Self-inflicted injury (SII), including suicide attempts and nonsuicidal self-harm (see Crowell et al., Reference Crowell, Beauchaine, McCauley, Smith, Vasilev and Stevens2008; Nock, Reference Nock2010), portends a wide range of adverse outcomes among affected adolescents, including social problems, academic underachievement, and poor psychological adjustment (for reviews, see Crowell, Derbidge, & Beauchaine, Reference Crowell, Derbidge, Beauchaine and Nock2014; Crowell, Kaufman, & Lenzenweger, Reference Crowell, Kaufman, Lenzenweger, Beauchaine and Hinshaw2013; Derbidge & Beauchaine, Reference Derbidge, Beauchaine, Lewis and Rudolph2014; Nock, Reference Nock2010). Most adolescents who engage in SII meet criteria for one or more psychiatric disorders, with risk dispersed roughly evenly across the internalizing and externalizing spectra (Crowell et al., Reference Crowell, Beauchaine, McCauley, Smith, Stevens and Sylvers2005; Nock, Joiner, Gordon, Lloyd-Richardson, & Prinstein Reference Nock, Joiner, Gordon, Lloyd-Richardson and Prinstein2006). Perhaps of more importance, SII is the single best predictor of eventual completed suicide (Hamza, Stewart, & Willoughby, Reference Hamza, Stewart and Willoughby2012).
Although some correlates of suicidal behavioral and nonsuicidal self-harm appear to be distinct, such as levels of hopelessness, perceived burden others, and reported goals of the behavior, most if not all self-injurers experience negative/irritable mood states, a desire to reduce/avoid emotional distress, and high levels of impulsivity (see Brown, Comtois, & Linehan, Reference Brown, Comtois and Linehan2002; Glenn & Klonsky, Reference Glenn and Klonsky2010; Horesh et al., Reference Horesh, Rolnick, Iancu, Dannon, Lepkifker and Apter1997; Joiner et al., Reference Joiner, Pettit, Walker, Voelz, Cruz and Rudd2002; Minkoff, Bergman, Beck, & Beck, Reference Minkoff, Bergman, Beck and Beck1973; Rudd, Joiner, & Rajad, Reference Rudd, Joiner and Rajad1996).
Obtaining accurate prevalence estimates of SII and related behaviors is difficult for a number of reasons, yet rates of self-injury appear to have increased over the past two decades, especially among adolescents (Centers for Disease Control [CDC], 2008; Nock, Reference Nock2008). Even in community samples, up to 45% of adolescents report having engaged in some form of self-harm (Lloyd-Richardson, Perrine, Dierker, & Kelley, Reference Lloyd-Richardson, Perrine, Dierker and Kelley2007). In clinical samples, prevalence rates of SII are higher still (Darche, Reference Darche1990). Furthermore, roughly 400,000 individuals in the United States receive medical attention for self-injury each year (CDC, 2006). Given the level of functional impairment associated with SII, and the risk it confers for psychiatric morbidity and mortality (Klonsky, May, & Glenn, Reference Klonsky, May and Glenn2013), it is recognized by both the National Center for Injury Prevention and Control (CDC, 2009), and the US Public Health Service (1999) as an urgent public health problem.
For at least some adolescents, SII is also a developmental precursor to borderline personality disorder (BPD; see Beauchaine, Klein, Crowell, Derbidge, & Gatzke-Kopp, Reference Beauchaine, Klein, Crowell, Derbidge and Gatzke-Kopp2009; Crowell, Beauchaine, & Linehan, Reference Crowell, Beauchaine and Linehan2009; Crowell et al., Reference Crowell, Kaufman, Lenzenweger, Beauchaine and Hinshaw2013). In our own research on SII among 14- to 18-year-old adolescent girls (Crowell et al., Reference Crowell, Beauchaine, McCauley, Smith, Stevens and Sylvers2005, Reference Crowell, Beauchaine, McCauley, Smith, Vasilev and Stevens2008, Reference Crowell, Beauchaine, Hsiao, Vasilev, Yaptangco and Linehan2012), a plurality already meet criteria for BPD, even though they are recruited based solely on endorsement of three or more episodes of self-injury in a 6-month interval, and not any other symptoms of the disorder. Even among adolescents who engage less frequently in self-injury, about 50% meet criteria for BPD (Nock et al., Reference Nock, Joiner, Gordon, Lloyd-Richardson and Prinstein2006). Moreover, roughly two-thirds of adults with BPD report initiating self-injury before age 18 (American Psychiatric Association, 2004). Thus, despite a paucity of longitudinal research on the development of BPD, SII appears to be an important etiological factor, at least for some individuals. Nevertheless, much remains to be learned about the development of SII and BPD. Of importance, furthered understanding of etiology may be a prerequisite of earlier identification, prevention, and more effective treatment for vulnerable individuals (see, e.g., Beauchaine, Neuhaus, Brenner, & Gatzke-Kopp, Reference Beauchaine, Neuhaus, Brenner and Gatzke-Kopp2008).
For most of the past two decades, etiological models of BPD have focused on emotion dysregulation as the principle, if not defining, feature of the disorder. According to Linehan's (Reference Linehan1993) theory, SII emerges as a means of coping with persistent emotional distress (see also Crowell et al., Reference Crowell, Beauchaine and Linehan2009; Kuo & Linehan, Reference Kuo and Linehan2009). Self-injury is therefore reinforced and maintained because it is effective, at least sometimes, in dampening emotional discomfort (see Johnson, Hurley, Benkelfat, Herpertz, & Taber, Reference Johnson, Hurley, Benkelfat, Herpertz and Taber2005; Nock, Reference Nock2009). Over time, the emotion regulatory function of self-harm becomes canalized, perhaps in part through endogenous opioid release (see Bresin & Gordon, Reference Bresin and Gordon2013).
Literature addressing the emotion dysregulation hypothesis of SII and BPD is voluminous and cannot be reviewed here. However, support for the hypothesis is extensive (see, e.g., Klonsky & Muehlenkamp, Reference Klonsky and Muehlenkamp2007), based on self-report (e.g., Nock, Reference Nock2008, Reference Nock2009, Reference Nock2010), peripheral psychophysiological (e.g., Crowell, Baucom, et al., Reference Crowell, Baucom, Yaptangco, Bride, Hsiao and McCauley2014; Crowell et al., Reference Crowell, Beauchaine, McCauley, Smith, Stevens and Sylvers2005, Reference Crowell, Beauchaine, McCauley, Smith, Vasilev and Stevens2008; Kuo & Linehan, Reference Kuo and Linehan2009), and neuroimaging data (Hazlett et al., Reference Hazlett, Zhang, New, Zelmanova, Goldstein and Haznedar2012; Lis, Greenfield, Henry, Guile, & Dougherty, Reference Lis, Greenfield, Henry, Guile and Dougherty2007). Nevertheless, we have argued that emotion dysregulation, in and of itself, is insufficient to account for the development of SII and BPD. Many if not most forms of psychopathology are characterized by emotion dysregulation (see, e.g., Beauchaine, Reference Beauchaine2001; Beauchaine, Reference Beauchaine2015a, Reference Beauchaine2015b; Beauchaine, Gatzke-Kopp, & Mead, Reference Beauchaine, Gatzke-Kopp and Mead2007; Beauchaine & Thayer, in press; Cole, Hall, & Hajal, Reference Cole, Hall, Hajal, Beauchaine and Hinshaw2013; Goldsmith, Pollak, & Davidson, Reference Goldsmith, Pollak and Davidson2008), but most are not characterized by self-injury.
