Advertisement

Genetic association of ADHD and major depression with suicidal ideation and attempts in children: the adolescent brain cognitive development (ABCD) study

Open AccessPublished:December 22, 2021DOI:https://doi.org/10.1016/j.biopsych.2021.11.026

      ABSTRACTS

      Background

      Suicide is among the leading cause of death in children and adolescents. There are well-known risk factors of suicide, including childhood abuse, family conflicts, social adversity, and psychopathology. While suicide risk is also known to be heritable, few studies have investigated genetic risk in younger individuals.

      Methods

      Using polygenic risk score (PRS) analysis, we examined whether genetic susceptibility to major psychiatric disorders is associated with suicidal behaviors among 11,878 children enrolled in the Adolescent Brain Cognitive Development study. Suicidal ideation (SI) and attempt (SA) data were assessed using the youth-report of the Kiddie Schedule for Affective Disorder and Schizophrenia (KSADS-5). After performing robust quality control of genotype data, unrelated individuals of European descent were included in analyses (N=4,344).

      Results

      Among eight psychiatric disorders we examined, depression PRSs were associated with lifetime SA both in the baseline (OR=1.55, 95% CI=1.10-2.18, p-value=1.27x10-2) and the follow-up year (OR=1.38, 95% CI=1.08-1.77, p-value=1.05x10-2), after adjusting for children’s age, sex, socioeconomic backgrounds, family history of suicide, and psychopathology. In contrast, attention deficits hyper-activity (ADHD) PRSs were associated with lifetime SI (OR=1.15, 95% CI=1.05-1.26, p-value=3.71x10-3), suggesting a distinct contribution of the genetic risk underlying ADHD and depression on suicidal behaviors of children.

      Conclusion

      The largest genetic sample of suicide risk data in US children suggests a significant genetic basis of suicide risk related to ADHD and depression. Further research is warranted to examine whether incorporation of genomic risk may facilitate more targeted screening and intervention efforts.

      Keywords

      INTRODUCTION

      Suicide is the second leading cause of death in children and adolescents worldwide (
      • Xu J.
      • Murphy S.L.
      • Kockanek K.D.
      • Arias E.
      Mortality in the United States, 2018.
      ,

      World Health Organization (2014): Global health estimates 2013: deaths by cause, age and sex, estimates for 2000–2012. Geneva: World Health Organization.

      , ). Tragically, more than a half of adolescents who die by suicide have previous records of suicide attempts and self-harming behaviors (
      • Appleby L.
      • Kapur N.
      • Shaw J.
      • Rodway C.
      • Turnbull P.
      • Ibrahim S.
      • et al.
      Suicide by children and young people.
      ), which start during childhood and persist over several years (
      • Hawton K.
      • Bale L.
      • Brand F.
      • Townsend E.
      • Ness J.
      • Waters K.
      • et al.
      Mortality in children and adolescents following presentation to hospital after non-fatal self-harm in the Multicentre Study of Self-harm: a prospective observational cohort study.
      ). Suicidality in children, here defined as an umbrella term that includes both suicidal ideation (SI) and suicide attempts (SA) (
      • Huber R.S.
      • Sheth C.
      • Renshaw P.F.
      • Yurgelun-Todd D.A.
      • McGlade E.C.
      Suicide Ideation and Neurocognition Among 9- and 10-Year Old Children in the Adolescent Brain Cognitive Development (ABCD) Study.
      ,
      • Janiri D.
      • Doucet G.E.
      • Pompili M.
      • Sani G.
      • Luna B.
      • Brent D.A.
      • et al.
      Risk and protective factors for childhood suicidality: a US population-based study.
      ,
      • Aguinaldo L.
      • Goldstone A.
      • Hasler B.P.
      • Brent D.A.
      • Coronado C.
      • Jacobus J.
      Preliminary Analysis of Low-Level Alcohol Use and Suicidality with Children in the Adolescent Brain and Cognitive Development (ABCD) Baseline Cohort.
      ), is also significantly associated with psychiatric disorders in later life, suggesting intrinsic etiologic connections between suicide risk and mental health that start at earlier developmental periods. Understanding the etiological basis of suicidality in children may facilitate prevention and early intervention strategies (
      • Kessler R.C.
      • Bossarte R.M.
      • Luedtke A.
      • Zaslavsky A.M.
      • Zubizarreta J.R.
      Suicide prediction models: a critical review of recent research with recommendations for the way forward.
      ,
      • O’Connor R.C.
      • Robb K.A.
      Identifying suicide risk factors in children is essential for developing effective prevention interventions.
      ).
      Twin, family, and adoption studies have consistently reported that suicidal behaviors are heritable (
      • Franklin J.C.
      • Ribeiro J.D.
      • Fox K.R.
      • Bentley K.H.
      • Kleiman E.M.
      • Huang X.
      • et al.
      Risk factors for suicidal thoughts and behaviors: A meta-analysis of 50 years of research.
      ,

      CC Zai VdL, J Strauss, RP Tong, I Sakinofsky, JL Kennedy (2012): Chapter 11. Genetic Factors and Suicidal Behavior. In: Dwivedi Y, editor. The Neurobiological Basis of Suicide. Boca Raton (FL): CRC Press/Taylor & Francis.

      ). The latest population-based Swedish cohort study of more than 2.7 million offspring reported that 70% of the correlation between maternal and offspring suicidal behaviors was attributed to genetic factors shared across the generations, while the remaining was due to adverse environmental factors specific to the exposure to maternal suicidal behavior (
      • O'Reilly L.
      • Kuja-Halkola R.
      • Rickert M.E.
      • Class Q.A.
      • Larsson H.
      • Lichtenstein P.
      • D'Onofrio B.M.
      The intergenerational transmission of suicidal behavior: an offspring of siblings study.
      ). Deciphering specific genetic risk mechanisms underlying suicide, however, has met with limited success. Several genome-wide association studies (GWAS) (
      • Levey D.F.
      • Polimanti R.
      • Cheng Z.
      • Zhou H.
      • Nunez Y.Z.
      • Jain S.
      • et al.
      Genetic associations with suicide attempt severity and genetic overlap with major depression.
      ,
      • Perlis R.H.
      • Huang J.
      • Purcell S.
      • Fava M.
      • Rush A.J.
      • Sullivan P.F.
      • et al.
      Genome-wide association study of suicide attempts in mood disorder patients.
      ,
      • Docherty A.R.
      • Shabalin A.A.
      • DiBlasi E.
      • Monson E.
      • Mullins N.
      • Adkins D.E.
      • et al.
      Genome-Wide Association Study of Suicide Death and Polygenic Prediction of Clinical Antecedents.
      ,
      • Mullins N.
      • Bigdeli T.B.
      • Børglum A.D.
      • Coleman J.R.I.
      • Demontis D.
      • Mehta D.
      • et al.
      GWAS of Suicide Attempt in Psychiatric Disorders and Association With Major Depression Polygenic Risk Scores.
      ,
      • Mullins N.
      • Kang J.
      • Campos A.I.
      • Coleman J.R.I.
      • Edwards A.C.
      • Galfalvy H.
      • et al.
      Dissecting the shared genetic architecture of suicide attempt, psychiatric disorders and known risk factors.
      ,
      • Willour V.
      • Seifuddin F.
      • Mahon P.B.
      • Jancic D.
      • Pirooznia M.
      • Steele J.
      • et al.
      Nonsuicidal self-injury, suicide ideation, and suicide attempts among sexual minority children.
      ) have reported significant genetic correlations of self-harm and SA with psychiatric disorders, with major depression showing the most extensive genetic overlap. Yet, few suicide-specific risk have been identified with conclusive evidence, possibly due to the extensive polygenic nature of suicidal behaviors and the challenges of identifying samples of adequate sizes given the relatively lower base rate of suicide compared to psychiatric illness.
      The primary aim of this study was to examine whether genetic risk for major psychiatric disorders is associated with SI and/or SA in children. Emerging evidence in psychiatric genomics supports genetic overlap between SA and major psychiatric disorders in adults. Yet, it is unknown whether genetic risk underlying these conditions is associated with suicidality in children. To test this hypothesis, we investigated data from the Adolescent Brain Cognitive Development (ABCD) study, a population-based longitudinal study of more than 11,000 US children enrolled at 9-10 years old. A wide range of measures encompassing social, familial, physical, mental, and behavioral aspects have been collected longitudinally since 2016, along with the genome-wide genotype data of the participants. As of now, several reports of suicidality data have been published using the initial release of the ABCD cohort (6-8, 20-23). These studies have confirmed distinct characteristics of children who reported suicidality including child psychopathology, family conflicts, and a parental history of suicide. Yet, few if any studies to date have incorporated genome-wide genetic data of ABCD participants in suicide research.
      Harnessing the power of the ABCD cohort, we specifically aimed: (
      • Xu J.
      • Murphy S.L.
      • Kockanek K.D.
      • Arias E.
      Mortality in the United States, 2018.
      ) to examine the association of common genetic variation underlying eight major psychiatric disorders with SI and/or SA in children; (

      World Health Organization (2014): Global health estimates 2013: deaths by cause, age and sex, estimates for 2000–2012. Geneva: World Health Organization.

