VU Research Portal

(1)

VU Research Portal

Childhood arrestees: neural correlates of antisocial and psychopathic development

Cohn, M.D.

2017

document version

Publisher's PDF, also known as Version of record

Link to publication in VU Research Portal

citation for published version (APA)

Cohn, M. D. (2017). Childhood arrestees: neural correlates of antisocial and psychopathic development.

General rights

Copyright and moral rights for the publications made accessible in the public portal are retained by the authors and/or other copyright owners and it is a condition of accessing publications that users recognise and abide by the legal requirements associated with these rights. • Users may download and print one copy of any publication from the public portal for the purpose of private study or research. • You may not further distribute the material or use it for any profit-making activity or commercial gain

• You may freely distribute the URL identifying the publication in the public portal ? Take down policy

If you believe that this document breaches copyright please contact us providing details, and we will remove access to the work immediately and investigate your claim.

E-mail address:

vuresearchportal.ub@vu.nl

(2)

509781-L-bw-Cohn 509781-L-bw-Cohn 509781-L-bw-Cohn 509781-L-bw-Cohn Processed on: 4-5-2017 Processed on: 4-5-2017 Processed on: 4-5-2017

Processed on: 4-5-2017 PDF page: 139PDF page: 139PDF page: 139PDF page: 139

9

M.D. Cohn, E. Viding, L. Pape, W. Van den Brink, T.A.H. Doreleijers, D.J. Veltman, E. McCrory*, A. Popma*. Under review. * Shared last author

(3)

140 Abstract

(4)

9

141 Introduction

Individuals who experience childhood adversity, which includes both deprivation and maltreatment, show an elevated risk for developing psychiatric disorders across the lifespan (Gilbert et al., 2009). This association pertains to a broad variety of disorders, including depression, anxiety disorder, post-traumatic disorder, substance abuse, conduct disorder and psychosis. However, there have been relatively few neuroimaging studies to date investigating the impact of childhood maltreatment on regional brain function (for review, see McCrory et al., 2011a). This is surprising, because investigating aberrant brain functions in maltreated individuals may provide important clues as to how early adverse experiences become biologically embedded in ways that engender latent vulnerability to psychiatric disorder (McCrory & Viding, 2015).

At the phenotypical level, emotion dysregulation is a prominent feature of several adult mental health outcomes associated with childhood adversity or maltreatment, including depression, borderline personality disorder and a range of anxiety disorders. The majority of extant functional magnetic resonance imaging (fMRI) studies of individuals who have experienced childhood adversity or maltreatment have therefore tended to focus on affect processing (e.g. McCrory et al., 2013) and response inhibition (e.g. Mueller et al., 2010). These studies have been relatively consistent in reporting that across both child and adult samples, a history of childhood maltreatment or neglect is associated with amygdala hyperactivity during the processing of negative facial affect, consistent with what has been observed in studies of depressed and anxious childhood and adult samples (e.g. Thomas et al., 2001; Straube et al., 2006). This pattern arises in maltreated individuals even in the absence of psychiatric disorder (Dannlowski et al., 2012). In addition, early life interpersonal trauma (Carrion et al., 2008), or a history of early caregiver deprivation (Mueller et al., 2010) has been reported to be associated with higher anterior cingulate cortex responses during tasks requiring response inhibition, suggesting an association between early adversity and alterations in top-down regulatory processes. It has been postulated that these neural response patterns reflect neural adaptations that potentiate hypervigilance to threat, which in turn contributes to underlying psychiatric vulnerability (McCrory et al., 2011b). However, such a relatively narrow paradigmatic focus on affect regulation is increasingly challenged by the study of a much wider set of processes that may underlie vulnerability to mental illness.

