Reactive attachment disorder (RAD), a severe social functioning disorder associated with early childhood maltreatment, has a prevalence of 1.4–2.4% in the general population Reference Minnis, Macmillan, Pritchett, Young, Wallace and Butcher1 and 19.4–40% among foster children. Reference Zeanah, Scheeringa, Boris, Heller, Smyke and Trapani2,Reference Lehmann, Havik, Havik and Heiervang3 According to DSM-5, 4 RAD is characterised by emotionally withdrawn/inhibited behaviours towards caregivers, which is different from disinhibited social engagement disorder (disinhibited type of RAD based on DSM-IV criteria), where the child exhibits indiscriminate social/disinhibited behaviours. Reference Gleason, Fox, Drury, Smyke, Egger and Nelson5 Children with RAD display various difficulties with cognition (learning and memory), behaviour, and social and emotional disturbances. Reference Guttmann-Steinmetz and Crowell6,Reference Kay and Green7 Complex neurodevelopmental problems, including multiple psychiatric comorbidities, Reference Kocovska, Puckering, Follan, Smillie, Gorski and Barnes8 such as substance use disorders, Reference Andersen and Teicher9,Reference Pechtel, Woodman and Lyons-Ruth10 are also prevalent. These individuals are likely to experience disturbed and developmentally inappropriate social interactions throughout adulthood. Reference Kocovska, Puckering, Follan, Smillie, Gorski and Barnes8,Reference van der Vegt, van der Ende, Ferdinand, Verhulst and Tiemeier11
Research on RAD has reported several deficiencies, including the absence of focused attachment behaviour towards a preferred caregiver, failure to seek and respond to comfort when distressed, reduced social and emotional reciprocity and emotion regulation disturbances (e.g. reduced positive affect and unexplained fearfulness or irritability). Reference O'Connor and Rutter12,Reference Green and Goldwyn13 Children with disrupted attachment, who had experienced early maternal withdrawal, developed a substance use disorder in young adulthood, Reference Pechtel, Woodman and Lyons-Ruth10 and childhood maltreatment is associated with increased severity of substance use disorder and may also increase the number or frequency of relapse. Reference Andersen and Teicher9 A brain imaging study has demonstrated strong links between substance use disorder and dysfunction in reward circuitry, and in particular dopamine function within cortico-striatal circuitry. Reference Dichter, Damiano and Allen14 Given an increased risk for drug and alcohol misuse, dopaminergic dysfunction may underlie RAD pathophysiology. Thus, identifying neural reward system functionality in RAD could help elucidate disease pathogenesis and inform preventive and therapeutic interventions. However, assessments of reward circuitry in RAD using functional magnetic resonance imaging (fMRI) have yet to be conducted (for a structural MRI study of RAD see Shimada et al Reference Shimada, Takiguchi, Mizushima, Fujisawa, Saito and Kosaka15 ).
Therefore, the present study investigated potential deficiencies in reward-related neural activity in children and adolescents with RAD using fMRI during performance of a monetary reward task. Reference Izuma, Saito and Sadato16 Furthermore, we investigated the association between neural activity in the reward system and clinical symptoms. Given the key role of the basal ganglia, including the striatum (nucleus accumbens (NAc), caudate and putamen), in reward processing, Reference Haber and Knutson17 we hypothesised that children and adolescents with RAD would display reduced neural activity in response to reward, and these decrements would be predictive of clinical RAD symptoms.
Sixteen right-handed, medication-naive children and adolescents with a clinical diagnosis of RAD (DSM-5 criteria), aged 10–15 years (mean 12.6 years), were recruited from the Department of Child and Adolescent Psychological Medicine at the University of Fukui Hospital from August 2013 to March 2015.
