Entertainment & Media

A general population twin study of conduct problems and the auditory P300 waveform

Description
Reduced amplitude of the P300 event-related potential has been consistently associated with a variety of externalising problems, including conduct disorder. The few available genetically-informative studies of these relationships, however, were
Published
of 9
All materials on our website are shared by users. If you have any questions about copyright issues, please report us to resolve them. We are always happy to assist you.
Related Documents
Share
Transcript
  A General Population Twin Study of Conduct Problemsand the Auditory P300 Waveform Eleonora Bertoletti  &  Giorgia Michelini  &  Sara Moruzzi  & Giuseppina Ferrer  &  Luigi Ferini-Strambi  &  Maria Antonietta Stazi  & Anna Ogliari  &  Marco Battaglia # Springer Science+Business Media New York 2013 Abstract  Reduced amplitude of the P300 event-related poten-tial has been consistently associated with a variety of externalising problems, including conduct disorder. The fewavailable genetically-informative studies of these relationships,however, were conducted among adolescents/adults (i.e., at anage when conduct disorder has typically already become man-ifest). Among 200 general population twins with a mean age of 9 years (range 6  –  14 years), we studied the relationship betweenthe P300 waveform elicited by an auditory oddball task and theDSM-oriented conduct problems scale of the Child Behavior Checklist 6  –  18. Conduct problems scores were negatively andsignificantly correlated ( r  = − 0.19,  p =0.01) with P300 ampli-tude; correlations between P300 amplitude and the other DSM-oriented Child Behavior Checklist scales were non-significant,except for oppositional defiant problems (  p =0.01). We foundmoderate heritability estimates for both P300 amplitude (0.58,CI:0.37;0.73) and conduct problems (0.52, CI:0.25;0.70). Bi-variate twin analyses indicated that the covariation betweenthese two phenotypes can be explained by additive geneticfactors only, with a genetic correlation of   − 0.33. An association between reduced P300 amplitude and conduct problems can besubstantiated already in childhood, at an age that precedes themosttypicalonsetofconductdisorder.Thisrelationshipappearstobegeneticinnature.ReducedP300amplitudecanrepresenta valuable marker for conduct problems,andcancontribute totheearly identification of children at high-risk for conduct disorder. Keywords  Event-relatedpotentials .P300 .CBCL .Conduct  problems .Twinstudies .Children Introduction Conduct disorder(CD)belongs tothe broad,heritable(Vidinget al. 2005) dimension of externalising psychopathology to-gether with oppositional defiant disorder and antisocial per-sonality, which is in turn associated with heightened risk for substance use disorders (Ruiz et al. 2008). Lifetime preva- lence of CD in the US is 9.5 %: while it can be observed inchildren before age 10, the onset iscommonly(75 % ofcases) between age 10 and 18 (Kessler et al. 2005). Early interven- tion on aggressive children may reduce the risk for CD inadolescence (Foster etal. 2006),further underlyingthe impor-tance of early recognition/interception of CD.In the development of biological markers of risk for externalising behaviour, event-related potentials (ERP) evoked by oddball paradigms yield a consistent picture that relatesexternalising behaviour to the P300 waveform both for visual(e.g., Iacono and McGue 2006) and auditory (Viana- Wackermann et al. 2007) tasks. In an oddball paradigm theP300 is thought to map the processing of a rare and unexpected E. Bertoletti ( * ) : G. Michelini : S. Moruzzi :  G. Ferrer  :  A. OgliariAcademic Centre for the Study of Behavioural Plasticity, Vita-SaluteSan Raffaele University, 20 via S. D ’ Ancona, 20127 Milan, Italye-mail: bertoletti.eleonora@hsr.it L. Ferini-StrambiSleep Disorders Centre, Vita-Salute San Raffaele University,20 via S. D ’ Ancona, 20127 Milan, ItalyM. A. Stazi National Centre for Epidemiology, Surveillance