      ) to examine the association of genetic risk for psychiatric disorders with known sociodemographic and clinical risk factors of suicide, including age, sex, parental education, household income, marital status, child psychopathology, and family history of suicide; and () to examine whether genetic risk for psychiatric disorders improves prediction performance of SI and/or SA in children independent of known suicide risk factors. To quantify children’s genetic liability to psychiatric disorders, we utilized polygenic risk scores (PRSs), which estimate the genome-wide aggregated effects of common genetic risk alleles in an individual based on independent genome-wide association studies of the target phenotype. Figure 1 summarizes the overview of our study design.
      Figure thumbnail gr1
      Figure 1Study outline of the polygenic risk score (PRS) analyses. Our study used ABCD data release 3.0, which included genetics and phenotypic data collected in the baseline and the first-year follow-up for 11,877 participants. Suicidal data were collected using the youth version of the computerized Kiddie Schedule for Affective Disorders and Schizophrenia for Diagnostic and Statistical Manual for Mental Disorders (KSADS-5). Using the individual item data, we generated three lifetime suicide-related measures, encompassing non-suicidal self-injuries (NSSI), suicidal ideation (SI), and attempts (SA). Well-known risk factors of suicide were included in the analysis using the ABCD survey data. For PRS data generation, eight psychiatric disorder GWAS of the largest available sample size were applied to 4,344 ABCD participants of European ancestry.

      METHODS

      The ABCD study participants. Our study used ABCD Data Release 3.0, which included data collected between September 2016 and February 2020 for 11,878 participants. The ABCD data were downloaded from the National Institute of Mental Health Data Archive (https:// data-archive.nimh.nih.gov/abcd). In this study, we focused on the baseline and the first-year follow-up data for which we had information of the full cohort. Detailed information about the selection of participants and assessment data has been published elsewhere (
      • Zucker R.A.
      • Gonzalez R.
      • Feldstein Ewing S.W.
      • Paulus M.P.
      • Arroyo J.
      • Fuligni A.
      • et al.
      Assessment of culture and environment in the Adolescent Brain and Cognitive Development Study: Rationale, description of measures, and early data.
      ,
      • Barch D.M.
      • Albaugh M.D.
      • Avenevoli S.
      • Chang L.
      • Clark D.B.
      • Glantz M.D.
      • et al.
      Demographic, physical and mental health assessments in the adolescent brain and cognitive development study: Rationale and description.
      ,
      • Luciana M.
      • Bjork J.M.
      • Nagel B.J.
      • Barch D.M.
      • Gonzalez R.
      • Nixon S.J.
      • et al.
      Adolescent neurocognitive development and impacts of substance use: Overview of the adolescent brain cognitive development (ABCD) baseline neurocognition battery.
      ).
      Demographic and family socioeconomic status. Demographic and family socioeconomic information was obtained from ABCD Parent Demographics Survey data. Indicator variables for four races (White, Black, Asian, Other) were created by combining the information from multiple race-related questions. Ethnicity represented Hispanic and non-Hispanic descendants. For parental education, five categories were defined following Huber et al. (
      • Huber R.S.
      • Sheth C.
      • Renshaw P.F.
      • Yurgelun-Todd D.A.
      • McGlade E.C.
      Suicide Ideation and Neurocognition Among 9- and 10-Year Old Children in the Adolescent Brain Cognitive Development (ABCD) Study.
      ): (
      • Xu J.
      • Murphy S.L.
      • Kockanek K.D.
      • Arias E.
      Mortality in the United States, 2018.
      ) Less than High school diploma or GED; (

      World Health Organization (2014): Global health estimates 2013: deaths by cause, age and sex, estimates for 2000–2012. Geneva: World Health Organization.

      ) High school diploma or GED; () Some college, including associate degrees; (
      • Appleby L.
      • Kapur N.
      • Shaw J.
      • Rodway C.
      • Turnbull P.
      • Ibrahim S.
      • et al.
      Suicide by children and young people.
      ) Bachelor's degree; and (
      • Hawton K.
      • Bale L.
      • Brand F.
      • Townsend E.
      • Ness J.
      • Waters K.
      • et al.
      Mortality in children and adolescents following presentation to hospital after non-fatal self-harm in the Multicentre Study of Self-harm: a prospective observational cohort study.
      ) Post-graduate degrees. Yearly household income was categorized into three groups: (
      • Xu J.
      • Murphy S.L.
      • Kockanek K.D.
      • Arias E.
      Mortality in the United States, 2018.
      ) Less than $50,000; (

      World Health Organization (2014): Global health estimates 2013: deaths by cause, age and sex, estimates for 2000–2012. Geneva: World Health Organization.

      ) Between $50,000 and $100,000; and () Greater than $100,000.
      Suicide risk outcome measure. Lifetime measures of SI, SA, and non-suicidal self-injuries (NSSI) were generated using the youth-report of the computerized Kiddie Schedule for Affective Disorders and Schizophrenia for Diagnostic and Statistical Manual for Mental Disorders (KSADS-5). The KSADS-5 survey included a suicide module, which consists of 35 items asking the participants’ experiences of self-injuries, passive or active thoughts of suicide, or suicide attempts at present or in the past. Table S1 summarizes the details of the youth-report-based KSADS-5 items used for generating the lifetime SI, SA, and NSSI measures. The parent-report of KSADS-5 was available only for the baseline. For comparison, we provided the case numbers of NSSI, SI, and SA based on the youth- and the caregiver-report and their concordance in Table S1.
      Risk factors of suicidality. In the ABCD study, the Child Behavior Checklist (CBCL) was used to assess dimensional psychopathology spectrums of children (
      • Achenbach T.
      International findings with the Achenbach System of Empirically Based Assessment (ASEBA): applications to clinical services, research, and training.
      ,

      Achenbach TM, & Rescorla, L.A. (2001): Manual for the ASEBA School-Age Forms & Profiles. In: University of Vermont RCfC, Youth, & Families., editor. Burlington, VT.