A diagnosis of RAD was confirmed during structured interviews by five licensed paediatric-psychiatric clinicians (S.T., S.M., H.K, H.Ko, A.T.) according to DSM-5 criteria. To exclude other psychiatric diseases (e.g. mood-related disorders, anxiety disorders, post-traumatic stress disorder (PTSD) and substance use disorder) and neurodevelopmental disorders, such as autism spectrum disorder and attention-deficit hyperactivity disorder (ADHD), the Mini-International Neuropsychiatric Interview for Children and Adolescents (MINI-KID) Reference Sheehan, Sheehan, Shytle, Janavs, Bannon and Rogers18 and an assessment module of DSM-IV ADHD taken from the Schedule of Affective Disorders and Schizophrenia for School-Age Children, Epidemiologic version (K-SADS-E) were administered to the participants.
All participants with RAD had a history of severe maltreatment when younger, including emotional abuse, neglect, physical abuse and/or sexual abuse. The participants with RAD were living within a stable placement (in a child welfare facility) even though they were not living with biological parents (for information about child welfare services in Japan see Suzuki et al Reference Suzuki and Tomoda19 ). Participants were all medication-naive, and those taking dopaminergic drugs (including treatment with methylphenidate and atomoxetine), selective serotonin reuptake inhibitors for mood and anxiety symptoms, or antipsychotic medications for aggressive behaviour and irritability were excluded from the study.
A group of 20 right-handed age- and gender-matched individuals considered as typically developing (TD; mean age 12.7 years) with no histories of abuse or neglect and meeting DSM-IV 20 criteria for any major Axis I or Axis II psychiatric disorder were included as a control group. The control sample was recruited from the general Fukui Prefecture community. Participants were excluded if they had a full scale intelligence quotient (FSIQ) <80 on the Wechsler Intelligence Scale for Children-Fourth Edition (WISC-IV). Reference Wechsler21
The study protocol, approved by the Ethics Committees from the University of Fukui, Japan (Assurance No. FU25-157), was conducted in accordance with the Declaration of Helsinki. All participants and their parent(s) or director of the respective child welfare facility gave written informed assent and consent for participation. This study is registered with the University Hospital Medical Information Network (UMIN000014655).
Psychiatric symptom measures
We obtained official information regarding participants’ maltreatment during early childhood from Child Protective Services. All psychiatric symptom measures were administered to the participants by five licensed paediatric-psychiatric clinicians (S.T., S.M., H.K, H.Ko, A.T.). To assess participants’ maltreatment in early childhood, the Child Abuse and Trauma Scale (CATS) Reference Sanders and Becker-Lausen22 was administered. Attachment-related symptoms were measured using the Internal Working Model Scale (IWMS) questionnaire. Reference Hazan and Shaver23 The IWMS consists of three subscales: secure attachment (e.g. ‘I can get close to others easily’), avoidant attachment (e.g. ‘I do not like being dependent on others’) and ambivalent attachment (e.g. ‘I get annoyed by others because I always want to be with them’). Each subscale scores can be treated as scale scores or the highest subscale score is useful as a categorical score. This questionnaire has been standardised, and its reliability and validity have been confirmed. Reference Sheehan, Sheehan, Shytle, Janavs, Bannon and Rogers18 Behavioural/emotional problems were assessed using the Child Behaviour Checklist (CBCL). Reference Achenbach24
Additional instruments used to exclude other psychiatric diseases included the Birleson Depression Self-Rating Scale for Children (DSRSC), Reference Birleson25 the Impact of Event Scale-Revised (IES-R), Reference Asukai, Kato, Kawamura, Kim, Yamamoto and Kishimoto26 the Autism-Spectrum Quotient (AQ) Reference Wakabayashi, Baron-Cohen, Uchiyama, Yoshida, Tojo and Kuroda27 and the Japanese version of the ADHD Rating Scale-IV (ADHD-RS-IV-J). Reference Yamazaki28 To assess RAD-related symptom severity, participants completed the self-report version of the Strength and Difficulties Questionnaire (SDQ), Reference Goodman29 a 25-item questionnaire assessing children's behaviour problems, as well as prosocial behavioural tendencies.
