FABBS Doctoral Dissertation Research Excellence Award

Katherine Lopez, Washington University in St. Louis, Department of Psychological and Brain Sciences

Cognitive, personality, and neural signatures of impulsivity in childhood suicide: An Adolescent, Brain, and Cognitive Development (ABCD) Study

Katherine Lopez, Washington University in St. Louis, Department of Psychological and Brain Sciences

Abstract

Research reveals a concerning rise in childhood suicide, with recent efforts leveraging on the adult literature to pinpoint candidate risk factors central to childhood suicide. A noteworthy area has examined the role of impulsivity in elevating suicide risk— a link warranting further study in childhood given known changes in self-control and brain maturity across development. Using the Adolescent Brain and Cognitive Development dataset, we examined a large, representative sample of children endorsing lifetime suicide ideation (N=849), suicide attempt (186), and no suicide history (10,754), delineating group differences in 1) cognitive impulsivity 2) personality impulsivity and 3) several brain organization properties relevant to self-control (i.e., whole-brain topology, network topology, and hubs). Through multilevel modeling, our findings revealed that suicide ideation was marked by broad elevations in trait impulsivity while attempt was better characterized by cognitive rigidity (girls only), elevations in urgency measures, and hub disruptions in the dorsolateral prefrontal cortex of the frontoparietal network. Hubs, however, did not emerge as a mechanism by which impulsivity distinguished ideators from attempters, suggesting that an interacting profile of biopsychosocial variables likely explain the ideation-to-attempt shift. Together, these findings corroborate the roles of impulsivity and frontoparietal regions in childhood ideation and attempt, extending adult-like profiles of suicide into earlier developmental periods.

Introduction/Background

Suicide has increased worldwide, with the United States having seen a 25% rise in deaths by suicide in the past decade (Curtin et al., 2016). Recent figures emphasize a disproportionate upturn in ideation and attempts among youth, with suicide representing the second leading cause of death among children and adolescents (Nock et al., 2013; Patton et al., 2009). Critically, these rates are projected to continue rising, underscoring the need to better understand how emotional, cognitive, and neurobiological risk factors of early development place youth at heightened risk for suicidal thoughts and attempts.

The role of impulsivity in suicide has drawn mixed attention from researchers interested in understanding whether disruptions in self-control heighten risk of ideation and/or attempt. Impulsivity— defined as a tendency for rash, poorly conceived behavior— has been proposed to play a precipitating role in the ideation-to-attempt shift (e.g., Baumeister, 1990), with a key implication that individuals with a history of suicide attempt might display greater evidence of impulsivity than those endorsing suicidal thoughts (i.e., no attempt history).

Evidence for this idea has been inconclusive, complicated by the various ways impulsivity has been operationalized across psychological disciplines (Bagge et al., 2013). There is now growing consensus that impulsivity can be best understood as a multifaceted construct, composed of cognitive (e.g., executive abilities), personality (e.g., trait impulsivity), and emotional processes (e.g., externalizing). Thus, if the broader construct of impulsivity were to play a role in the phenomenology of suicide, one might expect diverging relationships between the various components of impulsivity and suicidality. While early evidence favors this multifactorial view, there remains ongoing uncertainty on which aspects of impulsivity are most consistently related to the presentation of suicide. Given that impulsivity is among several factors considered in determining suicide risk, in conjunction with an increasing number of risk assessments being conducted in younger age groups (Bryan & Rudd, 2006), there remains a clear need to improve our understanding of distinct pathways that impulsivity confer risk to suicide, particularly in early developmental when processes of self- control, psychopathology, and related neural properties are known to change.

Notably, this research area is relevant to extant neuroimaging work, which links suicide to multimodal disruptions in key brain regions supporting self-control and executive abilities. These regions include the dorsolateral prefrontal cortex (DLPFC) of the frontoparietal network (FPN), the dorsal anterior cingulate cortex (dACC) and insula of the cingulo- opercular (CON), and the ventrolateral prefrontal cortex (VLPFC) of the ventral attention network (VAN; Corbetta, 1998; Dosenbach et al., 2007; Vossel, Geng, & Fink, 2014). These networks bear critical associations to impulsivity and demonstrate prototypical patterns of maturity that hold promise for refining our understanding of altered developmental trajectories in suicide. (J. R. Cohen & D’Esposito, 2016; Lopez, Kandala, Marek, & Barch, 2020; Marek et al, 2015).

In response, we leveraged on advancements in the suicide and developmental neuroscience literatures to examine the cognitive, personality, and neural substrates of impulsivity in a large, representative sample of children endorsing suicidal thoughts, past attempts, and no suicide history.

