Ancestry-specific polygenic scores and SNP heritability of 25(OH)D in African- and European-ancestry populations

Kathryn E. Hatchell, Qiongshi Lu, Scott J. Hebbring, Erin D. Michos, Alexis C. Wood, Corinne D. Engelman

Abstract

Vitamin D inadequacy, assessed by 25-hydroxyvitamin D [25(OH)D] concentrations, affects approximately 50% of adults in the United States and is linked to various adverse health outcomes. While genetic factors are known to influence 25(OH)D levels, traditional genome-wide association studies (GWAS) have only identified a small portion of these contributors. This study aimed to develop ancestry-specific polygenic scores (PGS) to better predict 25(OH)D levels and inform personalized supplementation. Using data from 9,569 European-ancestry and 2,761 African-ancestry participants across three cohorts, researchers created PGS models using weights from the TRANSCEN-D consortium. The results showed that individuals at the highest genetic risk had 25(OH)D levels 2.8–3.0 ng/ml lower than those at the lowest risk, requiring an estimated additional 467 to 500 IU of daily vitamin D intake to maintain equivalent blood concentrations. These findings suggest that PGS can serve as a powerful tool for precision nutrition and the prevention of vitamin D deficiency.

Introduction

Vitamin D deficiency is a pervasive public health issue, associated with a range of conditions from impaired bone health to increased cardiovascular risk. Although heritability estimates suggest that 20% to 40% of the variation in blood 25(OH)D levels is due to genetics, existing GWAS findings have historically explained less than 3% of this variance. This "missing heritability" indicates that many small-effect genetic variants remain unaccounted for in standard clinical models.

Furthermore, the majority of genetic research has focused on populations of European ancestry, creating a significant gap in our understanding of how these genetic factors manifest in other groups. This study addresses these gaps by developing and validating ancestry-specific polygenic scores for both African- and European-ancestry populations.

Methodology

To capture a more comprehensive set of genetic factors, the researchers employed a polygenic score (PGS) approach, which calculates a weighted sum of trait-associated alleles across the genome.

Data Sources and Weighting

The study utilized summary statistics from the TRANSCEN-D consortium—the largest multi-ethnic vitamin D GWAS to date—to provide weights for individual Single Nucleotide Polymorphisms (SNPs). The models were tested and validated using data from three major multi-ethnic cohorts:

  • ARIC (Atherosclerosis Risk in Communities)
  • MESA (Multi-ethnic Study of Atherosclerosis)
  • WHI (Women’s Health Initiative)

Optimization and Analysis

The researchers used PRSice v2 software to optimize the PGS by testing various p-value thresholds to determine which set of SNPs provided the best predictive power. Additionally, GCTA (Genome-wide Complex Trait Analysis) was used to estimate total SNP heritability, allowing the team to compare the variance captured by the new PGS against the total potential genetic variance explained by common SNPs.

Key Insights

The Importance of Ancestry Matching

A critical finding of the study was the necessity of ancestry-specific tools. When a PGS developed using European-ancestry weights was applied to African-ancestry individuals, the predictive power ($R^2$) dropped significantly (from 0.31% to 0.14%). This underscores the risk of exacerbating health disparities if genomic tools are not tailored to diverse populations.

Genetic Architecture Differences

The optimal PGS for African-ancestry populations required the inclusion of all measured SNPs (a p-value cut-off of 1.0), whereas the European-ancestry model was optimized at a much more stringent threshold ($3.5 \times 10^{-4}$). This suggests that in African-ancestry populations, the genetic influence on vitamin D is distributed across a vast number of variants with very small individual effects.

Results and Clinical Impact

Predictive Performance

The optimized PGS explained 1.0% of the variance in the European validation cohort and 0.3% in the African validation cohort. While these percentages appear modest, they represent a significant advancement in capturing the genetic components of vitamin D regulation compared to previous models.

Quantifying Supplementation Needs

The study translated these genetic scores into actionable clinical metrics. Individuals in the highest decile of genetic risk for low vitamin D had concentrations 2.8 to 3.0 ng/ml lower than those in the lowest risk decile. To bridge this gap, the researchers calculated that high-risk individuals would need an additional 467 to 500 IU of vitamin D daily to achieve the same blood levels as those with a low genetic risk profile.

Discussion and Limitations

While this study marks a step forward for precision medicine, several limitations remain. The African-ancestry tuning dataset was relatively small (n=57), which likely constrained the model's predictive power. Furthermore, the current PGS focuses on common SNPs (Minor Allele Frequency > 0.01), meaning rare variants that might contribute to heritability were not included.

Environmental factors also remain a challenge; while the study adjusted for UV radiation based on geography and season, these are proxies for actual sun exposure, which is modified by individual behaviors such as sunscreen use and time spent outdoors.

Conclusion

Ancestry-specific polygenic scores provide a more nuanced understanding of vitamin D metabolism than traditional genetic markers. By identifying individuals who are genetically predisposed to lower 25(OH)D levels, clinicians can move toward a precision medicine approach—prescribing specific vitamin D dosages tailored to an individual's genetic profile to effectively prevent inadequacy and its associated health risks.

Summary updated

BibTeX
@inproceedings{hatchell2019ancestry,
  author = {Kathryn E. Hatchell and Qiongshi Lu and Scott J. Hebbring and Erin D. Michos and Alexis C. Wood and Corinne D. Engelman},
  title = {{Ancestry-specific polygenic scores and SNP heritability of 25(OH)D in African- and European-ancestry populations}},
  booktitle = {Human},
  year = {2019},
  doi = {10.1007/s00439-019-02049-x},
}
Figure