Participants performed a block-design gambling task used in our previous studies Reference Izuma, Saito and Sadato16,Reference Mizuno, Yoneda, Komi, Hirai, Watanabe and Tomoda30 ( ). Participants were asked to choose one of three cards by pressing a button. Each card was randomly associated with 0, 30 or 60 yen. Three conditions of eight trials were performed (24 s). Unknown to participants, the total reward was predetermined. To investigate the effect of reward levels between groups, two reward conditions with different levels were designed. In the high monetary reward (HMR) condition, participants earned an average of 330 yen (range 270–390), which was consistently higher than the expected value of the 8 reward trials (240 yen). In the low monetary reward (LMR) condition, participants earned an average of 150 yen (range 90–210), which was consistently lower than the expected value. In the no monetary reward (NMR) condition, the outcome presented was always ‘XXX’ to control for effects other than reward level. The NMR condition or a fixation rest condition (24 s) was always inserted between the two reward conditions. During scanning, participants performed four sessions. Each session consisted of four blocks from each of the four conditions (HMR, LMR, NMR and fixation rest). The order of the four sessions was counterbalanced across participants.
fMRI acquisition and analysis
Functional images from each participant were acquired using a 3.0-T MR scanner (Discovery MR 750; General Electric Medical Systems, Milwaukee, Wisconsin, USA). For functional imaging, a series of 528 volumes (132 volumes per session) was acquired using interleaved T2-weighted, gradient echo, echo planar imaging (EPI) sequences. Each volume consisted of 44 transaxial slices with a thickness of 3.0 mm between slices, which included the entire cerebrum and cerebellum (repetition time (TR), 3000 ms; echo time (TE), 25 ms; flip angle (FA), 90°; field of view (FOV), 192 mm; in-plane matrix size, 64 × 64 pixels, voxel dimensions, 3.0 × 3.0 × 3.0 mm; slice gap, 0 mm).
Data were analysed using SPM8 (The Wellcome Trust Centre for Neuroimaging, London, UK) implemented in MATLAB 2014a (Mathworks, Natick, Massachusetts, USA; see supplemental file for more details). At the first level, individual task-related activation was evaluated. Three regressors for each condition (i.e. HMR, LMR and NMR) were modelled at the onset of each block (duration of 24 s), which were convolved with a canonical haemodynamic response function to obtain expected task-related signal change. The weighted sum of the parameters estimated during individual analyses consisted of ‘contrast’ images. Global signal changes were utilised to remove confounding factors such as scanner gain.
At the second level, contrast images corresponding to each condition for each participant were used for group analyses with a random-effects model to obtain population inferences. We utilised one-sample t-tests to depict relevant brain regions for each group and two-sample t-tests to compare activation between the RAD and TD group. To exclude the effects of age, gender, intellectual ability and depression, Reference Edmiston, Wang, Mazure, Guiney, Sinha and Mayes31–Reference Pizzagalli, Holmes, Dillon, Goetz, Birk and Bogdan33 we included these measures as regressors. Significant signal change for each contrast was assessed by means of t-statistics on a voxel-by-voxel basis. The statistical threshold was set at P<0.001 at the voxel level and P<0.05 with a family-wise error (FWE) correction for multiple comparisons at the whole-brain cluster level.
We depicted brain regions that responded to each reward condition, relative to the control condition, with the following contrasts: HMR v. NMR, LMR v. NMR and HMR v. LMR for each group. We then compared activations between the RAD and TD groups. Finally, we conducted a correlation analysis to examine whether brain activation patterns were related to clinical symptoms.
Sensitive period analysis
The presence of a potential ‘sensitive period,’ during which exposure to childhood maltreatment might be more strongly associated with alterations in striatal responses to HMR, was assessed using random forest regression with conditional inference trees (‘cforest’ in R package party Reference Strobl, Boulesteix, Zeileis and Hothorn34 ), as used in previous studies Reference Birleson25 (see supplemental file for more details). The analysis was conducted to evaluate the importance of potential predictors of left and right striatal responses.
