Differentiating clinically depressed adolescents with and without active suicidality: An examination of neurophysiological and self-report measures of reward responsiveness

Abstract

Background:

Suicidality in youth is a major public health problem, and objective methods for identifying those at greatest risk are critically needed. Suicidality has been associated with alterations in reward-related decision-making, but the extent to which measures of reward responsiveness (RR) can differentiate youth with and without suicidality in clinical samples remains unclear.

Methods:

We examined reliable and accessible neurophysiological (i.e., reward positivity [RewP] event-related potential) and self-report (Behavioral Activation System [BAS] subscales) measures of RR in relation to active suicidality in 58 clinically depressed adolescents (14-18 years old).

Results:

Logistic regression analysis indicated that active suicidality in depressed adolescents was associated with heightened RR at both the self-report and neurophysiological levels. A relatively more positive RewP to win and a more negative RewP to loss uniquely predicted active suicidality beyond demographic, clinical, and self-report measures.

Conclusions:

Results support the utility of neurophysiological measures in differentiating clinically depressed adolescents with and without suicidality. Although depression is commonly characterized by reduced RR, depressed adolescents with active suicidality exhibited relatively enhanced neural responses to reward and loss feedback. Results highlight the need for consideration of heterogeneity in RR in depression and research on personalized depression treatment.

Introduction

Suicide in youth is a global public health concern. Lifetime prevalence of suicidal ideation (SI) in adolescents is approximately 12% in the general population and as high as 56% in adolescents with major depressive disorder (MDD; Lewinsohn, Rohde, & Seeley, 1996; Nock et al., 2013). Rates of SI increase dramatically between the ages of 12 and 17 (Nock et al., 2013), marking adolescence as a high-risk developmental period. SI is a strong predictor of subsequent suicide attempts (Franklin et al., 2017; Prinstein et al., 2008). One study found a nearly eightfold increase in suicide attempts among those with a history of SI compared to the overall sample (Nock et al., 2013). Epidemiological evidence highlights the need for research improving the identification of those at risk for suicidality.

Demographic and clinical correlates of suicidality in adolescents have been identified, including gender differences and psychopathology, especially depression (Nock et al., 2013). Yet, a recent meta-analysis demonstrated that existing risk factors yield small effect sizes and limited predictive ability across time (Franklin et al., 2017). This may be due in part to a focus on only a few methods, particularly self-report (Franklin et al., 2017). There is a critical need for research using more objective methods, like neurophysiology (Glenn, Cha, Kleiman, & Nock, 2017; Nock et al., 2010).

Suicidality has been characterized by alterations in frontostriatal brain circuits (Dombrovski & Hallquist, 2017; Ho et al., 2018; Zhang, Chen, Jia, & Gong, 2014), which could underlie impairments in reward-related decision-making, including alterations in valuation, prediction, and learning observed in adult suicide attempters (Clark et al., 2011; Dombrovski et al., 2010; Liu, Vassileva, Gonzalez, & Martin, 2012). Although the literature is more limited, similar deficits have been observed in adolescent suicide attempters (Bridge et al., 2012). Value-based models of learning indicate that reward responsiveness (RR) and reward prediction errors are key mechanisms driving feedback-based learning (Jocham, Klein, & Ullsperger, 2011). Adolescence is thought to be a particularly high-risk period for alterations in decision-making and RR due to the linearly protracted development of the prefrontal cortex (PFC), heightened response to reward in the striatum compared to childhood, and age-related changes in corticostriatal connectivity (Blakemore & Robbins, 2012; Cohen et al., 2010; Hartley & Somerville, 2015; van den Bos, Cohen, Kahnt, & Crone, 2012). Thus, indicators of RR may be particularly relevant predictors of suicidality in adolescents.

At the neurophysiological level, RR can be reliably measured by the reward positivity (RewP). This event-related potential (ERP) derived from the electroencephalogram (EEG) is enhanced in response to reward relative to loss feedback and thought to reflect activation of reinforcement learning systems (Carlson, Foti, Mujica-Parodi, Harmon-Jones, & Hajcak, 2011; Holroyd & Coles, 2002; 2008). RewP has been associated with activation of the striatum and medial PFC (Carlson et al., 2011; Foti, Carlson, Sauder, & Proudfit, 2014; Foti, Weinberg, Dien, & Hajcak, 2011) and individual differences in reward learning (Bress & Hajcak, 2013). Importantly, RewP demonstrates reliability across development (Kujawa et al., 2018; Luking, Nelson, Infantolino, Sauder, & Hajcak, 2017), suggesting it could have utility for detecting suicidality risk across time.

