between those of affected and control children. However, during middle-to-late adolescence, unaffected offspring were not impaired compared to control offspring and differed from ADHD probands, whereas during late adolescence/early adulthood, all offspring groups performed
EP
equally. Affected and unaffected parents of families with ADHD showed increased RTV compared to controls, regardless of age (not significant after adjusting for IQ). There were
AC C
indications for shared familiality between RTV and ADHD as reflected by sibling crosscorrelations, and between RTM and ADHD, as reflected by sibling cross-correlations and a maternal parent-offspring relation (parent-of-origin effect). Conclusions: RTV and its familial characteristics are influenced by development during adolescence. Increased RTV in children with ADHD appears to reflect immaturities in their neurocognitive functioning. Maternal ADHD effects might be involved in transmission of RTM (not RTV), but overall RTM showed less compelling (familial) relationships with ADHD than RTV.
ACCEPTED MANUSCRIPT
AC C
EP
TE D
M AN U
SC
RI PT
Keywords: ADHD, motor timing, reaction time variability, familiality, development
ACCEPTED MANUSCRIPT Introduction
Attention deficit/hyperactivity disorder (ADHD) is a strongly heritable neurodevelopmental disorder1 defined by a persistent pattern of age-inappropriate impairing symptoms of inattention,
RI PT
hyperactivity, and impulsivity.2 Numerous studies have focused on the neurocognitive deficits underlying ADHD. Consistent evidence was found that ADHD is not only related to executive functioning (EF) deficits, such as deficits in working memory and motor inhibition3,4, but also
SC
strongly to deficits in a range of timing functions.5,6 These functions refer to a multidimensional construct of temporal information processing reflecting the ability to deal with timekeeping in
M AN U
behaviour and thoughts. This is often investigated by using motor timing paradigms6 including two components: central timekeeping and motor response variability.7 Both components have been proven to be related to ADHD, with patients showing premature responses when estimating short intervals (involving mainly subcortical brain regions), but showing overestimations when estimating long intervals (involving mainly cortical brain regions).8,9 Furthermore, patients with
TE D
ADHD often show a pattern of infrequent, abnormally slow responses, reflected in the long tail of the response time distribution as indicated by τ (tau)10,11, as well as a high variability in responding.12 Thus far, elevated response time variability (RTV) in general--including RTV in
EP
motor timing --emerges as the most universal finding in ADHD research,12-14 and it has even been proposed to be the true core deficit of ADHD.15,16
AC C
RTV is of particular interest to ADHD researchers, given its relation to a familial predisposition to the disorder, which has been suggested by impaired performance in unaffected (twin) relatives.17,18 Motor timing RTV, more than accuracy, even proved to be one of the strongest ADHD-intermediate phenotypes (i.e. a heritable vulnerability trait along the pathway between clinical phenotype and genotype, heightening the risk for developing ADHD19) of all neurocognitive intermediate phenotype candidates studied.20 This was indicated by two findings: unaffected siblings had RTV smaller than affected siblings but larger than healthy controls, and the presence of significant sibling (cross-)correlations. This represents a clear clue to the familial-
ACCEPTED MANUSCRIPT genetic underpinnings of ADHD.19 Motor timing RTV has also been linked to molecular genetics by demonstrating strong genetic linkage to chromosome 2q21.1 (logarithm of the odds [LOD] score 3.944).21 Furthermore, it was found to tap into unique genetic aspects of ADHD, possibly related to loci conferring risk for ADHD (12q24.3 and 17p13.3).14 Therefore, motor timing RTV could be of
RI PT
help in unravelling the genetic factors involved in ADHD. Previous support for the role of motor timing RTV as an intermediate phenotype mediating the link between genes and ADHD was based mostly on young sibling and twin designs. So far, two
SC
key aspects in the relation between ADHD and motor timing are still underexposed: the cosegregation of ADHD and motor timing deficits from parents to offspring, and the role of
M AN U
developmental factors. Previous studies have found that significant parent-offspring relations exist for both ADHD and motor timing separately. These studies, however, did not examine the cosegregation of these traits, i.e. did not study whether ADHD and motor timing deficits are transmitted together.18,22,23 Earlier findings suggested this may not necessarily be the case, since no association was found between parental ADHD symptoms and offspring RTV (on a 1 s interval
TE D
motor timing task) in a study including parents and childhood offspring from the IMAGE-sample that preceded the current follow-up study five to six years ago.24 Further, selective parent-of-origin effects were detected for related functions like motor control and long time interval reproduction.