We determined the importance of exposure during specific time periods including in utero exposure in mothers experiencing domestic violence and direct exposure to neglect or physical, emotional or sexual abuse from birth to 15 years of age, annually. Years were scored either 0 or 1 for exposure to any type of maltreatment during the year. Duration of abuse was included as an additional predictor to control for the potential confounding relationship between early exposure and longer duration.
Statistical analyses were performed using IBM SPSS 22 software package (SPSS, Chicago, Illinois, USA). Bonferroni adjustments for partial correlations were applied when such analyses were used. Analysis of variance (ANOVA) or analysis of covariance (ANCOVA) was used to compare participant characteristics across groups.
Demographics and assessment outcomes
Clinical characteristics of the RAD and TD groups are shown in Table 1. The two groups were well matched on gender and age. In comparison with the TD group, the RAD group showed lower FSIQ, F(1, 33)=7.30, P<0.05, and WISC subscale scores (Verbal Comprehension Index, F(1, 33)=4.58, P<0.05; Perceptual Reasoning Index, F(1, 33)=1.41, P=0.024; Working Memory Index, F(1, 33)=3.83, P=0.06; and Processing Speed Index, F(1, 33)=5.47, P<0.05). Regarding the IWMS, the RAD group showed significantly lower secure, F(1, 32)=6.24, P<0.05, whereas the RAD group showed significantly higher, avoidant and ambivalent scores, F(1, 32)=9.75, P<0.01 and F(1, 32)=9.14, P<0.01 respectively.
|TD group n=20||RAD group n=16||F||χ2||d.f.||P|
|Age, years: mean (s.d.)||12.7 (1.8)||12.6 (2.0)||0.047 a||1, 33||0.83|
|WISC-IV FSIQ, mean (s.d.)||103.5 (9.5)||94.5 (10.4)||7.30 a||1, 33||<0.05|
|Verbal Comprehension Index||106.3 (11.2)||97.6 (12.9)||4.58 a||1, 33||<0.05|
|Perceptual Reasoning Index||98.9 (13.8)||93.8 (11.0)||1.41 a||1, 33||0.24|
|Working Memory Index||102.7 (13.3)||94.0 (12.7)||3.83 a||1, 33||0.06|
|Processing Speed Index||102.4 (9.6)||95.4 (9.8)||5.47 a||1, 33||<0.05|
|SDQ total difficulties score, mean (s.d.)||7.2 (3.8)||16.1 (6.0)||23.5 b||1, 32||<0.001|
|CBCL total, mean (s.d.)||50.2 (7.8)||61.3 (10.3)||11.13 b||1, 32||<0.01|
|ADHD-RS total score, mean (s.d.)||1.6 (1.9)||8.3 (8.2)||8.94 b||1, 32||<0.01|
|AQ total score, mean (s.d.)||10.8 (3.1)||18.3 (8.4)||7.95 b||1, 32||<0.01|
|DSRSC, mean (s.d.)||6.5 (3.3)||14.6 (7.7)||17.88 b||1, 32||<0.001|
|IWMS-secure, mean (s.d.)||22.9 (6.6)||17.1 (5.1)||6.24 b||1, 32||<0.05|
|IWMS-avoidant, mean (s.d.)||13.6 (4.2)||15.6 (4.3)||9.75 b||1, 32||<0.01|
|IWMS-ambivalent, mean (s.d.)||14.8 (4.8)||19.0 (6.8)||9.14 b||1, 32||<0.01|
|CATS, mean (s.d.)||15.9 (6.4)||41.6 (27.2)||19.15 b||1, 31||<0.001|
|IES-R, mean (s.d.)||2.0 (4.6)||23.1 (18.4)||36.38 b||1, 32||<0.001|
|Male participants, n (%)||10 (50)||8 (50)||1.00||1||1.00|
|Right handedness, n (%)||20 (100)||16 (100)||1.00||1||1.00|
|Types of maltreatment, n (%)|
|Physical abuse||–||7 (44)|
|Emotional abuse||–||10 (63)|
|Sexual abuse||–||1 (6)|
|Number of types of maltreatment, mean (s.d.)||–||1.8 (0.8)|
|Duration (years) of maltreatment, mean (s.d.)||–||8.0 (4.5)|
TD, typically developing; RAD, reactive attachment disorder; WISC-IV, Wechsler Intelligence Scale for Children-Fourth Edition; FSIQ, full scale intelligence quotient; SDQ, Strengths and Difficulties Questionnaire; CBCL, Child Behaviour Checklist; ADHD-RS, ADHD Rating Scale-IV; AQ, autism spectrum quotient; DSRSC, Depression Self-Rating Scale for Children; IWMS, Internal Working Models Scale; CATS, Child Abuse and Trauma Scale; IES-R: Impact of Event Scale-Revised.