Based on this observation, and on other considerations outlined below, we have advanced a developmental model of SII and borderline personality development that extends Linehan's (Reference Linehan1993) theory in several important ways (see Beauchaine et al., Reference Beauchaine, Klein, Crowell, Derbidge and Gatzke-Kopp2009; Crowell et al., Reference Crowell, Beauchaine and Linehan2009; Crowell, Derbidge, et al., Reference Crowell, Baucom, Yaptangco, Bride, Hsiao and McCauley2014; Derbidge & Beauchaine, Reference Derbidge, Beauchaine, Lewis and Rudolph2014). Although space constraints preclude full articulation of our developmental model, it follows in large part from mounting empirical evidence for each of the following conjectures:
-
1. In childhood, vulnerability to development of SII and borderline pathology is conferred through trait impulsivity, a highly heritable individual difference (see, e.g., Krueger et al., Reference Krueger, Hicks, Patrick, Carlson, Iacono and McGue2002) that accounts for (a) overlap in biological correlates of BPD and other impulse control disorders (see Beauchaine et al., Reference Beauchaine, Klein, Crowell, Derbidge and Gatzke-Kopp2009), and (b) shared familial risk for BPD and antisocial personality among adult females and males, respectively (e.g., Goldman, D'Angelo, & DeMaso, Reference Goldman, D'Angelo and DeMaso1993). The assertion that impulsivity confers vulnerability to later SII is consistent with recent findings indicating heightened risk for self-injury and suicide attempts among adolescent girls who were diagnosed with attention-deficit/hyperactivity disorder (ADHD) in middle school (Hinshaw et al., Reference Hinshaw, Owens, Zalecki, Huggins, Montenegro-Nevado and Schrodek2012), with well replicated associations between trait impulsivity and both self-injury and BPD in adulthood (see Beauchaine et al., Reference Beauchaine, Klein, Crowell, Derbidge and Gatzke-Kopp2009; Glenn & Klonsky, Reference Glenn and Klonsky2010), and with positive associations between impulsivity scores and greater likelihood and severity of self-injurious behaviors (Dougherty et al., Reference Dougherty, Mathias, Marsh-Richard, Prevette, Dawes and Hatzis2009; Lynam, Miller, Miller, Bornovlova, & Lejuez, Reference Lynam, Miller, Miller, Bornovalova and Lejuez2011). According to this perspective, trait impulsivity is necessary but insufficient for development of SII and BPD.
-
2. Trait impulsivity is conferred through heritable deficiencies is mesolimbic dopamine (DA) responding (see Beauchaine et al., Reference Beauchaine, Klein, Crowell, Derbidge and Gatzke-Kopp2009). Among other functions, the mesolimbic DA system is responsible for hedonic capacity (e.g., Ashby, Isen, & Turken, Reference Ashby, Isen and Turken1999; Gorwood, Reference Gorwood2008), so blunted responding within the system results in anhedonia, or lower than normal levels of pleasure from pursuit of and consummation of incentives (see Neuhaus & Beauchaine, Reference Neuhaus, Beauchaine, Beauchaine and Hinshaw2013; Sagvolden, Johansen, Aase, & Russell, Reference Sagvolden, Johansen, Aase and Russell2005). Blunted reward responding also results in a chronically aversive, irritable mood, which motivates affected individuals to seek frequent and immediate rewards in order to increase mesolimbic DA levels and upregulate their mood state (see Laakso et al., Reference Laakso, Wallius, Kajander, Bergman, Eskola and Solin2003). It is important to note that mesolimbic DA dysfunction is not specific to impulsivity; attenuated reward responding is also characteristic of depression in adolescence and adulthood, and is a likely neural substrate of anhedonia and irritability among those with mood disorders (see, e.g., Forbes, & Dahl, Reference Forbes and Dahl2005, Reference Forbes and Dahl2011). Thus, reduced pleasure in response to incentives (i.e., anhedonia) and high levels of irritability characterize both depression and impulse control disorders. Attenuated tonic and phasic mesolimbic DA function are therefore transdiagnostic substrates of both impulsivity and depression (see Sauder, Beauchaine, Gatzke-Kopp, Shannon, & Aylward, Reference Sauder, Beauchaine, Gatzke-Kopp, Shannon and Aylward2012).Footnote 1 It follows that self-injuring adolescents, who are highly impulsive and often depressed, should exhibit (a) heterotypic comorbidity, which is the case empirically (see, e.g., Crowell, Baucom, et al., Reference Crowell, Baucom, Yaptangco, Bride, Hsiao and McCauley2014; Crowell et al., Reference Crowell, Beauchaine, McCauley, Smith, Stevens and Sylvers2005, Reference Crowell, Beauchaine, McCauley, Smith, Vasilev and Stevens2008), and (b) deficiencies in mesolimbic responding to incentives (see below), which have not been demonstrated but which we evaluate in this study.
-
3. Emotional lability and emotion dysregulation develop after preexisting impulsivity in the ontogenesis of SII and borderline pathology through operant reinforcement mechanisms expressed within families, as articulated originally by Linehan (Reference Linehan1993) and more recently by Beauchaine and Zalewski (in press). Invalidating and coercive family environments reinforce repeated escalation of anger, dyadic conflict, emotional lability, and emotion dysregulation (see Crowell, Baucom, et al., Reference Crowell, Baucom, Yaptangco, Bride, Hsiao and McCauley2014; Crowell, Derbidge, et al., Reference Crowell, Derbidge, Beauchaine and Nock2014). Although far less heritable than trait impulsivity (Goldsmith et al., Reference Goldsmith, Pollak and Davidson2008), emotional lability and emotion dysregulation assume traitlike qualities across development via protracted reinforcement (see also Beauchaine, Reference Beauchaine2015b; Beauchaine & Zalewski, in press; Linehan, Reference Linehan1993).
-
4. Enduring patterns of emotion dysregulation elicit failures of adaptive coping, interfere with healthy socioemotional development, and confer adverse evocative effects on extrafamilial interpersonal relationships. As a result of these mechanisms and others (see Derbidge & Beauchaine, Reference Derbidge, Beauchaine, Lewis and Rudolph2014), affected individuals engage in maladaptive social behaviors, affiliate with deviant peers, and are at high risk for developing SII and borderline personality traits (Crowell, Baucom, et al., Reference Crowell, Baucom, Yaptangco, Bride, Hsiao and McCauley2014; Crowell et al., Reference Crowell, Beauchaine and Linehan2009).
Our primary purpose in conducting this study was to test the second component of our developmental model (Item 2 above). Accordingly, we sought to (a) determine whether or not adolescent girls who engage in SII exhibit neural response patterns to reward that characterize impulsive and depressed samples, and (b) evaluate correlations between mesolimbic neural responses to incentives and symptoms of impulsivity and depression among adolescent girls with histories of SII.