and HealthPromotion, Istituto Superiore di Sanità, 299 Viale Regina Elena,00161 Rome, ItalyM. Battaglia ( * )Department of Psychiatry & Neuroscience, Laval University,Québec, Canada e-mail: Marco.Battaglia@fmed.ulaval.ca M. Battaglia Institut Universitaire en Santé Mentale de Québec, 2601 Chemin dela Canardiére, Quebec City, Québec G1J 2G3, Canada G. MicheliniMRC Social, Genetic and Developmental Psychiatry Centre,Institute of Psychiatry, King ’ s College London, De Crespigny Park,SE5 8AF London, United KingdomJ Abnorm Child PsycholDOI 10.1007/s10802-013-9836-7  stimulus, with amplitude reflecting the amount of attentionalresources allocated to the task, and latency reflecting the speedof stimulus classification (Polich 2007). Reduced P300 ampli- tude is thus generally thought to indicate inefficient deployment of neural resources (Gao and Raine 2009). Some data support a  functional link between reduced P300 in oddball tasks, alteredinformation processing, and externalising behavior. More spe-cifically, the interplay between: a) frontal cortical regions that driveattentiontowardsstimulusprocessing,andb)hippocampaltemporal/parietal junction regions governing memory updatingis thought to be important to explain the P300 waveform (Gaoand Raine 2009; Polich 2007). Consistent with this model (Gao and Raine 2009), functional MRI studies have found that theoddballtaskactivatestheanteriorcingulate,midlineandanterior frontal cortex, insular cortex, andinferior parietal regions (Kiehland Liddle 2003; Linden et al. 1999). In adolescents, reduced P300 amplitudes to rare stimuli have been found associated with heightened risk for externalising behaviour and increased substance abuse and antisocial behav-iour  (Patrick et al. 2006). In childhood, a similar association has  been reported for attention deficit/hyperactivity disorder (Aleksandrov et al. 2005; Barry et al. 2003), which may precede orco-occurwithCD(Waschbusch2002).Moreover,boysatrisk  for disinhibited conditions and reduced P300 at age 17 showedhigher risk for substance abuse at age 20 (Iacono et al. 2002), implying some predictive value for this psychophysiologicalmeasure. Neurofunctional correlates of reduced P300 amplitudeand clinical studies thus support the role of this measure as anindex of the faulty attentional and inhibitory mechanisms com-mon to externalising conditions, including ADHD or CD.Genetically-informative analyses of P300 amplitude showedmoderate heritability estimates (with a meta-heritability of 0.60,(vanBeijsterveldtandvanBaal2002),andsharedgeneticfactorswithexternalisingpsychopathologyinlateadolescence(Gilmoreet al. 2010). Moreover, healthy monozygotic (MZ) twin adoles-cents showed the same P300 amplitude decrement as their alcohol-abusing co-twins (Carlson et al. 2002), and an influence offamilialhistoryofalcoholismuponP300amplitudewasfound both among subjects aged 7  –  13 years (Ehlers et al. 2001) and amongadults (Bauer et al. 1994).Moreover,the P300amplitude elicited by oddball tasks has good temporal stability, with test   –  retest correlation coefficients ranging from 0.50 to 0.80 (Polich2007), and highest reliability values for Pz, compared to Cz andFz,electrodeinauditorytasksforbothadults(WalhovdandFjell2002) and adolescents (Williams et al. 2005). Altogether, these data point towards reduced P300 ampli-tude as a reliable and replicable measure, anda possible marker of the neurophysiological processes underlying vulnerability toa general factor of externalisation (Iacono and Malone 2011). While these relationships are supported among adolescents/ adults, it remains unknown whether the P300 amplitude corre-lates with CD/conduct problems also among children, and towhat extent genetic and environmental factors contribute toexplain such correlation. If reduced P300is a biological marker of externalisation, then it should not only persist after adoles-cence, but also be documentable in childhood, so to predate theonset of full-blown externalising conditions.We are not