(5)

142

investigated reward anticipation in individuals with a history of childhood adversity (Dillon et al., 2009; Mehta et al., 2010). Dillon and colleagues (2009) employed a paradigm in which cues signalling monetary rewards and losses were manipulated, in a group of young adults (n=13; 69% female) who experienced maltreatment as children. They found that globus pallidus responses during reward anticipation were lower in maltreated persons compared to non-maltreated controls (n=29; 45% female). Mehta and colleagues (2010), using a monetary incentive delay task, compared a sample of adolescents (n=12; 50% female) who had experienced severe deprivation early in their lives in Romanian institutions to a group of non-adopted controls (n=11; 45% females). Lower ventral striatum responses during reward anticipation were reported for the early adversity group. While these important early studies provide preliminary evidence for ventral striatum hypo-responsiveness during reward anticipation in individuals who had experienced either childhood maltreatment or adversity, replication in larger samples is warranted. Furthermore, to our knowledge, no previous functional imaging study has explored whether different forms of maltreatment differentially influence brain responses. This is despite suggestive evidence of distinct associations between maltreatment subtypes and psychiatric disorders (Bernstein et al., 1998) and neuropsychological measures, i.e., emotion recognition (Pollak et al., 2000). These findings at the behavioural and psychological level have prompted recommendations to investigate specific maltreatment subtypes’ dimensional contributions to developmental psychopathology (Manly, 2005).

The primary aim of the current study was to systematically investigate the relationship between reward anticipation in the ventral striatum and different types of childhood maltreatment (physical abuse, physical neglect, emotional abuse, emotional neglect and sexual abuse). The secondary aim was to explore the relation between regional brain function during other incentive processing events (i.e. reward outcome, loss anticipation and loss outcome) and childhood maltreatment subtypes. The study was conducted in a large sample of childhood arrestees. This population is at an increased risk of having experienced childhood maltreatment and also at an increased risk of a range of negative mental health outcomes (Kim-Cohen et al., 2003). Childhood maltreatment has been associated with juvenile offending in general (Fergusson & Lynskey, 1997) as well as with its earlier onset (Rivera & Widom, 1990). The absence of previous neuroimaging or neuropsychological studies on distinct effects of maltreatment subtypes largely precluded formulating specific predictions with respect to their direction. In view of the evidence pointing to differential outcomes in adulthood depending on maltreatment type experienced (Bernstein et al., 1998), we did hypothesize that there would be distinct effects for abuse and neglect (Pollak et al., 2000) on ventral striatum responsiveness during reward anticipation (Dillon et al., 2009; Mehta et al., 2010). Investigation of VS reactivity patterns during the other incentive processing epochs was exploratory.

Methods

Participants

(6)

9

143

years while the mean age at wave three was 13.1 (SD 1.5) years. For the current neuroimaging study (wave four; mean age 17.7 (SD 1.6) years), a subsample representative of a range of externalizing risk (total n=150) was recruited, with an even distribution of subjects with a low-risk, a medium-risk and a high-risk of antisocial behaviour (see online supplement for the sampling strategy). From this sample (n=150), a total of 25 participants were excluded from analyses for a variety of reasons, including: invalid MRI data (n=10); invalid task performance (performance deviating >3SD from the mean; n=9); drug use in the day before scanning (n=3); and missing questionnaire data (n=3). Analyses were performed on the remaining 125 participants.

Procedure

The study was approved by the IRB of the VU University Medical Center Amsterdam (VUmc). All participants (and their parents/custodians, if age of the participant was below 18) signed informed consent and were visited at home for behavioural testing, including the Childhood Trauma Questionnaire (see below). On a second occasion, participants were scanned using a Philips 3T Intera MRI-scanner at the Academic Medical Center Amsterdam (AMC) in The Netherlands.