EP
This suggests that parental gender may be a relevant factor in understanding the causal pathways underlying neurocognitive deficits in ADHD and should be included in data analyses.25
AC C
The second issue is the role of developmental factors. The vast majority of ADHD neurocognitive literature focuses on children rather than adults, and the role of adolescence in the developmental trajectory of ADHD from childhood to adulthood has been studied much less. This is particularly problematic given the remission of ADHD that is mainly driven by prefrontal cortex maturation in a proportion of cases during this exact period of development.26,27 The brain structures that appear to be mostly involved in timing perception networks are the cerebellum and basal ganglia.28,29 These brain structures undergo fewer developmental changes during adolescence and young adulthood than the prefrontal regions.30 However, subjects with ADHD show reductions
ACCEPTED MANUSCRIPT in cerebellar volume throughout adolescence.31 On a behavioural level, it is found that (motor) timing deficits and motor timing variability in ADHD are present throughout adolescence 32,33 and adulthood 34, but it is unclear to what extent the co-segregation of ADHD and motor timing deficits is influenced by age. Therefore, the two main aims of the current study were to clarify whether (a)
RI PT
motor timing impairments are present in adolescents and (young) adults with ADHD and their unaffected relatives to the same extent as has been reported in children, and (b) ADHD and motor timing share familial underpinnings as reflected by co-segregation from parent to offspring and
AC C
EP
TE D
M AN U
SC
cross-correlations between siblings.
ACCEPTED MANUSCRIPT Method Participants Participants were selected from the Dutch follow-up (NeuroIMAGE [2009-2012]) of the International Multicenter ADHD Genetics (IMAGE) study performed between 2003-2006 (as
RI PT
described previously).35 Inclusion criteria for all participants were then: age between 5-19 years, European Caucasian descent, IQ ≥ 70, and no diagnosis of autism, epilepsy, general learning
difficulties, brain disorders, or known genetic disorders (such as Fragile X syndrome or Down
SC
syndrome). All family members were invited for follow-up measurement, with a mean follow-up period of 5.9 years (SD=.72). No additional inclusion criteria were applied, except a minimum age
M AN U
of 8 years for neurocognitive testing (not interviews/questionnaires). In the current study, 259 families with members having ADHD and 98 control families with at least one family member completing neurocognitive assessment participated. Mean age was 17.3 years (SD=3.7) for offspring, and 48.6 years (SD=5.0) for parents. Table 1 provides offspring and parents’
TE D
characteristics.
[TABLE_1]
EP
ADHD Diagnoses
All participants were reassessed using a semi-structured diagnostic interview (Dutch translation
AC C
of the Kaufman Schedule for Affective Disorders and Schizophrenia for School-Age Children Present and Lifetime Version: K-SADS-PL) and Conners' ADHD questionnaires to determine current ADHD diagnoses. A diagnostic algorithm was applied to combine symptom counts on the K-SADS-PL and Conners’ questionnaire (Supplement 1, available online, provides a detailed description) for calculation of the diagnostic status of the three groups included in this study: affected, unaffected, and control.
Measures
ACCEPTED MANUSCRIPT Parent and Offspring ADHD Symptoms Current ADHD information was obtained using the Dutch version of the Conners’ Adult ADHD Rating Scale - Observer: Screen Version (CAARS-O:SV) for parents and the CPRS-R:L for offspring.36 For parents and offspring, the DSM Total Score was used as a measure of current
RI PT
ADHD severity.
Motor Timing Paradigm
SC
A self-paced time production paradigm was employed to measure motor timing in parents and offspring.37 Details on the task are provided elsewhere.20 Briefly, participants were instructed to
M AN U
produce an interval of 1,000 ms in a 40 trial computer task by pressing a button as soon as they thought a 1-s time interval had elapsed after hearing a brief tone.
Motor Timing Performance Measures
Central timekeeping was measured by calculating accuracy in terms of the median of production
TE D
times (RTM) in ms, thereby limiting the influence of outliers, in line with earlier studies using a similar paradigm.37 Furthermore, response time variability (trial-by-trial variability in ms) of response timing was calculated: √ (Σ (RTi - RTi-1)2 / (n -1)), where i = trial number, n = number of
EP
trials and RT = response time.38 This measure of response time variability (RTV) correlated highly with RT standard deviation (RT SD): r=.89 in offspring and r=.93 in parents. Because the short
AC C
interval duration (1 s) yielded a positive skew in RT distribution, RTV was included as it provides for a more pure measure of variability compared to the mean dependent RT SD. In addition, exGaussian analyses were performed using MATLAB in order to acquire τ, the exponential part of the distribution with higher τ representing more infrequent long response times.10,11
Intelligence Full-scale IQ was estimated for parents and offspring by combining scores on the two subtests of the Wechsler Intelligence Scale for Children (WISC)/ Wechsler Adult Intelligence Scale (WAIS-
ACCEPTED MANUSCRIPT III) that show the highest correlation with full-scale IQ score: Vocabulary and Block Design (.88 for WISC and .90 for WAIS).39,40 Individual correlations between IQ, motor timing variables, and ADHD are provided in Table S1, available online.