a Adjusted for gender.
b Adjusted for gender and FSIQ.
The RAD group showed significantly higher scores on the CATS, F(1, 31)=19.15, P<0.001, total behaviour problems on the SDQ, F(1, 32=23.5, P<0.001, and CBCL, F(1, 32)=11.13, P<0.01, in comparison to the TD group. The RAD group scored higher on certain items of the ADHD-RS-IV-J, F(1, 32)=8.94, P<0.01; AQ, F(1, 32)=7.95, P<0.01; DSRSC, F(1, 32)=17.88, P<0.001; and IES-R, F(1, 32)=36.38, P<0.001, measures. However, no participant met the criteria for PTSD, ADHD or autism spectrum disorder.
Response rates on the reward task did not differ between the RAD group (mean=99.7%, s.d.=0.4%) and the TD group (mean=99.9%, s.d.=0.1%). Moreover, there were significant differences in terms of reaction times (RTs), with the RAD group responding more slowly as compared with the TD group (). A two-way ANOVA revealed no significant main effect of reward condition, F(2, 68)=2.37, P=0.101, and no interaction with group and reward condition, F(2, 68)=0.16, P=0.850, whereas the main effect of group was significant, F(1, 34)=4.75, P=0.036. Furthermore, there was no correlation between RTs and FSIQ.
A whole-brain analysis with FWE-cluster correction (P<0.05) was conducted for the HMR and LMR conditions (HMR or LMR minus NMR; Fig. 1). The TD group, regardless of monetary reward level, exhibited significant activation in several brain regions, including the ventral striatum (caudate nucleus and NAc). In contrast, the RAD group did not show significant activation in the ventral striatum in any reward condition ( ). To further specify the brain regions involved in reward level-dependent differences, we contrasted the HMR condition with the LMR condition (HMR minus LMR) in each group. No significant differences were found with this analysis.
To identify potential deficiencies in reward-related activation in RAD, we used the between-group contrast (TD minus RAD) in the HMR (HMR minus NMR) and LMR (LMR minus NMR) conditions (Fig. 1). For the HMR condition, there was a significant decrease in activation in the bilateral ventral striatum (caudate nucleus and NAc) for the RAD group compared with the TD group (FWE corrected P=0.015 at the cluster level). Within the identified cluster of the ventral striatum, we extracted activation intensity (i.e. contrast estimates) at the peak Montreal Neurological Institute (MNI) coordinates (left, x = −8, y = 12, z = −4; right, x = 8, y = 14, z = −4). In the LMR condition, no differences were detected in the activation level of each region between the TD and RAD groups. There was no correlation between intensity and RTs in either sample. No differences were found for either reward condition when the RAD group was contrasted with the TD group (RAD minus TD).
Correlations between fMRI activation and clinical symptom measures
We calculated partial correlation coefficients between activation levels and relevant clinical measures, controlling for age, gender, the FSIQ and DSRSC (Table 2). Results revealed a negative association between the IWMS avoidant attachment score and bilateral ventral striatal activation (right: r=−0.481, P=0.0054; left: r=−0.469, P=0.0068), indicating that children with high avoidant tendencies had less striatal activity than children with low avoidant tendencies.
|Left ventral striatum||Right ventral striatum|
fMRI, functional magnetic resonance imaging; FSIQ, full scale intelligence quotient; DSRSC, Depression Self-Rating Scale for Children; SDQ, Strengths and Difficulties Questionnaire; CBCL, Child Behaviour Checklist; IWMS, Internal Working Models Scale; IES-R: Impact of Event Scale-Revised; CATS, Child Abuse and Trauma Scale.