      ). We examined the normalized t-scores of 11 CBCL syndromes representing: anxiety/depression, withdrawal/depression, somatic complaints, social behaviors, thought problems, attention issues, rule-breaking, aggression, internalizing domain, externalizing domain, and total problem scores. We also included a family history of suicide in the analyses. The parents’ report of the presence or absence of a blood relative of the youth ever having a suicide attempt or death by suicide was used to generate a measure of a family history of suicidality. The children’s socioeconomic backgrounds were added using household income, parental marital status, poverty, and highest parental education. To assess distinct characteristics of suicide risk factors between different groups (e.g., full cohort vs. genetics sample), we used Welch’s two-sample t-test for quantitative measures and proportion tests for categorial variables in R (v4.0.5).
      Genotyping, data quality control (QC), and imputation. Genotyping of ABCD samples was performed using the Affymetrix NIDA SmokeScreen Array at Rutgers. The array included 733,293 SNPs. We applied standard QC of genome-wide data as we have previously described (
      • Lee P.
      • Anttila V.
      • Won H.
      • Feng Y.A.
      • Rosenthal J.
      • Zhu Z.
      • et al.
      Genomic Relationships, Novel Loci, and Pleiotropic Mechanisms across Eight Psychiatric Disorders.
      ). The QC-ed data contained 516,598 genetic variants for 11,099 individuals. We excluded related individuals, retaining 9,109 independent participants. Principal component analysis with the 1000 Genomes Project samples identified 4,344 subjects of European ancestry. Imputation was conducted using the Michigan Imputation Server (v.1.5.7) and minimac (v4-1.0.2) based on the Human Reference Consortium panel. Haplotype phasing was conducted using eagle (v2.4) with r2 filtering of 0.8. Correlation of allele frequencies between the ABCD samples and the HRC was 0.984. QC-ed imputation data included 6.7 million SNPs.
      Polygenic risk scores. We generated PRSs for 8 psychiatric disorders: major depression (MD), bipolar disorder (BD), schizophrenia (SCZ), attention deficits hyperactivity disorder (ADHD), autism spectrum disorders (ASD), post-traumatic stress disorder (PTSD), anorexia nervosa (ANO), anxiety disorder (ANX). Table S2 summarizes the details of the GWAS datasets. PRSs were calculated using the PRSice-2 software (
      • Choi S.
      • Mak T.S.
      • O'Reilly P.F.
      Tutorial: a guide to performing polygenic risk score analyses.
      ). To the best of our knowledge, these summary statistics represent the largest publicly available GWAS for these disorders. First, linkage disequilibrium (LD)-independent SNPs were identified using clumping implemented in PLINK (window of 250kb, LD clump-r2 = 0.1). We used a standard weighted sum scoring approach, representing each child’s additive genome-wide genetic risk to a target phenotype. As expected, PRSs showed a normal distribution (shapiro.test in R p-value>0.05) and were standardized.
      Statistical analysis. To examine association of PRSs with lifetime suicide risk measures, we used multiple logistic regression glm in R. Each outcome measure was used as a binary dependent variable, while PRS was used as an independent variable along with ten principal components (PCs) of genetic ancestry as covariates. The effect of PRS on the outcome measure was presented using the odds ratio (OR), which was derived as the exponential of the logistic regression beta coefficient estimate. The OR represents the odds of inclusion in the suicidal group with an increase of one standard deviation change in the PRS. For multiple testing correction, we used the false discovery rate (FDR) q-value of 0.05, considering the number of youth-report-based outcome measures and investigation of both baseline and the first-year follow-up data for 8 PRSs. To measure the unique proportion of variance explained by PRSs, we calculated Nagelkerke’s pseudo-R2 (
      • Nagelkerke N.
      A note on a general definition of the coefficient of determination.
      ).
      To examine association of PRSs with child psychopathology, we calculated partial correlation between the two, adjusting for ancestry PCs, using the R ppcor package. Along with Pearson correlation estimates, t-score statistics were used to represent a standardized relationship of the two variables. For the family history of suicide, we conducted un-paired t-test to examine the differences of PRSs between the participants with and without the family history.
      Lastly, we assessed predictive improvement of PRSs on each outcome measure independent of known risk factors using two logistic regression models in R. The first model included known risk factors of suicide as independent variables (i.e., base model), while the second model included PRS as an additional variable (i.e., genetic risk model). In all analyses, we included ten genetic ancestry PCs as covariates to account for potential population sub-stratification. Along with Nagelkerke’s pseudo-R2, the likelihood ratio test was performed to assess whether a genetic risk model significantly improves the prediction of suicidality compared to a base model in R (v4.0.5).
      Sensitivity analysis. For PRSs of significant association with a target outcome measure, we conducted sensitivity analyses to assess whether the identified association varies as the p-value threshold for constructing PRS changes. Ten p-values in total, 5x10-8, 5x10-6, 5x10-4, 5x10-2, 0.1, 0.2, 0.3, 0.4, 0.5, and 1.0, were used in sensitivity analysis. We also tested a two PRS model to examine whether association of one PRS remains significant when the second PRS was added to the model. Here, in addition to PRSs for psychiatric disorder GWAS, we also assessed the impact of suicide PRS constructed based on the latest GWAS of SA (
      • Mullins N.
      • Bigdeli T.B.
      • Børglum A.D.
      • Coleman J.R.I.
      • Demontis D.
      • Mehta D.
      • et al.
      GWAS of Suicide Attempt in Psychiatric Disorders and Association With Major Depression Polygenic Risk Scores.
      ).
      Ethical approval. All caregivers and the participants of the ABCD study provided written informed consent for general research. The University of California at San Diego Institutional Review Board (IRB), which is responsible for the oversight of the ABCD Study, noted that analyses using publicly released ABCD Study data are not human subject research and did not require its approval. The present study obtained the IRB approval from the Massachusetts General Hospital (Boston, USA) as the secondary data analysis of the publicly available ABCD study (#2021P001872).

      RESULTS

       Prevalence and sociodemographic characteristics of suicidality in ABCD children

      Table 1 describes the demographic, socioeconomic, and family characteristics of ABCD children based on the youth report of lifetime SI and SA in the baseline. The average age of the children at the enrollment was 9.91 (± 0.62) years; 47.83% of participants were female. In the baseline, we had suicide-related survey data separately reported by children and caregivers (Table S1). Overall, 8.63% of children reported SI, while the incidence was 7.5% based on caregivers. Youth-report SAs were approximately three times more prevalent than when reported by caregivers (1.31% vs. 0.44%). Concordance between the youth- and the caregiver reports was low; approximately 25% of SI reported by the youth (=255/1,025) was recognized by the caregivers, while the rate reduced to 15% (=14/156) for SA.
      Table 1Demographic, socioeconomic, and family history of the ABCD participants based on youth-reported suicidality in the baseline
      DataMeasuresABCDSuicide IdeationSuicide Attempt
      All Samples (N = 11,878)Case No (N = 1,025)Control No (N = 10,853)Case No (N = 156)Control No (N = 11,722)
      AgeMean Age (SD)9.91 (0.62)9.91 (0.63)9.92 (0.62)9.93 (0.62)9.91 (0.62)
      SexMale (%)6,196 (52.17)595 (57.99)5,601 (51.62)89 (57.05)6,107 (52.1)
      RaceCaucasian (%)7,695 (64.79)637 (62.09)7,058 (65.04)85 (54.49)7,610 (64.93)
      African American (%)2,269 (19.1)199 (19.4)2,070 (19.08)47 (30.13)2,222 (18.96)
      Asian (%)1,113 (9.37)119 (11.6)994 (9.16)15 (9.62)1,098 (9.37)
      Other (%)800 (6.74)71 (6.92)729 (6.72)9 (5.77)791 (6.75)
      EthnicityHispanic (%)2,410 (20.29)195 (19.01)2,215 (20.41)40 (25.64)2,370 (20.22)
      Parental

      Education
      < HS Diploma/GED (%)593 (4.99)45 (4.39)548 (5.05)8 (5.13)585 (4.99)
      HS Diploma/GED (%)1,132 (9.53)88 (8.58)1,044 (9.62)22 (14.1)1,110 (9.47)
      Some College (%)3,079 (25.92)307 (29.92)2,772 (25.55)62 (39.74)3,017 (25.74)
      Bachelor's Degree (%)3,029 (25.5)258 (25.15)2,771 (25.54)37 (23.72)2,992 (25.53)
      Post Graduate Degree (%)4,044 (34.05)328 (31.97)3,716 (34.25)27 (17.31)4,017 (34.27)
      Household

      Income
      < $50 K (%)3,223 (27.14)297 (28.95)2,926 (26.97)76 (48.72)3,147 (26.85)
      > $50 K & < $100 K (%)4,089 (34.43)377 (36.74)3,712 (34.21)52 (33.33)4,037 (34.44)
      > $100 K (%)4,565 (38.44)352 (34.31)4,213 (38.83)28 (17.95)4,537 (38.71)
      Marital StatusParents Married (%)8,087 (68.09)651 (63.45)7,436 (68.53)77 (49.36)8,010 (68.34)
      Family HistorySuicide (%)1,827 (15.38)198 (19.3)1,629 (15.01)42 (26.92)1,785 (15.23)
      Demographic and family socioeconomic information was obtained from the ABCD Parent Demographics Survey data. Suicidality data were generated based on the “youth” version of the computerized Kiddie Schedule for Affective Disorders and Schizophrenia for Diagnostic and Statistical Manual for Mental Disorders (KSADS-5). Detailed information for individual KSADS-5 items is summarized in Table S1.
      Principal component analyses (PCA) revealed the diverse racial and ethnic backgrounds of the ABCD children (Figure S1). To minimize potential complications related to population stratification, our primary polygenic risk score (PRS) analysis focused on 4,344 children of European ancestry. Overall, the European genetics sample showed distinct socioeconomic status compared to the full cohort: higher parental education, higher household income, more married parents, and less poverty (all p< 2.2x10-16) (Table S3). This trend was consistent when we compared the European genetics sample with suicidality to the same group from the full cohort (Figure S2). For age, sex, family history of suicide, and child psychopathology measures, participants with suicidality did not differ significantly between the European genetics sample and the full cohort (Table S4).

