Skip to main content
NCBI home page
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2022 Sep 1.
Published in final edited form as: Behav Ther. 2021 Feb 2;52(5):1080–1092. doi: 10.1016/j.beth.2021.01.006

A Transdiagnostic Multi-level Examination of Interoceptive Processing in Individuals with a Remote History of Suicidal Behavior

Danielle C DeVille a,b, Sahib S Khalsa a,c; Tulsa 1000 Investigatorsa, Rachel C Lapidus a,b, Evan White a, Martin P Paulus a,c, Robin L Aupperle a,c
PMCID: PMC8403233  NIHMSID: NIHMS1697390  PMID: 34452663

Abstract

A developing area of research suggests that there may be a relationship between interoception and suicidal behavior. For example, it was recently reported that individuals who made a suicide attempt within the previous five years exhibit behavioral and neural abnormalities across multiple domains of interoception relative to non-attempters. This included increased tolerance for aversive sensations of pain and dyspnea, reduced heartbeat perception accuracy, and blunted insula activity during attention to cardiac sensations. However, the degree to which interoceptive deficits persist following a suicidal attempt is unknown. In the current study, we examined differences between individuals with a remote history of suicide attempts (greater than 5 years ago; N = 56) versus those with no history of attempts (N = 240). We found that remote suicide attempters demonstrated greater pain tolerance and lower ratings of stress during a cold-pressor challenge and lower ratings of suffocation during a breath-hold challenge as compared to non-attempters. In contrast, there were no group differences in breath hold duration, interoceptive accuracy on a heartbeat tapping task, or insula activation during cardiac attention. An exploratory resting-state functional connectivity analysis of individuals with suicide attempts in the past five years (N = 23), individuals with more remote histories of suicide attempts (N = 39), and non-attempters (N = 232) revealed preliminary and subtle evidence of differences in insula connectivity with areas of the temporal cortex in remote suicide attempters. Taken together, these findings suggest that blunted affective responses to aversive interoceptive sensations is an enduring characteristic of suicide attempters, even when assessed many years after a suicide attempt, whereas differences in the experience of non-aversive interoceptive sensations may be less persistent.

Keywords: interoception, suicide, pain perception, insula, breath-hold

Introduction

Interoception refers to the process by which the nervous system senses, interprets, and integrates signals originating from within the body (Khalsa et al., 2018). Interoceptive awareness is comprised of several facets, including (but not limited to) interoceptive attention (i.e., observation of internal sensations), interoceptive accuracy (i.e., the extent to which conscious perception of internal sensations align with physiological assessments), and the perceived intensity or magnitude of internal sensations (Khalsa et al., 2018; Khalsa & Lapidus, 2016). These various aspects of interoception occur both in and outside of conscious awareness, and therefore require multimodal assessment involving self-report, behavioral, and physiological forms of measurement. Neuroimaging assessments have identified the insular cortex as a potentially important hub within a larger prefrontal-cingulate-insular circuitry involved in interoceptive processing. Specifically, the dorsal mid-to-posterior insula is thought to be involved in the primary representation of visceral afferents from within the body whereas the more anterior regions are thought to be involved in the integration of interoception with cognition and emotion (Craig, 2003; Critchley et al., 2004; Hassanpour et al., 2018). There is an extensive body of literature linking aberrant interoception and/or insula functioning to psychopathology, including major depressive disorder (Avery et al., 2014; Wiebking et al., 2015), substance use disorders (Paulus & Stewart, 2014; Stewart et al., 2019), and eating disorders (Berner et al., 2018; Kerr et al., 2016; Khalsa et al., 2015), all of which are associated with increased risk for suicide (Harris & Barraclough, 1997; Nock et al., 2010).

In recent years, there has been increased interest in understanding the role of interoception in suicidal behaviors (DeVille et al., 2020; Dodd et al., 2018; Forrest et al., 2015; Smith, Forrest, & Velkoff, 2018). Interoception is an adaptive process that enables the preservation of the body’s internal milieu by supporting the perception of and response to aversive sensations from within the body during homeostatic threat (e.g., injury). Suicidal behaviors and other forms of self-directed physical harm defy this adaptive process. In a recent study, we examined a transdiagnostic sample of individuals who reported a history of suicide attempts within the past five years compared to a psychiatric reference sample of non-attempters matched for symptoms of psychopathology (DeVille et al., 2020). We noted that the suicide attempters exhibited increased tolerance for cold pain, increased tolerance for respiratory perturbation, reduced interoceptive awareness of heartbeat sensations, and blunted reactivity in the right mid and posterior insula during attention to cardiac sensations. The observed group differences appeared to be explained by suicide attempt history rather than severity of suicidal ideation (i.e., group differences remained when directly comparing ideators and attempters and when covarying for intensity of ideation). Based on these findings, we proposed that blunted sensitivity to interoceptive signals may result in an ability to ignore or override the cascade of interoceptive signals that protect against bodily harm during homeostatic threat, leading to a heightened capacity for suicidal behaviors (DeVille et al., 2020). This study provided supporting behavioral evidence for earlier work suggesting that individuals with a history of suicide attempts demonstrate interoceptive disturbances based on self-report measures such as the Interoceptive Deficits subscale of the Eating Disorders Inventory (Dodd et al., 2018; Forrest et al., 2015; Garner et al., 1983; Smith et al., 2018), and the Multidimensional Assessment of Interoceptive Awareness Scale (Mehling et al., 2012; Rogers et al., 2018).

However, this work leaves many unanswered questions. As prior behavioral and physiological work has focused on individuals with suicide attempts occurring within five years of assessment (DeVille et al., 2020), the persistence of interoceptive deficits in the years following a suicidal attempt is unknown. In a prior study, Forrest et al. (2015) examined self-reported interoceptive deficits among individuals with a history of suicide attempts, categorized by duration since most recent attempt (i.e., <1 month, 1–12 months, 1–5 years, and 5+ years, or non-attempters). They found that interoceptive deficits were most pronounced in the first five years following suicide attempts and leveled off thereafter, although subtle group differences between non-attempters and individuals with remote attempts (i.e., occurring more than 5 years ago) were maintained. The authors concluded that interoceptive deficits may be most pronounced around the time of the suicide attempt and become increasingly (although not completely) normal as time passes and the suicide attempt becomes more remote.

Examination of the extent to which behavioral and neural responses to aversive and non-aversive interoceptive stimuli endure in the years that follow a suicide attempt may provide further clarity about the nature of the relationship between interoception and suicidal behavior. If interoceptive deficits are more proximally associated with suicidal behavior, then some degree of normalization following suicide attempts would be expected. However, if interoceptive deficits are an enduring trait among individuals with heightened suicidal capacity (e.g., due to a genetic predisposition to heightened pain tolerance), it would be expected that interoceptive deficits would exist prior to the onset of suicidal behavior and persist thereafter. In the absence of longitudinal interoception data that examines pre-morbid interoceptive functioning among suicide attempters, examination of individuals several years after their suicide attempt may be revealing. In particular, if individuals with a remote history of suicidal behavior exhibit blunted interoceptive awareness relative to non-attempters that are not attributed to group differences in psychopathology (e.g., depression severity), then this may provide preliminary evidence suggestive of a broader, trait-like difference.

