Understanding Suicide in Sexual Minority Youth: Neural Reactivity to Social Cues as a Moderating Influence

Abstract

Objective:

Sexual minority youth (SMY) experience heightened suicide risk, partly due to stigma surrounding sexual orientation identity. Neurobiological characteristics can influence reactivity to social cues (e.g., perceived liking or rejection) and suicide risk. These effects are exacerbated during adolescence, a developmental period of heightened sensitivity to social contexts. In this cross-sectional study of youth at varying psychiatric risk, we examined if neural reactivity to social cues moderated the link between sexual minority status and suicidal ideation (SI), and if sexual orientation victimization experiences further influenced these effects.

Method:

75 youth (aged 14–22 years; 52% SMY, 48% heterosexual) reported depression, SI, and victimization, and completed a fMRI task involving viewing of unfamiliar face stimuli and receipt of social cues in rewarding and ambiguous contexts. Regions-of-interest analyses examined task-related neural reactivity in neural social regions. Moderation analyses were conducted using linear regressions.

Results:

SMY reported more severe depression, victimization, and SI (ps<.05). Left temporoparietal junction (TPJ) activation to social cues, regardless of the degree of valence and certainty, moderated the link between sexual minority status and SI, where SMY (vs. non-SMY) with dampened left TPJ activity had higher SI. Exploratory analyses indicated that these associations were not further influenced by victimization.

Conclusion:

Results indicate enhanced suicide risk in SMY with altered social processing in the TPJ—a key region of neural social systems—across contexts, regardless of victimization history. Findings suggest that individual differences in neural reactivity to social cues are critical for understanding SMY suicide risk and have potentially important clinical implications.

Suicide is the second leading cause of death among U.S. youth, with rates rising in the past five years and exacerbated by the COVID-19 pandemic (Bersia et al., 2022; Kann et al., 2018). It has been estimated that one in five youth has seriously considered suicide (Van Meter et al., 2023). Critically, sexual minority youth (SMY; individuals who experience same-sex attraction and/or identity) are at higher risk for suicide. Compared with heterosexual peers, SMY are up to four times more likely to contemplate and attempt suicide (di Giacomo et al., 2018; Marshal et al., 2011). SMY may be at greater risk for suicide due to unique social stressors (e.g., stigma) related to sexual orientation identity in addition to typical challenges during the vulnerable developmental period of adolescence (Feinstein, 2020; Forbes et al., 2021; Meyer, 2003). Despite societal attitudes shifting towards acceptance of sexual orientation identity, these suicide disparities remain persistent (Meyer et al., 2021). While SMY as a group are at greater risk for suicide, not all SMY experience suicidal thoughts and behaviors. Therefore, it is critical to pinpoint factors driving individual differences in suicide risk for SMY. In particular, more research examining neural mechanisms that contribute to variability in SMY suicide risk is needed. Doing so can improve understanding of who is most vulnerable, with implications for mitigating the occurrence and course of suicide.

Suicide is a complex behavior that is difficult to predict, although suicidal ideation (SI; i.e., thinking about, considering, and/or planning suicide) is recognized as an important precursor (Nock et al., 2008). Social disconnection, which is characterized by a disrupted sense of affiliation with others, is a key contributor of SI. For example, the Interpersonal Theory of Suicide (Van Orden et al., 2010) highlights two dimensions of social disconnection: thwarted belongingness (a lack of meaningful connection with others) and perceived burdensomeness (the belief that one is a liability to others), with meta-analyses supporting their associations with SI across diverse populations (Chu et al., 2017). SMY face higher risk for SI due to a greater propensity for experiencing social disconnection because of stigma and marginalization due to sexual orientation identity (Fulginiti et al., 2020). The Minority Stress Theory (Meyer, 2003) argues that distal (external factors targeting an individual; e.g., discrimination or violence) and proximal (internal factors targeting oneself; e.g., internalized stigma or expectations of rejection) stressors specific to sexual minorities contribute to hostile social environments that negatively impact mental health. In the context of suicide, social stressors such as discrimination, rejection, or victimization, may exacerbate feelings of social disconnection and increase SI risk in SMY (de Lange et al., 2022; Green et al., 2022). In contrast, positive and affirming social experiences may mitigate suicide risk (Green et al., 2021; Hatzenbuehler, 2011). Taken together, these findings suggest that social cues from the environment, including those perceived as threatening (e.g., victimization) and/or rewarding (e.g., acceptance), can influence SI risk in SMY.