aware of other studies that have confirmedattenuated P300 amplitude in young children prior to theonset of conduct problems, and since the majority of CDcases have their onset in adolescence, finding an associa-tion between CD and P300 in childhood would consider-ably enhance the value of this psychophysiological mea-sure. Here, by an auditory oddball paradigm applied togeneral population twins with a mean age of 9 years, weinvestigated how the P300 amplitude relates to conduct  problems, and the aetiological influences underlying suchrelationship. Materials and Methods ParticipantsThisstudyisbasedonthe Italian TwinRegistry(ITR,Pesenti-Gritti et al. 2008; Stazi et al. 2002), a database of all possible twins in the Italian general population initially set up in 2001.In2010,130newfamiliesoftwinsaged6  –  14intheprovincesof Milano and Monza were invited to enrol in the ITR and participateinthisstudy.Familiesweresubsequentlycontacted by mail or telephone to provide further information and set anappointment for ERP acquisition. Of these 130 families, 25(19 %) did not want to participate and 5 (3.8 %) agreed tocomplete only behavioural questionnaires, leaving 100 com- plete pairs aged 6  –  14 years (mean: 9.23±2.10 years) to take part in the present study. None of the participating children carried certified mental/  physical handicaps that would require special attention, suchas a remedial teacher or differential academic programmes.The procedures of this study were accepted by the ethicalcommittees of the participating institutions, and parentssigned an informed consent for all the participants.Zygosity DeterminationZygosity was assigned by the parent-rated Goldsmith ques-tionnaire (Goldsmith1991),which includesitems about phys- ical similarity, frequency of confusion of the twins by others,medical information and parental opinion on zygosity. Thisinstrument has an accuracy of zygosity determination of 94 %(van Beijsterveldt et al. 2004) when using the algorithm weemployed to score responses.Accordingtothequestionnaire,therewere38monozygotic(MZ) pairs (23 female and 15 male) and 62 dizygotic (DZ) pairs(17male,12female,33oppositesex).The MZ/DZ samesex/DZ opposite sex ratio was thus 1.3/1/1.1, close to the J Abnorm Child Psychol  expected 1/1/1 population distribution. Due to the smallsample size, the analyses were not stratified by age andgender, and DZ same and opposite sex pairs weregrouped into the DZ category.ERPAcquisition and AnalysisOur auditory oddball paradigm consisted of 400 stimuli pre-sented to both ears through earphones. Frequent standardtones of 1,000 Hz and rare target tones of 2,000 Hz were presented in a random sequence, with a probability of 80 %and 20 % respectively, in line with currently adopted proce-dures (Polich 1996). Both stimuli had an intensity of 90 dB, a  duration of 500 milliseconds and the inter-stimulus intervalwas 2,000 ms. Electroencephalogram (EEG) was continuous-ly recorded from Fz, Cz, Pz, C3, C4, T3, T4 electrodes, left (A1)andright(A2)mastoidsofthe10  –  20systemwiththeuseof silver   –  silver chloride electrodes referred to the occipital(Oz) electrode, with an amplifier (Neuroscan SynAmp, Neuroscan Labs, Sterling, VA, USA). The ground electrodewas positioned on the forehead. Electro-oculogram (EOG)was recorded from electrodes placed vertically above and below the right eye. Electrodes impedance was maintained below 5 k  Ω . EEG and EOG were amplified, analogically band-pass filtered (0.10  –  30 Hz), digitised and acquired at a 500-Hz sampling rate.EEG acquisition, processing and analysis were carried out using Scan 4.4 software (Neuroscan Labs). After the acquisi-tion, all channels were re-referenced to the average of the twomastoids.Forraretones,EEGandEOGepochsbetween  –  100and 2,000 ms from the stimulus onset were obtained. Epochswere rejected if affected by artefacts (i.e., amplitudes greater than +50  μ  V or smaller than  –  50  μ  V between  –  100 and600 ms). For each subject, an average of the ERP waveformsin