Neural responsiveness to reward and suicidal ideation in social anxiety and major depression before and after psychotherapy

Abstract

Suicidality is prevalent in Social Anxiety Disorder (SAD) and Major Depressive Disorder (MDD). Limited data indicate the reward positivity (RewP), a neurophysiological index of reward responsivity, and subjective capacity for pleasure may serve as brain and behavioral assays for suicide risk though this has yet to be examined in SAD or MDD in the context of psychotherapy. Therefore, the current study tested whether suicidal ideation (SI) relates to RewP and subjective capacity for anticipatory and consummatory pleasure at baseline and whether Cognitive Behavioral Therapy (CBT) impacts these measures. Participants with SAD (n=55) or MDD (n=54) completed a monetary reward task (gains vs. losses) during electroencephalogram (EEG) before being randomized to CBT or supportive therapy (ST), a comparator common factors arm. EEG and SI data were collected at baseline, mid-treatment, and post-treatment; capacity for pleasure was collected at baseline and post-treatment. Baseline results showed participants with SAD or MDD were comparable in SI, RewP, and capacity for pleasure. When controlling for symptom severity, SI negatively corresponded with RewP following gains and SI positively corresponded with RewP following losses at baseline. Yet, SI did not relate to subjective capacity for pleasure. Evidence of a distinct SI-RewP association suggests RewP may serve as a transdiagnositic brain-based marker of SI. Treatment outcome revealed that among participants with SI at baseline, SI significantly decreased regardless of treatment arm; also, consummatory, but not anticipatory, pleasure increased across participants regardless of treatment arm. RewP was stable following treatment, which has been reported in other clinical trial studies.

1. Introduction

Social anxiety disorder (SAD) and major depressive disorder (MDD) are among the most frequent and debilitating disorders in the U.S. (Kessler et al., 2005) where SAD is characterized by fear and avoidance of social situations and MDD is characterized by pervasive sadness and/or anhedonia (American Psychiatric Association, 2013). Yet, despite features that distinguish these disorders, studies indicate suicidal ideation (e.g., thoughts or wishes about ending own life; Posner et al., 2007) is common in both SAD and MDD. For example, in a National Comorbidity Survey study, 34.8% of individuals with SAD reported suicidal ideation (‘SI’) and associations between SAD and SI have been observed even when controlling for depression (Cox et al., 1994; Sareen et al., 2005). With regard to MDD, a recent meta-analysis showed the overall prevalence of SI was 37.7% (Cai et al., 2021). Collectively, findings show SI is prevalent in SAD and MDD suggesting features of these disorders are transdiagnostic.

As SI plays a robust role in suicide death (Hubers et al., 2018) and over 700,000 people die by suicide each year (World Health Statistics 2021), it is important to develop brain and behavioral assays that have the potential to identify individuals at risk for suicide, and/or serve as targets for intervention. Promising assays include those that involve response to positive experiences. For example, reduced capacity to experience pleasure (i.e., anhedonia; Snaith, 1993) is higher in individuals with SI compared to those without SI and the relationship between anhedonia and SI is significant even when controlling for depression and other psychiatric illness (Ducasse et al., 2018). Thus, anhedonia is a potential behavioral marker of SI.

As for a brain-based marker of SI, accruing findings indicate the reward positivity (RewP) is a candidate marker. Specifically, RewP is a frontocentral event-related potential (ERP) component derived from electroencephalograph (EEG) that occurs approximately 250-350 milliseconds (ms) following receipt of a reward and is proposed to be a measure of reward sensitivity (i.e., processing of reward relative to non-reward or loss) (Proudfit, 2015). RewP also has high clinical utility as EEG is more cost effective and portable than other neuroscience methods (e.g., functional magnetic resonance imaging). Yet, the RewP has largely been used in parallel lines of research such that studies of SI and internalizing psychopathologies have generally been investigated separately resulting in gaps in the literature.

To date, evaluation of SI-RewP relationships has predominately involved youth and findings are mixed. Specifically, in a study involving children, those with SI exhibited less ΔRewP (i.e., less amplitude following monetary gain) based on a difference score (gain minus loss) than children without SI even after controlling for symptoms (e.g., depression, anxiety) indicating the SI-RewP association was not significantly explained by symptom severity (Tsypes et al., 2019). In children of parents with a history of a suicide attempt, more reactivity to monetary loss minus gain was observed in those with a prior attempt compared to children of parents with no history of an attempt (Tsypes et al., 2017). In other words, the response to loss was driven by more feedback negativity (FN) or feedback-related negativity (FRN) (Hajcak et al., 2006; Yeung et al., 2005). Hereafter, we refer to the ERP following monetary loss as FN. Results suggest heightened neural response to loss may represent a distinctive pathway for suicide risk (Tsypes et al., 2017). In depressed adolescents, binary logistic regression analysis showed more raw RewP following monetary gain and less raw FN following monetary loss, evaluated separately, corresponded with adolescents with active suicidality but not non-suicidal adolescents (Pegg et al., 2020) and in a study that compared adults with and without a history of suicide attempt, raw RewP did not differ between groups (Tsypes et al., 2021). Methodological differences that may contribute to inconsistent results include samples of ‘suicidal ideators’ versus ‘suicide attempters’ as differences between these populations have been detected (e.g., increased pain persistence; Klonsky et al., 2018). However, despite inconsistencies, the majority of studies show SI is sensitive to the ERP following monetary gains or losses indicating it is a promising brain marker of SI.

Outside of research on suicidality, studies of internalizing psychopathologies have shown less RewP corresponds with greater depression severity (Bress et al., 2015; Foti & Hajcak, 2009; Liu et al., 2014; Nelson & Jarcho, 2021) and anxiety severity (Gu et al., 2020; Kessel et al., 2015) though the relationship between RewP and anxiety has not been as consistently observed as in depression (Bress et al., 2015; Burkhouse et al., 2017). The more consistent RewP finding in depression may relate to anhedonia being a core symptom of depression (American Psychiatric Association, 2013; Pizzagalli, 2014). Because we are not aware of a study that examined RewP-SI associations in adults with SAD or MDD, it is unclear as to whether RewP distinctly corresponds with SI (i.e., is not explained by symptom severity) in these clinical populations.

With regard to the impact of treatment on SI, standard Cognitive Behavioral Therapy (CBT), the ‘gold standard’ psychotherapy for SAD, MDD, and other internalizing psychopathologies (David et al., 2018) has been shown to reduce SI. For example, a meta-analysis revealed in-person CBT, but not internet-delivered CBT, significantly reduced SI and suicidal behavior in adults with depression and other psychiatric illness (Leavey & Hawkins, 2017). Less is known about the effects of CBT on SI in anxiety disorders. A study comprising treatment-seeking adults in an anxiety clinic showed SI significantly decreased following CBT across anxiety disorders (e.g., SAD, generalized anxiety disorder, panic disorder) though when evaluating SAD specifically, improvement in SI did not reach significance (Brown et al., 2018). Altogether, CBT is associated with reductions in SI though there have been few studies that have examined the impact of standard CBT on SI in individuals with SAD or MDD in an outpatient setting.