Additionally, the precise role of the insular cortex in the relationship between interoception and suicidal behaviors remains unclear. Although previous findings suggest that individuals with a history of suicide attempts exhibit blunted activity in the dorsal mid and posterior insula during attention to cardiac sensations (DeVille et al., 2020), the extent to which this reflects broader dysfunction of insula neurocircuitry is unknown. Prior anatomical investigations have found that suicide attempters have reduced insula volume (e.g., Giakoumatos et al., 2013; Soloff et al., 2012). Additionally, a retrospective analysis showed that a cohort of individuals who later died by suicide had reduced cerebral blood flow to the insula relative to non-suicidal individuals (e.g., Willeumier et al., 2011). These findings introduce the possibility of broader insula pathophysiology (i.e., not limited to its reactivity during specific tasks). Resting state fMRI provides a unique opportunity for assessing the functional coupling, or functional connectivity between specific regions of interest and the wider brain, independent of specific task demands or performance differences (Friston et al., 1993). Functional connectivity of the insula has not yet been explored in relation to suicidal behaviors.

The aim of this study was to examine whether individuals with a remote history of suicide attempts (i.e., attempts occurring more than five years ago) exhibit interoceptive deficits across behavioral and neural assessments of interoception. To this end, we examined the primary tasks reported in the DeVille et al (2020) investigation, including a breath hold task, cold pressor task, heartbeat perception task, and interoceptive attention neuroimaging task. However, in contrast with our prior study, we compared remote suicide attempters to non-attempters on these tasks. We hypothesized that remote attempters would exhibit increased tolerance for aversive interoceptive sensations (i.e., cold pain and breath hold), reduced heartbeat perception accuracy, and reduced insula response to interoceptive attention relative to non-attempters. In consideration of the findings of Forrest et al. (2015), we anticipated the possibility that the differences between remote attempters and non-attempters may be small. Additionally, we sought to clarify whether suicide attempters exhibit a general, task-independent dysfunction of insula connectivity. To this end, we conducted an exploratory analysis of mid and posterior insula functional connectivity across three groups: individuals with more recent attempts (i.e., within five years of study entry, as reported in DeVille et al., 2020), individuals with more remote attempts (i.e., greater than five years), and non-attempters. Due to the lack of available literature and the exploratory nature of this analysis, we did not have directional hypotheses regarding functional connectivity.

Methods and Materials

Participants

The participants in this study were drawn from a pre-existing dataset from the Tulsa-1000 study, a naturalistic investigation that aims to longitudinally follow individuals with mood, anxiety, substance use, and/or eating disorders. Participants within the Tulsa-1000 study were recruited regionally using media advertisements and advertisement at local mental health clinics. Additional recruitment and protocol information is provided in Victor et al. (2018). Participants were considered eligible for entry into the larger study if they fulfilled any of the following criteria: Patient Health Questionnaire (Kroenke et al., 2001) ≥ 10 and/or Overall Anxiety Severity and Impairment Scale (Campbell-Sills et al., 2009) ≥ 8, and/or Drug Abuse Screening Test (McCabe et al., 2006) score > 3, and/or Eating Disorder Screen (Morgan et al., 2000) score ≥ 2. A number of other self-report and clinician-administered assessments were included as part of this larger study, including the Columbia Suicide Severity Rating Scale (CSSRS; Posner et al., 2011). Participants’ reports of lifetime suicide attempt history on the CSSRS were used to guide selection for inclusion in the current study, and participants were separated into groups based on the reported date of their most recent suicide attempt. All participants in the larger study provided written informed consent and received financial compensation for their involvement, and all procedures were approved by an Institutional Review Board.

It was noted that 56 individuals from the larger study reported suicide attempts occurring more than 5 years ago (SA5+). In the analysis of the breath hold challenge, cold pressor challenge, heartbeat perception, and interoceptive attention imaging task, we compared these 56 SA5+ participants to 240 non-attempters (NA). Interoception task data from participants with suicide attempts occurring within the past 5 years were already analyzed in DeVille et al. (2020) and therefore were not included in the analysis of interoceptive task data.

As our prior investigation (i.e., DeVille et al., 2020) did not include an assessment of insula functional connectivity, we were able to include three groups in the resting state analysis. This consisted of: 1) individuals who reported having made a suicide attempt within 5 years of data collection (SA<5), 2) individuals who report a history of suicide attempts more than 5 years prior to data collection (SA5+), and 3) non-attempters (NA). Due to the strong influence of excess motion on resting state data quality, individuals with an overall censor fraction exceeding 15% were excluded for resting state analysis. The final sample for this analysis included 23 SA<5 participants, 39 SA5+ participants, and 232 NA participants. Detailed information regarding participant motion and data exclusion is provided in the Appendix. Additionally, a summary of participant demographics for the final resting state sample is also included in the Appendix.

Procedures

Breath-Hold Challenge

Each participant completed two inspiratory breath-hold trials, providing a brief measure of sensitivity to respiratory perturbation. Participants were seated in front of a computer screen, fitted with a respiration belt (Biopac Systems, Inc.), provided with a nose clip, and a tube-like breathing apparatus was placed into the mouth. During normal breathing, concentrations of oxygen and carbon dioxide were analyzed from their exhaled air, providing a baseline measurement. Participants were then instructed to inhale maximally and, at the end of inhalation, to begin holding their breath for as long as they were able to tolerate. Trial duration was limited to two minutes, with a two-minute rest period between each trial. Participants were not informed of the time limit. Participants were instructed to exhale into the breathing apparatus when they were no longer able to tolerate the breath-hold. Following each breath-hold, participants provided ratings of the task (i.e., intensity, unpleasantness, and difficulty) as well as ratings of associated psychological experiences (i.e., stress, required effort, breathlessness, urge to breathe, sensations of suffocation, fear of suffocation) on a visual analogue scale (VAS) ranging from Not at all (0) to Extremely (100).

Cold Pressor Challenge

Participants immersed their dominant hand in a circulating pool of water cooled to 6 degrees Celsius. They were asked to keep their hand submerged for as long as they could tolerate. Task duration was limited to two minutes although this was not disclosed to participants. During the task, participants made continuous real-time pain intensity ratings on a scale ranging from No pain) (0) to Worst pain imaginable (100). These ratings were used to calculate each participant’s peak pain rating, as well as the time to reach ratings of mild (25/100), moderate (50/100), and peak pain. Afterwards each participant provided VAS ratings of unpleasantness, difficulty, and stress ranging from Not at all (0) to Extremely (100).