response to rare stimuli was constructed from artefact-freeepochs for each electrode. A minimum of 50 artefact-freeepochs was required for inclusion into the analysis. A grandaverage was thencomputedfromthe individuals ’ averagesfor the rare stimuli (Fig. 1).On the basis of visual inspection of the grand average, a time window between 200 and 550 ms was identified: withinthis latency range, in line with numerous previous studies of children and adolescents (e.g., Anjana et al. 2010; Patrick et al. 2006), the P300 amplitude was defined and measuredas the distance of the most positive peak from the baseline.ERP data were available for 184 twins due to 9 children whorefused to undergo ERP recording, and 7 children whose data were excluded due to an insufficient number of artefact-freeepochs. Preliminary and model fitting analyses for P300 arereferred to the Pz electrode, where this waveform reachedmaximum amplitude and visibility (see Fig. 1), coherentlywith previous studies in developmental samples (Bauer andHesselbrock  1999; Patrick et al. 2006). Behavioural Task During ERPAcquisitionDuring the ERP recording children were seated in a quiet andlight-attenuated room and were instructed to keep a runningcount of the number of targets they heard in order to ensureattention. Immediately after the end of the ERP recordingsession they were asked about the total number of rare tonesthey had counted. Inaccuracy in this behavioural task wasestimated in absolute units of deviation from the correct number( n =80)ofraretones,sothatachildwhohadreported,for example, 78 tones, was considered as inaccurate as a childwho had reported 82 tones.Measurement of Children ’ s Behaviour The Child Behavior Checklist (CBCL) 6  –  18 (Achenbach andRescorla  2001) filled in by mothers was employed to assesschildren ’ s behavioural problems. The CBCL is a parent-ratedstandardised questionnaire encompassing 118 behavioural andemotional problem items. Respondents rate each item on a 3- pointscale:0= nottrue ,1=  somewhatorsometimestrue ,2= veryor often true . The Achenbach system of empirically basedassessment (ASEBA, Achenbach and Rescorla  2001) includes eight empirically-based syndrome scales obtained by factorialanalysis of the 118 problem items, and six DSM-oriented scales(DOSs) aimed at covering common childhood mental disorders.The six DOSs were generated after evaluation of the degree of consistency of the CBCL/6  –  18 items with corresponding DSM-IV criteria: affective problems, anxiety problems, somatic prob-lems, attention-deficit/hyperactivity problems, oppositional defi-ant problems and conduct problems by clinician consensus.Although the DOSs are not directly equivalent to any clinicaldiagnosis,theysatisfactorilypredictDSM-IVdiagnoses(Lengua et al. 2001), have good temporal stability (Nobile et al. 2013), simplercausalstructureandhigherheritabilitythantheempiricalscales (Spatola et al. 2007). Statistical AnalysesDescriptive statistics for CBCL DOS CP and P300 amplitudeand latency at Pz electrode were calculated for the wholesample. Before model fitting analyses CP and P300 valueswere then log-transformed to reduce skewness and kurtosisand approximate normal distribution. Model fitting analyseswere carried out with the Mx programme (Neale et al. 2003).We calculated phenotypic correlations between log-transformed P300 amplitude at Pz and both CP and other CBCL DOSs (affective problems, anxiety problems, somatic problems, attention-deficit/hyperactivity problems, opposi-tional defiant problems) through saturated models.For CP and P300 amplitude, the saturated model was alsoused to estimate the cross-twin/within-trait correlations (be-tween twin 1 and twin 2 for the same trait) and the cross-twin/  J Abnorm Child Psychol  cross-trait correlations (between one trait in twin 1 and theother trait in twin 2, and vice versa) in MZ and DZ pairs. Thesaturated model was specified by imposing different con-straints in successive steps: the means and variances of each phenotype were