RI PT
Procedure
The current study was part of a comprehensive assessment protocol (see
http://www.neuroimage.nl for an overview of the complete protocol and previous publications on
SC
one or more of the other cognitive tasks41,42). Participants were asked to withhold use of
psychostimulant drugs and other psychotropic medications for 48 hours and at least 3 days before
M AN U
assessment, respectively. After the study procedures had been explained, informed consent was signed by all participants, whereby parents signed for offspring .05), followed by a step down to the quadratic model, and so on (e.g. 44). If an interaction with age was present, parents or offspring were divided into three age groups for post
M AN U
hoc analyses (specified in the results section). Since IQ differed significantly between the three groups (for parents as well as offspring), analyses on all motor timing measures were conducted with and without IQ as a covariate. Effect sizes in the models were defined as standardized b (parent-offspring analyses), or Cohen’s d (main group comparisons).
To test whether parental ADHD and motor timing predicted offspring ADHD and motor timing,
TE D
the aforementioned model in GEE was used. In all parent-offspring analyses---executed for fathers and mothers separately--age and sex of the child, parental age, and their interactions with the main predictor were implemented in the model. If an interaction with age was present, a median split for
EP
age groups was applied for post hoc analyses (specified in the results section). Individual correlations between motor timing and ADHD symptomatology, and sibling (between first- and
AC C
second-born) and spouse (cross-)correlations for motor timing and ADHD symptomatology were calculated with correction for age difference (spouse and sibling) and sex difference (sibling).
ACCEPTED MANUSCRIPT Results
Offspring and Parent Motor Timing Group Differences All group results are provided in Table 2. Significant group x age interactions were found for
RI PT
RTV in offspring (group x age b=-6.6, p=.009 [affected vs. control]; group x age b=-7.4, p=.008 [unaffected vs. control]; group x age3 b=0.1, p=.009 [affected vs. control]). Post hoc analyses with RTV as dependent variable in three groups (children [aged ≤14 y, n=229], adolescents [aged 15-21
SC
y, n=494] and young adults [aged ≥22 y, n=85]) indicated that at childhood age, unaffected siblings performed in between affected siblings and controls (affected [m=298 SD=97] > unaffected [m=295
M AN U
SD=85] > control [m=244 SD=75]), but older unaffected siblings and affected siblings performed equally to controls in the adolescence and young adult time-window, respectively (adolescents: affected [m=269 SD=80] > unaffected [m=233 SD=73] = control [m=222 SD=76]; young adults: affected [m=226 SD=80] = unaffected [m=217 SD=79] = control [m=216 SD=70]). With regard to RTM and τ, no group x age or sex interactions appeared. On RTM, affected
TE D
siblings scored significantly lower (m=978 SD=71) than unaffected siblings (m=998 SD=75), but affected and unaffected siblings did not differ from controls (m=994 SD=65), indicating most pronounced time underestimations in affected siblings. Group comparisons with τ revealed affected
EP
siblings to show higher τ (m=0.10 SD=0.9) than controls (m=-0.21 SD=0.9). Main effects for IQ were found on all measures (p-values ranging from 3
Hyperactive-Impulsive
54.9 (11.1)
45.1 (7.1)
43.7 (5.1)
93.2**
1>2=3
Inattentive
26 (15)
-
-
Hyperactive-Impulsive
15 (8)
-
Combined
65 (37)
-
-
Subthreshold
68 (39)
-
-
ADOLESCENTS
n=406
n=209
Mean (SD) years of age
17.4 (3.4)
17.4 (4.0)
16.8 (3.8)
NS
9.6
13.4
17.1
7.1*
12=3
96.6 (16.3)
101.2 (15.4)
107.3 (14.5)
31.2**
12=3
67.2 (48.8)
48.8 (7.9)
46.6 (5.0)
283.5**
1>2=3
d
AC C
% Male
EP
%< 12 years of age
-
TE D
ADHD Diagnosis n(%)
M AN U
Conners’ Observer DSM-IV M(SD)
Estimated IQ M (SD)
e
n=193
Conners’ Parent DSM-IVd M(SD)
Inattention
Hyperactive-Impulsive
ADHD diagnosis DSM-IV n(%)
ACCEPTED MANUSCRIPT Inattentive
155 (37)
-
-
Hyperactive-Impulsive
37 (9)
-
-
Combined
153 (37)
-
-
Subthreshold
61 (17)
Note: NS = not significant. number of families with offspring=1, 2, 3 or 4=35, 199, 93, and 24, respectively.