Methods

In particular, we sought to delineate the neurocognitive signatures of childhood suicide by first and foremost (Aim 1) replicating profiles of executive and personality measures of impulsivity in much younger samples of children endorsing suicide ideation, attempt, and no suicide history. Second (Aim 2), we aimed to characterized key organizational properties of control networks across suicide cohorts to establish whether disruptions in control networks are altered in children endorsing suicide. Our final goal (Aim 3) was to probe brain-behavior relationships as a potential mechanism of suicide by examining whether network properties mediated the relationship between impulsivity and suicide status (i.e., ideation, attempt).

Participants were 11,782 children aged 9-10 years old from the Adolescent Brain and Cognitive Development (ABCD) study, a large-scale, multisite, longitudinal study examining cognitive, emotional, and neural development in adolescents across 21 sites in the United States (Barch et al., 2018). Data from the first wave of the ABCD dataset (Release 2.01) were examined and included twin pairs and siblings across sites. Participants included in the present analyses completed: (1) the Kiddie Schedule for Affective Disorders and Schizophrenia for DSM-5 (KSADS-5) to assess diagnostic status; (2) ABCD Youth NIH Toolbox to assess cognitive impulsivity; (3) UPPS-P- Impulsive Behavior Scale, Youth

Version (UPPS-P-Y) to examine personality impulsivity; and 4) at least one resting-state fMRI scan to examine brain organization of control networks. Specifically, 849 children endorsed suicidal thoughts, 186 children reported a previous history of attempt, while the majority of children (N= 10,754) noted no suicide history. Impulsivity, in particular, was measured cognitively, via executive abilities (i.e., cognitive flexibility (set shifting), updating (working memory), and cognitive inhibition), and from a personality perspective, using five dissociable factors of dispositional trait impulsivity (i.e., negative urgency, positive urgency, lack of perseverance, lack of planning, sensation seeking). Brain organization metrics included segregation and integration measures of four control networks along with hub regions of each of these networks. Networks include the frontoparietal FPN, involved in initiating and flexibly adjusting control, the cingulo-opercular network (CON), involved in the maintenance of tasks, and the dorsal attention and ventral attention networks (DAN, VAN), implicated in top-down and bottom-up attention processing, respectively (Corbetta, 1998; Dosenbach et al., 2007; Ptak, Schnider, & Fellrath, 2017; Vossel, Geng, & Fink, 2014).

Finally, to account for the nested structure of the ABCD dataset (e.g., children nested within families and 21 different acquisition sites), we employed hierarchical linear models for Aims 1 and 2. We used structural equation modeling for our mediation analysis (Aim 3). Of note, suicide status was defined as a three-level categorical variable, yielding two primary contrast variables: Control v. Ideation and Ideation v. Attempt. Each model controlled for the effects of age (in months), sex, IQ, SES, depressive symptoms, and hopelessness symptoms. Additionally, a subsequent ‘suicide status by gender’ interaction was performed to examine sex differences in cognitive- and personality- based impulsivity and False Discovery Rates (p<0.05) was applied within each behavioral domain to correct for multiple comparisons.

Results

Aim1: Suicide status in relation to executive functioning- While the Control v. Ideation group did not predict performance on any measure of executive functioning, the Ideation v. Attempt group contrast significantly predicted set-shifting performance, such that children with a history of suicide attempt displayed poorer set-shifting performance than children endorsing ideation (b= -3.75, T= -3.02, p=0.002, R2(fixed)= 0.11; Figure 2). Gender interactions indicated that only females displayed set shifting differences between ideators and attempters (b= 4.44, T= 2.69, p=0.007, R2(fixed)= 0.11). Suicide status in relation to trait impulsivity – Models predicting trait impulsivity revealed a two-fold profile. First, the Control v. Ideation group significantly predicted negative urgency (b= -1.17, T= -10.48, p=0.000, R2(fixed)= 0.05), positive urgency (b= -0.93, T= -7.44, p=0.000, R2(fixed)= 0.05), lack of perseverance (b= -0.67, T= -6.98, p=0.000, R2(fixed)= 0.04), and lack of planning (b= -0.86, T= -8.52, p=0.000, R2(fixed)= 0.11). For each of these effects, ideators displayed significantly elevated trait impulsivity scores than controls. Second, the Ideation v. Attempt group predicted negative urgency (b= 0.85, T= 3.51, p=0.000, R2(fixed)= 0.05) and positive urgency (b= 0.79, T= 2.91, p=0.003, R2(fixed)= 0.05), such that attempters displayed significantly elevated urgency scores than ideators. Notably, no sex differences were observed for any trait impulsivity measure. In short, findings indicate that ideators displayed broad elevations in four trait impulsivity measures while attempters showed further elevations in negative, positive urgency, and set shifting impairment in girls only.