Existing Neuroimaging Research on SII and Borderline Personality
To date, most empirical studies of SII and borderline personality development, whether behavioral or neurobiological, focus on the emotion dysregulation component of each. Although several neuroimaging studies of BPD have appeared in the literature, almost all have either evaluated structural compromises in brain regions involved in emotion regulation (e.g., Minzenberg, Fan, New, Tang, & Siever, Reference Minzenberg, Fan, New, Tang and Siever2008; Soloff et al., Reference Soloff, Pruitt, Sharma, Radwan, White and Diwadkar2012) or examined patterns of functional activation while participants engaged in emotion evocation or emotion discrimination tasks (e.g., Hazlett et al., Reference Hazlett, Zhang, New, Zelmanova, Goldstein and Haznedar2012; Koenigsberg et al., Reference Koenigsberg, Siever, Lee, Pizzarello, New and Goodman2009; Kraus et al., Reference Kraus, Valerius, Seifritz, Ruf, Bremner and Bohus2010). Thus, few studies have evaluated structural or functional associations with impulsivity or depression among self-injuring or borderline participants (for exceptions, see Völlm et al., Reference Völlm, Richardson, McKie, Elliott, Dolan and Deakin2007; Wolf et al., Reference Wolf, Thomann, Sambataro, Vasic, Schmid and Wolf2012). Furthermore, almost none have been conducted with adolescents who are at risk for BPD, as evidenced by a history of SII and/or endorsement of full criteria for the disorder. In one exception, Plener, Bubalo, Fladung, Ludolph, and Lulé (Reference Plener, Bubalo, Fladung, Ludolph and Lulé2012) reported stronger neural responses in the amygdala, hippocampus, and anterior cingulate cortex during emotion evocation among self-injuring adolescent girls compared with controls. Although consistent with findings obtained from adults with BPD, this study also focused on the emotion dysregulation component of SII, not on impulsivity or depression.
Neural Correlates of Trait Impulsivity
As described briefly above, contemporary theories of trait impulsivity all focus at least in part on the mesolimbic DA system, within which aberrant reward responding is observed in many, if not most, forms of externalizing psychopathology (see Gatzke-Kopp, Reference Gatzke-Kopp2011; Sagvolden et al., Reference Sagvolden, Johansen, Aase and Russell2005). Extensive neuroimaging research reveals reduced mesolimbic and/or mesocortical reactivity to monetary incentives among individuals with ADHD (see Bush, Valera, & Seidman, Reference Bush, Valera and Seidman2005; Carmona et al., Reference Carmona, Hoekzema, Ramos-Quiroga, Richarte, Canals and Bosch2011; Dickstein, Bannon, Castellanos, & Milham, Reference Dickstein, Bannon, Castellanos and Milham2006; Durston, Reference Durston2003), conduct disorder (CD; e.g., Rubia et al., Reference Rubia, Smith, Halari, Matukura, Mohammad and Taylor2009), substance use disorders (see, e.g., Martin-Soelch et al., Reference Martin-Soelch, Leenders, Chevalley, Missimer, Kunig and Magyar2001; Volkow, Fowler, & Wang, Reference Volkow, Fowler and Wang2004), and antisocial personality traits (e.g., Oberlin et al., Reference Oberlin, Dzemidzic, Bragulat, Lehigh, Talavage and O'Connor2012). Moreover, reduced mesolimbic DA transporter, D2 receptor, and/or D3 receptor binding are observed among adults with ADHD (Volkow et al., Reference Volkow, Wang, Kollins, Wigal, Newcorn and Telang2009) and alcoholism (e.g., Laine, Ahonen, Räsänen, & Tiihonen, Reference Laine, Ahonen, Räsänen and Tiihonen2001), and compromised functional connectivity between mesolimbic and mesocortical structures is observed among adolescents with ADHD and CD (e.g., Shannon, Sauder, Beauchaine, & Gatzke-Kopp, Reference Shannon, Sauder, Beauchaine and Gatzke-Kopp2009). This latter finding is of particular interest given the importance of top-down control by mesocortical structures over mesolimbic activity and reactivity in effective modulation of behavior and emotion (see Beauchaine, Reference Beauchaine2015b; Goldsmith et al., Reference Goldsmith, Pollak and Davidson2008). However, despite consistently high scores across all indices of externalizing behavior among girls who engage in SII (e.g., Crowell et al., Reference Crowell, Beauchaine, McCauley, Smith, Stevens and Sylvers2005, Reference Crowell, Beauchaine, McCauley, Smith, Vasilev and Stevens2008, Reference Crowell, Beauchaine, Hsiao, Vasilev, Yaptangco and Linehan2012), and common familial risk for borderline and externalizing symptoms among adult females and males, respectively (for a review, see Beauchaine et al., Reference Beauchaine, Klein, Crowell, Derbidge and Gatzke-Kopp2009), mesolimbic responding to incentives has yet to be evaluated among adolescent girls who engage in SII.
Neural Correlates of Depression and Heterotypic Comorbidity
As also noted above, adolescent girls who engage in SII are at exceedingly high risk for heterotypic comorbidity. Crowell et al. (Reference Crowell, Beauchaine, McCauley, Smith, Stevens and Sylvers2005) reported that girls who were recruited based solely on three or more self-harm episodes in the past 6 months, or five or more episodes in their lifetime, scored in the 83rd and 82nd percentiles on internalizing and externalizing behavior, respectively, vis-à-vis national norms. Two-thirds also met full diagnostic criteria for major depression.
In addition to its role in impulsivity and vulnerability to externalizing behaviors (see above), the mesolimbic DA system is clearly implicated in adolescent depression (see, e.g., Forbes et al., Reference Forbes, Christopher May, Siegle, Ladouceur, Ryan and Carter2006). This may help to account for high rates of heterotypic comorbidity of internalizing and externalizing disorders, in both self-injuring and other samples (see, e.g., Sauder at al., Reference Sauder, Beauchaine, Gatzke-Kopp, Shannon and Aylward2012). Similar to research outlined above conducted among externalizing samples, numerous studies demonstrate blunted mesolimbic responding to monetary incentives among depressed adolescents (see Forbes et al., Reference Forbes, Christopher May, Siegle, Ladouceur, Ryan and Carter2006; Forbes & Dahl, Reference Forbes and Dahl2011; Richards, Plate, & Ernst, Reference Richards, Plate and Ernst2013). Altered mesolimbic reward processing likely confers risk for heterotypic comorbidity through its effects on positive affectivity and hedonic capacity (see above; Forbes & Dahl, Reference Forbes and Dahl2005; Neuhaus & Beauchaine, Reference Neuhaus, Beauchaine, Beauchaine and Hinshaw2013). Blunted striatal responding to incentives is associated with lower self-reports of positive affect in naturalistic settings (Forbes et al., Reference Forbes, Hariri, Martin, Silk, Moyles and Fisher2009), and low levels of striatal DA (assessed with positron emission tomography) predict trait irritability (Laakso et al., Reference Laakso, Wallius, Kajander, Bergman, Eskola and Solin2003), which characterizes both internalizing and externalizing disorders. In contrast, infusions of DA into mesolimbic structures produce pleasurable affective states (see Berridge, Reference Berridge2003; Berridge & Robindon, Reference Berridge and Robinson2003). Finally, Internalizing × Externalizing Symptom interactions account for individual differences in gray matter densities in mesolimbic brain regions (Sauder et al., Reference Sauder, Beauchaine, Gatzke-Kopp, Shannon and Aylward2012). Observed heterotypic comorbidity of impulsivity and depression among self-injuring individuals, and an apparent common neurobiological substrate for internalizing and externalizing symptoms, provided an additional impetus for evaluating reward-related neural responding among self-injuring adolescent girls.
Following from the above discussion, we evaluated patterns of neural reactivity during anticipatory reward processing among self-injuring adolescent girls, compared to those observed among control adolescents. We focused specifically on anticipatory reward processing for several reasons. First, anticipatory and consummatory reward processing are separable processes, with distinct neural correlates (Liu, Hairston, Schrier, & Fan, Reference Liu, Hairston, Schrier and Fan2011). Second, prominent etiological models link behavioral impulsivity to deficits in reward anticipation (e.g., Tripp & Wickens, Reference Tripp and Wickens2008). Third, restricting analyses to only anticipatory reward processing increased our statistical power and reduced the familywise alpha error rate. Consistent with research outlined above, we predicted less neural reactivity in mesolimbic structures during anticipation of reward.