As to whether RewP improves with CBT, limited data indicate RewP is stable following standard treatment. Specifically, a study involving adults with clinical anxiety or depression showed RewP based on a difference score (monetary gain minus loss) did not significantly change following CBT or a selective serotonin reuptake inhibitor (SSRI) (Burkhouse et al., 2018). Also, in a study comprising adolescents with MDD, no pre-to-post CBT change in raw RewP following monetary gain or raw FN following loss, when evaluated separately, was observed (Webb et al., 2021). Yet, in young children with depression, RewP based on a residualized score (gain partialing out effect of loss) was found to increase following a parent-child interaction therapy that included an emotional development component (Barch et al., 2020). Thus, age and/or type of psychotherapy may factor into whether RewP is malleable.

Regarding anhedonia, CBT has been shown to improve subjective anhedonia in individuals with depression (Alsayednasser et al., 2022; Hanuka et al., 2022). We are not aware of a study that examined anhedonia following CBT in SAD even though anhedonia is also observed in SAD and other anxiety disorders (Brown et al., 1998; Kashdan, 2007; Kashdan et al., 2011). Since anhedonia is expected to cut across diagnostic boundaries, we would expect CBT to improve anhedonia regardless of principal diagnosis.

Therefore, the objective of the current study was to expand on previous work to fill important gaps in the literature. First, we examined SI and RewP in treatment-seeking participants with SAD or MDD. Because we are not aware of an SI study that directly compared SAD with MDD, we did not have a specific hypothesis as to whether diagnostic groups would significantly differ in level of SI.

Second, since anhedonia may serve as a behavioral marker of SI (Ducasse et al., 2018), we evaluated whether subjective capacity for pleasure differed between SAD and MDD and tested for SI-pleasure and RewP-pleasure relationships. We expected less subjective capacity for pleasure would correspond with more SI and less RewP. We use the term ‘pleasure’ instead of anhedonia as this study did not involve a comparator healthy control group. Therefore, it was not possible to verify that participants with SAD or MDD experienced anhedonia. As previous work indicates anhedonia is transdiagostic (Brown et al., 1998; Kashdan, 2007; Kashdan et al., 2011; Pizzagalli, 2014), we expected SAD and MDD groups would not differ in capacity for pleasure, which can be parsed into anticipatory (e.g., prediction of pleasure from future reward) and consummatory (e.g., experience of pleasure in the moment) components (Berridge & Robinson, 2003; Gard et al., 2006). Since RewP represents consummatory pleasure (i.e., response to feedback), only self-reported consummatory pleasure was used to examine an association with RewP.

Third, we tested if SI corresponded with RewP when controlling for social anxiety and depression symptoms. Due to mixed RewP findings (Tsypes et al., 2019; Tsypes et al., 2017; Pegg et al., 2020), we did not have a specific hypothesis as to the direction of the SI-RewP association at baseline. Lastly, we tested whether CBT improved SI, capacity for pleasure, and RewP. For comprehensiveness, we examined symptom severity before and after treatment even though social anxiety and depression were not the primary focus of the study. To determine the extent to which the ‘active ingredients’ in CBT may contribute to improvement, CBT was compared with Supportive Therapy (ST), a common factors psychotherapy. We expected symptom severity, including SI and capacity for pleasure would improve to a greater extent in participants randomized to CBT than ST. In contrast, we hypothesized RewP would be stable regardless of treatment arm.

2. Method

2.1. Participants and Procedure

This is a secondary analysis from data collected in a clinical trial that examined neural predictors and mechanisms of psychotherapy in treatment-seeking individuals with SAD or MDD (ClinicalTrials.gov Identifier: NCT03175068). A portion of this data has been published (Feurer et al., 2021; Sheena et al., 2021); however, the objectives of the prior studies differ from the current one. One-hundred nine participants (SAD n=55; MDD n=54) completed EEG during a validated monetary reward task before being randomized to CBT or ST. In addition to EEG before treatment, EEG was collected at mid-treatment (Week 6) and post-treatment (Week 12) to evaluate possible differences in the trajectory of RewP-related outcome.

All participants were required to be between 18 and 65 years old; exclusionary criteria included (a) current active suicidal ideation (i.e., endorse plan or intent) or current self-harming behavior, and (b) use of psychotropic medication in the last 6 weeks. While comorbidity was permitted, individuals with SAD could not have comorbid MDD and vice versa. See Supplementary Materials for recruitment details and all exclusionary criteria.

All study procedures were conducted at the University of Illinois at Chicago and approved by the university’s Institutional Review Board and complied with the Helsinki Declaration. All participants were compensated for their time.

Regarding data collection, EEG was not collected for all participants at all time points due to premature discontinuation or COVID-related shutdowns. Consequently, of the 109 participants, those who had viable data comprised 108 participants at baseline, 84 at mid-treatment (Week 6), and 80 at post-treatment (Week 12). Regarding self-reported capacity for pleasure, two participants did not complete the measure at baseline and one participant did not complete the measure following treatment due to human error.

2.2. Clinical Measures

After obtaining consent, participants completed the Structured Clinical Interview for DSM-5 (SCID-5; First et al., 2015) and interviewer-based Liebowitz Social Anxiety Scale (LSAS) (Liebowitz, 1987) and Hamilton Depression Rating Scale (HAMD) (Hamilton, 1960) to assess severity of social anxiety and depression, respectively. All interviewer-based measures were conducted by a trained staff member blinded to treatment arm. The LSAS consists of 24 items that assess anxiety/fear and avoidance regarding social and performance situations whereas the HAMD consists of 17 items that assess depressive and somatic systems. To evaluate inter-rater reliability, percent agreement based on 10% of randomly selected measures between two trained raters was calculated; results showed percent agreement was 89% for LSAS and 84% for HAMD. Social anxiety and depression were evaluated with these measures every two weeks (i.e., 7 time points) to evaluate trajectory of change over the course of treatment.