       Association of polygenic risk scores and suicidality

      For 4,344 children with genotype data, PRSs for eight psychiatric disorders were calculated using GWAS summary statistics representing the largest available sample size of the disorders (Table S2). We used a standard weighted sum scoring strategy, representing each child’s additive genome-wide genetic risk to a target phenotype. We first examined the association between the eight PRSs and each lifetime suicide risk measure reported by children, adjusting for age, sex, and ten PCs reflecting population sub-stratification. Figure 2 summarizes the results (full results at Table S5, S6). It is notable that depression PRSs showed significant association with SA both in the baseline (Odds Ratio (OR)=1.85, 95% CI=1.32-2.60, uncorrected p-value = 3.21x10-4, false discovery rate (FDR) q=5.14x10-3) and the follow-up year (OR=1.63, 95% CI=1.28-2.08, uncorrected p-value = 6.95x10-5, q=1.67x10-3), while ADHD PRSs showed significant association with SI in the baseline (OR=1.17, 95% CI=1.05-1.3, uncorrected p-value=4.75x10-3, q=3.80x10-2) and in the follow-up year (OR=1.23, 95% CI=1.12-1.34, uncorrected p-value=1.01x10-5, q=4.85x10-4). In contrast to youth report data, we found no statistically significant associations between PRSs and caregiver-reported outcome measures (Table S7).
      Figure thumbnail gr2
      Figure 2Logistic regression analysis results for testing the association between eight psychiatric disorder PRS and lifetime suicide risk outcome measures. OR represents the exponential of the logistic regression estimates. Error bars represent 95% confidence intervals of ORs. ADHD: attention-deficit hyperactivity disorder; PTSD: post-traumatic stress disorder; NSSI: non-suicidal self-injury; SA: suicide attempts; SI: suicidal ideation.

       Association of polygenic risk scores and known risk factors of suicide

      We next examined whether the above associations of ADHD and MD PRSs with suicide risk measures reflect etiologic relationships between genetic susceptibility of the disorders and known clinical risk factors for suicide, such as child psychopathology. Specifically, we hypothesized that depression PRSs may be associated with participants’ internalizing problems, while ADHD PRS may be associated with external troubles. Table S8 summarizes the association analysis results of ADHD and MD PRSs with eleven CBCL measures of children’s emotional and behavioral problems assessed in the ABCD study. Unexpectedly, MD PRSs were significantly associated with all domains of children’s problematic behaviors, encompassing somatic, internalizing, and externalizing domains. ADHD PRSs showed association with all examined measures except a few internalizing problems (FDR q>0.01). Overall, we found stronger correlations between PRSs and CBCL measures in year 1 compared to the baseline (one-sided paired t-test p=2.16x10-4, Figure 3). We also found significantly higher ADHD and MD PRSs of the participants when stratified by the family history of suicide or disadvantageous socioeconomic status (Figure 3).
      Figure thumbnail gr3
      Figure 3Association analysis results of major depression (MD) and ADHD polygenic risk scores with known risk factors of suicide. (A) Partial correlations were measured between PRS and 11 CBCL measures conditioning on age, sex, and 10 genetic principal components to control for potential population sub-stratification within Europeans. Y-axis represents the t-statistics of partial correlation measures. (B) T test results for comparing ADHD and major depression (MD) PRSs between two groups of participants stratified by various risk factors. T test results Participants were divided into two groups based on the family history of suicide, parent college education, and poverty. Poverty was set yes if the household income of the participants is less than $20,000 annually. The top panels display MD PRS scores on the y-axis, while the bottom ones show ADHD PRS scores.

       Independent predictive effects of polygenic risk scores on suicidality

      Considering the significant association of ADHD and MD PRSs with child psychopathology and the family history of suicide, we examined whether PRSs could improve prediction of suicidality in children, independent of known risk factors of suicide. Multiple logistic regression analysis confirmed that MD PRS is independently associated with youth-reported lifetime SA, after accounting for age, sex, family history of suicide, socioeconomic status, and child psychopathology using the CBCL Total Problem score (Table 2, S9). With the addition of MD PRSs, Nagelkerke’s R2 indicated an improvement of 1.87% (PRS OR=1.55, CI= 1.10-2.18, p-value=1.27x10-2) in the baseline and 1.17% (PRS OR=1.38, CI= 1.08-1.77, p-value=1.05x10-2) in the follow-up year for the prediction model. We also confirmed the independent contribution of ADHD PRS with SI symptom in the follow-up year. In the prediction model, child psychopathology was the most significant predictor of SI (OR=1.70, SE=0.05, p-value<2x10-16). Association of ADHD PRS followed in the second (OR=1.15, CI= 1.05-1.26, p-value=3.71x10-3). Nagelkerke’s R2 was estimated as 0.39% with the addition of ADHD PRSs. Using both ADHD and MD PRSs in the same model yielded similar results, with little improvement in prediction performance (Likelihood ratio test p-value > 0.05; Table S10).
      Table 2Logistic regression analysis results of PRS on suicidality while accounting for demographic, socioeconomic, and family risk factors
      OutcomeIndependent VariableORL95H95BETASEP valueSigR2
      Suicide Attempts (Baseline)Age0.920.661.29-0.080.176.32E-011.87%
      Sex1.020.512.010.020.359.62E-01
      Marital Status0.610.281.35-0.490.402.24E-01
      Parental Education0.850.591.21-0.170.183.56E-01
      Household Income0.700.471.04-0.360.207.53E-02
      Poverty0.840.292.47-0.170.557.54E-01
      Child Psychopathology2.671.893.770.980.182.21E-08***
      Family History of Suicide1.120.861.460.110.144.14E-01
      Major Depression PRS1.551.102.180.440.171.27E-02*
      Suicide Attempts (Year 1)Age1.100.871.410.100.124.26E-011.17%
      Sex1.040.641.710.040.258.65E-01
      Marital Status0.640.361.15-0.440.301.40E-01
      Parental Education0.830.641.07-0.190.131.47E-01
      Household Income0.880.661.17-0.130.153.69E-01
      Poverty0.940.382.33-0.060.468.96E-01
      Child Psychopathology2.612.033.350.960.137.73E-14***
      Family History of Suicide1.150.951.400.140.101.58E-01
      Major Depression PRS1.381.081.770.320.131.05E-02*
      Suicidal Ideation (Baseline)Age0.900.811.00-0.110.065.30E-020.13%
      Sex0.830.671.04-0.180.111.07E-01
      Marital Status0.750.561.00-0.290.154.78E-02*
      Parental Education0.950.841.07-0.060.063.65E-01
      Household Income0.890.781.01-0.120.077.58E-02.
      Poverty0.710.411.25-0.340.282.36E-01
      Child Psychopathology1.591.431.770.460.05<2e-16***
      Family History of Suicide1.000.901.110.000.059.69E-01
      ADHD PRS1.090.981.220.090.061.30E-01
      Suicidal Ideation (Year 1)Age0.980.891.07-0.020.056.46E-010.39%
      Sex0.980.821.19-0.020.108.70E-01
      Marital Status0.840.651.08-0.170.131.72E-01
      Parental Education0.970.881.08-0.030.055.83E-01
      Household Income0.920.821.03-0.080.061.46E-01
      Poverty0.920.571.49-0.080.247.43E-01
      Child Psychopathology1.701.551.870.530.05< 2E-16***
      Family History of Suicide0.990.901.08-0.010.058.20E-01
      ADHD PRS1.151.051.260.140.053.71E-03**
      BETA and SE represent the logistic regression beta coefficient and its standard error. Odds Ratio (OR) was calculated as the exponential of the logistic regression beta coefficient. L95 and H95 displays the 95% confidence interval of OR. P value represents the significance of estimated beta using Z-statistic. Significance column was marked following regression: p<1x10-3: ‘***’, 0.001<p<0.01: ‘**’, 0.01<p<0.05: ‘*’. R2 represents Nagelkerke’s pseudo R2 , which estimates the unique proportion of variance explained by PRSs.
      To assess the robustness of the main findings, we examined whether the independent predictive effects of PRSs are observed consistently as the p-value threshold for generating PRS changes. This sensitivity analysis tested ten different p-values, 5x10-8, 5x10-5, 5x10-3, 5x10-2, 0.1, 0.2, 0.3, 0.4, 0.5, and 1.0, in total. We observed significant associations of genome-wide ADHD and MD PRSs with SI and SA, respectively across multiple p-value ranges (Table S11). We also found independent effect of MD and ADHD PRSs when predicting new on-set cases in year 1 (e.g., participants who did not endorsed SI in the baseline but did in year 1) (Table S12) or when suicide PRSs were added in the prediction model (Table S13).