Assessments

The Childhood Trauma Questionnaire short version (CTQ; Bernstein & Fink, 1998; Bernstein et al., 2003) was administered to retrospectively assess maltreatment during childhood. The CTQ is a self-report instrument with 25 Likert-scale items reflecting maltreatment frequency (ranging from 1=never true to 5=very often true). Factor analyses have confirmed the five-factor structure of the CTQ: physical abuse, physical neglect, emotional abuse, emotional neglect and sexual abuse (Bernstein et al., 2003). Convergent and divergent validity of these scales have been reported (Bernstein et al., 1997). The factor structure and criterion validity of the Dutch CTQ have also been confirmed (Thombs et al., 2009). The CTQ cut-off values for ‘low to moderate’, ‘moderate to severe’ and ‘severe to extreme’ abuse or neglect of each subscale were used for descriptive ƉƵƌƉŽƐĞƐ͗шϵ͕шϭϯĂŶĚшϭϲĨŽƌĞŵŽƚŝŽŶĂůĂďƵƐĞ͖шϴ͕шϭϬĂŶĚшϭϯĨŽƌƉŚǇƐŝĐĂůĂďƵƐĞ͕шϲ͕шϴĂŶĚ шϭϯĨŽƌƐĞǆƵĂůĂďƵƐĞ͕шϭϬ͕шϭϱĂŶĚшϭϴĨŽƌĞŵŽƚŝŽŶĂůŶĞŐůĞĐƚ͖ĂŶĚшϴ͕шϭϬĂŶĚшϭϯĨŽƌƉŚǇƐŝĐĂů neglect (Bernstein & Fink, 1998). See online supplement section for full assessment details.

fMRI task

(7)

144

target duration time was calibrated throughout the task based on the success rate; all participants succeeded on approximately two thirds of the trials. After a jittered delay (700-2100 ms), participants received feedback (1920 ms), notifying them about their present trial and current total score, followed by an inter-trial interval of 4000 ms. During reward trials, participants won €0.50 if they responded to the target on time; during loss trials, participants lost nothing if they responded to the target on time, but lost €0.50 if they failed to do so. Finally, during neutral trials participants neither won nor lost money, irrespective of task performance.

All participants completed a full run of the MID before entering the scanner. Participants were aware that they would be rewarded with the monetary outcome of the experiment after the MRI, in addition to their financial remuneration for participation in the study. Analyses on the Monetary Incentive Delay paradigm in relation to disruptive and psychopathic traits are reported elsewhere (Cohn et al., 2015).

fMRI protocol

First, T1-weighted anatomical scans, consisting of 180 sagittal 1mm thickness slices, with an in-plane resolution of 1x1 mm (FOV 256x256 mm, TR 9.0 ms, TE 3.5 ms), were acquired using an 8-channel SENSE head-coil. Furthermore, 400 T2* weighted echo-planar images (EPI) were acquired during the MID task, each volume consisting of 38 ascending slices of 3 mm thickness and 2.29x2.29 in-plane resolution, parallel to the anterior commissure - posterior commissure line (FOV 220x220mm, TR 2300, TE 30 ms).

Statistical analysis

Functional MRI data were processed using SPM8. Preprocessing included realignment, unwarping, slice-time correction to the middle slice, normalization to MNI space based on the segmented anatomical scan, and 8mm FWHM smoothing. First level models included separate regressors for each trial type, i.e. anticipation period (up to target), target, and feedback period. Next, contrast images were computed to assess reward anticipation (reward trial anticipation > neutral trial anticipation).

To address our first aim and assess the association between CTQ scales and ventral striatum (VS) responsiveness during the reward anticipation epoch, we extracted the mean parameter estimates (representing percentage signal change) from 8 mm spheres centred on meta-analytic ventral striatum peak voxel coordinates from neuroimaging studies on reward processing (Liu et al., 2011) (left [x=-12 y=10 z=-6] and right [x=12 y=10 z=-4]). The ventral striatum indeed showed robust activation during reward anticipation (see Figure 1): right ventral striatum t124=14.5, pFWE<.001, maximal at [x=12, y=16, z=-6]; left ventral striatum t124=14.3,

pFWE<.001, maximal at [x=-10, y=12, z=-6]. This measure was then used as the dependent variable

(8)

9

145

hit. We report both region of interest analysis in the ventral striatum (see above) and whole-brain analyses at a statistical threshold of family wise error corrected p<.05. Finally, to control for the effects of influential outliers (z>3), we performed follow-up analyses using nonparametric correlation analyses (Spearman’s rho).