constrained to be equal for twin 1 and twin 2in a pair, and for MZ and DZ pairs. By constraining cross-twin/cross-trait covariances to be equal within each zygositygroup,weimplementedthe assumptionthatthere issymmetry between twin and co-twin, and that twins are from the same population. Sex and age effects were also calculated within a saturated model.MZ and DZ differences in cross-twin/within-trait and cross-twin/cross-trait correlations allow to decompose the total phe-notypic variance and covariance of two traits into proportionsdue togenetic factors (additive -A- and dominance -D-), sharedenvironmental factors (C) and non-shared environmental fac-tors(E,alsoencompassingerror). IfMZcross-twincorrelationsare twice the DZ correlations, then additive genetic effects arelikely to be important in determining the phenotypic covaria-tion, while MZ cross-twin correlations more than twice the DZcorrelations suggest dominance effects. If DZ cross-twin corre-lations are greater than half the MZ correlations, shared envi-ronmental effects are to be taken into account.To test the sources of covariance between CP and P300amplitudeweappliedabivariatetwinmodel,whichallowsfor decomposing the total phenotypic variance and covariance of two traits into proportions due to A, C or D, and E. A and Dareassumedtobecorrelatedat1.0inMZ(astheyshare100%of their DNA), while in DZ the A correlation is set at 0.5 (asthey share half of their segregating genes on average) and theD correlation at 0.25. C is assumed to be correlated at 1.0 for  bothMZandDZ,accordingtotheequalenvironmentassump-tion (Neale and Cardon 1992). E effects are, by definition, uncorrelated. As the classical twin approach does not allow toestimate C and D simultaneously (Hudziak et al. 2005) and is typically underpowered to assess D effects (Rietveld et al.2003), we applied an ACE model.A bivariate Cholesky model (Neale and Cardon 1992) was applied,withtheCPscoresenteredasthefirstvariableandtheP300 amplitude as the second variable. This model providesthe fullest explanation of data because it does not impose anyrestrictions onto the genetic and environmental contributionsto covariation. The full Cholesky model was successivelyfurther refined by stepwise deletion of variance componentsin progressively more parsimonious models. Selection of the best fitting model was made by the Akaike ’ s informationcriterion (AIC, Aiken and West  1991) which is calculatedas  − 2LL-2( Δ df): the lower the AIC value, the better the balance between explanatory power and parsimony.The best fit bivariate results were then represented as a Correlated Factors model, which is a standardization of theCholesky decomposition (Neale et al. 2003). The re- parameterization of the Cholesky model into a Correlated Factor model allowed for the decomposition of the phenotypic correla-tion between CP and P300 amplitude into additive genetic ( r  g ),shared environmental ( r  c ) and non-shared environmental ( r  e )correlations. These correlations provide an indication of thedegree of overlap in causal influences (Neale et al. 2003). Results Table 1 shows the descriptive statistics for CBCL DOS CPand ERP P300 amplitude and latency. The phenotypic corre-lation between P300 amplitude at Pz and CBCL DOS CPwas  − 0.19 (  p =0.01). The oppositional defiant problems(ODP) scale was the only other CBCL DOSs to show a significant correlation with P300 amplitude ( r   = − 0.19, Fig. 1 a  Grand average of  participants ’  ERP waveformsgenerated by the rare stimuli at Pzelectrode;  b  topographic map for P300 maximum peak J Abnorm Child Psychol   p  =0.01), while attention-deficit/hyperactivity problems( r   = − 0.14,  p =0.06), affective problems ( r   = − 0.02,  p  =0.79), anxiety problems ( r   = − 0.08,  p =0.28) and somatic problems ( r   = − 0.07;  p  =0.35) all yielded non-significant correlations with P300 amplitude at Pz. Since the CBCLDOS CP is made of 17 items (vs. only 5 items in theDOS ODP) and consistently yields higher Cronbachalpha values than the ODP