Heartbeat Perception Task

To assess cardiac interoception, participants performed a heartbeat tapping task across three interoceptive conditions. Across each condition, participants were instructed to press a key on a keyboard every time they felt their heartbeat, without taking their pulse. Each trial was 60 seconds in duration. In the first trial (‘guess’), subjects were instructed to tap every time they felt their heartbeat without taking their pulse. Guessing was encouraged if they felt unsure. In the next trial (‘no guess’), guessing was discouraged and participants were asked to tap only when they felt confident in feeling their heartbeat. In the final trial (‘perturbation’), participants were instructed to inhale deeply, hold their breath, and tap along with their perceived heartbeats while sustaining the breath-hold, without guessing. The breath-hold was expected to amplify cardiac sensations and presumably increase heartbeat perception accuracy. Heartbeat perception accuracy was calculated using a common accuracy metric (Schandry, 1981). Afterwards, participants provided VAS ratings ranging from Not at all (0) to Extremely (100) to indicate their perceived heartbeat intensity, confidence in their ability to accurately estimate their heartbeat, and their assessment of task-related difficulty.

Interoceptive Attention fMRI Task

The interoceptive attention task engages top-down attention directed toward naturally-occurring interoceptive sensations in order to amplify activity in brain regions underlying interoceptive processing. Previous studies have demonstrated that this task is effective at mapping the neural basis of interoceptive attention in healthy individuals as well as those with major depressive disorder, eating disorders, and substance use disorders (Avery et al., 2014; Kerr et al., 2016; Simmons et al., 2013; Stewart et al., 2019). The task consisted of two types of trials: interoceptive trials and exteroceptive trials. During the interoceptive trials, the words “HEART” or “STOMACH” were presented in a black font against a white background with each trial lasting 10 seconds. Due to our primary focus on interoceptive signals pertaining to the cardiovascular system and in an effort to mirror our prior findings (DeVille et al., 2020), trials involving stomach interoception were not examined in the current study. During the exteroceptive trials, the word “TARGET” was presented on the screen in black text against a white background. The color of the word periodically changed from black to various lighter shades of gray; throughout the duration of the 10-second trial, participants were instructed to focus on the intensity of the color change. Heart, stomach, and target stimulus trials were each presented 12 times. To ensure that participants remained attentive during the task, participants were asked to rate the intensity of the sensations from their heart or stomach or the intensity of the color change following approximately one-half of the trials. Participants performed this task over two scanning runs, each lasting six minutes.

Resting State

For the duration of the resting state scan, participants were instructed focus their gaze on a cross in the center of the screen, to stay awake, and to avoid thinking about anything in particular. The rest scan lasted eight minutes.

Statistical Approach

We conducted analyses of demographic, clinical and behavioral data using the R base statistical software package version 3.5.1 (R Core Team, 2018). Linear mixed effects (LME) analyses were conducted using the ‘lmerTest’ package (Kuznetsova et al., 2017). A marginal analysis of variance (ANOVA) was applied to each LME model to examine F-tests for interactions and main effects. In the event of significant interactions, fixed effect summaries were examined for clarification. The Kenward-Roger approximation of degrees of freedom was used for all LME analyses. To provide estimates of effect size for significant findings, R2 statistics for the ANOVA main effects were computed using the ‘r2glmm’ package in R (Jaeger, 2017) as described in (Edwards et al., 2008) and Cohen’s d was calculated for each level of the LME fixed effects using lme.dscore in the ‘EMAtools’ package (Kleiman, 2017). Figures were created using the ‘ggplot2’ package (Wickham, 2011). Neuroimaging data were analyzed using the Analysis of Functional Neuroimages (AFNI) program (Cox, 1996).

Data cleaning and pre-processing

Behavioral data pre-processing

Continuous data that were significantly skewed were log transformed. For the behavioral and self-report data, k-nearest neighbors imputation was used to impute missing data (Jadhav, Pramod, & Ramanthan, 2019; Troyanskaya et al., 2001). A small number of individuals were determined to have data missing from certain tasks on the basis of non-adherence to task instructions (i.e., manually taking their pulse during the heartbeat perception task; 14 non-attempters; 5 remote attempters), experimenter error (1 non-attempter, 2 remote attempters in the breath hold task), and refusal to complete the cold pressor task (1 non-attempter). Data for these individuals were not missing at random and therefore were not imputed. This resulted in a total of 293 participants for the breath-hold task (55 remote attempters and 238 non-attempters), 294 participants for the cold pressor task (55 remote attempters and 239 non-attempters), and 277 participants for the heartbeat perception task (51 remote attempters and 226 non-attempters). However, LME models are robust to small amounts of missing data. Between-group homogeneity of variance was confirmed using Levene’s test.

Neuroimaging data acquisition and pre-processing

Structural and functional magnetic resonance images were acquired using a General Electric Discovery MR750 3 Tesla MRI scanner and 8-channel head array coil. A 3D MPRAGE sequence obtained high-resolution anatomical images used as an anatomical reference for the functional data (field of view=240mm x 192mm, slices/volume (axial)=186, slice thickness = 0.9mm, image matrix=256x256, voxel volume=0.938x0.938x0.9mm, repetition time/echo time = 5/2.012ms, acceleration factor = 2, flip angle = 8°, inversion/delay time = 725/1400ms) using an 8-channel head coil. Functional data were collected as echo-planar image (EPI) volumes depicting blood oxygen-level dependent (BOLD) contrast (180 volumes per run across 2 runs for the interoceptive attention scan, and 240 volumes for the resting scan; slice thickness = 2.9mm, image matrix = 128x128, voxel volume=1.875x1.875x2.9mm, acquisition matrix=96x96, repetition time=2000ms, echo time=27ms, flip angle=78°) with a sensitivity encoding factor of 2.

Data preprocessing was conducted using afni_proc.py (Cox, 1996). Six motion parameters were estimated using 3dVolreg (3 translations and 3 rotations). The Euclidean norm of the derivative of the six parameters was taken as a measure of motion per repetition time. This was averaged across the entire scan to yield a single metric of overall motion. The first three volumes of the functional scans were discarded to allow the signal to reach T1 equilibrium, and a de-spiking algorithm was used to remove any transient signal spikes from the data. For each participant, the remaining volumes were corrected for differences in slice acquisition time; head motion was corrected by rigid body translation and rotation; the first volume of the functional run (before discarding three volumes) was co-registered to the anatomical coordinates of the participant’s structural scan by linear warping, then normalized to the Montreal Neurological Institute template and resampled to 2x2x2 mm3 voxels. The EPI data were then smoothed using a 4-mm full-width at half-maximum Gaussian kernel, and the value for each EPI volume was normalized to percent signal change using each voxel’s average signal across the time course.

For the task-based analysis of cardiac interoceptive attention, the imaging data were analyzed at the subject level using a multiple linear regression model, with regressors for each task condition (i.e., heart attention, stomach attention, exteroceptive attention, and response periods). To adjust the model for the shape and delay of the BOLD function, task regressors were constructed by convolution of a block function having a 5- or 10-second width (depending on the trial duration) beginning at the onset of occurrence of each condition. Nuisance regressors included each run mean, linear, quadratic, and cubic signal trends, and motion variables.