Consistent with Gray’s motivational theory (1990), neurophysiological indicators of RR likely reflect one facet of the larger behavioral activation system (BAS), which drives approach motivation and appetitive behavior. At the self-report level, dimensions of RR and approach motivation are commonly assessed using the Behavioral Inhibition System/Behavioral Activation System (BIS/BAS) scales (Carver & White, 1994). BAS consists of three facets: drive, fun-seeking, and RR (Carver & White, 1994). Elevated scores on BAS RR have been associated with better decision-making (Franken & Muris, 2005), whereas fun-seeking and drive have been associated with greater risky behavior, including substance use and impulsivity (Franken & Muris, 2006; Leone & Russo, 2009; Voigt et al., 2009). Further, some studies have evidenced associations between RewP and BAS, suggesting they may reflect aspects of the same biobehavioral system, assessed at distinct levels (Bress & Hajcak, 2013; Lange, Leue, & Beauducel, 2012).

Across levels of analysis (i.e., self-report, behavior, physiology, circuit), reduced RR has consistently emerged as a risk factor for emergence of depressive symptoms. For example, both reduced ventral striatum activation and RewP to rewards have been shown to prospectively predict depressive symptoms in youth (for reviews, Keren et al., 2018; Kujawa & Burkhouse, 2017). Similar patterns have emerged at the self-report level, with some evidence that low BAS scores are associated with depression in adults (McFarland, Shankman, Tenke, Bruder, & Klein, 2006; Kasch, Rottenberg, Arnow, & Gotlib, 2002; but also see Mellick, Sharp, & Alfano, 2014).

Recent research has begun to examine alterations in RR in suicidality. Tsypes and colleagues examined RewP in a community sample of 7- to 11-year-old children with a parental history of suicide attempt and recent SI. Children with a parental history of suicide attempt exhibited an enhanced RewP to rewards vs. losses compared to controls (Tsypes, Owens, Hajcak, & Gibb, 2017). Conversely, children who endorsed recent thoughts of death or SI exhibited a reduced RewP to rewards vs. losses (Tsypes, Owens, & Gibb, 2019). Supporting BAS alterations in suicidality, a recent study indicated that young children with suicidality exhibited greater fun-seeking compared to depressed children without suicidality (Luby, Whalen, Tillman, & Barch, 2019). Yet, research in adults suggests mixed results regarding the link between BAS and suicidality, and that it may depend on interactions with other variables, including behavioral inhibition (e.g., Rasmussen, Elliott, & O’Connor, 2012).

Research to date suggests both self-report and neurophysiological measures of RR may be associated with suicidality in youth, but several key questions remain. First, much of the literature has focused on children or adults. It remains unclear how alterations in RR might emerge in adolescence—a period characterized by marked developmental change in frontostriatal circuity (Cohen et al., 2010; Hartley & Somerville, 2015). Further, research on RewP has relied on community samples with low rates of clinical depression (Tsypes et al., 2017, 2019). Given evidence of reduced RR in depression, research is needed to evaluate the extent to which RR measures can differentiate clinical populations with and without suicidality. Finally, despite the potential of neuroscience methods to predict future behavior, to demonstrate clinical utility, research is needed to examine the extent to which neurophysiological measures improve prediction beyond self-report measures of related constructs.

We examined self-report (i.e., BAS subscales) and neurophysiological (i.e., RewP to rewards and losses) measures of RR in clinically depressed adolescents with and without active suicidality, and tested whether these measures improve classification beyond demographic and clinical indicators. We hypothesized that both self-report and neurophysiological measures of RR would account for unique variance beyond demographic/clinical measures, but neurophysiological measures may perform better than self-report. Given discrepancies in the direction of effects observed for RewP in children of suicide attempters vs. children with SI (Tsypes et al., 2017; 2019), we hypothesized that suicidality in depressed adolescents may be characterized by hypo- or hyper-responsiveness to rewards. We tested the ability to classify active suicidality, rather than passive thoughts of death, as thoughts of death likely reflect a distinct process from active SI (Yoder, Whitbeck, & Hoyt, 2008) and are highly prevalent in clinical samples (Scott et al., 2012).

