Document Type

Conference Proceeding

Publication Date


Publication Title

Eur Neuropsychopharmacol


Background: Suicide is a major public health concern that accounts for more than 800,000 deaths each year around the world. Known risk factors for suicide are mental health conditions and their related clinical variables. Among them, opioid use disorder (OUD) significantly increases risk for suicidal ideation, attempts, and death. In spite of their co-occurrence and reported association from observational findings, it remains to be elucidated whether suicidal behaviors are a cause or consequence of opioid use. Both suicide attempt and OUD are heritable (heritability at 0.55 and 0.34, respectively) and each has been previously investigated individually through genome-wide association studies (GWAS). When combined, existing GWAS data from both phenotypes also provides a means (i.e., Mendelian randomization [MR]) to infer a possible causal relationship between them outside of a clinical trial setting.

Methods: GWAS summary statistics for suicidal attempt were collected from two resources: 1) the UK Biobank analysis (Neale Lab, round 2) that was accessible through the IEU OpenGWAS project and 2) a Danish iPSYCH study (Erlangsen et al. 2020). Fixed-effect meta-analysis was used to combine the two datasets for suicide. Genetic variants present in both datasets showing no heterogeneity were retained. In addition, GWAS summary statistics for OUD were downloaded from Psychiatric Genomic Consortium (Polimanti et al. 2020) that studied both opioid dependence and opioid exposure. For both suicide attempt and OUD GWAS, studies included only individuals of principally European ancestry, and no overlapping individuals were identified between suicide attempt GWAS and OUD GWAS. Variants were also harmonized to have consistent strand and allele coding. Genotypes from 1000 Genome CEU population (n=183) and another subset of healthy controls (n =70,000) in the UK Biobank project were used as reference. MR inference was then conducted by TwoSampleMR and MR-PRESSO R packages.

Results: TwoSampleMR analyses indicated moderate but significant causal relationship from opioid dependence to suicide attempt (inverse-variance weighted [IVW]: OR=1.043; PIVW < 0.001); Weighted median: OR= 1.048, Pweighted-median =0.002; MR-Egger: OR =1.029, PMR-Egger =0.409). Heterogeneity tests showed no significant horizontal pleiotropy effects exist between the two phenotypes (P > 0.05 from MR-Egger intercept test and MR-PRESSO global test). As a comparison, no such causal relationship was found from opioid exposure to suicide attempt (IVW: OR = 0.975, PIVW = 0.591; weighted median: OR = 1.011, PIVW = 0.870; MR-Egger: OR =0.930, PMR-Egger = 0.486).

Discussion: In summary, this MR study reveals a possible causal effect of opioid dependence on suicidal attempt. Another larger OUD GWAS dataset from the Million Veteran Program is being used to replicate this finding. In addition, approaches for transcriptomic-wide association studies (e.g., SMR, FUSION and KGGSEE) will be adopted to infer potential molecular-level mechanistic explanations (i.e., shared causal SNPs and genes) for this cross-phenotype relationship.

Disclosure: Nothing to disclose.



First Page


Last Page