Although exposure to social stressors related to sexual orientation identity can increase SI among SMY, not all experience SI, suggesting that individual variability in reactivity to social contexts may moderate SMY mental health disparities. Indeed, neurobiological susceptibility models suggest that individual differences in neural sensitivity to social contexts can influence mental health outcomes, where those with altered threat sensitivity are more likely to develop psychopathology in response to negative social environments (Schriber & Guyer, 2016). For instance, Yang and colleagues found that victimization was positively associated with suicide risk among youth with high (vs. low) neural threat sensitivity to negative emotional pictures (Yang et al., 2023). In addition, daily experiences of victimization have been associated with more severe SI among youth with greater neural sensitivity to social rejection (Oppenheimer et al., 2020). Together, these findings highlight the critical moderating role of neurobiological processes for understanding individual differences in suicide risk. However, no study to date has examined these relationships in SMY nor considered neural response to positive social contexts.

Notably, SI typically emerge and are exacerbated during adolescence, a developmental period characterized by heightened sensitivity to social experiences and increased valuation and salience of peer social relationships (Jaworska & MacQueen, 2015). Compared with children and adults, adolescents are more self-conscious, spend more time with peers, place greater emphasis on peer acceptance and belonging, and engage in greater risk-taking with peers (Blakemore & Mills, 2014; Knoll et al., 2015). Neuroimaging studies utilizing functional magnetic resonance imaging (fMRI) suggest that these changes are likely due to converging development of neural circuitry involved in social, affective, and self-regulatory processing during adolescence. These include brain regions responsible for reward and threat processing (e.g., ventral striatum [VS], amygdala), as well as self-regulation (e.g., prefrontal cortex; Forbes et al., 2021). Moreover, as adolescents develop greater social orientation, these neural circuits are influenced by social brain regions (e.g., precuneus, temporoparietal junction [TPJ]) involved in understanding the self and others (Blakemore, 2008). Therefore, the salience of social cues is particularly elevated during adolescence. Critically, the context in which these social cues occur (e.g., being accepted vs. rejected) can alter activity in neural social circuits and influence suicide risk. For example, adverse events (e.g., peer victimization) have been associated with decreased response in the medial prefrontal cortex (mPFC; brain area relevant for social, cognitive, and affective processing) to reward (Casement et al., 2014). Furthermore, blunted activity in brain areas related to reward, social cognition, and pain during simulated peer rejection and acceptance have been associated with SI severity among depressed adolescents (Harms et al., 2019).

Adolescence also tends to be a time when sexual orientation identity develops and is communicated with others, which may heighten sensitivity to social contexts (e.g., rejection). Conformity to group norms is often emphasized during adolescence, which is challenging for SMY who may perceive themselves as different from heterosexual peers (Blakemore & Mills, 2014; Hall et al., 2021). Therefore, the salience of social cues is even greater for SMY during this developmental period. Compared with heterosexual peers, neural sensitivity to social cues may be altered in SMY given their unique experiences of social stress, especially those related to sexual orientation stigma. Prior research has found decreased reactivity in the TPJ—a brain area involved in understanding, evaluating, and attributing the beliefs and intentions of others (Aichhorn et al., 2009; Samson et al., 2004)—to social reward among SMY, which was associated with more severe depressive symptoms (Eckstrand et al., 2019). In a separate sample, history of victimization due to sexual orientation has also been associated with altered VS and mPFC activity to reward in youth, where greater victimization and mPFC activity were associated with more severe depression (Eckstrand, Silk, et al., 2022). These findings suggest that alterations in neural social circuitry are more common in SMY (vs. heterosexual peers) and may explain individual differences in risk for poorer mental health.

Given well-documented suicide disparities and associated behavioral risk factors in SMY, it is critical to understand the neural mechanisms underlying individual differences in suicide risk for this at-risk population. Prior research has suggested that neurobiological processes are implicated in the association between stress exposure and psychopathology, with individual differences in neural sensitivity to social contexts moderating the relationship between stress and mental health (Schriber & Guyer, 2016). Building on this literature, the present study investigated whether alterations in neural social circuitry are associated with suicide risk for SMY (vs. non-SMY). Specifically, we examined neural reactivity to social cues, in the form of unfamiliar faces, that varied on the degree of valence and certainty (rewarding vs. ambiguous). We hypothesized that neural reactivity to social cues would moderate the relationship between sexual minority status and SI (two-way interaction) in a sample of SMY and heterosexual youth at varying SI severity. Based on past research (Blakemore, 2008; Eckstrand, Lenniger, et al., 2022; Forbes et al., 2021; Schriber & Guyer, 2016), we selected the following regions-of-interest (ROIs) given their relevance for neural social reward/threat processing: mPFC, VS, amygdala, precuneus, and TPJ. Previous research has also suggested that sexual minorities tend to be more vigilant and learn to anticipate rejection during social interactions due to their minoritized identities (Feinstein, 2020; Meyer, 2003). Therefore, we examined whether moderation effects differed for rewarding and ambiguous contexts, with the expectation that social cues in an ambiguous context might be perceived as more threatening by SMY. Finally, we explored if prior exposure to sexual orientation victimization further influenced the strength of these associations (three-way interaction).