Materials and Methods

Participants

Participants were currently depressed adolescents (14-18 years old) recruited through pediatric and mental health clinics for an ongoing treatment study. Participants were recruited at two sites (Pennsylvania State College of Medicine and Vanderbilt University) because the lab relocated during the study. The majority of the sample (91.4%) was recruited at Vanderbilt. Exclusion criteria included antipsychotic medications/mood stabilizers, substance use disorders requiring treatment, intellectual or developmental disabilities, psychosis, or mania. Changes in treatment in the past 30 days resulted in delayed intake assessments until treatment stabilized. A total of 60 eligible participants were enrolled. Two participants withdrew prior to the EEG, leaving an analyzed sample of 58; M_age=15.90 (SD=1.48); 67.2% female; 87.9% White/Caucasian, 5.2% Hispanic/Latino, 5.2% Asian, 3.4% Black/African American, and 3.4% identified as mixed/other race.

Procedure

Procedures were approved by the Institutional Review Boards at both universities. Informed consent and assent were obtained from parents (for minors) and participants according to the Declaration of Helsinki. Participants completed interviews and questionnaires at an intake assessments, followed by the EEG assessment. Procedures and equipment were identical at both sites.

Measures

Diagnostic interviews.

Participants were interviewed using the DSM-V version of the Kiddie Schedule for Affective Disorders and Schizophrenia (KSADS; Kaufman et al., 2016). Depression modules and screening items for externalizing and anxiety disorders were also administered separately to parents of minors. Interviews were administered by masters-level research assistants or graduate students under the supervision of a licensed psychologist (AK). Summary ratings were derived based on a combination of adolescent and parent report (when available). A second interviewer coded audio recordings of a subset of interviews (n=7). Interrater reliability of MDD and persistent depressive disorder (PDD) diagnoses (κ=1.0) and suicidality items (ICCs=1.0) was excellent. Participants endorsing at least subthreshold active SI or suicidal acts were considered to have active suicidality. Subthreshold ratings were included in to capture potential underreporting of SI severity and because detection of true subthreshold levels is likely to be clinically meaningful given the potential risks of missed suicidality. Subthreshold active SI includes infrequent (i.e., less than once per month) or vague thoughts of suicide. Subthreshold suicidal acts include having a suicide plan but not following through or preparing without intent to die (e.g., holding pills in their hand). Due to the categorical nature of these 2 ratings and consistent with prior work (Tsypes et al., 2019), KSADS suicidality ratings were dichotomized into presence or absence of active suicidality.

Self-report measures.

Depressive symptoms were assessed dimensionally using the self-report Mood and Feelings Questionnaire (MFQ), a 33-item, well-validated measure of depressive symptoms in the past 2 weeks in youth (Angold et al., 1995; Messer et al., 1995). Participants also completed the Snaith-Hamilton Pleasure Scale (SHAPS; Snaith et al., 1995), a reliable measure of anhedonia, which is thought to be linked to RR. SHAPS is a 14-item measure of the experience of pleasure in the past few days (e.g., “I would be able to enjoy my favorite meal”) that is reverse scored, such that higher scores indicate more anhedonia (Snaith et al., 1995; Franken, Rassin, & Muris, 2007). Participants also completed the BIS/BAS, a 24-item measure of trait-like individual differences in biobehavioral motivational systems comprised of three BAS subscales: drive, fun-seeking, and RR (Carver & White, 1994). Drive assesses determination in pursuit of goals (e.g., “I go out of my way to get things I want”), fun-seeking measures desire to obtain new rewards/sensations (e.g., “I crave excitement and new sensations”), and RR measures positive emotions in response to rewards (e.g., “When good things happen to me, it affects me strongly”). The BIS/BAS has good test-retest reliability, internal consistency, and predictive power (Carver & White, 1994). MFQ, SHAPS, BAS drive, BAS fun-seeking, and BAS RR all had acceptable to excellent internal consistency (Cronbach’s α=.94, .79, .90, .78, and .76, respectively).

Monetary reward task.

The monetary reward task was adapted from an established ERP task (Bress, Meyer, & Proudfit, 2015; Kujawa et al., 2018). At the beginning of each trial, participants were presented with two doors and instructed to select a door that might have a prize behind it. Next, a fixation mark (+) appeared for 1000 ms, followed by feedback presented for 1500 ms. Participants were told they could win $0.50, indicated by a green “↑,” or lose $0.25, indicated by a red “↓”, on each trial. Next, a fixation mark appeared for 1000 ms and was followed by the message “Click for the next round”, which remained on screen until the participant responded and the next trial began. The task included 30 win and 30 loss trials in a random order. Participants completed 2 practice trials to familiarize them with the feedback cues. Participants were informed they could win up to $5, and all participants received the full $5.