Figure 1. Main task effects during reward anticipation (n=125). Statistical parametric map displaying a

one-sample t-test for main task effects during reward anticipation, overlaid on an anatomical template at y=10.

Results

Sample characteristics and task performance

The sample was characterized (see Table 1) by a mean IQ of 91, substantial variation in both internalizing and externalizing psychopathology, and a broad range of childhood adversity scores (ranging from 25 [no abuse or neglect] to 71 [severe abuse and neglect] out of a possible total of 125; see Table 1). CTQ subscales were correlated (r-values between .08 and .60) but showed sufficient unique variation to examine their distinct neurobiological correlates using multiple regression (maximal potential variance inflation factor [VIF]=1.4-3.4, i.e., VIF<10, cf. Kutner et al., 2004). CTQ total scores were moderately correlated with parent-reported (CBCL) and self-reported (YSR) internalizing (r=.31, p<.001;r=.24, p<.001) and externalizing problem scores (r=.30, p<.001; r=.44, p<.001).

Participants won an average of €2.60 (SD 1.20) during the MID task and were successful in 61% (SD 5%) of the trials, with no effect of trial type (p=.9). As expected, mean reaction time was shorter for reward and loss trials compared to neutral trials (Welch F2,242=26.8, p<.001). CTQ

(9)

146

Table 1. Socio-demographic and mental health characteristics of the current sample (n=125)

CTQ emotional abuse, mean (SD) 7.2 (2.9) CTQ physical abuse, mean (SD) 6.0 (2.4)

CTQ sexual abuse, mean (SD) 5.4 (1.6)

CTQ emotional neglect, mean (SD) 10.2 (4.2) CTQ physical neglect, mean (SD) 6.5 (2.1)

CTQ total score, mean (SD) 35.3 (9.5)

No abuse or neglect, no. (%) 52 (42%)

Low to moderate abuse or neglecta, no. (%) 40 (32%) Moderate to severe abuse or neglectb, no. (%) 16 (13%) Severe to extreme abuse or neglectc, no. (%) 17 (14%)

Age, years, mean (SD) 17.8 (1.5)

Male gender, no. (%) 106 (85%)

Low SES neighborhood, no. (%) 66 (54%)

Non-Western ethnicity, no. (%) 33 (26%)

IQ, mean (SD) 91.3 (14.1)

CBCL Internalizing, mean T-score (SD) 51.1 (10.9) YSR Internalizing, mean T-score (SD) 47.7 (9.8) CBCL Externalizing, mean T-score (SD) 52.1 (11.5) YSR Externalizing, mean T-score (SD) 52.7 (9.8)

ADHD no. (%) 41 (33%)

ODD no.(%) 16 (13%)

CD no. (%) 15 (12%)

PTSD no. (%) 2 (1.6%)

Notes. a_{Reaching the ‘low to moderate’ cut-off on any subscale;}b_{Reaching the ‘moderate to severe’ cut-off}

on any subscale; c Reaching the ‘moderate to severe’ cut-off on any subscale.

Abbreviations: SES: Socio-economic status, RPQ: Reactive Proactive aggression Questionnaire, ADHD: Attention Deficit/Hyperactivity Disorder, DBD: Disruptive Behaviour Disorders, PTSD: Post-Traumatic Stress Disorder, CBCL: Child Behavior Checklist, YSR: Youth Self Report, YPI: Youth Psychopathic traits Inventory.