* For Bonferroni correction at P<0.05 (individual test, P<0.007 without Bonferroni correction).
Sensitive period analysis: age of exposure
As illustrated in Fig. 2, the most important temporal predictor of right striatal activation to HMR was whether children and adolescents with RAD were exposed to maltreatment at 0–1 year of age. Monte Carlo re-randomisation tests indicated that the probability of obtaining an importance measure of this magnitude at any age was quite low (P=0.0004). The second most important age was 2–3 years, and the probability of obtaining two peaks with this combined degree of importance was P=0.001. Altogether, importance was high from 0 to 4 years of age. Total duration of exposure was the fourth most important predictor of right striatal responses. Predicted responses based only on these temporal factors were correlated with actual responses (r=0.571, P<0.011). Similarly, the most important predictor of left striatal activation to HMR was exposure to maltreatment at 0–1 year of age (P=0.0005). Total duration of exposure was the second most important temporal predictor, and the likelihood of obtaining two peaks of this combined height was low (P=0.003). The sensitive period for left striatal responses appeared to be shorter than for the right striatum. Predicted responses based only on these temporal factors were correlated with actual responses (r=0.462, P<0.05; ).
The present fMRI study investigated whether neural activity during reward processing is altered in children and adolescents with RAD. Results showed remarkably decreased activity in the ventral striatum (caudate and NAc) between the RAD and TD groups during the HMR condition. Additionally, activity in the identified regions during reward processing was negatively associated with avoidant attachment scores for both groups. Together, these results suggest that neural reward processing is severely reduced in RAD, and these alterations in neural processing may influence the emergence of an avoidant attachment style.
Our finding of reduced activation in the ventral striatum is largely consistent with previous reports, suggesting decreased striatal activity during reward anticipation in adults and adolescents with a history of childhood adversity. Reference Mehta, Gore-Langton, Golembo, Colvert, Williams and Sonuga-Barke36–Reference Hanson, Hariri and Williamson39 Specifically, two notable neuroimaging studies directly assessing reward system deficits during adolescence observed reduced activity to reward cues in the ventral striatum among Romanian adolescent adoptees who had experienced early global deprivation, Reference Mehta, Gore-Langton, Golembo, Colvert, Williams and Sonuga-Barke36 and individuals who had experienced emotional neglect in early childhood. Reference Hanson, Hariri and Williamson39 Whereas ventral striatal activity during reward processing was previously associated with depressive symptoms in maltreated adolescents, Reference Hanson, Hariri and Williamson39 the reduced reward-related activity in the ventral striatum was also observed in the RAD group compared with the TD group, despite controlling for related depression scale scores (i.e. DSRSC). That is, reward dysfunction (as well as depressive symptoms) may be associated with attachment problems in maltreated individuals. The present study also differed from these aforementioned reports in that participants were medication-naive and diagnosed with RAD. To the best of our knowledge, the present study is the first to demonstrate neural decrements in the bilateral ventral striatum in children and adolescents with RAD. Our findings suggest that these functional alterations may influence RAD pathogenesis and/or severity.
Children and adolescents with RAD are more likely to be at-risk for multiple psychiatric disorders in adulthood, including substance use disorder and depressive disorders. These disorders have been previously associated with reward circuitry dysfunction. For example, substance use disorder has been strongly linked to dysfunction in reward networks, particularly in dopamine function within cortico-striatal circuitry. Reference Dichter, Damiano and Allen14 Moreover, patients with major depressive disorder also show reduced striatal responses to rewards. Reference Pizzagalli, Holmes, Dillon, Goetz, Birk and Bogdan33 Thus, our findings may indicate that alterations in the striatal system represent an early stage in psychosis development.