Analysis of Behavioral Tasks

For the breath-hold challenge, cold pressor challenge, and heartbeat tapping task, linear mixed effects (LME) models were used to examine the relationship between group and the dependent measure for each task. A participant identifier was included as a random effect within each model and intensity of suicidal ideation was included as a covariate in each model. Individual fixed effects specifications are provided for each model below. Additionally, the relationship between group and task-related VAS ratings were examined using linear models, with the VAS rating modeled as the dependent variable, group included as a predictor variable, and intensity of ideation included as a covariate. Benjamini-Hochberg adjustments were applied within each set of analyses to account for repeated testing.

Breath-Hold and Cold Pressor Challenges

For the breath hold task, duration in seconds was the primary dependent measure. Group, trial repetition, and the interaction between group and trial were included as fixed effects. For the cold pressor challenge, duration in seconds the primary dependent measure. Group, timepoint (i.e., markers of mild pain, moderate pain, peak pain, and task discontinuation), and the interaction between group and timepoint were included as fixed effects.

Heartbeat Perception Task

Interoceptive accuracy during a heartbeat tapping task was included as the dependent measure, using the traditional formula drawn from heartbeat counting studies (Schandry, 1981). We focused on the no-guess and perturbation conditions of this task due to known limitations of the guessing condition (e.g., Desmedt et al., 2018). Group, condition (i.e., no-guess and perturbation), and the interaction between group and condition were included as fixed effects.

Analysis of Interoceptive Attention fMRI Task

We identified the insular cortex as the a priori region of interest due to its documented involvement in interoceptive attention (Avery et al., 2014; Kerr et al., 2016; Simmons et al., 2013), and focused our analysis on the right insula following previous observation of blunted activity in this region. A mask of the entire right insula was defined using the Human Brainnetome Atlas (Fan et al., 2016). The beta values derived from the contrast of cardiac interoception versus the exteroceptive control condition, reflecting the mean percent signal change during interoceptive attention relative to exteroception, were extracted for each subject and a compared using the AFNI program 3dttest++, with a voxel-wise threshold of p < .005 and a cluster-threshold of 20 voxels. Intensity of ideation was included as a covariate.

Analysis of Mid and Posterior Resting State Functional Connectivity

We selected the right dorsal granular insula and right dorsal dysgranular insula as seed regions for the resting state functional connectivity analyses (as defined by the Brainnetome Atlas). These regions are implicated in somatosensation and pain perception and overlap with the regions identified by DeVille et al. (2020). For each participant, the time course of the residual images from the preprocessing step were averaged across voxels within each seed region, and Pearson’s correlation coefficients were computed between the seed and all other voxels in the brain. We performed an r-to-Z transform on these values and used the AFNI program 3dMVM to identify regions exhibiting group differences in spontaneous signal fluctuations correlated with the seed region fluctuations. Scores on the SCOFF, DAST, OASIS, and PHQ were included as covariates. As with all other analyses, intensity of ideation was included as a covariate. We compared the Z-scores generated for the SA<5, SA5+, and NA groups. Two contrasts were examined: 1) differences between all attempters and non-attempters (i.e., SA<5 + SA5+ versus NA) and 2) differences between the two suicide attempter groups (i.e., SA<5 versus SA5+). A voxel-wise threshold of p < .005 was applied (with two-sided thresholding and second nearestneighbor clustering). A cluster threshold of 143 voxels for protection against false positives was determined through AFNI’s 3dClustSim using the spherical autocorrelation function parameters determined by 3dFWHMx. This approach to correction in accordance with the most recent recommendations for minimizing false discovery in neuroimaging analyses (Cox, Chen, Glen, Reynolds, & Taylor, 2017).

Results

Demographics and Clinical Characteristics

The clinical characteristics of the groups compared in the breath-hold, cold pressor, heartbeat perception and interoceptive attention task analyses are summarized in Table 1. There were no significant differences between groups on scores related to symptoms of anxiety, depression, substance use, or disordered eating; although individuals with a remote history of suicide attempts had significantly higher scores on the measure of lifetime suicidal ideation intensity. Clinical characteristics of the groups included in the resting state analyses are summarized in the Appendix. Diagnostic information for all groups is provided in the Appendix.

Table 1.

Demographics and Clinical Characteristics for Cold Pressor, Breath Hold, Heartbeat Perception, and Heartbeat Attention fMRI Task

SA5+ (n = 56) NA (n = 240) p
Demographics
 Age, years 37 (11) 35 (10) .09
 BMI 28.2 (4.7) 28.1 (5.1) .91
Sex, n (%) .37
 Male 15 (26.8%) 82 (34.2%)
 Female 41 (73.2%) 158 (65.8%)
Race/Ethnicity, n (%) .65
 Asian 0 (0%) 2 (<1%)
 Black 2 (3.6%) 18 (7.5%)
 Hispanic 4 (7.1%) 9 (3.8%)
 Native American 12 (21.4%) 41 (17.1%)
 White/Caucasian 35 (26.8%) 159 (66.2%)
 “Other” 3 (5.4%) 11 (4.6%)
Clinical Features
 C-SSRS Ideation Severity 3.54 (1.89) 0.99 (1.57) <.001
 C-SSRS Intensity of Ideation 12.5 (6.7) 5.1 (6.8) < .001
 PHQ-9 Suicidality Past 2 Weeks 0.29 (0.56) 0.24 (0.54) .56
 PHQ-9 Total 10.7 (5.4) 9.7 (6.3) .29
 OASIS 9.1 (3.7) 7.9 (4.6) .06
 DAST 3.6 (3.7) 3.0 (3.7) .33
 SCOFF 0.7 (1.0) 0.9 (1.2) .46
 % Unmedicated 32 (n=18) 45 (n=108) .11
Open in a new tab

Note: All values reported are in the format of Mean (SD) unless otherwise indicated. SA5+ = participants with suicide attempts occurring more than 5 years prior to study entry; NA = Non-attempters; BMI = Body Mass Index; PHQ-9=Patient Health Questionnaire; IOI = Intensity of Ideation; OASIS=Overall Anxiety Severity and Impairment Scale; DAST=Drug Abuse Screening Test; SCOFF=Eating Disorders Screening Tool. All scores on clinical measures reflect total scores. unless otherwise specified. For all clinical measures, higher numbers indicate greater endorsement of the construct assessed.

Breath-Hold Challenge

Attempt status was significantly associated with VAS ratings of suffocation, such that SA5+ group reported lower sensations of suffocation than those in the NA group, t(287) = −2.71, p = .007, corrected p = .045, d = −.24 (see Figure 1). All other VAS rating group comparisons were non-significant (corrected p ranging from .339 to .860). For the LME model examining breath-hold duration, there was a significant main effect of trial, F(1, 287.0) = 14.27, p < 0.001, R2 = 0.042, such that both groups held their breath for significantly longer in the second trial relative to the first trial. Although individuals in the SA5+ group held their breath approximately 4 seconds longer on average, there was no significant effect of group, F(1, 355.0) = 0.08, p = 0.78, and no group by trial interaction, F(1, 288.0) = 0.16, p = 0.69.

Figure 1. Behavioral and subjective responses to aversive interoceptive stimuli.