SI was examined with the self-report Inventory of Depression and Anxiety Symptoms – Second Version (IDAS-II) (Watson et al., 2012) suicidality subscale (IDAS-SS). The subscale involves a Likert-type scale (1 = ‘Not at all’, 5 = ‘Extremely’) and includes six items in total comprising suicidal thoughts (e.g., ‘I had thoughts of suicide’, ‘I thought that the world would be better off without me’) and thoughts of self-harm (e.g., ‘I thought about hurting myself’). The IDAS-SS has been used in outpatient (Hausman et al., 2020) and community (Bauer et al., 2019) samples and has been shown to have good convergent and discriminant validity (Watson et al., 2012). Higher scores reflect greater SI. The IDAS-SS had acceptable internal consistency at baseline (Cronbach’s alpha = 0.69). The IDAS-SS was also collected at mid-treatment (Week 6), and after treatment (Week 12).

Capacity for pleasure was evaluated with the Temporal Experience of Pleasure Scale (TEPS) (Gard et al., 2006), which consists of 18 items involving a Likert-type scale (1 = ‘very false for me’, 6 = ‘very true for me’) where 10 items assess anticipatory pleasure (TEPS-A) and 8 items assess consummatory pleasure (TEPS-C). Higher scores reflect greater pleasure. TEPS-A had acceptable internal consistency at baseline (Cronbach’s alpha = 0.76) whereas TEPS-C did not (Cronbach’s alpha = 0.64). The TEPS was also collected after treatment (Week 12).

2.3. Reward Task

Participants completed a validated computerized guessing task consisting of 40 trials (Burkhouse et al., 2017, 2018; Proudfit, 2015) at baseline, Week 6, and Week 12. During each trial, participants chose one of two doors that were shown side by side on a computer monitor until the participant made a decision. After a door was chosen, a fixation mark would appear for 1000 ms followed by feedback which was displayed on the screen for 2000 ms. Feedback was either a green arrow pointing upwards that signified a gain (i.e., ‘win’) of $0.50, or a red, downward pointing error that indicated a loss of $0.25. After feedback was presented, another fixation mark was presented for 1500 ms and participants were shown a screen displaying, ‘Click for the next round’, until the participant responded. In total, participants were randomly presented with 20 gain feedback trials and 20 loss feedback trials. The participants received winnings from the task in the form of money after completing the task. The winnings were collected at the time participants were compensated for all the parts of the study they completed.

2.4. EEG Data Acquisition and Preprocessing

Continuous EEG was recorded during the task using an elastic cap and the ActiveTwo BioSemi system (BioSemi, Amsterdam, Netherlands). Thirty-four standard electrode sites were used (including Fz and Iz), based on the 10/20 system, along with one electrode on each mastoid. Four facial electrodes were placed on participants for electrooculogram recordings generated from eye blinks and eye movements (two electrodes were placed 1 cm above and below the right eye to measure vertical eye blinks and movements, and two electrodes were placed 1 cm beyond the outer edge of each eye to measure horizontal eye blinks and movements). The data were digitized at 24-bit resolution with a sampling rate of 1024 Hz. All offline EEG processing was done using EEGLAB (Delorme & Makeig, 2004) and ERPLAB (Lopez-Calderon & Luck, 2014) in MatLab (MATLAB Version 9.3.0.713579 (R2027b), 2017). Offline, raw EEG were resampled at 250 Hz using a mild antialiasing filter. To remove line noise, data were cleanlined by adaptively estimating sine wave amplitude in a sliding window using the PREP eeglab plugin with default values (Bigdely-Shamlo et al., 2015). This function first applies a 1.0 Hz FIR zero-phase highpass filter before estimating line noise and second removes this noise from the unfiltered dataset. This approach preserves phase information by removing line noise without filtering (Bigdely-Shamlo et al., 2015). Next, a similar approach was adapted to prepare data for independent component analysis (ICA) used to remove muscle and electrooculogram artifacts. First, unfiltered data were saved. Second, to remove electrode drift, data were filtered using a 1.0 Hz FIR zero-phase highpass (Winkler et al., 2015). Third, bad channels were removed separately for scalp and electrooculogram channels using an automated procedure in EEGLAB (clean_rawdata, (Kothe & Makeig, 2013). Scalp and external channels were marked as abnormal if they exceeded 8 or 16 standard deviations of the mean channel line noise or flatlined over 5 or 10 seconds, respectively. Scalp channels that correlated less than 0.7 with neighboring scalp channels for over 5 second were also marked. Any channel marked for over 10% of the total data length were removed. Fourth, large transient artifacts were removed using an automated procedure in ERPLAB (Lopez-Calderon & Luck, 2014). Segments of continuous data were removed if any scalp electrode exceeded 500 μV in a 500 ms time window that shifted every 250 ms across the full continuous data length. Finally, ICA was performed. After ICA, the unfiltered data were loaded, identical scalp and external channels identified previously were removed, and the resulting ICA weights were applied to this minimally preprocessed data. ICA components corresponding to ocular and muscular artifacts were removed using visual inspection. Data were then re-referenced to the average of the left and right mastoid and the formerly bad channels were interpolated.

To prepare data for ERP analysis, data were Butterworth bandpass filtered from 0.1 to 30 Hz, segmented into 1000 ms epochs for each condition following stimulus onset, and baseline corrected using a 200 ms pre-stimulus interval. Artifactual epochs were identified and removed using an automated procedure if any electrode exceeded a voltage threshold of 200 uV in a 200 ms time window that shifted every 100 ms across the full length of each time epoch. After preprocessing, two participants were removed for excessive artifact rejection (greater than 50% rejected trials). All remaining participants contained at least 14 acceptable trials per condition after artifact rejection (average trials rejected=0.30, SD=0.76) in line with prior research recommending a minimum of 10 trials per condition to measure reliable RewP amplitudes (Levinson et al., 2017). The remaining ERP segments were then averaged for each participant so that each electrode had a single waveform for each condition.

ERPs were averaged across gain and loss trials, and the RewP represented the mean amplitude 250-350 ms following feedback at FCz, where monetary gain minus loss based on a subtraction score (ΔRewP) was maximal. ΔRewP has been used in studies of SI and treatment outcome (Tsypes et al., 2017; Tsypes et al., 2019; Burkhouse, 2018), therefore, to aid in comparison across studies, we elected to report ΔRewP. However, since ΔRewP does not permit for a straightforward interpretation and tends to have lower psychometric properties than residualized RewP, we also report residualized RewP, which has been used in a previous psychotherapy study (Barch, Whalen 2020). Therefore, for significant ΔRewP findings, results for residualized RewP are also reported. Residualized RewP was calculated by regressing ERPs for gain trials onto ERPs for loss trials. Residualized FN was calculated by regressing ERPs for loss trials onto ERPs for gain trials. When referring to residualized RewP or FN, unstandardized residuals are reported.