      DISCUSSION

      Despite significant heritability and familial aggregation of suicidal behaviors, defining features of genetic risk underlying suicide have been elusive, especially in children and adolescents. Recent GWAS now offer the possibility to quantify heritable risk for a range of phenotypes that are relevant to this construct. The potential for such information to serve as a tool for risk stratification is particularly important to study in relation to children, given the rising rates of suicide in youth and the dearth of empirical studies in this age group (
      • Ayer L.
      • Colpe L.
      • Pearson J.
      • Rooney M.
      • Murphy E.
      Advancing Research in Child Suicide: A Call to Action.
      ).
      In the largest US sample of children characterized with in-depth phenotype and genome-wide genetic variation data, our study shows robust evidence that common genetic variants underlying ADHD and MD are significant predictors of suicidality in children. Importantly, higher genetic susceptibility to MD is associated with increased SAs in children, while genome-wide genetic risk to ADHD is associated with SI in children, suggesting distinctive contributions of these clinical conditions to children’s suicide risk. These associations remain significant independent of established clinical, familial, and demographic risk factors of suicide.
      Our findings are in agreement with previous epidemiological studies that showed significant association between MD and suicide (
      • Franklin J.C.
      • Ribeiro J.D.
      • Fox K.R.
      • Bentley K.H.
      • Kleiman E.M.
      • Huang X.
      • et al.
      Risk factors for suicidal thoughts and behaviors: A meta-analysis of 50 years of research.
      ,
      • Ribeiro J.
      • Huang X.
      • Fox K.R.
      • Franklin J.C.
      Depression and hopelessness as risk factors for suicide ideation, attempts and death: meta-analysis of longitudinal studies.
      ). Genome-wide genetics studies have reported significant genetic overlap of MD PRS with a range of suicide risk phenotypes in adults, including SAs (
      • Mullins N.
      • Bigdeli T.B.
      • Børglum A.D.
      • Coleman J.R.I.
      • Demontis D.
      • Mehta D.
      • et al.
      GWAS of Suicide Attempt in Psychiatric Disorders and Association With Major Depression Polygenic Risk Scores.
      ), severity of SAs (
      • Levey D.F.
      • Polimanti R.
      • Cheng Z.
      • Zhou H.
      • Nunez Y.Z.
      • Jain S.
      • et al.
      Genetic associations with suicide attempt severity and genetic overlap with major depression.
      ), and suicide death (
      • Docherty A.R.
      • Shabalin A.A.
      • DiBlasi E.
      • Monson E.
      • Mullins N.
      • Adkins D.E.
      • et al.
      Genome-Wide Association Study of Suicide Death and Polygenic Prediction of Clinical Antecedents.
      ). Our results extend previous evidence showing that increased risk of SAs in children may also be driven at least in part by genetic susceptibility to depression. Moreover, compared to previous PRS studies of adults with suicidal behaviors, the variance explained for the ABCD participants was larger. In Mullins et al. (
      • Mullins N.
      • Bigdeli T.B.
      • Børglum A.D.
      • Coleman J.R.I.
      • Demontis D.
      • Mehta D.
      • et al.
      GWAS of Suicide Attempt in Psychiatric Disorders and Association With Major Depression Polygenic Risk Scores.
      ), depression PRS explained up to 0.42% of SAs in psychiatric adult patients. In Levey et al (
      • Levey D.F.
      • Polimanti R.
      • Cheng Z.
      • Zhou H.
      • Nunez Y.Z.
      • Jain S.
      • et al.
      Genetic associations with suicide attempt severity and genetic overlap with major depression.
      ), depression PRSs explained up to 0.7% of phenotypic variance for the severity of SA in adults, while our study shows the highest of 3.3%.
      We also found that higher genetic risk to ADHD is specifically associated with increased SI in children. ADHD is one of the most common child-onset psychiatric disorders, with significantly increased risk of SI and SA (
      • Shen Y.
      • Chan B.S.M.
      • Huang C.
      • Cui X.
      • Liu J.
      • Lu J.
      • et al.
      Suicidal behaviors and attention deficit hyperactivity disorder (ADHD): a cross-sectional study among Chinese medical college students.
      ,
      • Fuller-Thomson E.
      • Rivière R.N.
      • Carrique L.
      • Agbeyaka S.
      The Dark Side of ADHD: Factors Associated With Suicide Attempts Among Those With ADHD in a National Representative Canadian Sample.
      ,
      • Giupponi G.
      • Giordano G.
      • Maniscalco I.
      • Erbuto D.
      • Berardelli I.
      • Conca A.
      • et al.
      Suicide risk in attention-deficit/hyperactivity disorder.
      ). There are multiple mechanisms through which ADHD may increases the risk of suicide. Several groups suggested a mediating role of depression between ADHD and suicidality based on the elevated comorbidity of MD among youth with ADHD (
      • Yoshimasu K.
      • Barbaresi W.J.
      • Colligan R.C.
      • Voigt R.G.
      • Killian J.M.
      • Weaver A.L.
      • Katusic S.K.
      Psychiatric comorbidities modify the association between childhood ADHD and risk for suicidality: a population-based longitudinal study.
      ,
      • Biederman J.
      • Ball S.W.
      • Monuteaux M.C.
      • Mick E.
      • Spencer T.J.
      • McCREARY M.
      • Cote M.
      • Faraone S.V.
      New insights into the comorbidity between ADHD and major depression in adolescent and young adult females.
      ). Our findings however suggest a potentially distinct etiologic role of the two disorders in suicidality during development. Higher levels of impulsivity and irritability have also been attributed to increasing suicide risk in children with ADHD (
      • Garas P.
      • Balazs J.
      Long-Term Suicide Risk of Children and Adolescents With Attention Deficit and Hyperactivity Disorder—A Systematic Review.
      ). Further studies will be essential to clarify genetic relationships between ADHD and SI, and to identify potential environmental stressors that may trigger the transition from SI to the action.
      Other notable finding is that associations of ADHD and MD PRSs with children’s suicidality were only observed for the youth-report-based outcome measures. Considering the high level of discordance between the caregiver and the youth-report, more attention needs to be put into mental health assessments of children starting as early as elementary school ages. Furthermore, we found no association of psychiatric disorder PRSs with NSSI. While several studies have reported common genetic etiology between NSSI and SI (
      • Campos A.
      • Verweij K.J.H.
      • Statham D.J.
      • Madden P.A.F.
      • Maciejewski D.F.
      • Davis K.A.S.
      • et al.
      Genetic aetiology of self-harm ideation and behaviour.
      ,
      • Maciejewski D.
      • Creemers H.E.
      • Lynskey M.T.
      • Madden P.A.
      • Heath A.C.
      • Statham D.J.
      • et al.
      Overlapping genetic and environmental influences on nonsuicidal self-injury and suicidal ideation: different outcomes, same etiology?.
      ), our finding suggests that the genetic basis of NSSI is distinct from that of suicidal ideation and behaviors at least among children of this age group.
      Our study has several strengths. To the best of our knowledge, this is the first genetic data analysis of suicidal phenotypes reported for the ABCD study. Our conservative QC and focus on the participants of European ancestry assure that the results are robust to potential confounding from population genetic structure. Consistent association of PRSs across a range of SNP selection thresholds also substantiates that PRSs for ADHD and MD, one representing internalizing psychopathology and the other for externalizing problems, contribute to SI and SA in children. Our study is also based on the largest GWAS of ADHD and MD, each representing hundreds of thousands of cases and controls. Use of independent GWAS with sufficient statistical power is one of the most critical factors in polygenic scoring analysis (
      • Lewis C.
      • Vassos E.
      Polygenic risk scores: from research tools to clinical instruments.
      ).
      The present findings should be interpreted in light of several limitations. First, although the addition of ADHD and MD PRS clearly explains more variance of suicidality independent of other risk factors, clinical utility of the prediction model is still limited. The ORs of both child psychopathology and PRSs, the two strongest risk predictors, were modest, and overall discrimination of the prediction models remained poor. Our future research aims to improve the accuracy of the prediction models, which includes the investigation of additional predictor sets and advanced statistical analysis methods (
      • Kessler R.C.
      • Bossarte R.M.
      • Luedtke A.
      • Zaslavsky A.M.
      • Zubizarreta J.R.
      Suicide prediction models: a critical review of recent research with recommendations for the way forward.
      ). Secondly, our PRS genetics data analysis is based on ABCD participants of European ancestry thus may not be generalizable to other populations. Our decision to restrict genetic analyses to European ancestry was to ensure findings that are not confounded by population genetic structure (
      • Tucker G.
      • Price A.L.
      • Berger B.
      Improving the Power of GWAS and Avoiding Confounding from Population Stratification with PC-Select.
      ,
      • Martin A.
      • Kanai M.
      • Kamatani Y.
      • Okada Y.
      • Neale B.M.
      • Daly M.J.
      Clinical use of current polygenic risk scores may exacerbate health disparities.
      ). Furthermore, GWAS data of eight psychiatric disorders were based on European descendants, which may bias PRS analysis when applied to non-European individuals due to differences in causal variants, effect estimates, and LD structure between populations (
      • Lewis C.
      • Vassos E.
      Polygenic risk scores: from research tools to clinical instruments.
      ,
      • Martin A.
      • Kanai M.
      • Kamatani Y.
      • Okada Y.
      • Neale B.M.
      • Daly M.J.
      Clinical use of current polygenic risk scores may exacerbate health disparities.
      ,
      • Duncan L.
      • Shen H.
      • Gelaye B.
      • Meijsen J.
      • Ressler K.
      • Feldman M.
      • et al.
      Analysis of polygenic risk score usage and performance in diverse human populations.
      ). We call for more proactive efforts to create well-powered GWAS datasets for currently under-represented populations in genetic studies. Third, we have applied p-value-based selection strategies to generate PRS scores, which have room for improvement. While PRSs are powerful predictors over individual genome-wide significant variants, a considerable proportion of SNPs included in the calculation may not be related to a target phenotype and thus limit the statistical power. Improvement of PRS scores, for example, based on relevant biological knowledge or statistical techniques, merits further study. Fourth, we note that we controlled for child psychopathology in our analyses using the CBCL Total score, which includes items such as “deliberately harms self or attempts suicide” and “talks of killing self.” Our choice to include this score in our analysis promotes the generalizability of our findings, given that the CBCL is commonly used in both clinical and research settings; however, the inclusion of suicide-related items in these scales may have underestimated relationships in relevant analyses and speaks to the strength of the associations that we found. Finally, we note that our sample targeted a relatively narrow age range in children. Further studies of young children and adolescents are needed to clarify the relevance of polygenic risk scores to predictors at different age epochs (
      • Bridge J.
      • Goldstein T.R.
      • Brent D.A.
      Adolescent suicide and suicidal behavio.
      ,
      • Sheftall A.
      • Asti L.
      • Horowitz L.M.
      • Felts A.
      • Fontanella C.A.
      • Campo J.V.
      • Bridge J.A.
      Suicide in Elementary School-Aged Children and Early Adolescents.
      ).
      Despite these issues, our data advance the sparse empirical literature on suicide risk in children. There is a high level of interest in identifying risk factors for suicide that could open the door to targeted evidence-based prevention strategies (
      • Nierenberg A.
      Precision Psychiatry and a Vision for the Future.
      ,
      • Niculescu A.
      • Le-Niculescu H.
      • Levey D.F.
      • Phalen P.L.
      • Dainton H.L.
      • Roseberry K.
      • et al.
      Precision medicine for suicidality: from universality to subtypes and personalization.
      ). Despite their modest prediction and the fact that the majority of the variance in childhood suicidality is yet unaccounted for, our findings show that PRSs provide independent predictive value relevant to other risk variables of suicidality. The rationale for studying youth samples is augmented by the possibility, suggested indirectly by our data, that there may be a greater genetic contribution to children’s risk for suicidality compared to adults. The ABCD study provides a rich collection of longitudinal neuroimaging datasets. Our future research includes the investigation of these datasets to understand how genetic susceptibility to ADHD and major depression leads to structural or functional changes of brain development, which may contribute to youth suicidality (51, 52). In conclusion, coupled with the known genetic basis of suicidality and the growing evidence from other fields of medicine that polygenic scores may contribute to risk stratification, our data support further research into personalized suicide screening that incorporates genomic information.