In addition, we predicted attenuated reward responding within both the orbitofrontal cortex (OFC) and the amygdala. Imaging studies indicate consistently that incentives activate the OFC, amygdala, and striatum, and that these regions all play important roles in evaluation of reward cues (Cardinal, Parkinson, Hall, & Everitt, Reference Cardinal, Parkinson, Hall and Everitt2002; McClure, York, & Montague, Reference McClure, York and Montague2004). For example, the OFC represents stimulus reward values, and is critical in reward learning (Rolls, Reference Rolls2000). Furthermore, functional disruptions in this region are associated with impulsivity and suicidality (Oquendo & Mann, Reference Oquendo and Mann2000), and aggressive children and adults with ADHD show alterations in OFC activity during reward processing (Rubia et al., Reference Rubia, Smith, Halari, Matukura, Mohammad and Taylor2009; Wilbertz et al., Reference Wilbertz, Tebartz van Elst, Delgado, Maier, Feige and Philipsen2012). A number of studies also indicate that the amygdala links anticipation of reward with hedonic value (see Murray, Reference Murray2007). Moreover, the amygdala is linked both anatomically and functionally to the OFC and the striatum (Baxter & Murray, Reference Baxter and Murray2002; Camara, Rodriquez-Fornells, & Münte, Reference Camara, Rodriguez-Fornells and Münte2008; Cardinal et al., Reference Cardinal, Parkinson, Hall and Everitt2002; Holland & Gallagher, Reference Holland and Gallagher2004), damage to the amygdala is associated with increased impulsivity among both animals and humans (Bechara, Damasio, Damasio, & Lee, Reference Bechara, Damasio, Damasio and Lee1999; Winstanley, Theobald, Cardinal, & Robbins, Reference Winstanley, Theobald, Cardinal and Robbins2004), and functional activation within the amygdala correlates with impulsivity (Brown, Manuck, Flory, & Hariri, Reference Brown, Manuck, Flory and Hariri2006). It is important to note that the OFC and amygdala are also implicated in mood regulation and aggression (Banks, Eddy, Angstadt, Nathan, & Phan, Reference Banks, Eddy, Angstadt, Nathan and Phan2007; Davidson, Putnam, & Larson, Reference Davidson, Putnam and Larson2000), and therefore may represent a second neural substrate of impulsivity and depression among self-injuring adolescents.
Method
Participants
Self-injuring (n = 22) and control (n = 22) adolescent females between ages 13 and 19 years were enrolled in the study. Among these, 3 participants from the self-injuring group and 2 participants from the control group were excluded due to motion-related functional magnetic resonance imaging (fMRI) artifacts and/or acquisition errors (see below). In addition, 1 participant from the control group was excluded because she was taking a selective serotonin reuptake inhibitor (SSRI) for depression, even though she did not report a current or lifetime mood disorder. Thus, 19 participants per group were included in all analyses. Twenty-four participants identified as Caucasian, 6 as Hispanic Caucasian, and 8 as biracial (4 mixed African American and Caucasian, 1 Filipino and Caucasian, 1 Japanese/Korean/Caucasian, and 2 Hispanic/non-Caucasian). Males were not included given lower prevalence rates of self-injury and an inadequate sample size for evaluating sex effects. Participants were recruited from previous studies, direct mailings to families, and advertisements/letters sent to mental health providers, community centers, and pediatrician's offices. Study procedures were approved by the Seattle Children's Hospital Institutional Review Board. Preliminary screening interviews with a trained research assistant took place by phone. Based on results of these interviews, selected respondents were invited to the lab for more extensive interviews, and then for a scanning session if they met inclusion criteria described below.
Phone screen
Callers completed the preliminary phone screen, which included questions that assessed lifetime and current self-injurious behavior; previous psychiatric diagnoses; current major depressive disorder (MDD); possible mental retardation; IV drug use; current medications; handedness; and MRI safety screening. Potential control group participants were screened out if they endorsed DSM-IV (APA, 1994) criteria for a depressive disorder, or if they reported any lifetime self-injury event. Those who reported a history of bipolar disorder or schizophrenia, possible mental retardation, IV drug use, left-handedness, and/or MRI contraindications (e.g., metal implants or braces) were excluded from both groups.
Lab interview
Respondents who met preliminary phone screen criteria were evaluated during a subsequent lab visit. Current disorders, including CD, generalized anxiety disorder, panic disorder, obsessive–compulsive disorder, posttraumatic stress disorder, social phobia, schizophrenia, dysthymia, bipolar disorder, anorexia, and bulimia were assessed using the Youth's Inventory (YI; Gadow et al., Reference Gadow, Sprafkin, Carlson, Schneider, Nolan and Mattison2002), a child self-report measure, and the Adolescent Symptom Inventory (ASI; Gadow & Sprafkin, Reference Gadow and Sprafkin1997), a parent-report measure. Although the YI and ASI are not structured interviews, both assess full criterion sets for DSM-IV disorders, and yield screening cutoffs that are consistent with DSM-IV diagnoses (see below). Given high rates of comorbidity of self-injury, major depression, and substance use, both current and lifetime diagnoses of MDD and substance abuse/dependence were assessed using the Diagnostic Interview Schedule for Children—Computerized Version (DISC-C; Shaffer, Fisher, Lucas, Mina, & Schwab-Stone, Reference Shaffer, Fisher, Lucas, Mina and Schwab-Stone2000). To be considered for the self-injury group, respondents were required to report at least three self-injury episodes in the last year or five or more in their lifetime. At least one of these episodes had to have a lethality rating of 2 or higher on the Lifetime Suicide Attempt and Self-Injury Interview (L-SASI Count; Linehan & Comtois, Reference Linehan and Comtois1996). Examples include taking 6–10 pills (or fewer if the medication is potentially lethal), cigarette burns, jumping from places up to 10 feet, taking heroin at ≥1.5 times the dependent dose, and combining drugs and/or alcohol. Potential self-injuring participants who were taking SSRIs (n = 5) were allowed to participate. Given the large number of self-injuring individuals who are prescribed SSRIs, excluding these respondents could have compromised generalizability. Qualifying participants were invited to participate in a more comprehensive lab visit screen to assess their eligibility more fully. Respondents who reported a lifetime history of SII and/or suicide attempts, or who met diagnostic criteria for any of the above disorders, as assessed by the ASI, YI, or parent or youth report on the DISC-C, were excluded from the control group. Among respondents who were invited to the lab, five were screened out based on their interview results.
Measures
L-SASI
The L-SASI (formerly the Lifetime Parasuicide Count; Linehan & Comtois, Reference Linehan and Comtois1996) is a structured interview designed to collect information about lethality, suicidal intent, level of medical treatment received for, and specific circumstances surrounding adolescents' first, most recent, and most severe SII episode. With interviewer guidance, participants count lifetime events of different forms of self-injury, self-injury with intent to die, self-injury with ambivalence, self-injury without suicidal intent, and self-injury requiring medical treatment. In addition, the highest lethality event in each category of SII is assessed. Lethality rankings range from 1 = very low (e.g., head banging) to 6 = severe (e.g., asphyxiation). There are no psychometric studies of the L-SASI. However, items are identical to a subset of those on a longer measure, the Suicide Attempt Self-Injury Interview, which has very good interrater reliability and adequate validity (Linehan, Comtois, Brown, Heard, & Wagner, Reference Linehan, Comtois, Brown, Heard and Wagner2006). Although the L-SASI was designed for use with adults, it has since been used successfully with adolescents (e.g., Crowell et al., Reference Crowell, Beauchaine, Hsiao, Vasilev, Yaptangco and Linehan2012, Reference Crowell, Kaufman, Lenzenweger, Beauchaine and Hinshaw2013).
