TL;DR
This study investigates the genetic underpinnings of suicide attempts (SA) in Korean patients with bipolar disorder (BD) using polygenic risk scores (PRS). They found that a PRS for SA derived from European ancestry data could predict SA in Koreans, and surprisingly, a PRS for obsessive-compulsive disorder (OCD) was associated with repeated SA, suggesting shared genetic pathways.
What Problem Does This Paper Address?
Suicide is a critical public health issue, with individuals with bipolar disorder (BD) facing a significantly higher risk of suicide attempts (SA) and completed suicides compared to the general population. While genetic factors are known to play a role in suicidal behavior, the specific genetic architecture, especially how it overlaps with other psychiatric disorders, remains complex and not fully understood. Previous large-scale genetic studies have primarily focused on European ancestry and often treated SA as a single phenotype. This study aims to dissect the genetic architecture of SA and repeated attempts in a Korean BD population by leveraging polygenic risk scores (PRS) derived from existing genome-wide association studies (GWAS) of SA and other psychiatric disorders. They specifically investigate whether genetic predispositions for SA and related disorders like BD, schizophrenia (SCZ), major depressive disorder (MDD), and obsessive-compulsive disorder (OCD) contribute to SA in this specific patient group.
Methods
The researchers employed polygenic risk score (PRS) analysis, a method that aggregates the effects of many common genetic variants across the genome to estimate an individual's genetic predisposition to a particular trait. They used the PRSice-2 software to calculate PRSs for suicide attempt (PRS-SA) and four psychiatric disorders (PRS-BD, PRS-SCZ, PRS-MDD, PRS-OCD). These PRSs were derived from existing GWAS summary statistics.
For statistical analysis, they used univariate logistic regression to assess the association between each PRS and lifetime SA, with Nagelkerke's R² used to evaluate model performance. Multivariate logistic regression was then used to adjust for covariates like age, sex, diagnosis (BD-I vs. BD-II), and principal components to account for population stratification. They also conducted univariate linear regression to explore the relationship between the number of attempts and PRSs. To investigate the association with repeated attempts, they used logistic regression to compare multiple attempters with single attempters.
Datasets
The study utilized two main types of data:
- Participant Data: 383 Korean patients diagnosed with bipolar I (BD-I) or bipolar II (BD-II) disorder from Samsung Medical Center and Seoul National University Bundang Hospital. These participants underwent clinical assessments for lifetime SA and the number of attempts.
- GWAS Summary Statistics:
- For PRS-SA: Data from the GWAS of suicide attempt in European ancestry (EUR) patients with BD, specifically from the study by Mullins et al. (2019). They also explored PRS-SA from SCZ and MDD samples from the same GWAS for exploratory purposes.
- For PRS-BD, PRS-SCZ, PRS-MDD, PRS-OCD: Summary statistics from the Psychiatric Genomics Consortium (PGC) for these disorders.
Key Insights and Ideas
- Cross-Ancestry Generalizability of PRS-SA: The study demonstrates that a PRS for SA, initially developed from European ancestry data, can be generalized to predict SA in a Korean population with BD. This is a significant finding, as PRS performance can sometimes decrease across different ethnic groups.
- Unexpected Link Between OCD and Repeated SA: Perhaps the most striking finding is the association between the PRS for OCD and repeated suicide attempts. This suggests a shared genetic pathway between OCD and the more severe phenotype of multiple SAs, which was not observed for PRS-SA in relation to repeated attempts.
- Refined Phenotyping is Crucial: The study highlights the importance of looking beyond a general SA phenotype. The PRS-OCD's association with repeated attempts, rather than just any SA, underscores the need for more granular sub-phenotypes in genetic research for suicide.
- BD-II Subgroup Specificity: The significant associations for PRS-SA and PRS-OCD were primarily observed in the BD-II subgroup, which also had a higher rate of SA. This suggests potential differences in the genetic architecture of SA between BD-I and BD-II, warranting further investigation with larger samples.
Key Results
- The PRS for SA (derived from EUR BD data) was significantly associated with lifetime SA in Korean BD patients, increasing the odds of SA by 1.36-fold per one SD increase in PRS (Nagelkerke's R² = 2.73%, p=0.007). This association remained significant after adjusting for covariates.
- The PRS for OCD was also significantly associated with lifetime SA (Nagelkerke's R² = 2.72%, OR=1.36, p=0.007).
- Crucially, the PRS for OCD was significantly associated with repeated attempts, showing a 1.53-fold higher odds of multiple attempts vs. single attempts (Nagelkerke's R²=4.91%, p=0.043).
- PRSs for BD, SCZ, and MDD did not show significant associations with lifetime SA in the overall sample.
- In subgroup analysis, PRS-SA and PRS-OCD were significantly associated with lifetime SA only in the BD-II subgroup.
Limitations and Future Work
The authors acknowledge several limitations:
- Limited Sample Size: The study's sample size may have been underpowered to detect smaller effect sizes, potentially leading to missed associations.
- Ethnic Differences in Reference Data: Using PRS derived from European ancestry data for a Korean population might reduce explanatory power.
- Lack of Clinical Correlates: The study did not evaluate diverse clinical correlates of SA that could mediate the observed associations.
- Retrospective Assessment: SA was assessed retrospectively, which carries inherent recall biases.
- Observation Period: The observation period for lifetime SA was not fixed, and the possibility of false negatives in non-attempters exists.
Future work should focus on:
- Larger, Ethnically Diverse Samples: Accumulating larger datasets from diverse ethnic backgrounds is crucial for validating findings and increasing statistical power.
- Refined Phenotypes: Developing and utilizing more specific sub-phenotypes of suicide (e.g., intent, lethality, method) will be key to uncovering more precise genetic architectures.
- Investigating BD Subtypes Separately: Future studies should consider performing separate analyses for BD-I and BD-II due to potential differences in genetic underpinnings.
- Integrating Genetic and Clinical Factors: Combining polygenic risk with clinical risk factors is essential for developing robust suicide prediction models.
Summary updated