2.5. Treatment

CBT consisted of manualized treatment for SAD or MDD, which included psychoeducation, strategies to alter maladaptive thoughts and behaviors (i.e., in vivo exposures for SAD, behavioral activation for MDD), and relapse prevention (Beck et al., 1979; Hope et al., 2006; Martell et al., 2010). ST comprised ‘common’ psychotherapy elements, including empathic responding and elicitation of affect and validation when appropriate (Markowitz et al., 2008). Both CBT and ST were delivered in 12 one-hour, weekly sessions by clinical psychologists trained in both CBT and ST to reduce potential therapist confounds. Thus, the same psychologist provided CBT and ST. Results concerning the psychologists’ fidelity and adherence to the treatment protocol indexed with the Cognitive Therapy Rating Scale (CTRS) (Young & Beck, 1980) indicated they followed CBT and ST protocols. See Supplementary Materials for details.

2.6. Analytic Strategy

All analysis was two-tailed with alpha level set to .05 and performed in the Statistical Package for the Social Sciences (SPSS; Chicago, IL version 27).

2.6.1. Baseline Characteristics

A series of independent t-tests and chi-square analyses were used to evaluate ΔRewP and potential clinical and demographic differences when comparing diagnostic (SAD/MDD) or SI (endorsed/did not endorse SI) groups. Similar analyses were performed to determine if there were differences between participants who discontinued treatment prematurely relative to those who did not. To evaluate a distinct relationship between SI severity and ΔRewP, a partial correlation was conducted controlling for social anxiety (LSAS) and depression (HAMD). The same analysis was performed to determine whether there were distinct SI-capacity for pleasure (TEPS-A, TEPS-C) relationships. For comprehensiveness, we also tested for a distinct ΔRewP-capacity for pleasure (TEPS-C) association.

2.6.2. Treatment outcomes

Since diagnosis of SAD or MDD was an inclusion criterion, LSAS and HAMD served as primary symptom measures for SAD and MDD, respectively. Consistent with our previous study (Feurer et al., 2021), the primary outcome measure was an internalizing symptom composite score created by summing patients’ LSAS and HAMD scores. Prior to summation, the LSAS and HAMD scores were transformed using the proportion of maximum scaling (POMS) method (i.e., LSAS and HAMD scores were divided by the observed maximum score) (Little et al., 2015) since standardized z scores are not recommended for longitudinal studies (Moeller, 2015). For comprehensiveness, we also examined diagnosis-specific outcomes. Here, the POMS-transformed LSAS was the outcome measure for the SAD group and the POMS-transformed HAMD was the outcome measure for the MDD group. In addition to these measures, SI (IDAS-SS), capacity for pleasure (TEPS), and ΔRewP were examined.

Linear mixed models (LMM) were used to evaluate treatment outcome for all measures except TEPS as it was only collected before and after treatment. For LMM, all main effects and interactions between treatment arm, groups (diagnostic group, SI group), and time were entered as fixed effects, and intercept and time were entered as random effects. For TEPS, a mixed analysis of variance (ANOVA) was performed where treatment arm and groups (diagnostic group, SI group) were entered as between-subjects factors and time (baseline, Week 12) was a within-subjects factor.

Since we expected ΔRewP to be stable across three timepoints (i.e., baseline, Week 6, Week 12), ΔRewP was also submitted to intraclass correlation coefficients (ICC) with corresponding 95% confidence intervals (CIs). ICCs reflect the ratio of between-subjects variance to total variance and are the appropriate metric for assessing test-retest reliability when observations are not independent (Shrout & Fleiss, 1979). ICCs range from −1 to 1 and are interpreted as follows: 0–0.2 as poor, 0.3–0.4 as fair, 0.5–0.6 as moderate, 0.7–0.8 as strong, and >0.8 as almost perfect (Sundvall et al., 2013). Specifically, the ICC model was a two-way mixed model with estimates for absolute agreement and 95% CIs. Absolute agreement between activations is an index of the degree to which scores are identical over time.

3. Results

3.1. Baseline characteristics

As anticipated, social anxiety (LSAS) was greater in participants diagnosed with SAD compared to those with MDD [t(107)=13.32, p<.001, d=2.55] and depression (HAMD) was greater in participants diagnosed with MDD relative to those with SAD [t(107)=6.42, p<.001, d=1.23]. Concerning demographic characteristics, average age across participants was 27.88 (SD=9.24) years and participants with SAD were younger than those with MDD [t(107)=3.44, p=.001, d=.66]. However, no other group effects emerged. See Supplementary Materials and Table 1 for details.

Table 1.

Clinical, neurophysiological, and demographic characteristics for diagnostic groups: Values reflect the mean and standard deviations are in parentheses

	SAD (n = 55)	MDD (n = 54)
Liebowitz Social Anxiety Score	82.76 (17.72)	37.02 (18.13)
Hamilton Depression Rating Score	8.64 (4.34)	14.21 (4.83)
IDAS-II Suicidality Subscale	7.54 (2.50)	8.05 (2.58)
TEPS-A	4.22 (.69)	3.94 (.84)
TEPS-C	4.51 (.69)	4.40 (.79)
ΔRewP at FCz (gain-loss)	4.01 (4.35)	3.71 (4.97)
Residualized RewP at FCz for gain	.09 (4.34)	−.10 (4.93)
Residualized FN at FCz for loss	−.35 (3.63)	.36 (4.29)
Age	25.00 (6.57)	30.81 (10.62)
Gender (%)	%	%
Female	65.5	70.4
Male	32.7	29.6
Not Reported	1.8	0.0
Ethnicity (% Hispanic/Latinx)	27.3	29.6
Racial Identity (%)	%	%
White	41.8	50.0
Black	18.2	7.4
Asian	23.6	14.8
Native American or Alaskan Native	1.8	0.0
Multi-Racial/Another Identity	10.9	26.0
Not Reported	3.6	1.9
Comorbid Diagnoses (%)	%	%
Generalized anxiety disorder	41.8	38.9
Insomnia	23.6	37.0
Hypersomnolence	12.7	14.8
Specific Phobia	9.1	5.6
Persistent Depressive Disorder	0.0	35.2
Panic Disorder	7.3	0.0
Posttraumatic Stress Disorder	1.8	5.6
Attention-Deficit/Hyperactivity Disorder	3.6	1.9
Adjustment Disorder	3.6	0.0
Alcohol Use Disorder (mild)	0.0	1.9
Binge Eating Disorder	1.8	0.0

Note. SAD=Social Anxiety Disorder; MDD=Major Depressive Disorder;

IDAS-II=Inventory of Depression and Anxiety Symptoms – Second Version; TEPS-A=Temporal Experience of Pleasure Scale, anticipatory pleasure score; TEPS-C=Temporal Experience of Pleasure Scale, consummatory pleasure score; FCz=frontocentral event-related potential component based on difference score; residualized RewP at FCz for gain (partialing out effect of loss); residualized FN at FCz for loss (partialing out effect of gain)

As expected, level of SI (IDAS-SS) did not differ between SAD and MDD groups [t(107)=1.04, p=.29, d=.20] nor did anticipatory pleasure (TEPS-A) [t(105)=1.82, p=.07, d=.35] or consummatory pleasure (TEPS-C) [t(105)=.78, p=.43, d=.15]. For ΔRewP, there was no difference between MDD and SAD groups at baseline [t(106)=.33, p=.73, d=.06]. See Table 1 for details.