Suicidal Ideation Questionnaire (SIQ)
The SIQ (Reynolds, Reference Reynolds1987, Reference Reynolds1988) was used to screen adolescents at each study visit for levels of suicidal ideation. There are two versions of this measure based on grade level. The SIQ is composed of 30 items and is used for students in Grades 10–12. The SIQ-JR is composed of 15 items and is designed for use by students in Grades 7–9. In both versions, items are rated on 7-point scales (0 = I never had this thought, 6 = I had this thought every day), which are summed. Internal consistencies are high for both the SIQ (α = 0.97) and the SIQ-JR (α = 0.93–0.94). Scale scores correlate moderately with constructs such as depression and hopelessness (0.52–0.70).
Structured Clinical Interview for DSM-IV Axis II
The Structured Clinical Interview for DSM-IV Axis II (First, Gibbon, Spitzer, Williams, & Benjamin, Reference First, Gibbon, Spitzer, Williams and Benjamin1997) is a widely used semistructured interview that assesses DSM-IV Axis II psychopathology. We used it only to assess BPD symptoms. The 15 BPD items evidence strong interrater reliability (κ = 0.91; Lobbestael, Leurgans, & Arntz, Reference Lobbestael, Leurgans and Arntz2010; Maffei et al., Reference Maffei, Fossati, Agostoni, Barraco, Bagnato and Deborah1997) and internal consistency (α = 0.71–0.94; Maffei et al., Reference Maffei, Fossati, Agostoni, Barraco, Bagnato and Deborah1997).
Youth Self-Report
The Youth Self-Report (Achenbach, Reference Achenbach1991) is a widely used and well-validated self-report measure of internalizing and externalizing behaviors, designed for use with adolescents ages 11–18 years old. One-week test–retest reliabilities in the validation sample were 0.80 for the broadband internalizing factor, and 0.81 for the broadband externalizing factor. These scales served as primary independent measures in our imaging analyses (see below).
DISC-C
The DISC-C (Schaffer et al., Reference Shaffer, Fisher, Lucas, Mina and Schwab-Stone2000) is a commonly used, highly structured interview that assesses a wide range of DSM-IV psychiatric disorders. We used the computerized version for ease of administration and scoring. In the validation sample, the 1-year test–retest reliability for self-reported MDD was 0.92.
YI
The YI (Gadow et al., Reference Gadow, Sprafkin, Carlson, Schneider, Nolan and Mattison2002) is a 120-item checklist of DSM-IV symptoms that yields both dimensional scores and diagnostic cut-offs. Items are rated on a 4-point scale (0 = never, 1 = sometimes, 2 = often, and 4 = very often). In the validation sample, 2-week test–retest reliabilities ranged from 0.54 to 0.92.
ASI
The ASI (Gadow & Sprafkin, Reference Gadow and Sprafkin1997) is a 120-item parent report checklist of DSM-IV symptoms that yields both dimensional scores and diagnostic cutoffs. Items are rated on a 3-point scale (0 = never, 1 = sometimes, 2 = often, and 3 = very often). The ASI exhibits moderate to high concurrent validity with the Child Behavior Checklist, with correlations ranging from 0.41 to 0.80 for similar emotional and behavioral problems.
Kaufman Brief Intelligence Test, Second Edition
Adolescent's IQs were assessed using the Kaufman Brief Intelligence Test, Second Edition (Kaufman & Kaufman, Reference Kaufman and Kaufman2004), a brief measure of verbal and nonverbal cognitive ability with excellent psychometric properties. Test–retest reliabilities range from 0.88 to 0.92, with strong internal consistencies for composite IQ scores (α = 0.89 to 0.96). Those with composite IQ scores less than 85 (n = 0) were excluded.
Mock scanning
Remaining participants, all of whom met inclusion/exclusion criteria, participated in a mock scanning session at the end of their lab visit in order to teach them the behavioral tasks (described below), and to assess their comfort level and ability to remain still during scanning. During mock scanning, a sticker was affixed to the forehead of participants. Next, participants were given an opportunity to practice the monetary incentive delay (MID) task (Knutson, Adams, Fong, & Hommer, Reference Knutson, Adams, Fong and Hommer2001), which is described in detail below. Although no participants were excluded based on mock scanning, five were lost following the mock scanning session due to attrition (e.g., moving away, too busy, scheduling conflicts, etc.). Adolescents were paid $25 for participating in this first lab visit.
MRI scanning
Participants returned for a second lab visit to complete the MRI scanning protocol. Following safety screening by a trained MR technician, participants completed the anatomical MRI scan before completing two tasks during functional scanning. The procedure took about 60 min, including 3 min for the structural scan, 3 min for B0 fieldmap acquisition, and 17 min for the reward task, described below.
The MID task (Knutson, Adams, et al., Reference Knutson, Adams, Fong and Hommer2001) requires participants to respond to stimuli on either the left or the right side of a screen. Prior to each trial, a visual cue is presented indicating whether participants can win or lose money in the subsequent trial. On “win” trials, participants see one of three circle cues indicating that they can win $0.20, $1.00, or $5.00 for responding correctly and within a time limit, with errors resulting in no monetary gain or loss. In “loss” trials, square cues indicate that errors will result in loss of $0.20, $1.00, or $5.00, with correct responses resulting in no monetary gain or loss. Participants are required to respond while target stimuli are present on the screen, and target duration is titrated to each individual's performance, to keep accuracy rates at approximately 66%. In addition, there are several control trials (triangle cues) in which participants neither win nor lose money, regardless of their responses. Following each trial, participants receive visual feedback regarding both their accuracy on the previous trial and the monetary outcome (gain, loss, or no change, reported numerically). During all phases of the task, with the exception of cue and target, total amount of money earned over the course of the task is displayed in the upper right-hand corner of the screen. At the end of the task, participants receive the actual dollar amount earned.
Three runs were completed by each participant, each of which consisted of 18 reward trials and 18 loss trials, split evenly among the three cue amounts ($0.20, $1.00, or $5.00). In addition, there were 12 control trials in which participants neither won nor lost money. Each trial lasted 6 s, consisting of one of seven cues (250 ms), fixation delay (2000–2750 ms), target (210–490 ms), a second delay (360–640 ms), and feedback (1650 ms). Intervals between cue presentations (5.5–8.75 s, M = 7 s), and intervals between cue and feedback (3.00–3.75 s, M = 3.4 s), were varied across trials to ensure optimal power to detect event-related signal differences during fMRI acquisition (Dale, Reference Dale1999).
fMRI data acquisition, preprocessing, and analysis
Structural and functional MRI scans were performed on a 3 Telsa Philips Achieva MR System (version 2.63, Philips Medical Systems, Best, The Netherlands) with dual Quasar gradients (80 mT/m at a slew rate of 110 mT/m/s; or 40 mT/m at a slew rate of 220 mT/m/s) and an eight-channel SENSE head coil. High resolution three-dimensional FFE T1-weighted magnetization prepared–rapid gradient echo fast imaging sequences generated 200 contiguous axial slices spanning the entire brain (repetition time [TR] = 7.7 ms; echo time [TE] = 3.6 ms; flip angle = 8°; field of view [FOV] = 220 × 220 × 200; matrix size = 220 × 205; voxel dimension = 1 × 1.07 × 1 mm; SENSE factor = 1). Total scan time for anatomical images was approximately 3 min. Structural data were used for image registration. Whole-brain T2*-weighted images were acquired using a single-shot gradient-recalled echo-planar imaging (EPI) sequence (TR = 2000 ms; TE = 21 ms; flip angle = 76°; FOV = 210 mm; matrix size = 72 × 72; in-plane resolution = 3 × 2.92 mm, slice thickness = 3 mm). Forty-seven or 48 contiguous 3-mm axial slices were collected per image volume using an ascending slice acquisition. Three functional runs of 174 dynamics were collected for the MID task, with a total functional scan time of approximately 17 min. A matching B0 field map using a fast field echo sequence was acquired to correct for distortions in the EPI data due to magnetic field nonhomogeneities (TR = 563 ms; TE = 2.8; flip angle = 90°; FOV = 210 × 210 × 141; matrix size = 72 × 72; in-plane resolution = 3 × 2.92; slice thickness = 3 mm). Forty-seven or 48 contiguous 3-mm axial slices spanning the entire brain were collected per image.