      Uncited reference

      • Assari S.
      Racial Variation in the Association between Suicidal History and Positive and Negative Urgency among American Children.
      ,
      • Assari SB S.
      • Bazargan M.
      • Caldwell C.H.
      African Americans’ Diminished Returns of Parental Education on Adolescents’ Depression and Suicide in the Adolescent Brain Cognitive Development (ABCD) Study.
      ,
      • DeVille D.
      • Whalen D.
      • Breslin F.J.
      • Morris A.S.
      • Khalsa S.S.
      • Paulus M.P.
      • Barch D.M.
      Prevalence and Family-Related Factors Associated With Suicidal Ideation, Suicide Attempts, and Self-injury in Children Aged 9 to 10 Years.
      ,
      • Schmaal L.
      • van Harmelen A.L.
      • Chatzi V.
      • Lippard E.T.C.
      • Toenders Y.J.
      • Averill L.A.
      • et al.
      Imaging suicidal thoughts and behaviors: a comprehensive review of 2 decades of neuroimaging studies.
      ,
      • Hibar D.
      • Stein J.L.
      • Renteria M.E.
      • Arias-Vasquez A.
      • Desrivières S.
      • Jahanshad N.
      • et al.
      Common genetic variants influence human subcortical brain structures.
      .

      ACKNOWLEDGEMENTS

      We thank the Adolescent Brain Cognitive Development (ABCD) research team for their great efforts in collecting data. We also thank the ABCD research participants and their families for their continued support of the ABCD study. The ABCD Study was supported by the National Institutes of Health under award numbers: U01DA041022, U01DA041025, U01DA041028, U01DA041048, U01DA041089, U01DA041093, U01DA041106, U01DA041117, U01DA041120, U01DA041134, U01DA041148, U01DA041156, U01DA041174, U24DA041123, and U24DA041147. We also thank the participants who donated DNAs for various genetics research of common and complex human diseases. We appreciate the clinical and scientific teams which processed, analyzed, and publicly shared the summary statistics of GWAS datasets for pain and depression including the Psychiatric Genomics Consortium and the UK Biobank. We are deeply indebted to the investigators for their dedications in genetics research and the open data science policy. For this research, PHL was supported by National Institute of Mental Health (NIMH) R00 MH101367 and R01 MH119243. Statistical analyses were carried out on the Partner’s Research Computing Cluster servers and high-performance computing clusters hosted by the Broad Institute of MIT and Harvard.
      DISCLOSURES
      All authors declare that they have no known competing financial interests or personal relationships that could have influenced the work reported in this paper. For financial disclosure, Dr. Perlis has received fees for consulting or service on scientific advisory boards for Genomind, Psy Therapeutics, Outermost Therapeutics, RID Ventures, and Takeda. He has received patent royalties from Massachusetts General Hospital. He holds equity in Psy Therapeutics and Outermost Therapeutics. In the past 3 years, Dr. Kessler was a consultant for Datastat, Inc., Holmusk, RallyPoint Networks, Inc., and Sage Pharmaceuticals. He has stock options in Mirah, PYM, and Roga Sciences. Dr. Fava has lifetime research support from Abbott Laboratories; Acadia Pharmaceuticals; Alkermes, Inc.; American Cyanamid; Aspect Medical Systems; AstraZeneca; Avanir Pharmaceuticals; AXSOME Therapeutics; BioClinica, Inc; Biohaven; BioResearch; BrainCells Inc.; Bristol-Myers Squibb; CeNeRx BioPharma; Cephalon; Cerecor; Clarus Funds; Clexio Biosciences; Clintara, LLC; Covance; Covidien; Eli Lilly and Company;EnVivo Pharmaceuticals, Inc.; Euthymics Bioscience, Inc.; Forest Pharmaceuticals, Inc.; FORUM Pharmaceuticals; Ganeden Biotech, Inc.; Gentelon, LLC; GlaxoSmithKline; Harvard Clinical Research Institute; Hoffman-LaRoche; Icon Clinical Research; Indivior; i3 Innovus/Ingenix; Janssen R&D, LLC; Jed Foundation; Johnson & Johnson Pharmaceutical Research & Development; Lichtwer Pharma GmbH; Lorex Pharmaceuticals; Lundbeck Inc.; Marinus Pharmaceuticals; MedAvante; Methylation Sciences Inc; National Alliance for Research on Schizophrenia & Depression (NARSAD); National Center for Complementary and Alternative Medicine (NCCAM); National Coordinating Center for Integrated Medicine (NiiCM); National Institute of Drug Abuse (NIDA); National Institutes of Health; National Institute of Mental Health (NIMH); Neuralstem, Inc.; NeuroRx; Novartis AG; Organon Pharmaceuticals; Otsuka Pharmaceutical Development, Inc.; PamLab, LLC.; Pfizer Inc.; Pharmacia-Upjohn; Pharmaceutical Research Associates., Inc.; Pharmavite® LLC; PharmoRx Therapeutics; Photothera; Premiere Research International; Protagenic Therapeutics, Inc.; Reckitt Benckiser; Relmada Therapeutics Inc.; Roche Pharmaceuticals; RCT Logic, LLC (formerly Clinical Trials Solutions, LLC); Sanofi-Aventis US LLC; Shenox Pharmaceuticals, LLC; Shire; Solvay Pharmaceuticals, Inc.; Stanley Medical Research Institute (SMRI); Synthelabo; Taisho Pharmaceuticals; Takeda Pharmaceuticals; Tal Medical; VistaGen; Wyeth-Ayerst Laboratories. Dr. Fava also has equity holdings at Compellis and Psy Therapeutics. The remaining authors report no biomedical financial interests or potential conflicts of interest.

      Supplementary Material

      REFERENCES

        • Xu J.
        • Murphy S.L.
        • Kockanek K.D.
        • Arias E.
        Mortality in the United States, 2018.
        NCHS Data Brief. 2020; : 1-8
      1. World Health Organization (2014): Global health estimates 2013: deaths by cause, age and sex, estimates for 2000–2012. Geneva: World Health Organization.

      2. WHO|Suicide data. World Health Organization (https://www.who.int/mental_health/prevention/suicide/suicideprevent/en/).

        • Appleby L.
        • Kapur N.
        • Shaw J.
        • Rodway C.
        • Turnbull P.
        • Ibrahim S.
        • et al.
        Suicide by children and young people.
        National Confidential Inquiry into Suicide and Homicide by People with Mental Illness (NCISH). Manchester: University of Manchester. 2017;
        • Hawton K.
        • Bale L.
        • Brand F.
        • Townsend E.
        • Ness J.
        • Waters K.
        • et al.
        Mortality in children and adolescents following presentation to hospital after non-fatal self-harm in the Multicentre Study of Self-harm: a prospective observational cohort study.