Figure 1

Open in a new tab

Note: Individuals with a remote history of suicide attempts (SA5+) reported higher pain tolerance and lower stress ratings during the cold pressor task than non-attempters (NA). Similarly, remote attempters report lower feelings of suffocation following the breath hold task. The mean values observed among individuals with more recent attempts (i.e., within five years) as previously reported by DeVille et al (2020) are denoted with a diamond symbol for reference.

Cold Pressor Challenge

Regarding the linear model examining VAS ratings, the SA5+group provided lower ratings of stress related to the cold-pressor task than the NA group, t(289) = −3.11, p = 0.002, corrected p = .008, d = −.41 (see Figure 1). There were no significant group differences in ratings of difficulty, unpleasantness, and pain (corrected p ranging from .175 to .188). For the LME model examining cold pressor duration, there was a significant main effect of timepoint, F(3, 867.4) = 346.6, p < .001, R2 = 0.51, which was qualified by a significant interaction between timepoint and group, F(3, 867.8) = 2.8, p = .041, R2 = 0.01. Examination of the LME fixed effects revealed that, while there were no significant group differences in the amount of time elapsed prior to reaching mild, moderate, and peak pain, remote suicide attempters kept their hands submerged in the cold water for significantly longer than non-attempters (i.e., approximately 7 seconds longer on average; t(867.1) = 2.2, p = 0.027, d = .15).

Given that group differences in task duration were observed for the cold pressor analysis but not the breath-hold analysis, we examined differences in task aversiveness. In a post-hoc analysis comparing subjective responses to the two tasks, both groups reported that the cold pressor task was significantly more difficult, t(288) = 9.23, corrected p < 0.001, d = 1.09, stressful, t(288)=8.88, corrected p < .001, d = 1.21, and unpleasant, t(288)=10.23, corrected p < .001, d = 1.05, relative to the breath hold task.

Heartbeat Perception and Interoceptive Attention Tasks

For the LME model examining heartbeat perception accuracy, the effects of condition, F(1, 263.0) = .27, p = .61, and group, F(1, 384.5) = .03, p = .862, were non-significant, and there was no significant interaction between group and condition, F(1, 263.0) = 0.28, p = 0.59. Similarly, there were no group differences in VAS ratings of overall difficulty, intensity, or confidence regarding the heartbeat perception task (corrected p ranging from .114 to .777). For the interoceptive attention fMRI task, we found no significant group differences between the SA5+ and NA groups within the right insula during the cardiac attention condition (relative to exteroceptive attention).

Insula Functional Connectivity

Across the two contrasts explored, we identified a number of regions with reduced functional connectivity to the dorsal dysgranular insula and dorsal granular insular seeds; however, none of the clusters identified were large enough to exceed a cluster correction threshold of 143 voxels (1,144 mm3) determined by 3dClustSim. As there are no published studies on insula functional connectivity in suicide attempters, we present the uncorrected maps in the Appendix to guide future work (e.g. hypothesis generation) in this area.

Supplemental Analyses

Sensitivity analyses without covariates showed results that were consistent with the primary analyses. Moreover, analyses of the resting state data using contrasts with un-pooled variances were largely consistent with the original results. This is detailed in the Appendix. Post-hoc correlations examining relationships between duration since participants’ most recent suicide attempt and scores on interoceptive measures were non-significant (all p-values > 0.10).

Discussion

This study aimed to examine interoceptive processing differences between individuals with a remote history of suicide attempts relative to non-attempters and yielded two main results. Relative to non-attempters, individuals with a remote history of suicide attempts exhibited increased tolerance for pain sensations and reduced feelings of suffocation during a breath-hold challenge. However, there were no differences between remote attempters and non-attempters in the cardiac interoception tasks. Taken together, we obtained partial support for our hypotheses, such that remote attempters exhibited an enduring blunted affective response to aversive interoceptive sensations, but no differences in responses to non-aversive interoceptive sensations.

Regarding the response to aversive interoceptive stimuli, we found that individuals with a remote history of suicide attempts exhibited a heightened tolerance for pain in the cold pressor task, despite the fact that over five years had passed following the occurrence of suicidal behavior. As we expected, the difference in task duration was smaller than what has been previously observed in individuals with more recent suicide attempts (i.e., remote attempters tolerated the cold pressor for approximately 7 seconds longer than non-attempters, whereas it has previously been shown that individuals with more recent attempts tolerated a cold pressor task for approximately 17 seconds longer than non-attempters; DeVille et al., 2020). With respect to the breath hold task, although the overall pattern exhibited by the remote attempters was in the same direction as what has previously been shown in more recent attempters (DeVille et al., 2020), the difference was rather small and not statistically significant. Remote attempters held their breath for about four seconds longer than non-attempters on average, whereas in prior research the more recent attempters sustained the breath hold for approximately 10 seconds longer than non-attempters (DeVille et al., 2020). Regarding their subjective appraisal of the aversive tasks, remote attempters reported significantly lower ratings of stress following the cold pressor challenge and lower sensations of suffocation following the breath-hold challenge. This suggests that, to some extent, suicidal behavior may be associated with an enduring ability to withstand pain sensations and/or an enduring reduction in the perceived salience of homeostatic threat. However, the group differences in task appraisal and task performance were relatively small, this was expected given prior research. In our sample, an extensive period of time (i.e., ranging from 5 to over 40 years) had passed between their most recent suicide attempt and the time of measurement. Given the findings of Forrest et al. (2015) suggesting that the overall magnitude of this deficit normalizes with time, we examined post-hoc correlations between the amount of time since the most recent suicide attempt and performance on the aversive interoceptive tasks. We did not observe a relationship between the amount of time since participants’ most recent suicide attempt and interoceptive task performance. Prospective research that follows individuals in the immediate aftermath of a suicide attempt may be helpful for clarifying the extent to which interoceptive deficits change over time.

Given that both the cold pressor and breath hold are designed to assess aversive aspects of interoception, it is interesting that both were not sensitive to behavioral differences (i.e., duration of stimulus exposure prior to task discontinuation). Post-hoc analyses indicated that across both groups, the cold pressor task was rated as more difficult, stressful, and unpleasant than the breath hold task. In this context, the breath hold challenge provides a weaker or less salient perturbation than the cold pressor challenge. As a result, this task may be less sensitive to subtle group differences in the ability to withstand aversive interoceptive sensations. If replicated, this finding may need to be taken into consideration for future work on the topic of homeostatic perturbation and suicidal behavior. For example, it may be beneficial to increase the unpleasantness of the breath hold task by increasing the number of trials administered. Additionally, exploration of other forms of respiratory perturbation may be worthy of consideration (e.g., resistive breathing loads [see Berner et al. (2018)] or pharmacological modulation [see Hassanpour et al. (2018)]).

In contrast to what has been previously observed among more recent attempters, there were no significant differences between remote attempters and non-attempters on heartbeat perception accuracy, nor were there differences in insula activation during attention to naturally-occurring heart sensations. It is possible that differences in the processing and awareness of cardiac signals (and non-aversive interoceptive stimuli more generally) are more transient characteristic of individuals who have made a suicide attempt, whereas differences in response to aversive stimuli are more enduring. However, there were no significant correlations between the amount of time following participants’ most recent suicide attempts and their cardiac interoception measures. Thus, further research is needed to clarify the relationship between cardiac interoception and suicidal behaviors across time.