Data were analyzed using statistical parametric mapping software (SPM8; http://www.fil.ion.ucl.ac.uk/spm/software/spm8/). Initial preprocessing included slice time correction, motion correction, and deformation of field nonhomogeneity using SPM's unwarp procedure in conjunction with B0 fieldmaps. Subsequently, data were visually inspected and analyzed to detect excessive rapid motion using ArtRepair5, an artifact detection and repair tool (Mazaika, Hoeft, Glover, & Reiss, Reference Mazaika, Hoeft, Glover and Reiss2009). Two participants were excluded from the control group and three from the self-injuring group due to EPI acquisition artifacts and/or excessive inter-EPI motion of greater than 0.5 mm over more than 15% of EPI's collected. An additional eight functional runs were excluded (5 control and 3 self-injuring) under the same criteria. Remaining participants (n = 19 self-injury, n = 19 control) were normalized to the MNI template using diffeomorphic anatomical registration through expotentiated lie algebra (Ashburner, Reference Ashburner2007), a high dimensional warping process. Finally, data were smoothed using an 8-mm Gaussian kernel.
To assess the effect of differing reward values, reward cues ($0.20, $1.00, or $5.00) were modeled separately. For each participant individually, reward cues were modeled relative to baseline (cues indicating no opportunity to win or lose) using a standard hemodynamic response function. First-level general linear models were created for each participant, and included the above described conditions of interest (e.g., reward cues) and noninterest (e.g., loss cue/feedback). Individual EPI scans with rapid motion of >0.5 mm were modeled as nuisance covariates, and data were detrended using an AR(1) model with a high-pass filter of 128 s.
To evaluate the within- and between-groups effects associated with anticipatory reward processes, a second-level mixed effects general linear model was performed using a Group (SII, control) × Reward Magnitude ($0.20, $1.00, $5.00) factorial design. Group × Reward Magnitude interaction effects were included. SSRI use was included initially as a covariate, but did not alter results and was subsequently removed from the final model.Footnote 2 A region of interest (ROI) analysis rather than whole-brain analysis was performed due to limited statistical power associated with the small sample size, the complexity of the task design, and clear a priori hypotheses regarding the regions implicated in reward processing. We selected ROIs based on theoretical considerations discussed above, and based on findings from previous publications in which the MID and related tasks were used (Knutson, Adams, et al., Reference Knutson, Adams, Fong and Hommer2001; Knutson, Fong, Adams, Varner, & Hommer, Reference Knutson, Fong, Adams, Varner and Hommer2001; Forbes et al., Reference Forbes, Christopher May, Siegle, Ladouceur, Ryan and Carter2006; Liu et al., Reference Liu, Hairston, Schrier and Fan2011). Specifically, we evaluated activation within the striatum (caudate+putamen) and OFC (orbital regions of the superior, middle, and frontal gyri), based on anatomical labels from the Automatic Anatomical Labeling Atlas (Tzourio-Mazoyer et al., Reference Tzourio-Mazoyer, Landeau, Papathanassiou, Crivello, Etard and Delcroix2002). In addition, we examined activation within the bilateral amygdalae. Main effects and interactions were modeled separately for each region, and corrected for multiple comparisons (p < .05) using a small volume correction based on the size of the ROI (ROIs were examined bilaterally, so there were three comparisons). Clusters of activation that did not exceed statistical thresholds or were less than 0.5 cm3 were excluded. To evaluate effects for each participant separately, SPM's built-in principle eigenvariates approach was used to calculate average beta values across all voxels within each area of functional activation.
Results
Demographics and self-reports of psychopathology
As reported in Table 1, the self-injuring and control groups did not differ on age or IQ. However, consistent with previous research (e.g., Crowell et al., Reference Crowell, Beauchaine, McCauley, Smith, Stevens and Sylvers2005, Reference Crowell, Beauchaine, Hsiao, Vasilev, Yaptangco and Linehan2012), self-injuring participants scored higher on all self-report measures of internalizing and externalizing psychopathology. Although not reported, parent-report measures demonstrated similar between-groups differences. Given that agreement between adolescents and parents on measures of psychopathology is modest at best (e.g., Seiffge-Krenke & Kollmar, Reference Seiffge-Krenke and Kollmar1998; Sourander, Helstela, & Helenius, Reference Sourander, Helstela and Helenius1999), we analyzed data from both informants.
Demographic variables, reports of self-injury and suicidal ideation, and self-reported internalizing and externalizing scale score by group
Note: KBIT, Kaufman Brief Intelligence Test, Second Edition (KBIT-2; Kaufman & Kaufman, Reference Kaufman and Kaufman2004); SIQ, Suicide Ideation Questionnaire (Reynolds, Reference Reynolds1987, Reference Reynolds1988), standardized by grade level: SIQ-Jr (Grades 7–9) raw scores ≥ 31 and SIQ (Grades 10–12) raw scores ≥ 41 indicate significant clinical concern regarding suicide risk; L-SASI, Lifetime Suicide Attempt Self-Injury Count (Linehan & Comtois, Reference Linehan and Comtois1996); SCID-II BPD, Structured Clinical Interview for DSM-IV Axis II (First et al., Reference First, Gibbon, Spitzer, Williams and Benjamin1997) borderline personality disorder; SII, self-inflicted injury; SA, suicide attempt; ADHD-combined, attention-deficit/hyperactivity disorder–combined; DISC, Diagnostic Interview Schedule for Children.
a t tests were conducted on normally distributed data; Mann–Whitney U tests were conducted on skewed data.
b Due to technical difficulties with the computerized questionnaire, Youth's Inventory data were lost for one participant who had already been screened into the control group and completed the scanning procedure.
c Means, standard deviations, and t tests are reported for log-transformed data.
*p ≤ .05. **p ≤ .01. ***p ≤ .001.
Behavioral responding to incentives
No group differences were observed during the reward task in reaction times (M control = 281.4 ms, M si = 267.3 ms), t (36) = 1.75, d = 0.58; proportion of accurate responses (M control = 0.64, M si = 0.64), t (36) = 0.74, d = 0.25; or amount of money earned (M control = $49.84, M si = $46.22), t (36) = 0.86, d = 0.29.
Neural responding to incentives
Analyses of participants' neural responses yielded main effects of both group and reward magnitude (see Table 2). Compared to controls, self-injuring adolescents showed significantly less activation to reward cues in the striatum (putamen), amygdala, and OFC bilaterally (see Figure 1). There were no regions of greater activation for the SII group. Attenuated neural responding among SII participants was observed across all cue values and all ROIs. Nevertheless, both groups showed greater activation for higher cue values relative to lower cue values (see Figure 2). The main effect of reward magnitude predicted activation only in the striatum, specifically the caudate nucleus. There were no significant Group × Reward Magnitude interactions.Footnote 3
(Color online) Striatal, orbitofrontal, and amygdalar regions of interest in which controls exhibited greater neural reactivity to reward cues.
