TL;DR
This study investigates whether polygenic risk scores (PRSs) for various psychiatric traits can predict general and specific psychopathology in children. The findings reveal that PRSs are more broadly associated with general psychopathology or a combination of general and specific symptoms, rather than predicting a single specific symptom domain exclusively. Combining multiple PRSs shows promise for improving predictive accuracy.
What Problem Does This Paper Address?
Psychiatric disorders are known to have a significant genetic component. Genome-wide association studies (GWASs) have identified genetic variants associated with various psychiatric conditions. Polygenic risk scores (PRSs) are derived from these GWASs to quantify an individual's genetic predisposition to a particular trait or disorder. However, a critical question remains: how specific are these PRSs in predicting psychopathology, especially in childhood, where symptoms can be fluid and co-occur? This paper addresses the gap in understanding the specificity of PRSs for different psychiatric domains in children, aiming to determine if they predict general psychopathology, specific symptom clusters (internalizing/externalizing), or both. This is crucial for developing early detection and intervention strategies.
Methods
The researchers computed 16 PRSs based on existing GWAS data for psychiatric phenotypes, neuroticism, and cognitive ability. These PRSs were then analyzed in relation to psychopathology factors derived from a bifactor model. This model distinguishes a general psychopathology factor from specific internalizing and externalizing factors. The study employed two main analytical approaches:
- Separate PRS Models: Each PRS was individually regressed against the general, internalizing, and externalizing psychopathology factors. This allowed for the assessment of each PRS's unique contribution.
- Mutually Adjusted PRS Model: All PRSs were entered simultaneously into a regression model to determine their independent contributions to general psychopathology, controlling for each other.
Analyses were conducted across three population-based cohorts and meta-analyzed. Statistical adjustments were made for age, sex, and ancestry. Multiple testing adjustments were applied to account for the large number of statistical tests performed.
Datasets
The study utilized data from three large, population-based, prospective cohorts:
- ALSPAC (Avon Longitudinal Study of Parents and Children): UK-based.
- Generation R (GenR): Netherlands-based.
- MAVAN (Maternal Adversity, Vulnerability, and Neurodevelopment): Canada-based.
The combined sample size for meta-analyses was 9,247 school-aged children. The PRSs themselves were derived from GWASs of mostly adult populations, as detailed in the paper's appendices. No new benchmark datasets were introduced.
Key Insights and Ideas
- Nonspecificity is the Norm: The most striking finding is that PRSs for psychiatric and cognitive traits are generally not specific to a single psychopathology domain in childhood. Instead, they tend to predict general psychopathology or a combination of general and specific symptoms. This highlights the complex genetic architecture underlying psychopathology, where many genetic variants influence multiple traits.
- General Psychopathology as a Dominant Predictor: PRSs appear to capture a broad genetic predisposition that underlies a wide range of psychiatric symptoms, rather than pinpointing a single disorder. This aligns with the concept of a general psychopathology factor ("p-factor").
- Combining PRSs Enhances Prediction: While individual PRSs have limited predictive power for specific symptoms, combining multiple PRSs significantly improves the prediction of general psychopathology. This suggests a polygenic approach, integrating information from various genetic predispositions, is more effective.
- Cognitive Ability as a Key Predictor: Notably, the PRS for cognitive ability emerged as a strong predictor of both general and specific psychopathology. This underscores the intricate link between cognitive function and mental health.
- Implications for Research and Clinical Practice: The findings have significant implications. Researchers using PRSs as genetic instruments in Mendelian randomization studies need to be cautious, as PRSs may affect outcomes through pathways not directly related to the disorder they were computed for. Clinicians should be aware of this nonspecificity when interpreting PRS results and consider using a combination of PRSs for more comprehensive risk assessment.
Key Results
- General Psychopathology: Seven PRSs (cognitive ability, ADHD, major depression, neuroticism, schizophrenia, insomnia, and depressive symptoms) were significantly associated with general psychopathology. These PRSs jointly explained approximately 2.02% of the variance in general psychopathology.
- Specific Psychopathology:
- Externalizing: Two PRSs (ADHD and cognitive ability) were associated with specific externalizing psychopathology.
- Internalizing: Five PRSs (neuroticism, generalized anxiety, ADHD, cognitive ability, and schizophrenia) were associated with specific internalizing psychopathology.
- Independence of Effects: Most of the seven PRSs associated with general psychopathology showed independent contributions, with the exception of major depression and depressive symptoms, whose effects were not distinct when considered together with other PRSs.
- Limited Specificity: Only one PRS (generalized anxiety) was found to be exclusively associated with specific internalizing psychopathology without also being associated with general psychopathology.
Limitations and Future Work
- Limited Explained Variance: While combining PRSs improved prediction, the total explained variance in general psychopathology (around 2%) remains modest. This suggests that current GWASs, even with large sample sizes, capture only a fraction of the genetic contribution to psychopathology.
- Adult GWASs for Child Phenotypes: The PRSs were derived from GWASs of adult populations. The authors suggest that the development of more powerful GWASs specifically for childhood psychiatric traits could lead to more precise and clinically useful PRSs.
- Ancestry Limitations: The study focused on participants of European ancestry due to challenges in applying PRSs derived from European-ancestry GWASs to other populations. Future research should explore these associations in diverse populations to assess generalizability and potential exacerbation of health disparities.
- Age-Specific Effects: While the study included children, the age range varied (4-8 years). Genetic effects can change with age, and future studies could explore age-specific genetic influences on psychopathology.
- Future Directions: The authors emphasize the need for researchers and clinicians to consider related traits when computing PRSs and to combine multiple PRSs for improved prediction. They also highlight the potential of using general psychopathology-adjusted GWASs to develop more domain-specific PRSs.
Summary updated