Aim 2: Suicide status relationships to segregation, integration, and hubness- Both the Control v. Ideation and Ideation v. Attempt groups failed to predict differences in whole brain topology (i.e., modularity and global efficiency), network-level topology of control networks, and hubs. However, the Ideation v. Attempt group did significantly predict hubness of the DLPFC of FPN (b= 0.03, T= 2.37, p=0.02, R2(fixed)= 0.09), such that child attempters displayed higher levels of hubness in the DLPFC of FPN.

Aim 3: Mediation suicide status effects on executive functioning and trait impulsivity through brain organization—Having found group differences between ideators and attempters in set-shifting, negative urgency, positive urgency, and DLPFC hubness, we sought to determine whether hubness disruptions might be a plausible mechanism by which impulsivity might elevates risk for transitioning from ideation to attempt. Findings, however, did not show that DLPFC hubness mediated the relationship between any measure of impulsivity (i.e., set-shifting, negative urgency, positive urgency) and suicide status.

Discussion

Our findings suggest that suicide ideation is marked by global elevations in trait impulsivity dimensions in the context of intact cognition, whereas suicide attempt was better described by further elevations in urgency, along with cognitive inflexibility in girls only. Behavioral profiles observed in child attempters were accompanied by hubness alterations in the DLPFC. Together, these findings resemble neurobehavioral profiles found in the adolescent and adult, overall extending adolescent- and adult-like profiles of suicide into much younger periods of development.

Towards suicide ideation- The global nature of trait impulsivity elevations in ideators is most for noteworthy for its associations to key latent factors of suicide ideation— low affectivity and externalizing tendencies. Negative urgency, for example, has been linked to aspects of low positive affectivity, including perceived burdensomeness, especially in the presence of low distress tolerance (Anestis & Joiner, 2011) . Because coping continues to develop well into adolescence (Power, T. G., 2004), children with high urgency may show particular vulnerability to various forms of low positive affectivity. By contrast, elevations in lack of perseverance and planning may be related to the type of externalizing tendencies linked with suicidal thoughts. For example, reduced perseverance and planning may help promote shifts from passive thoughts about death to more active thoughts of ideation (Miller et al., 2019). In this way, elevations in a number of trait impulsivity dimensions may bring into bearing several key variables that underlie suicidal thoughts.

From ideation to attempt- Children with a history of past attempts displayed even further elevations in negative and positive urgency, with girls also showing impairments in set-shifting Cognitive inflexibility, in particular, is theorized to disrupt the ease with which individuals effectively switch from one coping mechanism to another (Wenzel & Beck, 2008). It is, therefore, possible that once children have formulated thoughts of suicide, perseveration on these suicide-related thoughts may hinder one’s ability to generate alternative coping strategies that may be more adaptive. This cognitive constriction, coupled with greater emotional bias to painful experiences via elevated urgency, may collectively represent one pathway that by which children, over time, begin to increase their ‘acquired capability for suicide’(Joiner & Silva, 2012). Interestingly, only girl attempters displayed cognitive inflexibility, similar to female-dominant ruminative tendencies seen during the onset of menses. Because rumination has been found to mediate the relationship between cognitive inflexibility and future suicide attempt, one speculation is that girls’ cognitive vulnerability to suicide may be hormonally meditated (Owens et al., 2020). Notably, child attempter also displayed altered DLPFC hubness of the FPN, one of the most frequently reported neural marker associated with suicide and known for its role in the implementation of cognitive control. Despite this however, we did not find that DLPFC disruptions mediated impulsivity and suicide status. Thus, although DLPFC emerges as a robust site of multimodal dysfunction in suicide, these hub disruptions do not appear to exacerbate existing suicidal thoughts by undermining cognitive flexibility, leaving mechanistic questions about the neural substrates of impulsivity in suicide open for further research.

Impact Statement

Because our findings are based on a large and representative sample of children across the US, replicating distinct profiles of impulsivity reported in adolescents and adults, the generalizability of these findings hold particular relevance for how impulsivity is being screened and used to assess suicide risk across a variety of settings. Our findings are in agreement with the growing consensus that impulsivity is not a unitary construct, and that diverse measures of impulsivity are likely needed to properly assess suicide risk. In particular, cognitive- and personality-based impulsivity relate differently to ideation and attempt, with further differences observed across gender. Thus, the procedures that we use to screen and establish risk may require a more standardized process that takes into consideration key demographic variables, including age, gender, and based on other work, ethnicity and socioeconomic status.