Neural responding among participants in the self-injury and control groups across all reward values and regions of interest. Values reflect beta weight estimates with arbitrary units of measurement.
Areas of significant activation for main effects of group and reward magnitude
Note: MNI, Standard Montreal Neurological Institute stereotaxic coordinates (x, y, z); OFC, orbitofrontal cortex. Negative z scores reflect greater activation for the SSI group relative to controls. Region of interest analyses were corrected for multiple comparisons using a false-discovery rate small volume correction for the spatial extent of the anatomical region of interest. Contiguous areas of functional activation that spanned hemispheres are italicized across adjacent rows.
We conducted follow-up analyses to assess whether functional activation within regions that differentiated between SII and control participants was associated with symptoms of impulsivity (ADHD) and/or depression. Significant correlations were found across groups for symptoms of both ADHD and MDD in both the striatum and amygdalae bilaterally (see Figure 3). Activation within the striatum was correlated negatively with ADHD combined symptoms, as reported by both parents (r = –.35, p = .03) and adolescents (r = –.42, p = .01). Similar effects were found for symptoms of MDD for both parents (r = –.31, p = .05) and adolescents (r = –.49, p = .002). Amygdala activation was also correlated negatively with both ADHD combined symptoms (parent report, r = –.41, p = .01; adolescent report, r = –.38, p = .02) and MDD symptoms (parent report, r = –.40, p = .01; adolescent report, r = –.46, p = .004). There were no associations between OFC activation and either parent or self-reports of ADHD or MDD, all ps > .23.Footnote 4
Correlations between neural reactivity in each region of interest and symptoms of (left) attention-deficit/hyperactivity disorder and (right) major depression across regions of interest. Values reflect beta weight estimates with arbitrary units of measurement. ADHD-C, attention-deficit/hyperactivity disorder, combined subtype; MDD, major depressive disorder; Sx, number of diagnostic symptoms met.
Discussion
Our primary objective in conducting this study was to test the hypothesis that central nervous system reward dysfunction is observed among self-injuring adolescent girls, and is expressed similarly to that observed in both impulse control and depressive disorders. Such findings support our developmental theory of SII and borderline pathology in which mesolimbic DA dysfunction predisposes to trait anhedonia and trait irritability, which characterize both impulse control and depressive disorders, and likely underlie patterns of heterotypic comorbidity observed among those who self-injure (see Kaufman, Crowell, & Stepp, Reference Kaufman, Crowell, Stepp, Beauchaine and Hinshaw2015). We evaluated neural responding in brain regions associated previously with reward processing during anticipation of incentives. Consistent with our hypotheses, self-injuring adolescents exhibited less neural responding to reward in the striatum (putamen), OFC, and bilateral amygdalae compared with controls. Furthermore, activation in both the striatum and amygdalae was correlated negatively with parent and adolescent reports of impulsivity (ADHD) and depression (MDD). We consider these findings in further detail below.
Trait impulsivity is highly heritable (Krueger et al., Reference Krueger, Hicks, Patrick, Carlson, Iacono and McGue2002) and is mediated by feedforward and feedback dopaminergic neural networks originating in the striatum and projecting to the prefrontal cortex (see Beauchaine & McNulty, Reference Beauchaine and McNulty2013; Gatzke-Kopp & Beauchaine, Reference Gatzke-Kopp, Beauchaine, Coch, Dawson and Fischer2007). Although early models of trait impulsivity assumed dopaminergic hyperactivity to reward (e.g., Quay, Reference Quay1993), more recent empirical findings provide convincing evidence for both hypoactivation within and altered connectivity between these structures among those with impulse control disorders (see Beauchaine & Gatzke-Kopp, Reference Beauchaine and Gatzke-Kopp2012; Gatzke-Kopp, Reference Gatzke-Kopp2011; Plichta & Scheres, Reference Plichta and Scheres2014; Sagvolden et al., Reference Sagvolden, Johansen, Aase and Russell2005; Shannon et al., Reference Shannon, Sauder, Beauchaine and Gatzke-Kopp2009). According to prevailing contemporary theories, reduced tonic mesolimbic activity confers vulnerability to impulsivity because it is experienced as an aversive, irritable mood state (e.g., Laakso et al., Reference Laakso, Wallius, Kajander, Bergman, Eskola and Solin2003), which affected individuals are motivated to upregulate through reward-seeking behavior. However, because these individuals also exhibit reduced phasic mesolimbic responding to incentives, they experience limited hedonic value from ordinary rewards, and therefore engage in increasingly frequent reward-seeking behavior, including searches for larger and larger rewards (see Sagvolden et al., Reference Sagvolden, Johansen, Aase and Russell2005). Consistent with this theory, neuroimaging studies demonstrate reduced reward responding in mesolimbic and/or mesocortical structures across impulse control disorders including ADHD, CD, substance use disorders, and gambling addiction (see, e.g., de Ruiter et al., Reference de Ruiter, Veltman, Goudriaan, Oosterlaan, Sjoerds and van den Brink2008; Dom, Sabbe, Hulstijn, & Van Den Brink, Reference Dom, Sabbe, Hulstijn and Van Den Brink2005; Rubia, Reference Rubia2011).
As noted above, abnormalities in DA-mediated reward function also play a role the etiology of depression (see Forbes & Dahl, Reference Forbes and Dahl2011; Nestler & Carlezon, Reference Nestler and Carlezon2006). Until recently, most neurobiological models of depression focused on serotinergic neural networks. However, an ever growing number of studies point toward mesolimbic reward dysfunction as a neural substrate of anhedonia, irritability, and low positive affectivity in both depression and impulsivity. Among both healthy and depressed individuals, anhedonia is associated with hypoactivity within the ventral striatum (Gorwood, Reference Gorwood2008). Moreover, blunted neural responding to incentives is observed among adolescents and adults with depression (Forbes et al., Reference Forbes, Christopher May, Siegle, Ladouceur, Ryan and Carter2006; Knutson, Bhanji, Cooney, Atlas, & Gotlib, Reference Knutson, Bhanji, Cooney, Atlas and Gotlib2008; Smoski et al., Reference Smoski, Felder, Bizzell, Green, Ernst and Lynch2009). Furthermore, increased neural activation in the striatum following treatment is associated with reduced depressive symptoms (Dichter et al., Reference Dichter, Felder, Petty, Bizzell, Ernst and Smoski2009, Schlaepfer et al., Reference Schlaepfer, Cohen, Frick, Kosel, Brodesser and Axmacher2007).
To our knowledge, ours' is the first study to examine central nervous system reward processing among self-injuring adolescents, and only one study has done so among those with BPD. In that study, participants with BPD exhibited reduced prefrontal and striatal responding to monetary incentives, and prefrontal responding was correlated negatively with self-reported impulsivity (Völlm et al., Reference Völlm, Richardson, McKie, Elliott, Dolan and Deakin2007). However, not all individuals in the clinical group met criteria for BPD, and instead were diagnosed with antisocial personality disorder (ASPD). Although it may be tempting to eschew results of this study based on inclusion of ASPD, BPD and ASPD share a number of etiological influences, and may reflect sex-moderated manifestations of a single pathology (see Beauchaine et al., Reference Beauchaine, Klein, Crowell, Derbidge and Gatzke-Kopp2009). Accordingly, results are likely relevant to the present discussion.