Although Forrest et al. (2015) demonstrated that self-reported deficits in interoceptive awareness persist to a small extent more than five years following suicidal behavior, the assessment of interoceptive awareness used in their study (i.e., the interoceptive deficits subscale of the Eating Disorders Inventory (EDI)) focuses primarily on the experience of interoceptive sensations with regard to emotional arousal and homeostatic need (i.e., hunger and fullness sensations). As a self-report measure, the EDI also does not account for the unconscious or implicit aspects of interoception (e.g., accuracy, attention, adaptive responses to perturbation; Khalsa & Lapidus, 2016). Further, the lack of significant findings between groups in insula activation during focused attention on naturally-occurring cardiac sensations does not preclude the possibility that remote attempters exhibit aberrant insula functioning in other settings or during other tasks. It may still be worth examining insula responses to aversive and/or painful interoceptive stimuli among remote attempters, given the differences they exhibited in response to homeostatic perturbation.

While there is some indication that individuals with certain psychiatric disorders exhibit altered functional connectivity of the insula (Avery et al., 2014; Hu et al., 2019; Wang et al., 2018), this has not been previously explored in relation to suicide. Although the regions identified in the resting state functional connectivity analyses did not survive cluster correction, we present our uncorrected findings of this exploratory analysis to guide future work (e.g., hypothesis generation) in this area. As an example, more recent attempters had evidence of stronger connectivity between the insula and the right middle temporal gyrus and the left inferior temporal gyrus when compared to remote attempters. The findings of reduced insula to temporal and insula to parahippocampal connectivity, assuming they can be confirmed in a larger sample using conventional statistical thresholds, are interesting in light of prior work implicating these regions in suicidal behavior (see Schmaal et al., 2020 for review). However, these preliminary findings should be interpreted cautiously given the potential risk of false positives in this analysis.

Limitations and Future Directions

In the absence of longitudinal data, we are unable to determine whether abnormal interoception in suicide attempters reflects an overarching trait that increases susceptibility to suicide or if it is reflective of a change in self-perception following a suicide attempt. Although it is plausible that individuals with particularly blunted response to homeostatic threat and/or low interoceptive awareness may be likely to make multiple suicide attempts and/or engage in more lethal forms of suicidal behavior, we did not account for number of previous attempts or the lethality of past suicidal behaviors in the current study. This may be an important area of investigation in future studies on interoception and suicide. While a strength of the current study was the inclusion of multiple methods of assessing interoception, there are potentially important aspects of interoception that we were unable to address (e.g., neural responses to aversive interoceptive sensations). Future longitudinal research is warranted to determine whether this represents an increased propensity for suicide (i.e., one that exists prior to attempts), whether blunted response to homeostatic threat in the years following suicidal behavior is associated with subsequent re-attempts, and whether interventions aimed at modifying interoceptive processing may influence suicide risk.

Conclusions

In the current study, we found that individuals who attempted suicide between five and 40 years ago exhibit heightened pain tolerance, lower stress levels during the experience of pain, and reduced feelings of suffocation during an inspiratory breath hold challenge relative to non-attempters. These findings support the assertion that blunted affective responses to aversive interoceptive stimuli endure in the years following suicide attempts.

Supplementary Material

appendix

Highlights.

  • Interoception was assessed in remote suicide attempters and non-attempters.

  • Interoception was measured using a battery of behavioral measures.

  • Remote suicide attempters had higher pain tolerance on a cold pressor task.

  • Remote suicide attempters reported less suffocation on a breath hold task.

  • Suicide attempters may exhibit persistent aversive interoceptive processing deficits

Acknowledgments

“Tulsa 1000 Investigators” includes the following additional contributors: Jerzy Bodurka, Ph.D., Justin S. Feinstein, PhD., Rayus Kuplicki, Ph.D., Jonathan Savitz, Ph.D., Jennifer L. Stewart, Ph.D., and Teresa A. Victor, Ph.D. The authors thank the assessment team and MRI technologists for their help with data collection, and Austin Lignieres, James Hale, and Max Paulus for assisting with physiological data inspection and correction.

Funding:

This work was supported in part by The William K. Warren Foundation, by National Institute of Mental Health grants K23MH112949 and K23MH108707, and the National Institute of General Medical Sciences Center grant 1P20GM121312. The funders had no involvement in the study design, collection, analysis, or interpretation of the data.