b
F-statistic; group differences for % male were tested with χ² tests
c
1=Affected 2=Unaffected 3=Control
d
T-scores based on American norm group; Dutch T scores are found to be lower (Dutch [Adult] Twin Register)
e
11% (n=23) remitted cases
SC
EP
TE D
M AN U
* pU C .003
.33
U>C .005
.32
-
-
-
-
-
-
-
-
-
-
-
-
A = U = C†
A = U = C₪
A>C C@
d
p post hoc
A>U .008
.40
A=U >.45
-
.61a
A>C .45
-
U>C .04
.26
U=C >.45
-
U=C .25a
-
AC .002
.31
A=U .19
-
U=C .10
-
Note: RTV is reflected trial by trial. A = ADHD-affected group; C = healthy control group; U = ADHD-unaffected group (related to
EP
affected siblings or offspring); τ = infrequent long response times.
a
Post hoc split up in three age groups for RTV group results. For RTV, these post hoc p-values and effects sizes in adolescents
column represent age group 15-21 years.
AC C
d
.49a
M AN U
RTM
Agea
Group
TE D
RTV
Adolescents
RI PT
Parents
* p = .01; ‡ p = .004; † p = .25; ₪ p = .29; # p = .012 (group x age); $.02 (group x age₪); @p = .002
ACCEPTED MANUSCRIPT
Table 3 Significant Familial Effects of Attention-Deficit/Hyperactivity Disorder (ADHD) Symptoms and Motor Timing Measures and Significant Shared Familial Relations Between ADHD Symptoms and Motor Timing Measures; Expressed in Correlations or Standardized b’s (Parent-Offspring Relations). ADHD Totala
RTV
Tau
Parent-offspring relation (father)
ns
.18
ns
.27
Parent-offspring relation (mother)
ns
.27
ns
.25 c
Sibling correlation
ns
.24
ns
.13
Spouse correlation
-.11
.20
.13
.20
RI PT
RTM
Shared Familiality
M AN U
Mother-offspring cross-relationb
SC
Familiality
maternal ADHDoffspring motor timing
Sibling cross-correlation motor timing sibling 1ADHD sibling 2
.
.24 d
ns
ns
-
.18
.12
ns
-
TE D
Note: significant correlations are indicated in bold, and reported effects did not differ between ADHD and healthy control families, and only between fathers and mothers in case of an underscored effect size. RT=reaction time; RTM=reaction time median; RTV=consecutive response time variability. ADHD scores obtained from observer reports.
b
Shared familial effects apply to mother-offspring relations only and were not significant for fathers.
c
b is main effect; b in interaction with maternal RT median=0.09 (p=.001).
d
b is main effect for young mothers (median split at age 47.6); b in interaction with maternal age = -0.02 (p=.007).
AC C
EP
a
ACCEPTED MANUSCRIPT Figures Figure 1: Adolescent group x age interactions on motor timing reaction time consecutive variability. Note: A = participant with attention-deficit/hyperactivity disorder (ADHD); C = control participant; U = unaffected sibling. a p values between .001 and .036 b p (A>U/C) .45
AC C
EP
TE D
M AN U
SC
RI PT
Figure 2: Schematic and simplified representation of cross-sectional group patterns for trial by trial variability, reaction time median, and infrequent long response times on a 1s motor timing task. The three measurements during adolescence represent findings from the current study. Childhood measurements are based on previous results from the IMAGE-sample that studied the current NeuroIMAGE cohort five to six years ago (τ not included there;20). Note: a. shows trial by trial reaction time variability; b. shows reaction time median; c. shows infrequent long motor response times (t). ADHD = attention-deficit/hyperactivity disorder.
ACCEPTED MANUSCRIPT Supplement 1 Diagnostic Algorithm for Attention-Deficit/Hyperactivity Disorder (ADHD) in the NeuroIMAGE Sample Participants were administered the Dutch translation of the Schedule for Affective Disorders and Schizophrenia for School-Age Children - Present and Lifetime Version (K-SADS-PL),1 which is compatible with the DSM-IV-TR.2 Both parents and children, if ≥ 12 years old, were interviewed separately and were initially only administered the ADHD screening interview. Participants with elevated screen scores were administered the full
RI PT
ADHD section. For participants using medication, ratings concerned functioning of medication. Furthermore, each child was assessed with a parent-rated questionnaire (Conners’ Parenting Rating Scale – Revised: Long version [CPRS-R:L]) combined with either a teacher-rating (Conners’ Teacher Rating Scale – Revised: Long version [CTRS-R:L]) applied for children