EEG Data Collection and Processing

Continuous EEG was recorded using a 32-electrode BrainProducts actiCHAmp system (Munich, Germany). Facial electrodes to measure electrooculogram (EOG) from eye movements and blinks were attached above and below one eye and on either side of each eye. Per the BrainProducts system design, online data acquisition was referenced to a scalp electrode, and then re-referenced to the linked mastoids (TP9/TP10) offline. Impedances were lowered below 30 kΩ. Data were sampled at 1000 Hz.

Data processing was performed using BrainVision Analyzer (Brain Products, Munich, Germany). Data were band-pass filtered with cut-offs of 0.1 and 30 Hz, and segmented from −200 ms to 800 ms after feedback. Ocular correction was conducted using Gratton’s algorithm (Gratton, Coles, & Donchin, 1983). Semiautomatic artifact rejection criteria were a voltage step greater than 50 μV between sample points, maximum voltage difference of 175 μV within trials, a minimal allowed amplitude of −200 μV and maximal allowed amplitude of 200 μV, and minimum voltage difference of 0.5 μV within 100 ms intervals. Data were visually inspected to remove remaining artifacts. Participants had a minimum of 26 artifact-free trials per condition at Cz (M=29.86-29.91). Segments were averaged separately for each condition and baseline corrected to the −200 ms window. Consistent with prior work, RewP was scored 250-350 ms after feedback at Cz (Luking et al., 2017; Nelson, Perlman, Klein, Kotov, & Hajcak, 2016; Pegg et al., 2019). Given evidence of specific links between suicide risk and RewP to loss (Tsypes et al., 2017; 2019), the RewP to rewards and losses were examined as unique predictors of suicidality. For comparison, the RewP to rewards minus losses difference score was also examined (see Table 1 and Supporting Information). RewP to win and loss feedback had good to excellent split-half reliability at Cz (Spearman-Brown coefficients=.89 and .91, respectively).

Table 1.

Bivariate correlations between study variables.

Variables	M (SD)	1	2	3	4	5	6	7	8	9	10	11
1. Age	15.90 (1.48)	--
2. Gender†	--	−.27*	--
3. MFQ	33.48 (15.07)	.10	.06	--
4. SHAPS	3.09 (2.97)	.09	−.32*	.22	--
5. BAS Drive	10.05 (2.77)	−.07	.01	.18	−.14	--
6. BAS Fun-seeking	10.62 (2.59)	.01	.08	−.02	−.21	.32*	--
7. BAS RR	14.90 (2.55)	−.03	.09	−.11	−.50***	.36**	−.33*	--
8. RewP win	14.71 (7.17)	.20	−.03	.14	.14	−.12	−.13	−.13	--
9. RewP loss	12.06 (6.64)	.13	−.09	.15	.11	−.10	−.24	−.10	.89***	--
10. ΔRewP	2.64 (3.33)	.18	.11	−.01	.09	−.05	.20	−.09	.39**	−.09	--
11. Active suicidality†	--	−.18	−.01	.39**	.38**	.04	.14	−.22	.07	−.10	.35**	--

Note:

^*p < .050

^**p < .010

^***p < .001

^†Biserial correlations; BAS=Behavioral Activation System; MFQ=mood and feelings questionnaire; RewP=reward positivity; RR=reward responsiveness; SHAPS=Snaith-Hamilton Pleasure Scale; ΔRewP = RewP to wins minus RewP to losses

Data Analysis

Bivariate/biserial correlations were conducted to examine descriptive associations between study variables. To examine demographic, clinical, self-report, and RR variables that differentiate depressed adolescents with and without active suicidality, a hierarchical binary logistic regression was conducted. Demographic variables (age and gender) and depressive and anhedonic symptoms (MFQ and SHAPS) were entered in the first block. In the second block, BAS subscales were entered to evaluate the unique variance accounted for by self-report RR measures. In the third block, RewP to win and loss were entered to evaluate the unique variance accounted for by neurophysiological RR measures above and beyond self-report RR. Comparable results were observed when RewP was examined as a difference score, tested using a mixed-design ANOVA, and removing covariates from the logistic regression (see Supporting Information).