Abbreviations

ANOVA

Analysis of variance

BMI

Body mass index

BOLD

Blood oxygen level dependent

DAST

Drug Abuse Screening Test

EPI

Echo-planar image

fMRI

Functional magnetic resonance imaging

IOI

Intensity of ideation

NA

Non-attempter

OASIS

Overall Anxiety Severity and Impairment Scale

PHQ-9

Patient Health Questionnaire

SA<5

Suicide attempt within 5 years of study entry

SA5+

Suicide attempt greater than 5 years prior to study entry

SCOFF

Eating Disorders Screening Tool

VAS

Visual analogue scale

Footnotes

Declarations of Interest: None

Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

References

  1. Avery JA, Drevets WC, Moseman SE, Bodurka J, Barcalow JC, & Simmons WK (2014). Major depressive disorder is associated with abnormal interoceptive activity and functional connectivity in the insula. Biological Psychiatry, 76(3), 258–266. 10.1016/j.biopsych.2013.11.027 [DOI] [PMC free article] [PubMed] [Google Scholar]
  2. Berner LA, Simmons AN, Wierenga CE, Bischoff-Grethe A, Paulus MP, Bailer UF, Ely AV, & Kaye WH (2018). Altered interoceptive activation before, during, and after aversive breathing load in women remitted from anorexia nervosa. Psychological Medicine, 48(1), 142–154. 10.1017/S0033291717001635 [DOI] [PMC free article] [PubMed] [Google Scholar]
  3. Campbell-Sills L, Norman SB, Craske MG, Sullivan G, Lang AJ, Chavira DA, Bystritsky A, Sherbourne C, Roy-Byrne P, & Stein MB (2009). Validation of a brief measure of anxiety-related severity and impairment: The overall anxiety severity and impairment scale (OASIS). Journal of Affective Disorders, 112(1–3), 92–101. 10.1016/j.jad.2008.03.014 [DOI] [PMC free article] [PubMed] [Google Scholar]
  4. Cox RW (1996). AFNI: Software for analysis and visualization of functional magnetic resonance neuroimages. Computational Biomedical Research, 29(3), 162–173. 10.1006/cbmr.1996.0014 [DOI] [PubMed] [Google Scholar]
  5. Cox RW, Chen G, Glen DR, Reynolds RC, & Taylor PA (2017). FMRI clustering in AFNI: false-positive rates redux. Brain connectivity, 7(3), 152–171. 10.1089/brain.2016.0475 [DOI] [PMC free article] [PubMed] [Google Scholar]
  6. Craig AD (2003). Interoception: The sense of the physiological condition of the body. Current Opinion in Neurobiology, 13(4), 500–505. https://doi.org/10/cfrtgc [DOI] [PubMed] [Google Scholar]
  7. Critchley HD, Wiens S, Rotshtein P, Ohman A, & Dolan RJ (2004). Neural systems supporting interoceptive awareness. Nature Neuroscience, 7(2), 189–195. 10.1038/nn1176 [DOI] [PubMed] [Google Scholar]
  8. Desmedt O, Luminet O, & Corneille O (2018). The heartbeat counting task largely involves non-interoceptive processes: Evidence from both the original and an adapted counting task. Biological Psychology, 138, 185–188. 10.1016/j.biopsycho.2018.09.004 [DOI] [PubMed] [Google Scholar]
  9. DeVille DC, Kuplicki R, Stewart JL, Tulsa I, Aupperle RL, Bodurka J, Cha YH, Feinstein J, Savitz JB, Victor TA, Paulus MP, & Khalsa SS (2020). Diminished responses to bodily threat and blunted interoception in suicide attempters. eLife, 9, e51593. 10.7554/eLife.51593 [DOI] [PMC free article] [PubMed] [Google Scholar]
  10. Dodd DR, Smith AR, Forrest LN, Witte TK, Bodell L, Bartlett M, Siegried N, & Goodwin N (2018). Interoceptive deficits, nonsuicidal self-injury, and suicide attempts among women with eating disorders. Suicide & Life-Threatening Behavior, 48(4), 438–444. 10.1111/sltb.12383 [DOI] [PubMed] [Google Scholar]
  11. Edwards LJ, Muller KE, Wolfinger RD, Qaqish BF, & Schabenberger O (2008). An r2 statistic for fixed effects in the linear mixed model. Statistics in Medicine, 27(29), 6137–6157. 10.1002/sim.3429 [DOI] [PMC free article] [PubMed] [Google Scholar]
  12. Fan L, Li H, Zhuo J, Zhang Y, Wang J, Chen L, Yang Z, Chu C, Xie S, Laird AR, Fox PT, Eickhoff SB, Yu C, & Jiang T (2016). The human brainnetome atlas: A new brain atlas based on connectional architecture. Cerebral Cortex, 26(8), 3508–3526. 10.1093/cercor/bhw157 [DOI] [PMC free article] [PubMed] [Google Scholar]
  13. Forrest LN, Smith AR, White RD, & Joiner TE (2015). (Dis)connected: An examination of interoception in individuals with suicidality. Journal of Abnormal Psychology, 124(3), 754–763. 10.1037/abn0000074 [DOI] [PubMed] [Google Scholar]
  14. Friston KJ, Frith CD, Liddle PF, & Frackowiak RS (1993). Functional connectivity: The principal-component analysis of large (PET) data sets. Journal of Cerebral Blood Flow and Metabolism, 13(1), 5–14. 10.1038/jcbfm.1993.4 [DOI] [PubMed] [Google Scholar]
  15. Garner DM, Olmstead MP, & Polivy J (1983). Development and validation of a multidimensional eating disorder inventory for anorexia nervosa and bulimia. International Journal of Eating Disorders, 2(2), 15–34. https://doi.org/bmm8pv [Google Scholar]
  16. Giakoumatos CI, Tandon N, Shah J, Mathew IT, Brady RO, Clementz BA, Pearlson GD, Thaker GK, Tamminga CA, Sweeney JA, & Keshavan MS (2013). Are structural brain abnormalities associated with suicidal behavior in patients with psychotic disorders? Journal of Psychiatric Research, 47(10), 1389–1395. 10.1016/j.jpsychires.2013.06.011 [DOI] [PMC free article] [PubMed] [Google Scholar]
  17. Harris EC, & Barraclough B (1997). Suicide as an outcome for mental disorders: A meta-analysis. The British Journal of Psychiatry, 170(3), 205–228. 10.1192/bjp.170.3.205 [DOI] [PubMed] [Google Scholar]
  18. Hassanpour MS, Simmons WK, Feinstein JS, Luo Q, Lapidus RC, Bodurka J, Paulus MP, & Khalsa SS (2018). The insular cortex dynamically maps changes in cardiorespiratory interoception. Neuropsychopharmacology, 43(2), 426–434. 10.1038/npp.2017.154 [DOI] [PMC free article] [PubMed] [Google Scholar]
  19. Hu L, Xiao M, Ai M, Wang W, Chen J, Tan Z, Cao J, & Kuang L (2019). Disruption of resting-state functional connectivity of right posterior insula in adolescents and young adults with major depressive disorder. Journal of Affective Disorders, 257, 23–30. 10.1016/j.jad.2019.06.057 [DOI] [PubMed] [Google Scholar]
  20. Jadhav A, Pramod D, & Ramanathan K (2019). Comparison of performance of data imputation methods for numeric dataset. Applied Artificial Intelligence, 33(10), 913–933. 10.1080/08839514.2019.1637138 [DOI] [Google Scholar]
  21. Jaeger B (2017). R2glmm: Computes r squared for mixed (multilevel) models (Version 0.1.2) [Computer software] https://CRAN.R-project.org/package=r2glmm
  22. Kerr KL, Moseman SE, Avery JA, Bodurka J, Zucker NL, & Simmons WK (2016). Altered insula activity during visceral interoception in weight-restored patients with anorexia nervosa. Neuropsychopharmacology, 41(2), 521–528. 10.1038/npp.2015.174 [DOI] [PMC free article] [PubMed] [Google Scholar]