At the time of study entry, 59 participants endorsed SI (SI group, 54.1%) and 50 participants did not (non-SI group, 45.9%). As hypothesized, the SI group had a higher baseline level of SI (IDAS-SS) than the non-SI group [t(107)=8.95, p<.001, d=1.72]. For distribution of SI (yes/no) groups, there was no difference between SAD and MDD groups [χ²(1)=3.36, p=.06, φ=.17] and SI/non-SI groups did not differ in level of social anxiety or depression (lowest p=.13, highest d=.28). When assessing pleasure (TEPS), the SI group exhibited less anticipatory pleasure than the non-SI group [t(105)=1.99, p=.048, d=.38] yet consummatory pleasure did not differ between groups [t(105)=1.90, p=.06, d=.36]. See Table 2 for SI/non-SI group details.

Table 2.

Clinical, neurophysiological, and demographic characteristics for participants with and without suicidal ideation collapsed across principal diagnosis: Values reflect the mean and standard deviations are in parentheses

	SI (n = 59)	Non-SI (n = 50)
Liebowitz Social Anxiety Score	60.44 (31.29)	59.70 (26.55)
Hamilton Depression Rating Score	12.13 (4.89)	10.60 (5.83)
IDAS-II Suicidality Subscale	9.32 (2.62)	6.00 (0.00)
TEPS-A	3.94 (.79)	4.24 (.75)
TEPS-C	4.33 (.76)	4.60 (.70)
ΔRewP at FCz (gain-loss)	2.86 (4.54)	5.02 (4.51)
Residualized RewP at FCz for gain	−.91 (4.51)	1.06 (4.55)
Residualized FN at FCz for loss	1.11 (4.13)	−1.29 (3.38)
Age	26.10 (8.19)	29.98 (10.03)
Gender (%)	%	%
Female	62.7	74.0
Male	35.6	26.0
Not Reported	1.7	0.0
Ethnicity (% Hispanic/Latinx)	33.9	22.0
Racial Identity (%)	%	%
White	47.5	44.0
Black	10.2	16.0
Asian	11.9	28.0
Native American or Alaskan Native	1.7	0.0
Multi-Racial/Another Identity	25.5	10.0
Not Reported	3.4	2.0
Comorbid Diagnoses (%)	%	%
Generalized Anxiety Disorder	47.5	32.0
Insomnia	28.8	32.0
Hypersomnolence	15.3	12.0
Specific Phobia	8.5	6.0
Persistent Depressive Disorder	23.7	10.0
Panic Disorder	3.4	4.0
Posttraumatic Stress Disorder	6.8	0.0
Attention-Deficit/Hyperactivity Disorder	1.7	4.0
Adjustment Disorder	1.7	2.0
Alcohol Use Disorder (mild)	1.7	0.0
Binge Eating Disorder	1.7	0.0

Note. SI=participants endorsed suicidal ideation at baseline (pre-treatment), Non-SI=participants did not endorse suicidal ideation at baseline (pre-treatment). IDAS-II=Inventory of Depression and Anxiety Symptoms – Second Version; TEPS-A=Temporal Experience of Pleasure Scale, anticipatory pleasure score; TEPS-C=Temporal Experience of Pleasure Scale, consummatory pleasure score; FCz=frontocentral event-related potential component based on difference score; residualized RewP at FCz for gain (partialing out effect of loss); residualized FN at FCz for loss (partialing out effect of gain)

Concerning demographic characteristics, participants in the SI group were younger than those in the non-SI group [t(107)=2.22, p=.02, d=.42]. However, there was no difference in gender distribution between groups [χ²(2)=2.15, p=.34, φ=.14], ethnicity [χ²(1)=1.88, p=.17, φ=.13], or race [χ²(6)=9.48, p=.14, φ=.29]. See Table 2 for details. The distribution of participants with SI or non-SI did not differ between those randomly assigned to treatment with CBT (SI group, n=32; non-SI group, n=26) versus ST (SI group, n=27; non-SI group, n=24) [χ²(1)=.05, p=.81, φ=.02].

When evaluating ΔRewP, it was found to be lower in the SI relative to non-SI group [t(106)=2.46, p=.01, d=.47]. See Figure 1 for depiction of ΔRewP and SI/non-SI groups. For residualized RewP, the SI group exhibited less response for gains compared to the non-SI group [t(106)=2.26, p=.02, d=.43]. For residualized FN following loss, the SI group showed less FN than the non-SI group [t(106)=3.27, p=.001, d=.63]. See Table 2 for details.

Figure 1.

Response-locked ERP waveform (FCz) following gain (i.e., win), loss, and the gain minus loss difference wave (ΔRewP) for non-suicidal ideation group (n=50) (top left panel) and suicidal ideation group (n=59) (bottom left panel) at baseline; topographic scalp map of neural activity depicting the gain minus loss (ΔRewP) difference 250-350 ms after feedback presentation for non-suicidal ideation group (top right panel) and suicidal ideation group (bottom right panel) at baseline.

3.2. Intent-To-Treat

Participants who completed all 12 sessions of treatment (n=89) were similar to non-completers (n=19) in terms of baseline clinical and demographic measures, except for anticipatory pleasure (TEPS-A) where completers exhibited less TEPS-A (M=4.01, SD=.78) than non-completers (M=4.43, SD=.68). Importantly, the number of participants who completed treatment did not significantly differ between CBT versus ST. See Supplementary Materials for details.

3.3. Partial Correlations

Since SI/non-SI groups differed in age, we controlled for age in addition to symptom severity (LSAS, HAMD). Results showed the SI-ΔRewP relationship was significant across participants with SAD or MDD (r=−.23, p=.01). See Figure 2 for scatterplot. To evaluate whether findings were maintained when only controlling for symptom severity, the same analysis was repeated without controlling for age and results remained significant (r=−.20, p=.03). For residualized RewP following monetary gain, a negative relationship with SI was observed (r=−.23, p=.01) and for residualized FN following monetary loss, a positive-SI relationship was detected (r=.27, p=.005) across participants with SAD or MDD. See Figure 3 for scatterplots.

Figure 2.