Methods

This study is a secondary analysis of data collected as part of a larger study. Importantly, the aims and hypotheses of the present research do not overlap with other publications using this dataset (Eckstrand, Silk, et al., 2022; Gupta et al., 2024).

Participants

Seventy-five youth (aged 14–22 years) at varying SI severity were selected from a larger study examining anhedonia development in adolescents. The study included two groups: heterosexual youth selected based on presence/absence of familial risk for anhedonia-related psychiatric disorders (depression, bipolar disorder, schizophrenia, schizoaffective disorder) and youth selected based on identification as sexual minority (i.e., not heterosexual). The final sample for this study was 39 SMY and 36 non-SMY¹ (matched based on age, sex assigned at birth, and race/ethnicity). Included participants had no missing behavioral and fMRI data, no excessive motion during fMRI scan (M=0.52mm/0.37°, SD=0.23mm/0.24°), and >50% VS coverage². Participants were recruited from the local community through flyers and research registries, other research studies, and clinics/community settings with higher proportion of SMY. Participants and their parents and/or siblings in the familial-risk sample completed diagnostic interviews to determine and confirm psychiatric risk classification and eligibility. As part of the larger study, youth were included in the study if they: 1) were not taking psychotropic medications for ≥2 weeks prior to study participation (besides a stimulant medication, which was allowed only up to 36 hours before the MRI scan); 2) did not meet criteria for a lifetime DSM-IV mood or psychotic disorder, 3) did not use nicotine daily, and 4) had no MRI contraindications.

Sample characteristics are reported in Table 1. Gender identity data was collected later in the longitudinal study, so this information is not available for all participants³. Although we attempted to match groups based on age, the SMY group was significantly older. This was likely the result of primarily recruiting from extant studies that were conducted in older SMY. Study procedures were approved by the university IRB and participants (and their parent/guardian if age<18) provided informed consent.

Table 1.

Final Sample Demographics and Primary Outcome Variables

Sample Characteristic	Total Sample (n=75)	SMY (n=39)	Non-SMY (n=36)	Tests of Significance
Age (M, SD)	17.36 (2.30)	18.67 (2.22)	15.94 (1.37)	t(64.11) = 2.72, p < .001
Sex Assigned at Birth (% Female)	61%	68%	56%	χ2 (1, n = 75) = 0.98, p = .324
Sexual Orientation				χ2 (4, n = 75) = 75.00, p < .001
100% Homosexual (%)	15%	28%	0%
Mostly Homosexual (%)	8%	15%	0%
Bisexual (%)	19%	36%	0%
Mostly Heterosexual (%)	11%	21%	0%
100% Heterosexual (%)	48%	0%	100%
Race				χ2 (3, n = 75) = 1.82, p = .611
% Asian	9%	13%	6%
% Black/African American	28%	31%	25%
% White	55%	49%	61%
% Other or Multiracial	8%	8%	8%
Ethnicity (% Hispanic)	7%	8%	7%	χ2 (1, n = 68) = 0.07, p = .794
Clinical & Social Characteristics (M, SD, Range)1
Sexual Orientation Victimization	8.40 (2.68) Range: 6—18	9.26 (3.25) Range: 6—18	7.47 (1.40) Range: 6—12	F(1, 72) = 12.16, p = .001, η2 = .14
Depression	16.74 (11.12) Range: 0—40	20.13 (11.10) Range: 2—40	13.07 (10.04) Range: 0—39	F(1, 72) = 8.47, p = .005, η2 = .11
Suicidal Ideation2	20.97 (25.87) Range: 0—117	24.23 (23.86) Range: 0—103	17.43 (27.79) Range: 0—117	F(1, 72) = 5.96, p = .017, η2 = .08

Note. SMY = Sexual minority youth; Depression was measured with the Center for Epidemiological Studies Depression Scale (Radloff, 1977); Suicidal ideation was measured with the Suicidal Ideation Questionnaire (Reynolds, 1987);

¹Tests of significance controlled for age as covariate;

²Scores reported here are untransformed values for ease of interpretation.