  23. Khalsa SS, Adolphs R, Cameron OG, Critchley HD, Davenport PW, Feinstein JS, Feusner JD, Garfinkel SN, Lane RD, Mehling WE, Meuret AE, Nemeroff CB, Oppenheimer S, Petzschner FH, Pollatos O, Rhudy JL, Schramm LP, Simmons WK, Stein MB … Interoception Summit Participants. (2018). Interoception and mental health: A roadmap. Biological Psychiatry: Cognitive Neuroscience and Neuroimaging, 3(6), 501–513. 10.1016/j.bpsc.2017.12.004 [DOI] [PMC free article] [PubMed] [Google Scholar]
  24. Khalsa SS, Craske MG, Li W, Vangala S, Strober M, & Feusner JD (2015). Altered interoceptive awareness in anorexia nervosa: Effects of meal anticipation, consumption and bodily arousal. International Journal of Eating Disorders, 48(7), 889–897. 10.1002/eat.22387 [DOI] [PMC free article] [PubMed] [Google Scholar]
  25. Khalsa SS, & Lapidus RC (2016). Can interoception improve the pragmatic search for biomarkers in psychiatry? Frontiers in Psychiatry, 7. 10.3389/fpsyt.2016.00121 [DOI] [PMC free article] [PubMed] [Google Scholar]
  26. Kleiman E (2017). EMAtools: Data management tools for real-time monitoring/ecological momentary assessment data (Version 0.1. 3) [Computer program] https://cran.r-project.org/web/packages/EMAtools/index.html
  27. Kroenke K, Spitzer RL, & Williams JB (2001). The PHQ-9: Validity of a brief depression severity measure. Journal of General Internal Medicine, 16(9), 606–613. 10.1046/j.1525-1497.2001.016009606.x [DOI] [PMC free article] [PubMed] [Google Scholar]
  28. Kuznetsova A, Brockhoff PB, & Christensen RHB (2017). Lmertest package: Tests in linear mixed effects models. Journal of Statistical Software, 82(13). 10.18637/jss.v082.i13 [DOI] [Google Scholar]
  29. McCabe KM, Boyd CJ, Cranford JA, Morales M, & Slayden J (2006). A modified version of the drug abuse screening test among undergraduate students. Journal of Substance Abuse Treatment, 21, 297–303. 10.1016/j.jsat.2006.04.010 [DOI] [PMC free article] [PubMed] [Google Scholar]
  30. Mehling WE, Price C, Daubenmier JJ, Acree M, Bartmess E, & Stewart A (2012). The multidimensional assessment of interoceptive awareness (maia). PLoS One, 7(11), e48230. 10.1371/journal.pone.0048230 [DOI] [PMC free article] [PubMed] [Google Scholar]
  31. Morgan JF, Reid F, & Lacey JH (2000). The SCOFF questionnaire: A new screening tool for eating disorders. Western Journal of Medicine, 172(3), 164–165. 10.1136/ewjm.172.3.164 [DOI] [PMC free article] [PubMed] [Google Scholar]
  32. Nock MK, Hwang I, Sampson NA, & Kessler RC (2010). Mental disorders, comorbidity and suicidal behavior: Results from the national comorbidity survey replication. Molecular Psychiatry, 15(8), 868–876. 10.1038/mp.2009.29 [DOI] [PMC free article] [PubMed] [Google Scholar]
  33. Paulus MP, & Stewart JL (2014). Interoception and drug addiction. Neuropharmacology, 76, 342–350. 10.1016/j.neuropharm.2013.07.002 [DOI] [PMC free article] [PubMed] [Google Scholar]
  34. Posner K, Brown GK, Stanley B, Brent DA, Yershova KV, Oquendo MA, Currier GW, Melvin GA, Greenhill L, Shen S, & Mann JJ (2011). The Columbia-Suicide Severity Rating Scale: Initial validity and internal consistency findings from three multisite studies with adolescents and adults. American Journal of Psychiatry, 168(12), 1266–1277. 10.1176/appi.ajp.2011.10111704 [DOI] [PMC free article] [PubMed] [Google Scholar]
  35. R Core Team. (2018). R: A language and environment for statistical computing. [Computer software manual]. https://www.R-project.org/.
  36. Rogers ML, Hagan CR, & Joiner TE (2018). Examination of interoception along the suicidality continuum. Journal of Clinical Psychology, 74(6), 1004–1016. 10.1002/jclp.22564 [DOI] [PubMed] [Google Scholar]
  37. Schandry R (1981). Heart beat perception and emotional experience. Psychophysiology, 18(4), 483–488. 10.1111/j.1469-8986.1981.tb02486.x [DOI] [PubMed] [Google Scholar]
  38. Schmaal L, van Harmelen AL, Chatzi V, Lippard ETC, Toenders YJ, Averill LA, Mazure CM, & Blumberg HP (2020). Imaging suicidal thoughts and behaviors: A comprehensive review of 2 decades of neuroimaging studies. Molecular Psychiatry, 25(2), 408–427. 10.1038/s41380-019-0587-x [DOI] [PMC free article] [PubMed] [Google Scholar]
  39. Simmons WK, Avery JA, Barcalow JC, Bodurka J, Drevets WC, & Bellgowan P (2013). Keeping the body in mind: Insula functional organization and functional connectivity integrate interoceptive, exteroceptive, and emotional awareness. Human Brain Mapping, 34(11), 2944–2958. 10.1002/hbm.22113 [DOI] [PMC free article] [PubMed] [Google Scholar]
  40. Smith A, Forrest L, & Velkoff E (2018). Out of touch: Interoceptive deficits are elevated in suicide attempters with eating disorders. Eating disorders: The Journal of Treatment & Prevention, 26(1), 52–65. 10.1080/10640266.2018.1418243 [DOI] [PubMed] [Google Scholar]
  41. Soloff PH, Pruitt P, Sharma M, Radwan J, White R, & Diwadkar VA (2012). Structural brain abnormalities and suicidal behavior in borderline personality disorder. Journal of Psychiatric Research, 46(4), 516–525. 10.1016/j.jpsychires.2012.01.003 [DOI] [PMC free article] [PubMed] [Google Scholar]
  42. Stewart JL, Khalsa SS, Kuplicki R, Puhl M, Tulsa 1000 Investigators, & Paulus MP (2019). Interoceptive attention in opioid and stimulant use disorder. Addiction Biology, e12831. 10.1111/adb.12831 [DOI] [PMC free article] [PubMed] [Google Scholar]
  43. Troyanskaya O, Cantor M, Sherlock G, Brown P, Hastie T, Tibshirani R, … & Altman RB (2001). Missing value estimation methods for DNA microarrays. Bioinformatics, 17(6), 520–525. 10.1093/bioinformatics/17.6.520 [DOI] [PubMed] [Google Scholar]
  44. Victor TA, Khalsa SS, Simmons WK, Feinstein JS, Savitz J, Aupperle RL, Yeh H, Bodurka J, & Paulus MP (2018). Tulsa 1000: a naturalistic study protocol for multilevel assessment and outcome prediction in a large psychiatric sample. BMJ Open, 8. 10.1136/bmjopen-2017-016620 [DOI] [PMC free article] [PubMed] [Google Scholar]
  45. Wang C, Wu H, Chen F, Xu J, Li H, Li H, & Wang J (2018). Disrupted functional connectivity patterns of the insula subregions in drug-free major depressive disorder. Journal of Affective Disorders, 234, 297–304. 10.1016/j.jad.2017.12.033 [DOI] [PubMed] [Google Scholar]
  46. Wickham H (2011). GGPlot2. Wiley Interdisciplinary Reviews: Computational Statistics, 3(2), 180–185. 10.1002/wics.147 [DOI] [Google Scholar]
  47. Wiebking C, de Greck M, Duncan NW, Tempelmann C, Bajbouj M, & Northoff G (2015). Interoception in insula subregions as a possible state marker for depression - An exploratory fMRI study investigating healthy, depressed and remitted participants. Frontiers in Behavioral Neuroscience, 9, 82. 10.3389/fnbeh.2015.00082 [DOI] [PMC free article] [PubMed] [Google Scholar]
  48. Willeumier K, Taylor DV, & Amen DG (2011). Decreased cerebral blood flow in the limbic and prefrontal cortex using spect imaging in a cohort of completed suicides. Translational Psychiatry, 1(8), e28. 10.1038/tp.2011.28 [DOI] [PMC free article] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

appendix
NIHMS1697390-supplement-appendix.pdf (1.2MB, pdf)

RESOURCES