Scatterplot illustrating relationship between suicidal ideation and FCz based on difference for gains (i.e., win) minus loss (ΔRewP) controlling for anxiety (LSAS), depression (HAMD), and age in years (i.e., residuals) at baseline.

Suicidal ideation = Inventory of Depression and Anxiety Symptoms – Second Version suicidality subscale; LSAS = Liebowitz Social Anxiety Scale; HAMD = Hamilton Depression Rating Scale; gray circles and gray fit line = social anxiety disorder; black triangles and black fit line = major depressive disorder

Figure 3.

Scatterplot illustrating relationship between suicidal ideation and residualized RewP at FCz for gain (i.e., win) controlling for anxiety (LSAS), depression (HAMD), and age in years (i.e., residuals) at baseline (left panel). Scatterplot illustrating relationship between suicidal ideation and residualized FN at FCz for loss controlling for anxiety (LSAS), depression (HAMD), and age in years (i.e., residuals) at baseline (right panel).

Suicidal ideation = Inventory of Depression and Anxiety Symptoms – Second Version suicidality subscale; LSAS = Liebowitz Social Anxiety Scale; HAMD = Hamilton Depression Rating Scale; gray circles and gray fit line = social anxiety disorder; black triangles and black fit line = major depressive disorder

When evaluating subjective capacity for pleasure, partial correlations showed no relationship between SI and anticipatory (TEPS-A) (r=−.18, p=.06) or consummatory (TEPS-C) pleasure (r=−.04, p=.67). Also, no ΔRewP-consummatory pleasure relationship emerged (r=−.05, p=.56).

3.4. Treatment outcome

Composite score. Since age differed between diagnostic groups (SAD/MDD), age was included in the linear mixed model (LMM) evaluating change in symptom severity. LMM consisting of the composite POMS score (LSAS, HAMD) revealed a main effect for time [t(1, 115.36)=11.44, p<.001, r_{effect size}=.73] such that symptoms decreased over time (see Figure 4). There was also a main effect for diagnostic group [t(1, 101.94)=3.37, p=.001, r_{effect size}=.32] where the SAD group exhibited greater symptom severity (M=0.78, SE=0.03) than the MDD group (M=0.63, SE=0.03). No other main effects emerged (lowest p=.58, highest r_{effect size}=.05). Results also revealed a time x treatment arm interaction [t(1, 114.09)=2.11, p=.04, r_{effect size}=.19] such that the rate of symptom improvement was greater throughout CBT [t(1, 61.93)=10.67, p<.001, r_{effect size}=.80] than ST [t(1, 51.82)=5.67, p<.001, r_{effect size}=.62]. No other interactions were observed (lowest p=.50, highest r_{effect size}=.06). See Table 3 for CBT and ST details.

Figure 4.

Bars depicting average change in symptom severity collected every two weeks across treatment for patients assigned to CBT (dark gray bars) or ST (light gray bars); error bars = standard error of the mean.

Composite Symptom Score = Hamilton Depression Rating Scale and Liebowitz Social Anxiety Scale CBT = cognitive behavioral therapy; ST = supportive therapy

Table 3.

Clinical and neurophysiological characteristics in the context of treatment: Values reflect mean and standard deviations are in parentheses

Measures	CBT (n = 57)*	ST (n = 51)*
Composite Score
Baseline (T1)	.88 (.24)	.88 (.24)
Week 2 (T2)	.78 (.26)	.84 (.27)
Week 4 (T3)	.70 (.30)	.72 (.26)
Week 6 (T4)	.63 (.29)	.72 (.29)
Week 8 (T5)	.61 (.32)	.67 (.30)
Week 10 (T6)	.52 (.26)	.61 (.33)
Week 12 (T7)	.48 (.30)	.60 (.32)
IDAS-II Suicidality Subscale
Baseline	7.86 (2.57)	7.72 (2.53)
Week 6	6.98 (2.26)	6.83 (1.66)
Week 12	6.54 (1.32)	7.12 (2.10)
TEPS-A
Baseline	4.04 (.76)	4.14 (.80)
Week 12	4.40 (.64)	4.31 (.78)
TEPS-C
Baseline	4.42 (.75)	4.50 (.74)
Week 12	4.79 (.79)	4.78 (.70)
ΔRewP at FCz (gain-loss)
Baseline	3.47 (4.49)	4.30 (4.80)
Week 6	3.64 (3.69)	2.56 (4.18)
Week 12	3.17 (4.08)	2.60 (4.36)

Note.

^*Reflects the number of participants randomly assigned to treatment arm; CBT = Cognitive Behavioral Therapy; ST = Supportive Therapy; Composite score = Liebowitz Social Anxiety Scale and Hamilton Depression Rating Scale using the proportion of maximum scaling; T = Time; IDAS-II = Inventory of Depression and Anxiety Symptoms – Second Version; TEPS-A = Temporal Experience of Pleasure Scale, anticipatory pleasure score; TEPS-C = Temporal Experience of Pleasure Scale, consummatory pleasure score; FCz = frontocentral event-related potential component based on difference score

Disorder-specific scores. When evaluating diagnostic groups separately, LMM results showed a main effect for time for each group such that symptoms decreased over time. However, there was no main effect for treatment arm or interaction with treatment arm in either group. See Supplementary Materials for details.

Suicidal ideation. Since age differed between the SI/non-SI groups at baseline, age was included in the LMM to evaluate change in SI.¹ SI/non-SI groups were collapsed across diagnostic group as the SAD and MDD groups did not differ in level of SI. Results revealed a main effect for time [t(1, 285.38)=4.17, p<.001, r_{effect size}=.24] such that SI significantly decreased over the course of treatment. There was also a main effect for group [t(1, 124.25)= 7.65, p<.001, r_{effect size}=.57] where the SI group exhibited greater SI (M= 8.13, SE=0.18) than the non-SI group (M=6.08, SE=0.20). No other main effects were detected (lowest p=.69, highest r_{effect size}=.04). There was also a significant time x SI group interaction [F(1, 282.08)=20.15, p<.001, r_{effect size}=.26] such that SI decreased over time for participants in the high SI group [t(1, 158.01)=−4.70, p<.001, r_{effect size}=.35] but not the non-SI group [t(1, 126.00)=1.83, p=.07, r_{effect size}=.16]. No other interactions were observed (lowest p=.22, highest r_{effect size}=.07). See Table 3 for CBT and ST details.

RewP. Age was also included LMM when evaluating ΔRewP². Results showed a main effect of SI/non-SI group [t(1, 89.90)=2.05, p=.043, r_{effect size}=.21] such that the SI group exhibited less ΔRewP than the non-SI group. No other main effects were observed (lowest p=.17, highest r_{effect size}=.14) and no significant interactions emerged (lowest p=.15, highest r_{effect size}=.14). For ICC, results revealed the average measure was .40 (95% CIs = [.13; .60], p=.004). Thus, ICC results indicate temporal reliability for ΔRewP was in the fair range. See Table 3 for CBT and ST details.