Procedure

Data reported here were collected at baseline assessment for the familial-risk sample and at the sole time-point for the SMY sample. Participants completed questionnaires assessing demographics, exposure to identity victimization events, depression, and SI. If participants endorsed imminent SI (e.g., current SI plan, intent, or attempt) during the study visit, a trained study staff conducted a suicide risk assessment, created a safety plan, and provided mental health resources and referrals. These procedures were supervised by a licensed psychologist on the study team. Following, participants underwent fMRI scanning during a task involving viewing of face stimuli and receipt of social cues (see Measures).

Measures

Sexual Orientation Identity

Sexual orientation was assessed with a single-item measure as in the National Longitudinal Study of Adolescent Health, which defined sexual orientation in terms of sexual attraction: “100% Heterosexual (Straight),” “Mostly Heterosexual (Straight but somewhat attracted to people of the same sex),” “Bisexual (Attracted to men and women equally),” “Mostly Homosexual (Gay, but somewhat attracted to the opposite sex),” “100% Homosexual (Gay or Lesbian),” “Don’t Know,” or “Not Listed.” No participant indicated “Don’t Know” or “Not Listed.” Participants who endorsed any option besides “100% Heterosexual (Straight)” were classified as SMY.

Identity Victimization Exposure

A 23-item questionnaire (Dermody et al., 2016) assessed exposure to various victimization events (1. bullied, 2. hit/beaten up, 3. treated unfairly, 4. called hurtful or insulting names, 5. witnessing others being hit/beaten up, and/or 6. witnessing others being called hurtful or insulting names) based on sexual orientation, race, gender, and/or body size over the past six months. For each item, participants responded on a 4-point scale (“1-Never” to “4-Many Times”). Frequency of exposure to sexual orientation victimization events was based on the total score across the six items on sexual orientation (Cronbach’s α=.85).

Depression

The 20-item Center for Epidemiological Studies Depression Scale (Radloff, 1977) assessed past-week depressive symptoms. Participants responded on a 4-point scale (“0-Rarely or none of the time (less than 1 day)” to “3-Most or all of the time (5–7 days)”). Items were summed to create an index of depression (Cronbach’s α=.88). A score of ≥16 indicates risk for clinical depression. The sample’s mean was 16.74 (SD=11.12), of which 48% scored above the clinical cut-off.

Past-Month Suicidal Ideation

The 30-item Suicidal Ideation Questionnaire (SIQ; Reynolds, 1987) assessed past-month frequency of SI, ranging from passive thoughts about death to serious thoughts of actively engaging in suicidal behaviors. Participants responded on a 7-point scale (“0-never had this thought” to “6-almost every day”). An index of SI was obtained by calculating the sum across items (Cronbach’s α=.97). A score of 41 has traditionally been used as an indicator of those at clinical risk and 16% of the sample scored above this cut-off.

Past-Year Suicidal Ideation and Behaviors

To further characterize severity and history of SI, four items from the Youth Risk Behavior Survey (YRBS; details in supplement) were used to index high-risk suicidal thoughts and behaviors in the past year: 1) serious thoughts about attempting suicide (yes/no), 2) making a suicide plan (yes/no), 3) frequency of suicide attempts (from “0 times” to “6 or more times”), and 4) injuries from suicide attempt(s) requiring medical attention (“I did not attempt suicide during the past 12 months,” “yes,” or “no”).

Other Relevant Clinical Variables

In the larger study, participants also completed self-report measures of anhedonia, anxiety, impulsivity, desire for emotional closeness, and substance use items from the 2015 Youth Risk Behavior Survey (YRBS). Given their relevance to SI, we conducted supplemental analyses that included these variables as additional covariates in our primary analyses but they did not significantly impact the results. Details are described in the supplement (Table S4).

fMRI Task

Neural reactivity to social cues was examined using a block-design fMRI task (Figure 1; Healey et al., 2014). Across eight blocks, participants viewed faces of other age-matched unfamiliar youth (50% female) whom they previously rated in terms of likeability before scanning and were informed whether: 1) they were rated positively by these youth (rewarding condition; four blocks) or 2) they were not rated yet, which could be interpreted in different ways (ambiguous condition; four blocks). Scanning was performed in a Siemens 3T trio scanner and fMRI acquisition parameters followed current convention: MPRAGE structural images with high-resolution T1-weighted images with 1mm isometric voxels (TR/TE/flip angle=1,500ms/3.19ms/8°; FOV=256×256; 176 continuous slices) and field maps (2.3mm isotropic voxels; TR/TE1/TE2/flip angle=550ms/4.92ms/7.38ms/50°; FOV=220×220; bandwidth 380Hz/Px) were obtained. SPM12 was used to conduct preprocessing (image realignment/correction/transformation) and analysis (details in online supplement).

Figure 1.