Results

Participant Characteristics

Most participants met criteria for both current MDD and PDD (62.1%), indicating high rates of chronic depression; 24.1% met for MDD only; and 13.8% for PDD only. Average age of onset for the first depressive episode was 13.66 years (SD=2.55). Average duration of the current MDD/PDD episode was 116.18 weeks (SD=126.95). Common comorbid disorders included generalized anxiety disorder (32.8%), social anxiety disorder (36.2%), and attention deficit hyperactivity disorder (15.5%). Most of the sample endorsed subthreshold or threshold current passive thoughts of death (77.6%; KSADS item 4a). Rates of current SI (39.7%; KSADS item 4b) and current suicidal acts (5.2%; KSADS item 4c) were lower.

Descriptive Statistics and Correlations

Descriptive statistics and correlations between variables are presented in Table 1. Male participants tended to be older than female participants and had higher anhedonia. Self-reported RR subscales were associated with each other. Self-reported BAS RR was negatively correlated with anhedonia. RewP difference was positively correlated with suicidality. Active suicidality was also associated with greater depressive and anhedonic symptoms.

Logistic Regression Analysis

Results of the hierarchical logistic regression are reported in Table 2. The base model (i.e., age, gender, depressive symptoms, and anhedonia) explained 37.8% of the variance and correctly classified 74.1% of participants. In the second block in which the BAS subscales were added, the model explained 45.1% of the variance and correctly classified 81.0% of participants. The addition of BAS subscales explained 7.3% of variance. Finally, in the third block in which RewP was added, the final model explained 64.5% of the variance and correctly classified 82.8% of participants. The sensitivity and specificity of the final model were 79.2% and 85.3%, respectively. The addition of RewP to win/loss explained 19.4% of variance. Rates of active suicidality was higher in participants who were younger and had higher depressive and anhedonic symptoms. Further, active suicidality was characterized by a relatively more positive RewP to win and a relatively more negative RewP to loss (Figures 1 and 2).

Table 2.

Hierarchical binary logistic regression of demographic, clinical, and self-report and neurophysiological reward responsiveness variables classifying clinically depressed adolescents with and without active suicidality

Variables	Block 1				Block 2				Block 3
	OR	B (SE)	Wald	p	OR	B (SE)	Wald	p	OR	B (SE)	Wald	p
Block 1 – Demographics
Age	0.65	−0.43 (0.24)	3.26	.071	0.62	−0.48 (0.25)	3.60	.058	0.44	−0.83 (0.32)	6.73	.009
Gender (male)	1.09	0.08 (0.79)	0.01	.916	0.82	−0.21 (0.81)	0.06	.801	0.38	−0.98 (0.95)	1.07	.301
MFQ	1.06	0.06 (0.02)	5.92	.015	1.07	0.07 (0.03)	5.97	.015	1.11	0.11 (0.04)	8.21	.004
SHAPS	1.32	0.27 (0.12)	5.07	.024	1.31	0.27 (0.14)	3.93	.047	1.44	0.36 (0.16)	4.91	.027
Block 2 – Self-report RR
BAS Drive					0.96	−0.05 (0.14)	0.10	.751	1.08	0.08 (0.17)	0.21	.650
BAS Fun-seeking					1.34	0.29 (0.15)	3.89	.049	1.25	0.23 (0.17)	1.67	.196
BAS RR					0.88	−0.13 (0.17)	0.53	.465	0.82	−0.20 (0.22)	0.87	.352
Block 3 – Neural RR
RewP to win									1.71	0.54 (0.20)	7.13	.008
RewP to loss									0.51	−0.67 (0.24)	7.72	.005
	X 2			p	X 2			p	X 2			p
Model Chi-square	19.10			.001	23.64			.001	37.83			<.001
Block Chi-square	19.10			.001	4.54			.208	14.19			.001
	R2				R2				R2
Nagelkerke’s R2	.378				.451				.645
Change in Nagelkerke’s R2					.073				.194

Note: MFQ=Mood and Feelings Questionnaire; RewP=reward positivity; RR=reward responsiveness; SHAPS=Snaith-Hamilton Pleasure Scale

Figure 1.

ERP waveforms at Cz and scalp distributions 250-350 ms after feedback depicting responses to monetary reward and loss feedback in depressed adolescents with (top) and without (bottom) active suicidality. Scalp distributions depict the win minus loss difference. (32-channel montage with linked mastoid reference.) Note: ERP=event-related potential; RewP=reward positivity

Figure 2.