Regarding the effect of reward level, we also found significant activation in the ventral striatum for LMR conditions for the TD group but not the RAD group; however, differences between groups were only marginally significant based on a between-groups contrast. These findings suggest consistently reduced activation in the ventral striatum, regardless of reward level; conversely, our previous study suggested specific dysfunction within a ‘low reward’ level in ADHD. Reference Mizuno, Yoneda, Komi, Hirai, Watanabe and Tomoda30 Children and adolescents with RAD exhibited extremely low activity in the ventral striatum during both high- and low-value reward trials. These results suggest that reward sensitivity neural processing is severely impaired in children and adolescents with RAD. The decreased reward sensitivity and severe symptoms observed among children and adolescents with RAD may contribute to decreased neural processing in the ventral striatum, leading to diminished motivation for coping with daily life activities.
Although the neural mechanisms underlying RAD due to early life adversity remain unclear, it is likely that child maltreatment affects both anatomical alterations and behavioural dysfunction. Although anatomical changes associated with reward processing in RAD remain unknown, structural neuroimaging studies suggest that exposure to early life adversity is associated with smaller caudate volumes. Reference Dannlowski, Stuhrmann, Beutelmann, Zwanzger, Lenzen and Grotegerd40,Reference Cohen, Grieve, Hoth, Paul, Sweet and Tate41 Hence, for children and adolescents with RAD, there may be a possible association between decreased reward sensitivity and brain structure. Similarly, a previous fMRI study assessing the effects of early-life maltreatment on reward-related learning Reference Dillon, Holmes, Birk, Brooks, Lyons-Ruth and Pizzagalli37 showed significantly decreased left basal ganglia activity during reward anticipation, which was associated with anhedonia symptoms in adulthood. Together, these studies suggest that dopaminergic dysfunction may be a key RAD symptom, which may disrupt the utilisation and modulation of attentional processes typically employed for endogenous attentional control. Reference Pessoa and Engelmann42
As hypothesised, the present study identified deficiencies in reward-related ventral striatal activity in RAD samples. Conversely, functional alterations associated with reward processing were not observed in other brain regions linked with the reward-related dopaminergic system, such as the prefrontal cortex, the cingulate cortex and the limbic system, which have been implicated as important for reward processing. Reference Dichter, Damiano and Allen14,Reference Izuma, Saito and Sadato16 Although there were no significant between-group differences in reward-related activity, only the TD (not RAD) group exhibited significant reward-related activity (i.e. HMR or LMR minus NMR) in the prefrontal and anterior cingulate cortices ( ). Therefore, the RAD group may not only exhibit ventral striatal dysfunction but also a trend towards overall dopaminergic system dysfunction.
We observed negative associations between function in the ventral striatum and avoidant attachment tendencies. A growing body of evidence suggests an association between reward circuits and reinforcing processes in social approach and bonding. Reference Aron, Fisher, Mashek, Strong, Li and Brown43 In securely attached individuals, social interactions seem to be generally associated with more positive emotion experiences and stronger signals of reward. Conversely, these positive responses are much weaker or even absent in individuals with an avoidant attachment style. Reference Vrticka, Andersson, Grandjean, Sander and Vuilleumier44 Thus, reduced striatal responses to rewards in RAD could be explained by a lack of sensitivity towards social reward (e.g. bonding to a caregiver), leading to deficits in forming secure attachments with significant others.