Description of fMRI task design and procedure. Photos with green-colored background indicate rewarding social cues while those with white-colored background indicate ambiguous cues. Task procedure: Before scanning, participants rated 40 photos (50% female) of other age-matched youth based on how much they would like these peers. Unbeknownst to participants, these photos were obtained from a database of face stimuli. As part of a cover story, participants had their photos taken and were told that they would also be rated by other unfamiliar “peers.” A personalized stimulus set of 32 photos based on these ratings was created for each participant: 16 photos with rewarding social cues (i.e., informed that they were rated positively by other peers) and 16 with ambiguous cues (i.e., informed that peers had not rated them yet, which could be interpreted in different ways). To increase salience of social cues, rewarding cues comprised the eight highest-rated (mutual liking) and eight lowest-rated (received liking) photos (50% female) and ambiguous cues comprised 16 moderately-rated photos (50% female) for each participant. Task design: Stimuli were presented in a block design comprising four rewarding and four ambiguous blocks. Each photo was presented three times over these eight blocks, provided either rewarding or ambiguous cues but not both, and order of stimulus presentation was counterbalanced. Each block consisted of 12 photos and lasted for 84-s. To reduce habituation and predictability of the block design, 2 of the 12 photos in each block were of the opposite stimulus type (e.g., 2 ambiguous stimuli in the rewarding block). Each photo was presented for 3-s with a jittered inter-stimulus interval (1, 3, 5, or 7 seconds), and blocks were separated by an interval of 8-s. Participants were instructed to press a button each time they saw a photo to ensure that they were paying attention during the task. After the scan, participants were debriefed on task deception and informed that the faces were derived from a stimuli database and that their photos were not actually rated by peers.

Data Analytic Strategy

Neuroimaging Analyses⁴

First-level analyses examined participants’ neural reactivity to social cues generally (combining across rewarding and ambiguous task conditions) and exploratory analyses examined neural reactivity during each task condition respectively. A fixed-effect general linear model (GLM) was performed for each participant, which included the rewarding blocks, ambiguous blocks, and baseline fixation. Neural response to social cues was determined for Rewarding+Ambiguous>Baseline, Rewarding>Baseline, and Ambiguous>Baseline. Volumes with high motion and artifacts were identified using ART (average image intensity deviated >3SD from the mean intensity or where movement exceeded 0.5mm in translation or 0.01° in rotation from the previous image; Chai et al., 2014), which was used to create an additional regressor in the first-level GLM to reduce motion-related noise. The six realignment parameters determined during preprocessing were entered as covariates to control for head movement. A 128s high-pass filter and autoregressive modelling were implemented during fitting.

To examine task effects across the full sample, one-sample t-tests were used to examine brain activation across the whole brain and in our selected ROIs across both and for each task condition (vs. baseline). Information regarding task-based response is in Table S1. Brain activity in eight ROIs involved in social reward/threat processing were examined: 1) mPFC, 2) precuneus, 3) left and 4) right VS, 5) left and 6) right amygdala, and 7) left and 8) right TPJ. VS, mPFC, and TPJ masks were derived from the Neurosynth meta-analytic database (using the association test map at 50% threshold; Yarkoni et al., 2011), and precuneus and amygdala masks were defined in WFU PickAtlas toolbox; Maldjian et al., 2003). Neural activity to social cues was determined by extracting the principal eigenvariate in each of these eight ROIs (Figure S1). Principal eigenvariate values were winsorized to within 3SDs of the sample mean to reduce influence of outliers (1.2% of all values). These values were used in subsequent analyses.

Preliminary Analyses

First, we examined the distribution of primary outcome variables for normality and outliers, and applied appropriate transformations as needed. Then, we examined group differences in demographic and outcome variables using chi-square tests of independence for categorical variables and t-tests for continuous variables. If groups differed on a variable, it was included as a potential covariate in subsequent analyses. Bivariate correlations between variables were also examined to identify multicollinearity and potential covariates.

Moderation Analyses

Statistical analyses were conducted using PROCESS macro in SPSS. Primary analyses were conducted using neural reactivity during both rewarding and ambiguous task conditions as the moderator. For testing two-way moderation, sexual minority status (1=SMY; 0=non-SMY) was entered as the predictor, neural reactivity to social cues—as indexed by the principal eigenvariate value for each ROI—was entered as the moderator, and SI was entered as the outcome. Age and depressive symptoms were included as covariates. Note that moderation analyses tested the effects of each ROI separately. Therefore, a total of eight moderation models were examined (given eight ROIs). In models where significant moderation effects emerged, follow-up analyses explored whether these effects differed by task condition (see Supplemental Materials). The Johnson-Neyman technique was used to probe significant moderation effects by determining the region(s) of significance, i.e., value(s) of neural activity at which there were significant differences in SI due to sexual minority status (Bauer & Curran, 2005). In models with significant two-way interaction effects, sexual orientation victimization was included as another moderator (three-way interaction) in a second set of models to explore whether it influenced the strength of these associations. Given multiple hypothesis tests, we applied the Benjamini-Hochberg α-correction to control for the false discovery rate (set at .05) when examining statistical significance for each regression model.