Scatterplots of RewP to win and loss (in microvolts) in each group group. The active suicidality group exhibited a relatively more positive RewP to wins and relatively more negative RewP to losses compared to depressed adolescents without suicidality. Note: RewP=reward positivity

Discussion

We examined whether self-report and neurophysiological RR differentiated clinically depressed adolescents with and without active suicidality. Active suicidality was characterized by heightened RR at both levels, and neurophysiological measures accounted for unique variance beyond self-reported RR. Although bivariate correlations between suicidality and RewP to each condition were not significant, active suicidality was characterized by a relatively more positive RewP to rewards and a relatively more negative RewP to losses when accounting for the other condition. That is, the active suicidality group exhibited greater differentiation in the neural response to rewards vs. losses compared to depressed adolescents without suicidality. Results suggest that reliable and accessible neural measures, such as EEG, may improve prediction of suicidality risk and provide support for recommendations to incorporate objective measures in suicide research to better understand combinations of factors that best characterize risk (Glenn et al., 2017; Nock et al., 2010).

Although depression has been characterized by reduced RR (e.g., Bress et al., 2015; McFarland et al., 2006), depressed adolescents with active suicidality demonstrated enhanced RR compared to participants without suicidality. Specifically, those with active suicidality showed a relatively more positive RewP to rewards and more negative RewP to losses, similar to prior findings of greater differentiation in RewP responses to feedback in children of suicide attempters (Tsypes et al., 2017), and reported greater fun-seeking. Interestingly, the magnitude of RewP in the active suicidality group was comparable to that previously observed in healthy adolescents (Kujawa et al., 2018), whereas depressed adolescents without active suicidality tended to exhibit blunted RewP. Suicidality in the context of depression may not be associated with abnormally elevated RR compared to adolescents in general, but instead may be characterized by enhanced RR compared to the pattern typically observed in depression. One possibility for consideration in future research is that depressed adolescents with intact desire for new rewards (reflected by BAS fun-seeking) and neural processing of rewards (reflected by RewP) may experience hopelessness, as they may wish to obtain rewards in an environment where rewards are limited due to stress and/or loss. From an evolutionary perspective, downregulation of the reward system may be adaptive to persevere through depression (e.g., Nusslock & Miller, 2016). It should also be noted that RewP reflects activation of broad neural systems (Carlson et al., 2011; Foti et al., 2014; Foti et al., 2011), and heterogeneous disruptions in these systems may manifest as an enhanced RewP. Here, enhanced RewP could reflect atypical decision-making underlying suicidal thoughts and actions. Consistent with this possibility, BAS fun-seeking also differentiated active suicidality and is related to impulsivity and risk-taking behavior (Franken & Muris, 2006; Leone & Russo, 2009; Voigt et al., 2009).

Findings provide preliminary support for the utility of assessing BAS in clinical practice, as BAS fun-seeking differentiated active suicidality prior to accounting for neurophysiological measures. These results are consistent with previous findings of associations between BAS fun-seeking and suicidality in children (Luby et al., 2019), possibly reflecting impulsivity or impaired decision-making (Jollant et al., 2005). Compared to neurophysiological measures, self-reported RR is more easily assessed in practice, but results indicate that integration of neurophysiological assessments in the future could further enhance practice.

Although adolescents with active suicidality reported greater desire to obtain new rewards and showed an enhanced RewP, suicidality was also characterized by greater anhedonic symptoms. Our measure of anhedonia assessed experiences of pleasure in the past few days, rather than motivation and drive (Treadway & Zald, 2011), and was not correlated with RewP or BAS fun-seeking. Depressed adolescents with active suicidality may exhibit low pleasure but enhanced trait-like desire to obtain new rewards and heightened neural RR. These results challenge conceptualizations of RewP as reflecting anhedonia, and suggest that research on personalized treatment for depression characterized by relatively reduced vs. intact or enhanced RR is needed.

Several limitations and future directions should be noted. Longitudinal research is needed to test RR as a prospective predictor of suicidality. The sample was relatively small, yet unique because all participants had current clinical depression with early onset and high chronicity on average. Although the sample size is consistent with or larger than prior studies of neural measures and suicidality (e.g., Tsypes et al., 2019), large samples will be needed to develop norms of RR and translate research to practice.

Conclusion

The current study is the first to examine multiple measures of RR and active suicidality in depressed adolescents, providing support for the clinical utility of neurophysiological measures in identifying adolescents with suicidality.

Data Availability Statement

Data supporting these findings are available from the corresponding author by request.