We also investigated the existence of potential ‘sensitive periods,’ during which exposure to childhood maltreatment might be more strongly associated with alterations in striatal responses to HMR. We observed that a significant temporal predictor of bilateral striatal activation to HMR was whether children and adolescents with RAD were exposed to maltreatment at 0–1 year of age. Identifying sensitive periods when adversity has a particularly harmful impact on striatal activity is pivotal for linking childhood experiences to later psychopathology. Interestingly, we have previously discussed possible evidence for sensitive periods during which childhood maltreatment influences regional brain development, indicating a potential causal mechanism. Reference Andersen, Tomoda, Vincow, Valente, Polcari and Teicher45,Reference Pechtel, Lyons-Ruth, Anderson and Teicher46 Timing of exposure may be extremely important, and overall exposure levels across childhood may be misleading whether the maltreatment occurred before or after a sensitive period. Thus, the present findings support the assertion that brain regions may have unique windows of vulnerability towards traumatic stress. In fact, the present results (vulnerability from 0 to 1 year of age) also indicate that very early maternal care and post-birth child care system interventions are extremely important for preventing RAD and minimising long-term consequences of abuse.
The present study has a few notable limitations. First, the main limitation was the small sample size in both groups. Although the current sample size was similar to those in previous brain imaging studies, Reference Mehta, Gore-Langton, Golembo, Colvert, Williams and Sonuga-Barke36,Reference Bruce, Fisher, Graham, Moore, Peake and Mannering47,Reference Carrion, Garrett, Menon, Weems and Reiss48 the small sample may preclude relevant analyses on reward activity related to RAD-related changes in brain functioning associated with the experimental task. Hence, further, large-scale longitudinal studies are needed to address these issues for better elucidating whether particular genetic factors or social support could be a protective factor against adversity-induced reward system dysfunction. Reference Charney49,Reference Kundakovic, Gudsnuk, Herbstman, Tang, Perera and Champagne50 Second, there were substantial IQ differences between the groups, with the RAD group showing a lower FSIQ compared with the TD group, which has been reported elsewhere. Reference Kocovska, Puckering, Follan, Smillie, Gorski and Barnes8,Reference Smyke, Zeanah, Gleason, Drury, Fox and Nelson51 Although we applied statistical methods that controlled for IQ effects, the present results may not be clearly generalisable to an IQ-matched RAD group. Third, there is a potential limitation regarding RAD diagnostics. As previously suggested, Reference Minnis, Macmillan, Pritchett, Young, Wallace and Butcher1,Reference Kay and Green7,Reference Zeanah and Gleason52 a child is typically given a ‘suspected’ RAD diagnosis because such diagnostics are not fully clear. A more robust diagnosis needs to be based on ecological observations of the child's interactions, as well as clinical interviews and questionnaire data. Reference Minnis, Macmillan, Pritchett, Young, Wallace and Butcher1 Furthermore, it has been suggested that the RAD phenotype in the DSM-5 may be further encompassed by a broader phenotype. Reference Kay and Green7 Last, since we did not control for the amount or type of maltreatment, results described herein might identify neurobiological differences that are not only specifically associated with RAD, but might also apply to other forms of maltreatment. However, despite these limitations, the present study sheds light on the neural underpinnings of dysfunctional reward processing as a plausible, core RAD symptom.
The current study provides novel evidence that children and adolescents with RAD have decreased striatal activity during reward processing. The pattern of neural activity was similar to the profile we have observed in maltreated individuals tested with reward task. Such neurobiological decrements in RAD may lead to an increased risk for later dependency. Given that child maltreatment is a key factor for RAD, these results provide important insights towards better understanding the aetiology of this prevalent and debilitating disorder.
This work was supported by a Grant-in-Aid for Scientific Research (B), Young Scientists (B) and Challenging Exploratory Research (Houga) from the Ministry of Education, Culture, Sports, Science and Technology (MEXT) of Japan (KAKENHI: grant number 24300149 and 25560386 to A.T., grant number 25750405 to T.F.). This work was also partially supported by a Grant-in-Aid for Scientific Research from the Japan–US Brain Research Cooperation Program (grant number 230201 to A.T.).
We express our sincere appreciation to the participants and their families who participated in this research. We would also like to thank Dr Kei Mizuno for his helpful discussions and excellent technical advice in MRI data analyses, and Ms Reiko Fujisawa, Mr Hiroaki Naruse, Ms Kumi Yasuda and Ms Minaho Nishizato for their assistance in assessing participants and data analyses.