        Lancet Child Adolesc Health. 2020; 4: 111-120
        • Huber R.S.
        • Sheth C.
        • Renshaw P.F.
        • Yurgelun-Todd D.A.
        • McGlade E.C.
        Suicide Ideation and Neurocognition Among 9- and 10-Year Old Children in the Adolescent Brain Cognitive Development (ABCD) Study.
        Archives of Suicide Research. 2020; : 1-15
        • Janiri D.
        • Doucet G.E.
        • Pompili M.
        • Sani G.
        • Luna B.
        • Brent D.A.
        • et al.
        Risk and protective factors for childhood suicidality: a US population-based study.
        Lancet Psychiatry. 2020; 7: 317-326
        • Aguinaldo L.
        • Goldstone A.
        • Hasler B.P.
        • Brent D.A.
        • Coronado C.
        • Jacobus J.
        Preliminary Analysis of Low-Level Alcohol Use and Suicidality with Children in the Adolescent Brain and Cognitive Development (ABCD) Baseline Cohort.
        Psychiatry Research. 2021; 299
        • Kessler R.C.
        • Bossarte R.M.
        • Luedtke A.
        • Zaslavsky A.M.
        • Zubizarreta J.R.
        Suicide prediction models: a critical review of recent research with recommendations for the way forward.
        Mol Psychiatry. 2020; 25: 168-179
        • O’Connor R.C.
        • Robb K.A.
        Identifying suicide risk factors in children is essential for developing effective prevention interventions.
        The Lancet Psychiatry. 2020; 7: 292-293
        • Franklin J.C.
        • Ribeiro J.D.
        • Fox K.R.
        • Bentley K.H.
        • Kleiman E.M.
        • Huang X.
        • et al.
        Risk factors for suicidal thoughts and behaviors: A meta-analysis of 50 years of research.
        Psychol Bull. 2017; 143: 187-232
      3. CC Zai VdL, J Strauss, RP Tong, I Sakinofsky, JL Kennedy (2012): Chapter 11. Genetic Factors and Suicidal Behavior. In: Dwivedi Y, editor. The Neurobiological Basis of Suicide. Boca Raton (FL): CRC Press/Taylor & Francis.

        • O'Reilly L.
        • Kuja-Halkola R.
        • Rickert M.E.
        • Class Q.A.
        • Larsson H.
        • Lichtenstein P.
        • D'Onofrio B.M.
        The intergenerational transmission of suicidal behavior: an offspring of siblings study.
        Transl Psychiatry. 2020; 173
        • Levey D.F.
        • Polimanti R.
        • Cheng Z.
        • Zhou H.
        • Nunez Y.Z.
        • Jain S.
        • et al.
        Genetic associations with suicide attempt severity and genetic overlap with major depression.
        Transl Psychiatry. 2019; 9: 22
        • Perlis R.H.
        • Huang J.
        • Purcell S.
        • Fava M.
        • Rush A.J.
        • Sullivan P.F.
        • et al.
        Genome-wide association study of suicide attempts in mood disorder patients.
        Am J Psychiatry. 2010; 167: 1499-1507
        • Docherty A.R.
        • Shabalin A.A.
        • DiBlasi E.
        • Monson E.
        • Mullins N.
        • Adkins D.E.
        • et al.
        Genome-Wide Association Study of Suicide Death and Polygenic Prediction of Clinical Antecedents.
        Am J Psychiatry. 2020; 177: 917-927
        • Mullins N.
        • Bigdeli T.B.
        • Børglum A.D.
        • Coleman J.R.I.
        • Demontis D.
        • Mehta D.
        • et al.
        GWAS of Suicide Attempt in Psychiatric Disorders and Association With Major Depression Polygenic Risk Scores.
        American Journal of Psychiatry. 2019; 176: 651-660
        • Mullins N.
        • Kang J.
        • Campos A.I.
        • Coleman J.R.I.
        • Edwards A.C.
        • Galfalvy H.
        • et al.
        Dissecting the shared genetic architecture of suicide attempt, psychiatric disorders and known risk factors.
        medRxiv. 2020; (2020.2012.2001.20241281)
        • Willour V.
        • Seifuddin F.
        • Mahon P.B.
        • Jancic D.
        • Pirooznia M.
        • Steele J.
        • et al.
        Nonsuicidal self-injury, suicide ideation, and suicide attempts among sexual minority children.
        J Consult Clin Psychol. 2021; 89: 73-80
        • Assari S.
        Racial Variation in the Association between Suicidal History and Positive and Negative Urgency among American Children.
        J Educ Cult Stud. 2020; 4: 39-53
        • Assari SB S.
        • Bazargan M.
        • Caldwell C.H.
        African Americans’ Diminished Returns of Parental Education on Adolescents’ Depression and Suicide in the Adolescent Brain Cognitive Development (ABCD) Study.
        Eur J Investig Health Psychol Educ. 2020; 10: 656-668
        • DeVille D.
        • Whalen D.
        • Breslin F.J.
        • Morris A.S.
        • Khalsa S.S.
        • Paulus M.P.
        • Barch D.M.
        Prevalence and Family-Related Factors Associated With Suicidal Ideation, Suicide Attempts, and Self-injury in Children Aged 9 to 10 Years.
        JAMA Netw Open. 2020; 3
        • Zucker R.A.
        • Gonzalez R.
        • Feldstein Ewing S.W.
        • Paulus M.P.
        • Arroyo J.
        • Fuligni A.
        • et al.
        Assessment of culture and environment in the Adolescent Brain and Cognitive Development Study: Rationale, description of measures, and early data.
        Dev Cogn Neurosci. 2018; 32: 107-120
        • Barch D.M.
        • Albaugh M.D.
        • Avenevoli S.
        • Chang L.
        • Clark D.B.
        • Glantz M.D.
        • et al.
        Demographic, physical and mental health assessments in the adolescent brain and cognitive development study: Rationale and description.
        Dev Cogn Neurosci. 2018; 32: 55-66
        • Luciana M.
        • Bjork J.M.
        • Nagel B.J.
        • Barch D.M.
        • Gonzalez R.
        • Nixon S.J.
        • et al.
        Adolescent neurocognitive development and impacts of substance use: Overview of the adolescent brain cognitive development (ABCD) baseline neurocognition battery.
        Developmental Cognitive Neuroscience. 2018; 32: 67-79
        • Achenbach T.
        International findings with the Achenbach System of Empirically Based Assessment (ASEBA): applications to clinical services, research, and training.
        Child Adolesc Psychiatry Ment Health. 2019; 13
      4. Achenbach TM, & Rescorla, L.A. (2001): Manual for the ASEBA School-Age Forms & Profiles. In: University of Vermont RCfC, Youth, & Families., editor. Burlington, VT.