scale (Nakamura et al.2009), we limited our genetic analyses to the conduct  problems phenotype. There was no significant correla-tion between inaccuracy in counting the rare tones dur-ing the behavioural performance and either the CP scalescores ( r   = − 0.03;  p =0.70), or P300 amplitude ( r   =0.11;  p =0.21).Results of model comparisons under the saturated modelrevealed that means and variances could be equated across twin1/twin 2 and MZ/DZ twins, without significant fit deterioration.This suggests the absence of a significant effect of twin order (whichwascasuallyattributedinthissample)and/orzygosityonmeans and variances. The saturated model revealed no effect of age on either CP or P300 amplitudes, no effect of sex on P300amplitude, and a significant effect of sex (  p =0.01) on CP.Table 2 shows the twin correlations between CP and P300amplitudederivedfromthesaturatedmodel.TheMZcross-twin/ within-trait correlations were almost twice than DZ correlationssuggesting that genetic factors influence the variation of pheno-types.Alsocross-twin/cross-traitcorrelationsweregreaterinMZthan DZ pairs, indicating that genetic influences are important inexplaining the phenotypic covariance of CP and P300.Table 3 shows the bivariate model fitting results. The first model is the full ACE model; in model 2 we dropped allshared environmental effects. Model 3, in which commonand specific genetic paths were dropped, showed a clear deterioration of the AIC index. In model 4 we removed theunique environment influences on covariation, which yieldedan improvement of fit index. By AIC the best-fitting modelwas model 5, whereby the variances of CP and P300 ampli-tude were explained by A and E influences, and their covari-ance by genetic factors only.Table4showstheestimatesofgenetic,sharedenvironmen-tal and non-shared environmental contributions to the vari-ance and the covariance of CP and P300 amplitude derivedfrom the Cholesky best fitting model. The extent to which CPandP300amplitude share the samegenetic effectsisindicated by the genetic correlation between CP and P300 amplitude,with  r  g  of   − 0.33 (95 % CI:  − 0.59;  − 0.05) in the CorrelatedFactor model (Fig. 2), indicating that these two phenotypesshare one third of the genetic liability. Table 1  Age, CBCL DOS conduct problems scores and P300 amplitude and latency values (means ±  SD ) in the sample, and divided by sex andzygosityEntire sample Boys Girls MZ DZAge (years) 9.23±2.10 9.11±2.18 9.35±2.03 9.15±1.79 9.28±2.28Conduct problems 1.06±1.35 1.48±1.39 0.65±1.20 0.86±1.24 1.19±1.41 n =200  n =98  n =102  n =76  n =124P300 amplitude ( μ  V) 7.39±4.80 6.79±4.11 7.97±5.36 8.01±5.76 7.00±4.08 n =184  n =91  n =93  n =70  n =114P300 latency (ms) 354.28±48.94 354.18±49.22 354.39±48.93 358.82±45.58 351.44±50.92 n =184  n =91  n =93  n =70  n =114 Table 2  Twin correlations for CBCL DOS conduct problems and P300amplitude (95 % Confidence Intervals in parentheses)MZ DZCross-twin/within trait Conduct problems 0.54 0.23(0.21;0.72) (0.00;0.43)P300 0.59 0.25(0.36;0.74) (0.00;0.49)Cross-twin/cross-trait Conduct problems − − 0.13  − 0.07P300 ( − 0.31;0.71) ( − 0.25;0.11) Table 3  Multivariate ACE model fitting resultsVAR  a  COVAR   b − 2LL c DF d AIC e 1 ACE ACE 609.644 369  − 128.3562 AE AE 609.644 372  − 134.3563 CE CE 616.559 372  − 127.4414 ACE AC 610.364 370  − 129.636 5 AE A 610.364 373  − 135.636 a  elementswhich aretakeninto accountin each modeltoexplain variance  b elements which are taken into account in each model to explaincovariance c minus twice the log-likelihood d degrees of freedom e  χ 2 − 2( ∆ df)The best-fitting model is printed in boldface typeJ Abnorm Child Psychol
Search
Similar documents
View more...
Related Search
We Need Your Support
Thank you for visiting our website and your interest in our free products and services. We are nonprofit website to share and download documents. To the running of this website, we need your help to support us.

Thanks to everyone for your continued support.

No, Thanks