Distinct associations between fronto-striatal glutamate concentrations and callous-unemotional traits and proactive aggression in disruptive behavior

Abstract

Disruptive behavior is associated with societally and personally problematic levels of aggression and has been linked to abnormal structure and function of fronto-amygdala-striatal regions. Abnormal glutamatergic signalling within this network may play a role in aggression. However, disruptive behavior does not represent a homogeneous construct, but can be fractionated across several dimensions. Of particular interest, callous-unemotional (CU) traits have been shown to modulate the severity, neural and behavioural characterisation, and therapeutic outcomes of disruptive behaviour disorders (DBDs) and aggression. Further, individuals showing disruptive behavior differ to the extent that they engage in subtypes of aggression (i.e., proactive [PA] and reactive aggression [RA]) which may also represent distinct therapeutic targets. Here we investigated how glutamate signalling within the fronto-amygdala-striatal circuitry was altered along these dimensions in youths showing disruptive behavior (n = 140) and typically developing controls (TD, n = 93) within the age-range of 8–18 years. We used proton magnetic resonance spectroscopy (¹H-MRS) in the anterior cingulate cortex (ACC), striatum, amygdala and insula and associated glutamate concentrations with continuous measures of aggression and CU-traits using linear mixed-effects models. We found evidence of a dissociation for the different measures and glutamate concentrations. CU traits were associated with increased ACC glutamate (‘callousness’: b = .19, t (108) = 2.63, p = .01, r = .25; ‘uncaring’: b = .18, t (108) = 2.59, p = .011, r = .24) while PA was associated with decreased striatal glutamate concentration (b = −.23, t (28) = -3.02, p = .005, r = .50). These findings suggest dissociable correlates of CU traits and PA in DBDs, and indicate that the ACC and striatal glutamate may represent novel pharmacological targets in treating these different aspects.

Introduction

Disruptive behavior and aggression are among the most common problems in children and adolescents (American Psychiatric Asso, 2013). Disruptive behavior disorders (DBDs) include individuals diagnosed with oppositional defiant disorder (ODD) and conduct disorder (CD) and are often characterised by aggressive and/or antisocial behavior. DBDs exert a very significant burden to the affected individual and society and have a 5 to 10-fold increased risk of subsequent substance abuse, criminality, unemployment and early death (Odgers et al., 2007, Piquero et al., 2011). It has become increasingly recognised that DBDs include a heterogeneous population with different behavioral phenotypes. To date, phenotypic measures that have been found to most consistently predict worse outcome include: i) high or low on prosocial emotions - more commonly referred to as callous unemotional (CU) traits; and ii) proactive (PA) or reactive aggression (RA), where PA includes ‘cool’, purposeful aggression executed in a planned manner to achieve a specific goal, and RA involves ‘hot’, emotionally charged aggression, triggered with limited planning in response to real or perceived provocation. Other measures that predict poorer outcome include co-occurring mental health problems, particularly the presence of Attention Deficit Hyperactivity Disorder (ADHD). For example, response to established treatments is poorer (Hawes and Dadds, 2005, Waschbusch et al., 2007) and severity of antisocial behaviour and long-term prognosis is worse in youths with high CU traits (Frick and Marsee, 2006, Lynam et al., 2007), PA (Fite, Rubens, Preddy, Raine, & Pardini, 2014) and ADHD (August et al., 1983, Farrington et al., 1990, Frick and Ellis, 1999, Schachar et al., 1981). These behavioral phenotypes are also associated with neurobiological correlates and specific cognitive impairments involved with processing emotions and learning/decision making. For example, compared to children with low CU traits, those with high CU traits have reduced amygdala activation to fearful facial expressions (Carré et al., 2013, Jones et al., 2013, Marsh et al., 2008, Marsh et al., 2013, Viding et al., 2012, White et al., 2012), reduced anterior cingulate cortex (ACC), medial prefrontal cortex (mPFC), but increased insula response to facial expression of individuals in pain (Adolphs et al., 2005, Lockwood, 2016, Marsh et al., 2013). Increased striatal and ventromedial PFC (vmPFC) activity in response to punishment has also been reported (Finger et al., 2011). Overall, evidence from imaging studies indicates that PA and high CU traits have distinct neural correlates involving dysfunctional frontal-amygdala-striatal circuitry (Blair, 2013). Literature focussing on the distinct neural correlates of RA and PA is limited. However, studies from our own group suggest aggression subtype specific differences in brain structure and resting state functional connectivity patterns within the fronto-striatal and default mode network circuitry (Naaijen et al.; Werhahn et al., under review).

Glutamate is the major excitatory neurotransmitter in the frontal-amygdala-striatal network (Gleich et al., 2015, Naaijen et al., 2015, Pittenger et al., 2011) and increasing evidence suggests a role for glutamate (and the inhibitory neurotransmitter GABA) in aggression. Glutamate can be quantified in vivo with proton magnetic resonance spectroscopy (¹H-MRS). Previous ¹H-MRS studies have mainly focused on the ACC, a region highly relevant for aggression and disruptive behaviour due to its role in cognitive control. These studies reported increased Glx (composite signal of glutamate and glutamine) in the ACC of children with emotional dysregulation and aggressive traits (Wozniak et al., 2012), and a positive association between impulsivity and glutamate of the ACC in females with ADHD and Borderline Personality Disorder (Ende et al., 2016). In addition, differences in glutamate concentration in this network have been reported in other neurodevelopmental disorders associated with aggression (e.g., ADHD and autism spectrum disorders), although the direction of findings has been inconsistent (Naaijen et al., 2015) due to methodological issues and sample selection. Further, findings from animal models suggest that CU-like traits and aggression may arise in part due to an excitatory/inhibitory imbalance within the ACC (Jager et al., 2017), associated with altered glutamate and/or GABA levels. However, we are unaware of any studies that have analysed brain glutamate concentrations in individuals with DBDs. Therefore, we used ¹H-MRS to study glutamate concentrations in the ACC/mPFC, amygdala, insula and striatum, in a multi-center study of children and adolescents with and without DBDs and/or aggressive behavior, while controlling for the effect of ADHD, and analyzed the relationships with CU traits and RA/PA. We hypothesised that glutamate concentrations would differ between children with and without DBDs in the regions analysed. Further, whilst the direction of change was difficult to predict a priori, we hypothesised that glutamate concentrations within this network would differentiate continuous measures of aggression from each other.