Capacity for Pleasure. Age was included as a covariate and mixed ANCOVA results for anticipatory pleasure (TEPS-A) revealed age was significant [F(1,79)=6.39, p=0.01, η_p²=.07]; no other main effects or significant interactions were observed (lowest p=.08, highest η_p²=.03). Follow-up simple effects analysis showed older participants exhibited less anticipatory pleasure (M=3.87, SD=.94) than younger participants (M=4.28, SD=.62). See Supplemental Materials for details.

When the same ANCOVA was performed for consummatory pleasure (TEPS-C), there was a main effect for time [F(1,79)=7.53, p=0.007, η_p²=.08]; no other main effects were detected and there were no significant interactions (lowest p=.11, highest η_p²=.03). Collapsing across groups, treatment arm, and age, a follow-up paired t-test showed baseline TEPS-C (M=4.44, SD=.74) increased after completing treatment (M=4.79, SD=.75) [t(87)=5.84, p<.001, d=.62]. See Table 3 for CBT and ST details.

4. Discussion

The current study examined whether reward sensitivity, indexed with ΔRewP, and subjective capacity to experience pleasure corresponded with suicidal ideation (SI) in patients with SAD or MDD. As expected, more SI was associated with less ΔRewP, controlling for symptom severity and age. Yet, there was no relationship between SI and subjective capacity to experience pleasure; there was also no relationship between ΔRewP and self-reported consummatory pleasure. Regarding diagnostic status, SAD and MDD groups did not significantly differ in ΔRewP, SI, or subjective capacity for pleasure. Concerning treatment outcomes, social anxiety and depression based on a composite score decreased following psychotherapy though the rate of symptom improvement was greater in CBT than ST whereas the decrease in SI was similar between treatment arms. For subjective consummatory pleasure, there was a pre-to-post increase in capacity for pleasure regardless of treatment arm; however, self-reported anticipatory pleasure did not change following treatment. As hypothesized, ΔRewP was not significantly altered following CBT or ST. Altogether, hypotheses were partially supported.

To our knowledge this is the first study to evaluate SI in unmedicated patients with SAD (without comorbid MDD) against patients with MDD (without SAD comorbidity). The finding that SAD and MDD groups endorsed comparable levels of SI at baseline suggests SI cuts across these diagnostic boundaries, which is consistent with evidence SI is transdiagnostic (Glenn et al., 2018). Even so, it will be important to replicate findings in individuals with active SI (i.e., participants who endorse plan and intent) as the current study excluded individuals with active SI or self-harming behaviors. Thus, the potential to detect differences between diagnostic groups may have been reduced. Beyond SI, evidence SAD and MDD groups did not differ in ΔRewP or subjective capacity to experience pleasure suggests features of reward sensitivity may be similar in these disorders.

With regard to ΔRewP, the hypothesis that SI would correspond with ΔRewP at baseline, when controlling for symptom severity, was supported. Therefore, the relationship may be distinct to SI. As for the direction of SI-RewP relationships, residualized scores showed both RewP and FN distinctly corresponded with SI. Specifically, greater SI corresponded with less RewP following gains (i.e., partialing out effect of loss) and greater SI also corresponded with less FN following loss (i.e., partialing out effect of gain). Taken together, less reward and loss sensitivity may serve as vulnerability markers for SI. Previous reports in youth partially support our results--namely, a study involving children found those with SI exhibited less ΔRewP than children without SI after controlling for anxiety and depression symptoms (Tsypes et al., 2019). Though in a separate study, more negative ΔFN for loss-gain was observed in children of parents who attempted suicide compared those whose parents had no such history (Tsypes et al., 2017). Mixed findings may be due in part to methodological differences between studies, therefore, it will be important to replicate results in a larger sample. However, despite inconsistencies, findings extend prior work and suggest SI relates to neural response following favorable or unfavorable feedback.

The clinical inference is that different pathways may lead to SI. Less response to monetary gain may reflect attenuated response to natural rewards (e.g., pleasurable activities), which could have deleterious downstream effects. For example, hedonic capacity is intertwined with approach-related behaviors (e.g., Treadway et al., 2009) and motivation (Germans & Kring, 2000). Thus, reduced engagement in potentially enjoyable activities due to hedonic deficiency may facilitate feelings of despair or disconnection from others that, in turn, increases risk for suicidality (Van Orden et al., 2010). Reduced reactivity to loss on the other hand may reflect greater tolerance for pain. According to the interpersonal psychological theory of suicide, painful or provocative experiences (e.g., trauma, non-suicidal self-injury) increase pain tolerance and capacity for suicide (Joiner, 2005; Van Orden et al., 2010). However, since we did not directly evaluate pain tolerance, it will be important for future studies to provide support for this conjecture. Due to the complexity of suicide, there are likely multiple pathways, therefore, further study is needed to understand how neurophysiological response to monetary gains and losses relate to SI and the extent to which it factors into suicide attempts.

Our hypotheses that SI alone and ΔRewP alone would each correspond with subjective capacity for pleasure at baseline controlling for symptom severity were not supported. Findings suggest individual differences in subjective capacity for pleasure may not serve as a robust behavioral assay for SI or relate to variance in neural response following feedback. It is possible that restricted range reduced our ability to detect results, therefore, it will be important to replicate findings in a larger study designed to test these effects.

When comparing SI/non-SI groups regardless of diagnostic status at baseline, the SI group exhibited less ΔRewP relative to the non-SI group; also residualized RewP and FN scores showed the SI group exhibited less RewP for gains and less FN for loss than the non-SI group. The SI group also exhibited less anticipatory pleasure (TEPS-A) than the non-SI group, yet, no group effects for consummatory pleasure (TEPS-C) were observed. Evidence of less TEPS-A in the SI group is consistent with reports individuals with SI have reduced capacity to experience pleasure compared to those without SI (Ducasse et al., 2018). While it is not clear why level of TEPS-C was similar between SI/non-SI groups, findings highlight the relevance of evaluating components of pleasure. Again, by excluding individuals with active SI or self-harming behaviors, we may have reduced our ability to detect effects.

Beyond SI, we explored whether overall ΔRewP differed between SAD and MDD groups at baseline, and results revealed ΔRewP and residualized scores did not differ between diagnostic groups. Findings build on previous reports that indicate RewP relates to depression (Bress et al., 2015; Foti & Hajcak, 2009; Liu et al., 2014; Nelson & Jarcho, 2021) and anxiety (Gu et al., 2020; Kessel et al., 2015). Even so, our study did not focus on characterizing relationships between ΔRewP and anxiety or depression severity; nor did the study focus on determining whether ΔRewP was aberrant (i.e., different from a healthy control group). Therefore, conclusions about ΔRewP and diagnostic groups are limited.