Although sufficient evidence exists to suggest that anticipation and consummation of incentives are mediated by different neural processes (e.g., Dillon et al., Reference Dillon, Holmes, Jahn, Bogdan, Wald and Pizzagalli2008; Knutson, Fong, et al., Reference Knutson, Fong, Adams, Varner and Hommer2001; Liu et al., Reference Liu, Hairston, Schrier and Fan2011; Schultz, Reference Schultz2000), few studies have examined these processes separately in clinical populations. However, doing so may be particularly important when evaluating reward-related responding among impulsive children and adolescents, because anatomical regions associated with anticipation of reward and consummation of reward develop at different rates (Bjork et al., Reference Bjork, Knutson, Fong, Caggiano, Bennett and Hommer2004; Geier & Luna, Reference Geier and Luna2009; Geier, Terwilliger, Teslovich, Velanova, & Luna, Reference Geier, Terwilliger, Teslovich, Velanova and Luna2010), and trait impulsivity may be mediated more by mesolimbic dysfunction in childhood but more by mesocortical dysfunction by late adolescence and adulthood (see Beauchaine & McNulty, Reference Beauchaine and McNulty2013). Although nearly all studies of anticipatory reward processing among externalizing adults and youth have found decreased reward-related activity, increased neural activity during consummatory processing, especially in prefrontal regions, is sometimes observed among adults with ADHD and alcohol dependence (Strohle et al., Reference Strohle, Stoy, Wrase, Schwarzer, Schlagenhauf and Huss2008; Wrase et al., Reference Wrase, Schlagenhauf, Kienast, Wustenberg, Bermpohl and Kahnt2007). In contrast, studies that parse anticipatory and consummatory reward processes among adolescent and adult depressed individuals are much more consistent, demonstrating hypoactivity across both phases (Forbes et al., Reference Forbes, Christopher May, Siegle, Ladouceur, Ryan and Carter2006; Pizzagalli et al., Reference Pizzagalli, Holmes, Dillon, Goetz, Birk and Bogdan2009; Smoski et al., Reference Smoski, Felder, Bizzell, Green, Ernst and Lynch2009). Finally, among externalizing samples, group differences in reward reactivity may also be associated more strongly with anticipation than with consummation (Strohle et al., Reference Strohle, Stoy, Wrase, Schwarzer, Schlagenhauf and Huss2008).
Given that comorbid externalizing problems are often observed in SII, we focused on the anticipatory phase of reward processing, which differentiates both internalizing and externalizing samples from controls. Our results are consistent with theoretical expectations and previous empirical work outlined above on reward processing among individuals with impulse control disorders and depression. Self-injuring adolescents showed less reactivity to reward cues in both striatal and OFC regions, and activation within the striatum was correlated negatively with internalizing and externalizing symptoms. These results may reflect a common neural vulnerability to impulsivity and depression, and provide insight into much higher than expected rates of heterotypic comorbidity (Sauder et al., Reference Sauder, Beauchaine, Gatzke-Kopp, Shannon and Aylward2012).
Self-injuring participants also exhibited less activation relative to controls in the amygdala, which is implicated in tracking hedonic value of reward (Baxter & Murray, Reference Baxter and Murray2002; Murray, Reference Murray2007; Liu et al., Reference Liu, Hairston, Schrier and Fan2011). Although etiological models implicate the amygdala in the pathogenesis of both depression and BPD, nearly all theoretical and empirical work on the amygdala has focused exclusively on its processing of emotion, not reward. Such studies typically find increased amygdala reactivity in depression and BPD, particularly to negative emotional stimuli. In the present study, self-injuring adolescents showed reduced reactivity that was reward specific, with no group differences to loss cues. Activation within the amygdala was also correlated negatively with symptoms of both ADHD and depression. Few studies have examined amygdala reactivity to reward among impulsive or depression samples (for exception, see Forbes et al., Reference Forbes, Christopher May, Siegle, Ladouceur, Ryan and Carter2006; Stark et al., Reference Stark, Bauer, Merz, Zimmermann, Reuter and Plichta2011). Results of the current study further implicate this region in development of SII and related disorders (e.g., BPD and MDD), and suggest that amygdala abnormalities among those with these disorders may extend beyond processing of negative emotion.
Recent work by Rubia and others suggests that the amygdala and orbital/medial PFC are part of a separable “hot” network of executive functioning responsible for decisions related to motivation and affective experience (Castellanos, Sonuga-Barke, Milham, & Tannock, Reference Castellanos, Sonuga-Barke, Milham and Tannock2006; Rubia, Reference Rubia2011). Although dysfunction is observed in this network among children with ADHD, more severe deficits are observed among those with CD, which is characterized by greater mood lability, frequent displays of anger, and aggression (Cubillo, Halari, Smith, Taylor, & Rubia, Reference Cubillo, Halari, Smith, Taylor and Rubia2012; Rubia, Reference Rubia2011; Rubia et al., Reference Rubia, Smith, Halari, Matukura, Mohammad and Taylor2009). Amygdala hyperactivation and corresponding reductions in OFC activity, and connectivity between these regions, is also associated with affective instability among adults with BPD, a disorder that is also marked by anger and aggression (Linehan, Reference Linehan1993; Lis et al., Reference Lis, Greenfield, Henry, Guile and Dougherty2007; New et al., Reference New, Hazlett, Buchsbaum, Goodman, Mitelman and Newmark2007; Soloff et al., Reference Soloff, Meltzer, Becker, Greer, Kelly and Constantine2003).
Although our findings are consistent with the hypothesis that impulsivity is an etiological factor in the development of SII, several caveats should be considered. First, only main effects are reported. Given the small sample and study design, statistical power was inadequate for testing Group × Condition interactions. The small sample size may also have affected our ability to detect associations between reward-related responding and self-reported internalizing and externalizing symptoms. Second, we did not have a depressed control group. Thus, we do not know what neural processes, if any, differentiate self-injuring adolescents, almost all of whom are depressed, from depressed adolescents who do not self-injure.
As is often the case in studies of SII, there was also considerable heterogeneity within the self-injuring sample. Numbers of self-injury episodes varied substantially, as did ratings of lethality and intent. Moreover, although a history of SII was required, current self-injury was not. In addition, our findings may not generalize to self-injuring males. Neurobiological correlates of externalizing psychopathology differ for girls versus boys (see Beauchaine, Hong, & Marsh, Reference Beauchaine, Hong and Marsh2008), which illustrates why further research should be conducted to determine if reward-processing abnormalities observed among self-injuring adolescent girls extend to males who exhibit similar behaviors. Given the importance of developmental influences on the etiology and maintenance of nonsuicidal self-injuries (Beauchaine et al., Reference Beauchaine, Klein, Crowell, Derbidge and Gatzke-Kopp2009; Crowell et al., Reference Crowell, Beauchaine and Linehan2009; Crowell, Baucom, et al., Reference Crowell, Baucom, Yaptangco, Bride, Hsiao and McCauley2014; Derbidge & Beauchaine, Reference Derbidge, Beauchaine, Lewis and Rudolph2014), future longitudinal research is necessary to assess interrelations among impulsivity, depression, self-injury, and borderline personality development. Finally, future research is needed to address potential links between reward anticipation and emotion processing and regulation. Such research may be required to fully characterize relations among impulsivity, emotion deregulation, and self-injury.
Ours is the first study to examine neural correlates of reward responding among self-injuring adolescents. Reduced striatal and prefrontal activation were observed during reward anticipation in the self-injuring group, consistent with previous findings from those with impulse control disorders and depression. We hope future research with larger samples expands on our findings, toward disentangling what are clearly complex associations between neurobiological function, trait impulsivity, hedonic capacity, affectivity, and the development of debilitating conditions such as SII and borderline personality traits.
