Enhancing the Equity and Accuracy of Polygenic Risk Scores Through Inclusion of Diverse Genetic Data

Polygenic Risk Scores (PRSs) are promising tools in personalized medicine, offering insights into an individual’s predisposition to various diseases. However, a pressing issue with PRSs is their limited generalizability across diverse populations. The study titled "Inclusion of Variants Discovered from Diverse Populations Improves Polygenic Risk Score Transferability" by Cavazos and Witte systematically addresses this challenge, shedding light on biases inherent in European-derived PRSs and proposing strategies to enhance their accuracy and equity.

The Current Challenge
PRSs have primarily been developed using data from individuals of European ancestry, representing a stark overrepresentation compared to their global population percentage. This imbalance results in PRSs that perform well in European populations but falter in non-European groups, especially those of African ancestry. This discrepancy is attributed to differences in linkage disequilibrium (LD), allele frequencies, and population-specific genetic variants. Moreover, PRS accuracy decreases linearly with increasing genetic divergence from European ancestry.

Key Findings from Simulations
The authors employed simulations to explore the relationship between ancestry and PRS performance. Using coalescent modeling, they simulated European and African genotypes and created admixed populations with varying proportions of European and African ancestry. Their findings were compelling:

Bias in European-Derived PRSs:
European-derived PRSs exhibited high accuracy in European populations but significantly reduced accuracy in African and admixed populations.
For instance, accuracy (measured by Pearson’s correlation) dropped by 41% in African populations compared to Europeans.

Improvement with African-Specific Data:
Including variants discovered from African ancestry GWAS significantly improved PRS accuracy across diverse populations.
African-specific variants provided better LD tagging of causal variants, highlighting the need for genetic studies in underrepresented populations.

The Role of Admixed Populations:
PRS performance in admixed individuals was influenced by the proportion of European ancestry. The accuracy increased by 1.34% for every 10% increase in European ancestry.

Empirical Validation with UK Biobank Data
The study validated these findings using UK Biobank data, examining traits like hemoglobin A1c, asthma, and prostate cancer. They tested various PRS construction methods, including:
European and African ancestry-specific GWASs.
Fixed-effects meta-analyses combining both populations.

For African ancestry individuals, PRSs that incorporated African-specific variants and weights outperformed those derived solely from European data. Notably, a combined PRS approach, integrating data from both European and African populations, yielded the most balanced and accurate predictions across all populations.

Practical Implications
The findings underscore the critical need to include diverse populations in genomic studies. By leveraging African ancestry data, PRSs can achieve:

Enhanced Accuracy: Improved LD tagging and detection of population-specific variants lead to better prediction models.
Reduced Bias: A multi-ancestry approach diminishes disparities in PRS accuracy among different populations.

Future Directions
Expand Diverse Cohort Recruitment:
Large-scale initiatives like H3Africa and All of Us are pivotal for increasing representation in genomic datasets.

Optimize PRS Construction:
Approaches combining ancestry-specific PRSs, such as linear mixtures of population-derived scores, show promise in addressing current limitations.

Integrate Advanced Methods:
Incorporating whole-genome sequencing in diverse populations can overcome the limitations of imputation-based genotyping arrays.

Conclusion
The study by Cavazos and Witte provides a roadmap for improving PRS transferability and equity. By prioritizing the inclusion of underrepresented populations in genomic research, we can harness the full potential of PRSs, ensuring that advancements in precision medicine benefit individuals across all ancestries.

References:
Cavazos, T. B., & Witte, J. S. (2021). Inclusion of variants discovered from diverse populations improves polygenic risk score transferability. Human Genetics and Genomics Advances, 2(1).