Results

Preliminary Analyses

Sample characteristics and means (SDs) are reported in Table 1. SMY did not differ from non-SMY based on sex assigned at birth and race/ethnicity but were older (p<.001). A positive skew was observed for SIQ scores (skewness=1.93) and a square-root transformation was applied to reduce skewness (skewness=0.60)⁵. Depression was related to SI (r=.58, p<.001) and victimization (r=.25, p=.031). SI was not related to victimization (r=.17, p=.146) and age (r=−.03, p=.819).

Ten (13%) participants endorsed past-year suicidal behavior on the YRBS: seven indicated seriously considering attempting suicide (6 SMY, 1 non-SMY), three indicated making a suicide plan (2 SMY, 1 non-SMY), four indicated attempting suicide (1 SMY, 3 non-SMY), and one SMY indicated that their suicide attempt led to an injury requiring medical attention.

Group Differences in Mental Health Outcomes

One-way ANCOVAs revealed group differences in SI, depressive symptoms, and sexual orientation victimization (Table 1). Controlling for age, SMY reported greater SI (p=.017), depression (p=.005), and sexual orientation victimization (p=.001). Notably, SMY had mean depressive symptoms above the clinical cut-off (≥16). Descriptive data indicated that SMY were twice as likely as non-SMY (21% vs 11%) to have SIQ scores ≥41. SMY were also more likely (67% vs 28%) to be at clinical risk for depression (≥16 points).

Primary Analyses⁶

Neural Reactivity to Social Cues as a Moderator

Across both rewarding and ambiguous task conditions, left TPJ activation moderated the relationship between sexual minority status and SI, B=−1.08, p=.001, f²=0.16 (Table 2). The Johnson-Neyman technique revealed that SMY (vs. non-SMY) reported higher SI at lower levels of left TPJ activity, principal eigenvariate<−.34 arbitrary units, p<.05; range: −3.88—4.67. To illustrate the moderation effect, we plotted simple slopes at ±1 SD of TPJ activation (Figure 2). No other significant moderation effects were observed for brain activity in the other ROIs.

Table 2.

Two-way (Sexual Minority Status × Left TPJ Activity to Social Cues) Interaction in Predicting Suicidal Ideation (n=75)

Predictor	B	SE	t	p	B-H	95% CI	R 2
Rewarding + Ambiguous > Baseline							.53	F(5, 69) = 15.67, p < .001
Age	−.16	.12	−1.30	.199	.025	−.40 to .09
Depression	.16	.02	7.66	<.001	.010	.12 to .21
Sexual Minority Status	.42	.58	0.73	.471	.050	−.74 to 1.59
Left TPJ Activity	.74	.25	3.04	.003	.017	.26 to 1.23
Sexual Minority Status × Left TPJ Activity	−1.08	.32	−3.34	.001	.013	−1.72 to −.43

Note. TPJ = Temporoparietal junction; B-H = Benjamini-Hochberg α-correction (set at .05) and bolded p-values survived this correction; Depression was measured using the Center for Epidemiological Studies Depression Scale (Radloff, 1977).

Figure 2.

This figure describes differences in suicidal ideation (SI) based on sexual minority status plotted as a function of left temporoparietal junction (TPJ) response to social cues.

Two-way Interaction of Sexual Minority Status with Neural Reactivity to Social Cues (Rewarding + Ambiguous; plotted at ±1 SD) in Predicting SI. Significant group differences in SI observed only at low (−1 SD) TPJ activation.

Note. SI was measured with the Suicidal Ideation Questionnaire (Reynolds, 1987) and square-root transformed for analyses. Regions of significance were determined using Johnson-Neyman technique. For ease of interpretation, simple slopes were plotted at ±1 SD of TPJ activity to depict group differences in SI. Alternate Johnson-Neyman plots including the full range of TPJ activity are in the supplement (Figure S3). SMY = Sexual minority youth

Follow-up analyses examining moderation by left TPJ activation during each task condition are described in the Supplemental Materials. In general, we found a similar pattern of findings where left TPJ activation during the rewarding (p=.001) and ambiguous (p=.003) task condition, respectively, moderated the relationship between sexual minority status and SI. These results, including simple slopes plots, are described in Table S2 and Figure S2.