        • Lee P.
        • Anttila V.
        • Won H.
        • Feng Y.A.
        • Rosenthal J.
        • Zhu Z.
        • et al.
        Genomic Relationships, Novel Loci, and Pleiotropic Mechanisms across Eight Psychiatric Disorders.
        Cell. 2019; 179: 1469-1482
        • Choi S.
        • Mak T.S.
        • O'Reilly P.F.
        Tutorial: a guide to performing polygenic risk score analyses.
        Nat Protoc. 2020; 15: 2759-2772
        • Nagelkerke N.
        A note on a general definition of the coefficient of determination.
        Biometrika. 1991; 78: 691-692
        • Mullins N.
        • Bigdeli T.B.
        • Børglum A.D.
        • Coleman J.R.I.
        • Demontis D.
        • Mehta D.
        • et al.
        GWAS of Suicide Attempt in Psychiatric Disorders and Association With Major Depression Polygenic Risk Scores.
        Am J Psychiatry. 2019; 176: 651-660
        • Ayer L.
        • Colpe L.
        • Pearson J.
        • Rooney M.
        • Murphy E.
        Advancing Research in Child Suicide: A Call to Action.
        J Am Acad Child Adolesc Psychiatry. 2020; 59: 1028-1035
        • Ribeiro J.
        • Huang X.
        • Fox K.R.
        • Franklin J.C.
        Depression and hopelessness as risk factors for suicide ideation, attempts and death: meta-analysis of longitudinal studies.
        Br J Psychiatry. 2018; 212: 279-286
        • Shen Y.
        • Chan B.S.M.
        • Huang C.
        • Cui X.
        • Liu J.
        • Lu J.
        • et al.
        Suicidal behaviors and attention deficit hyperactivity disorder (ADHD): a cross-sectional study among Chinese medical college students.
        BMC Psychiatry. 2021; 21: 258
        • Fuller-Thomson E.
        • Rivière R.N.
        • Carrique L.
        • Agbeyaka S.
        The Dark Side of ADHD: Factors Associated With Suicide Attempts Among Those With ADHD in a National Representative Canadian Sample.
        Arch Suicide Res. 2020; (Online ahead of print.:1-19)
        • Giupponi G.
        • Giordano G.
        • Maniscalco I.
        • Erbuto D.
        • Berardelli I.
        • Conca A.
        • et al.
        Suicide risk in attention-deficit/hyperactivity disorder.
        Psychiatr Danub. 2018; 30: 2-10
        • Yoshimasu K.
        • Barbaresi W.J.
        • Colligan R.C.
        • Voigt R.G.
        • Killian J.M.
        • Weaver A.L.
        • Katusic S.K.
        Psychiatric comorbidities modify the association between childhood ADHD and risk for suicidality: a population-based longitudinal study.
        J Atten Disord. 2019; 23: 777-786
        • Biederman J.
        • Ball S.W.
        • Monuteaux M.C.
        • Mick E.
        • Spencer T.J.
        • McCREARY M.
        • Cote M.
        • Faraone S.V.
        New insights into the comorbidity between ADHD and major depression in adolescent and young adult females.
        J Am Acad Child Adolesc Psychiatry. 2008; 47
        • Garas P.
        • Balazs J.
        Long-Term Suicide Risk of Children and Adolescents With Attention Deficit and Hyperactivity Disorder—A Systematic Review.
        Front Psychiatry. 2020; 11: 557909
        • Campos A.
        • Verweij K.J.H.
        • Statham D.J.
        • Madden P.A.F.
        • Maciejewski D.F.
        • Davis K.A.S.
        • et al.
        Genetic aetiology of self-harm ideation and behaviour.
        Sci Rep. 2020; 10: 9713
        • Maciejewski D.
        • Creemers H.E.
        • Lynskey M.T.
        • Madden P.A.
        • Heath A.C.
        • Statham D.J.
        • et al.
        Overlapping genetic and environmental influences on nonsuicidal self-injury and suicidal ideation: different outcomes, same etiology?.
        JAMA Psychiatry. 2014; 71: 699-705
        • Lewis C.
        • Vassos E.
        Polygenic risk scores: from research tools to clinical instruments.
        Genome Med. 2020; 12: 44
        • Tucker G.
        • Price A.L.
        • Berger B.
        Improving the Power of GWAS and Avoiding Confounding from Population Stratification with PC-Select.
        Genetics. 2014; 197: 1045-1049
        • Martin A.
        • Kanai M.
        • Kamatani Y.
        • Okada Y.
        • Neale B.M.
        • Daly M.J.
        Clinical use of current polygenic risk scores may exacerbate health disparities.
        Nat Genet. 2019; 51: 584-591
        • Duncan L.
        • Shen H.
        • Gelaye B.
        • Meijsen J.
        • Ressler K.
        • Feldman M.
        • et al.
        Analysis of polygenic risk score usage and performance in diverse human populations.
        Nat Commun. 2019; 10: 3328
        • Bridge J.
        • Goldstein T.R.
        • Brent D.A.
        Adolescent suicide and suicidal behavio.
        J Child Psychol Psychiatry. 2006; 47: 372-394
        • Sheftall A.
        • Asti L.
        • Horowitz L.M.
        • Felts A.
        • Fontanella C.A.
        • Campo J.V.
        • Bridge J.A.
        Suicide in Elementary School-Aged Children and Early Adolescents.
        Pediatrics. 2016; 138e20160436
        • Nierenberg A.
        Precision Psychiatry and a Vision for the Future.
        Psychiatric Annals. 2021; 51 (206-206)
        • Niculescu A.
        • Le-Niculescu H.
        • Levey D.F.
        • Phalen P.L.
        • Dainton H.L.
        • Roseberry K.
        • et al.
        Precision medicine for suicidality: from universality to subtypes and personalization.
        Mol Psychiatry. 2017; 22: 1250-1273
        • Schmaal L.
        • van Harmelen A.L.
        • Chatzi V.
        • Lippard E.T.C.
        • Toenders Y.J.
        • Averill L.A.
        • et al.
        Imaging suicidal thoughts and behaviors: a comprehensive review of 2 decades of neuroimaging studies.
        Mol Psychiatry. 2020; 25: 408-427
        • Hibar D.
        • Stein J.L.
        • Renteria M.E.
        • Arias-Vasquez A.
        • Desrivières S.
        • Jahanshad N.
        • et al.
        Common genetic variants influence human subcortical brain structures.
        Nature. 2015; 520: 224-229