CTQ scores and monetary incentive processing: whole-brain analyses

No significant associations were found between CTQ scores and brain activation during any monetary incentive processing epoch (i.e., reward anticipation, reward outcome, loss anticipation and loss outcome) at the whole-brain level.

CTQ scores and reward anticipation in ventral striatum

(10)

9

147

continuous measures of mental health problems (CBCL and YSR) could account for these results. The inclusion of these additional variables in the model did not reduce the beta values associated with the CTQ subscales. However, CBCL externalizing and YSR internalizing scores were also significantly associated with left VS responses during reward anticipation. In an additional analysis to explore the role of a psychiatric diagnosis of DBD, participants who met criteria for either Oppositional Defiant Disorder or Conduct Disorder (n=31) were excluded. The pattern of findings with respect to the association between CTQ scales and VS responses was unchanged. Finally, additional analyses were conducted in order to control for the possible contribution of individual differences in behavioural performance and the contribution of realignment parameters in first level models. However, the main findings remained unaltered.

Table 2. Relation between CTQ subscales and ventral striatum responses during reward anticipation in the right (upper panel) and left (lower panel) ventral striatum (n=125).

Unstandardized coefficients

Standardized coefficient

B SE Beta t P-value

Right ventral striatum model

Step 1 (R2=.15, F(2,122)=11.0, p<.001)

Constant 0.45 0.11 4.3 <.001

CTQ Physical abuse 0.053 0.013 .37 4.1 <.001 CTQ Physical neglect -0.053 0.015 -.32 -3.6 <.001

Step 2

No new variables were added to the model

Left ventral striatum model

(11)

(12)

9

149

Figure 2. Association between CTQ subscales and ventral striatum responsiveness during reward anticipation

(n=125). Upper frames: statistical parametric map, overlaid on an anatomical template at y=10 (a) and partial regression plots for left (b) and right (c) ventral striatum responsiveness during reward anticipation, indicating positive associations with physical abuse. Lower frames: statistical parametric map, overlaid on an anatomical template at y=10 (d) and partial regression plots for left (e) and right (f) ventral striatum responsiveness during reward anticipation, indicating negative associations with physical neglect.

CTQ scores and ventral striatum responsiveness during other MID conditions

Table 3 shows that some of the CTQ scales were also related to VS responses during other incentive processing epochs. Physical abuse was positively correlated with VS activation during loss anticipation. Furthermore, sexual abuse was positively associated with bilateral VS responses during anticipation of loss. All findings remained unaltered after controlling for demographics and mental health covariates, except for the association between sexual abuse and right VS responses during loss anticipation (new E=.17, p=.07).

Non-parametric analyses

(13)

(14)

9

151 Discussion

This is the first brain imaging study to assess the impact of childhood maltreatment on reward processes in a community sample of at-risk adolescents; to our knowledge it is also the first study that systematically explores the distinct and unique contributions of maltreatment subtypes on brain function in late adolescence. Using an established fMRI incentive processing paradigm, we found that physical abuse was positively associated with VS responsiveness during reward and loss anticipation, that physical neglect was negatively associated with VS responsiveness during reward anticipation, and that sexual abuse was positively associated with VS responses during loss anticipation. None of these findings could be attributed to the presence of dimensional (CBCL, YSR) or categorical (DISC-IV) indicators of current psychopathology. Collectively, these findings indicate that childhood maltreatment is associated with aberrant reward processing in line with what has been found in a sample of adolescents who had experienced extreme neglect in an institutional context (Mehta et al., 2010). They also provide preliminary evidence that maltreatment subtypes are likely to have a differential impact on regional brain function, which may predispose for distinct psychiatric outcomes.