Exploratory Analyses

Sexual Orientation Victimization as an Additional Moderator

Analyses testing three-way moderation including sexual orientation victimization as an additional moderator in the above models were not significant (Table S3). Therefore, the effects of left TPJ response to social cues on the relationship between sexual minority status and SI were not influenced by sexual orientation victimization exposure. Given our modest sample size, however, we were likely underpowered to test three-way moderation effects.

Discussion

There is a critical need to examine neurobiological processes to improve understanding of disparities in suicide for SMY, a population at notably high risk for suicide (Eckstrand, Lenniger, et al., 2022; Forbes et al., 2021). The present research is the first neuroimaging study to demonstrate that suicide risk in SMY, as indexed by SI, varies with function in neural social circuitry. Consistent with our broad hypothesis, neural reactivity to social cues, such as unfamiliar face stimuli, moderated the relationship between sexual minority status and SI: More severe SI was observed among SMY with dampened left TPJ activation to social cues compared with heterosexual peers. Interestingly, follow-up analyses indicated that this pattern of findings was not specific to either the rewarding or ambiguous task condition, suggesting consistency across some social contexts. Furthermore, moderation effects remained significant even after controlling for relevant covariates such as age and mental health symptoms (e.g., depression). Contrary to our expectations, no significant moderation effects were observed for brain activity in the other ROIs. There was also no additional moderation by prior sexual orientation victimization, although the analyses may be underpowered due to a modest sample size. Together, these findings suggest that neural processing of non-specific social cues, regardless of the degree of valence and certainty, play an important role in explaining individual differences in suicide risk among SMY, where those with blunted neural response to social cues in the left TPJ may be at higher risk.

Well-known disparities in mental health, including suicide, among SMY have persisted despite shifting societal attitudes (Meyer et al., 2021). Consistent with prior work, we found that SMY reported higher levels of depression, SI, and victimization than heterosexual peers (di Giacomo et al., 2018; Marshal et al., 2011). There was also a greater proportion of SMY reporting more severe SI and depressive symptoms compared with heterosexual youth. Importantly, this study used a matched-samples design that helped to control for potential confounding effects of other demographic variables relevant to experiences of discrimination and mental health (e.g., race/ethnicity). Our findings add to extant literature highlighting prevailing SMY mental health disparities and signal the need for further research and intervention with this at-risk population.

Results from the current study provided support for altered neural social circuitry as a moderator of SMY mental health disparities. This finding is consistent with past research demonstrating that individuals with altered neural sensitivity to social contexts are more susceptible to developing psychopathology in response to negative social environments (Schriber & Guyer, 2016). That we did not find moderation effects in the other ROIs related to reward (e.g., VS) and threat (e.g., amygdala) processing, however, suggests that altered functioning specific to social brain regions, such as the TPJ, may uniquely confer vulnerability for suicide among SMY. Relatedly, dampened TPJ response during viewing of negative emotional pictures has also been observed in a small sample of adolescents⁷ with SI and history of victimization (Miller et al., 2018). This overlapping pattern of findings highlights the importance of considering individual differences in neural social circuitry, such as TPJ activity, for understanding and mitigating suicide risk in vulnerable populations.

The left TPJ is a key neural region involved in social cognition and associated with the ability to perceive differences in the beliefs and intentions of the self and others (Aichhorn et al., 2009; Samson et al., 2004). Moreover, it has been implicated in social decision-making and trust (Engelmann et al., 2019). In the context of our study, left TPJ activity is especially relevant as we utilized a task involving expectations/responses to social cues across rewarding and ambiguous contexts. Prior work with different samples have found that lower TPJ activation to social reward among adolescents is associated with lower self-reported emotional closeness and positive affect during everyday social interactions, as well as more severe depressive symptoms (Eckstrand et al., 2019; Flores et al., 2018). Results from the present study extend these findings, suggesting that the TPJ is also relevant for understanding SMY suicide risk. Dampened reactivity in the left TPJ may reflect greater disengagement or diminished interest in social cues. However, compared with heterosexual youth, such disengagement may be more pernicious for SMY because they already tend to experience considerable social disadvantage due to sexual orientation stigma. As a result, SMY with lower TPJ response to social cues may be at greater risk for experiencing social disconnection (e.g., lack of affiliation), and in turn, heightened SI (Fulginiti et al., 2020).