As for treatment outcome, a composite score (combining LSAS and HAMD) showed symptom severity significantly decreased over the course of psychotherapy, though rate of symptom improvement was greater in CBT than ST, providing further support for CBT as first-line psychotherapy (David et al., 2018). While findings are consistent with our hypothesis, patients randomized to ST also showed reduction in symptom severity, which underscores the role factors common to psychotherapies may play in improvement (e.g., therapeutic alliance, mitigation of isolation) (Cuijpers et al., 2019). Also, when restricting analysis to SAD and MDD groups, only a main effect for time was found. The lack of interaction with treatment arm within each diagnostic group is likely due to insufficient power.

SI also significantly decreased, regardless of treatment arm, but only among participants who endorsed SI, which may be due to a floor effect. Thus, contrary to the expectation that SI improvement would be greater following CBT than ST, treatments yielded comparable results. Again, excluding participants with active SI and self-harm behaviors may have reduced our ability to detect differences.

Also, although we were not powered to test a diagnostic group (SAD/MDD) interaction with treatment arm (CBT/ST), evidence SI significantly decreased following treatment across participants suggests SI in individuals with SAD or MDD improved with psychotherapy, which is consistent with prior research (Leavey & Hawkins, 2017). Nonetheless, it will be important to test for interactions between diagnostic groups and treatment arm in a large sample in the future given reports SI in SAD may not significantly improve after completing CBT (Brown et al., 2018).

As expected, ΔRewP was stable over the course of treatment though ICC results indicated temporal reliability was only in the fair range, pointing to subtle alterations in ΔRewP. Findings are in line with prior studies in adults that showed treatment with standard CBT, SSRI, or a novel behavioral intervention for internalizing psychopathologies did not significantly alter RewP (Brush et al., 2022; Burkhouse et al., 2018; Webb et al., 2021). The lack of change is also consistent with reports that individual differences in reward sensitivity, represented by RewP, are relatively stable over time (Weinberg et al., 2015).

Given evidence of stability of RewP and its association with SI, it will be important for studies to test whether RewP prospectively predicts SI recurrence in SAD or MDD after completing treatment. It will also be important to test whether psychosocial treatments that directly aim to improve the reward system (Craske et al., 2019; Taylor et al., 2017) modulate RewP. Lastly, findings may not generalize to youth as residualized RewP has been shown to increase in young children after completing a psychosocial intervention (Barch et al., 2020). Thus, further study is needed to determine the malleability of RewP across development.

Concerning self-reported capacity for pleasure, anticipatory pleasure did not increase after completing CBT or ST. Rather, a main effect for age emerged such that older participants exhibited less anticipatory pleasure than younger ones indicating age may impact this component of pleasure. However, we hesitate to interpret this finding as there is not strong support in the literature that anticipatory pleasure decreases over time in adults with MDD or SAD. Also, our study sample comprised relatively young adults and the groups used to detect a significant difference was imbalanced with far more participants in the younger cohort than older cohort. Therefore, further study is needed to determine whether age interacts with anticipatory pleasure in MDD and SAD. In contrast to anticipatory pleasure, consummatory pleasure increased following CBT or ST regardless of principal diagnosis or assignment to SI/non-SI group suggesting this component of pleasure may benefit from psychotherapy regardless of diagnostic status. Again, it will be important to replicate findings before drawing firm conclusions.

4.1. Limitations

Findings need to be interpreted in consideration of important limitations. First, results may not generalize to individuals who are receiving pharmacotherapy or who differ in demographic or clinical characteristics, including those with active SI or recent history of self-injurious behavior or suicide attempt. Second, there was no waitlist condition or non-psychotherapy intervention (e.g., medication, neuromodulation), which are needed to ascertain the extent to which psychotherapy impacted findings. Third, there was no healthy control group, therefore, it is not possible to determine whether neural activity or subjective capacity for pleasure were atypical or normative. Fourth, studies that compared CBT to other psychotherapies (e.g., supportive therapy, interpersonal therapy, psychodynamic) show the effect sizes for anxiety disorders and depression favoring CBT are small/non-significant to moderate (e.g., Braun et al., 2013; Tolin, 2010; Zhang et al., 2022). Therefore, we were underpowered to test for interactions between treatment arm and principal diagnosis. Moreover, meta-analytic conclusions have called for greater sample sizes in suicide research that use psychophysiological methodologies such as ERPs (Gallyer et al., 2021). Fifth, age significantly differed between diagnostic groups and SI/non-SI groups, therefore, we cannot rule out the possibility that age influenced findings. Sixth, the Snaith–Hamilton Pleasure Scale (Snaith et al., 1995) has been frequently used to evaluate relationships with SI and psychotherapy outcomes (Alsayednasser et al., 2022; Ducasse et al., 2018; Hanuka et al., 2022). Differences in measures used to assess capacity for pleasure does not permit direct comparison between studies. Seventh, the internal consistency for subjective consummatory pleasure (TEPS-C) was not acceptable; therefore, interpretation of this measure is questionable. Lastly, even though fidelity/adherence findings indicate psychologists followed CBT and ST protocols, we cannot be rule out that bias on the part of the psychologist favoring CBT or ST may have influenced treatment outcome.

4.2. Conclusions

In conclusion, preliminary results suggest SI cuts across SAD and MDD. Also, correlational analysis at baseline shows SI negatively relates with residualized RewP following monetary gains and SI positively relates with residualized FN following monetary loss, when controlling for symptom severity and age. Thus, RewP and FN may serve as transdiagnostic neuromarkers for SI. In contrast, subjective capacity for pleasure did not relate to SI. Therefore, further study is needed to identify a potential behavioral marker for SI. In terms of treatment outcome, findings based on a composite score also indicate depression and social anxiety symptoms improve following psychotherapy, though to a greater extent in CBT than ST. There was also evidence capacity for consummatory pleasure improves regardless of treatment arm. Lastly, among participants who endorsed SI prior to treatment, SI improved following psychotherapy regardless of treatment arm. In contrast, reward responsivity indexed with ΔRewP was not significantly impacted by CBT or ST. Since ΔRewP was stable following treatment, ΔRewP may represent a vulnerability for suicide risk even when symptom severity and SI improves with treatment.

Highlights.

Patients with social anxiety or depression had similar levels of suicidal ideation

Smaller RewP following monetary gains correlates with more suicidal ideation

Suicidal ideation significantly decreased after completing psychotherapy

RewP response was stable after completing psychotherapy

RewP response may be a risk factor for suicidality