The ventral striatum is at the heart of the brain’s dopaminergic reward and saliency signalling system (Haber & Knutson, 2010) and has been implicated in anticipatory reward signalling once learning has occurred (Knutson et al., 2001), coding specific aspects of reward anticipation, such as its salience, uncertainty and magnitude (Yacubian et al., 2007; Cooper & Knutson, 2008). Both enhanced and reduced VS responsiveness have been associated with psychiatric disorders, including depression, anxiety disorders, disruptive behaviour disorders and substance abuse (Bjork et al., 2010a; Figee et al., 2011; Hommer et al., 2011; Stoy et al., 2012). Several theories have been postulated to tentatively account for this association, and provide a mechanistic account as to how aberrant reward processing may increase vulnerability to mental health problems. Given the role of the VS in hedonic experience (see above), its hyporesponsiveness has been conceptualized as representing an anhedonic state (Pechtel & Pizzagalli, 2011) – relevant to depression and the negative symptoms of schizophrenia – but also to compensatory reward-seeking behaviour (Pechtel & Pizzagalli, 2011) such as in delinquency or substance abuse. VS hyperresponsiveness, on the other hand, has been linked to enhanced reward delay discounting (i.e. focusing on immediate reward; Hariri et al., 2006) which may also help explain its association with disruptive behaviour and substance use (Bjork et al., 2010a; Hommer et al., 2011), and, when high salience is erroneously attributed to innocuous stimuli, to the positive symptoms of schizophrenia (Howes & Kapur, 2009) (and potentially post-traumatic stress disorder). These mechanisms are not static and may shift during development (Galvan et al., 2006) and disease progression.

(15)

152

alcoholics (Yau et al., 2012). As our findings were not explained by current psychopathology or substance use, we suggest that the positive association between physical abuse and VS responsiveness during reward and loss anticipation may contribute to increased vulnerability to future substance use/risky behaviour, possibly as a result of high reward delay discounting (Hariri et al., 2006) due to higher anticipatory reward responses. While it is possible to speculate that higher VS responses during loss anticipation (both in relation to physical and sexual abuse) index positive hedonic experience, leading to increased risk-taking tendencies, it seems equally, if not more, likely that this pattern reflects some form of hypervigilance to potential threat, as has previously been reported in individuals exposed to childhood maltreatment (McCrory et al., 2013).

Second, the association between physical neglect and lower VS responsiveness during reward anticipation suggests increased uncertainty about the chance of actually receiving a possible reward. Individuals who have experienced physical neglect are likely to have grown up in an environment where there was indeed uncertainty that their basic needs would be met and where rewarding experiences were relatively rare. We hypothesize that VS activation in neglected individuals may reflect the systematic encoding of reward experience during development, which in turn acts as a latent vulnerability factor increasing risk of psychiatric disorder (McCrory & Viding, 2015). For example, reduced VS responses during reward anticipation have been linked to depression (Stoy et al., 2012). It is unlikely that the specific patters of atypical VS response reported here predispose to circumscribed clinical outcomes given the plethora of outcomes associated with neglect; rather, alterations in neural patterns of reward processing may represent the biological embedding of a more generic psychiatric vulnerability that may manifest depending on the nature of future stressor exposure.

These findings should be interpreted in the context of several limitations. First, we used a retrospective self-report of childhood maltreatment. While the CTQ is one of the most widely used measures of childhood maltreatment and has been validated extensively, it has been speculated (Dannlowski et al., 2012) that specific neural abnormalities may increase the risk of recall bias and thus influence their association with childhood maltreatment. As in the study by Dannlowski et al. (2012), our sample was well characterized in terms of current psychopathology, and controlling for these factors did not alter the main findings. However, longitudinal studies, following maltreated children into adulthood, are warranted to be confident about the direction of the association between brain dysfunction and reports of childhood maltreatment. Second, in the current study, we focused on the ventral striatum, while a range of other brain regions have also been implicated in reward processing (Liu et al., 2011). Finally, our cross-sectional design, as well as the age of the sample at the time of scanning, precludes an explicit test of the hypothesis that altered ventral striatum activity partly mediates the association between childhood maltreatment and adult mental health.