Notably, we found moderation effects of left TPJ response on SI across both rewarding and ambiguous social contexts. These findings suggest that the observed effects reflect altered neural processing of non-specific social cues related to approval and inclusion, such as faces, regardless of valence or certainty. This is largely consistent with prior work suggesting that, due to sexual orientation stigma, sexual minorities learn to be more vigilant and anticipate negative social interactions such as rejection, regardless of whether they actually had such experiences (Meyer, 2003). Moreover, heightened sensitivity to rejection is common among sexual minorities and associated with negative mental health outcomes, including suicide (Feinstein, 2020; Mereish et al., 2019). Therefore, dampened left TPJ response to social cues may reflect a defensive response via cognitive disengagement due to anticipated social rejection. While such disengagement can be protective in the face of ambiguous or threatening social cues, doing so in potentially positive contexts may lead to missed opportunities for rewarding social interactions and increasing social affiliation with peers. In turn, the lack of social connection may heighten suicide risk for SMY. An important limitation, however, is that our fMRI task only included social stimuli. In order to test whether our findings actually reflect non-specific social processing, future research would need to include and compare with non-social stimuli (e.g., monetary reward/loss; Eckstrand et al., 2019).

Unexpectedly, there was no additional moderation by sexual orientation victimization. This is surprising since victimization has been associated with altered neural social circuitry (Casement et al., 2014), as well as with neural response to monetary reward and depression in SMY (Eckstrand, Silk, et al., 2022). One potential explanation for this null finding is that the average frequency of victimization in our sample was relatively low. This may be because we only assessed victimization experiences that occurred within the past six months rather than lifetime exposure or during vulnerable developmental periods (e.g., mid-adolescence), which could have a greater impact on brain development and social cognition (Forbes et al., 2021). Due to the modest sample size, we were also likely underpowered to detect significant three-way moderation effects. Alternatively, our findings could reflect that the link between victimization and SI is related to other neural systems relevant for processing social threat (Oppenheimer et al., 2020; Yang et al., 2023). Future research could consider including an ostensibly “threatening” condition to explore this possibility. Our findings also suggest the need to consider the influence of internal/proximal stressors—rather than only violent forms of discrimination—on SI among SMY. Indeed, past research has found that proximal minority stressors, such as expectations of rejection or internalized homophobia, are associated with SI among SMY (Goldbach et al., 2017; Lea et al., 2014). Therefore, future research should consider incorporating more comprehensive measures of minority stressors and examine their interactions with neurobiology.

The current study had several strengths, including a diverse sample of youth at varying SI risk, neuroimaging methods, and well-validated self-report measures. In addition, rates of clinically-significant SI and suicidal behaviors in our sample were consistent with prior estimates in community populations (Van Meter et al., 2023). Notably, our regression models accounted for a significant proportion of the variance (up to 53%) in SI. Nevertheless, our findings should be interpreted with the following limitations. First, there was relatively limited variability in SI, with participants reporting lower scores on average. Although this is perhaps unsurprising because suicide is a low base-rate behavior, it may limit generalization of our findings to individuals at higher risk for suicide (O’Connor & Portzky, 2018). Furthermore, we did not comprehensively examine participants’ history of suicidal and self-harm behaviors or medication usage. Future research should consider examining whether our findings might replicate in samples with more severe suicide risk. Next, we relied on a one-time self-report assessment of SI, which prevented examination of changes in SI over time. Given the time-varying nature of SI and suicidal behavior, future research should consider examining how individual differences in neural social circuitry influence SI in daily life, where we would also expect variability in the types of social interactions. Additionally, we only examined sexual orientation victimization which is but one form of stigma. Future studies should assess other dimensions of minority stress (e.g., rejection sensitivity, internalized homophobia) and intersectional identities (e.g., SMY who are also racial/ethnic minority, sex differences), and how they interact with neural social circuitry in predicting suicide risk in SMY. In addition, it would be important to explore the influence of resilience factors (e.g., social support). Finally, information regarding variability in gender identity of our sample was unknown due to limited available data. Given the high rates of SI among gender minority youth, this is an important future consideration.

In conclusion, despite its limitations, the present study is a critical first step towards understanding neurobiological factors influencing SMY suicide risk, with important clinical implications. We found that, relative to heterosexual peers, SMY with altered neural social circuitry experienced heightened SI. Importantly, these effects were evident across rewarding and ambiguous social contexts and were not further influenced by sexual orientation victimization exposure. Taken together, our findings suggest that individual differences in neural reactivity to social cues are a helpful indicator that should be considered in the understanding, prevention, and intervention of suicide and related outcomes.

Open Practices Statement.

Data or materials for the study are available upon request. This study was not preregistered.

Availability of data and materials (data transparency)

Data or materials for the study are available upon request.

Code availability (software application or custom code)

Data analysis code for the study is available upon request.