{
  "abstract": "Background Coronary heart disease (CHD) remains the leading global cause of mortality and has a substantial genetic component. Polygenic risk scores (PGSs) have emerged as a promising tool for predicting CHD risk; however, most PGSs have been developed using predominantly European ancestry genome-wide association studies (GWAS). This raises concerns regarding their predictive accuracy and transferability to South Asian populations, who experience a disproportionately high burden of CHD (HR~2.03 between European and South Asian groups in general).Methods We evaluated the predictive performance of six published CHD PGSs derived from three independent studies in South Asian participants from the LOLIPOP cohort. Logistic regression models were fitted with and without lifestyle risk factors, adjusting for age and ancestry principal components. Predictive performance was assessed using area under the receiver operating characteristic curve (AUC), odds ratios (ORs) per standard deviation increase in PGS, and pseudo-R 2 values. Comparisons were made between CHD cases and controls.Results Substantial heterogeneity in predictive performance was observed across PGSs. Scores developed using Bayesian approaches, including Clumping and Shrinkage, LDpred, and LDpred2, demonstrated superior discrimination compared with pruning-and-thresholding methods. However, LDpred and LDpred2 models showed evidence of overfitting, likely reflecting mismatches between training and target populations. Overall, PGS performance was attenuated relative to that reported in European cohorts. Incorporation of lifestyle factors alongside genetic risk consistently improved model performance compared to using either genetic profile or lifestyle factors alone.Conclusion Existing CHD PGSs exhibit limited cross-ancestry transferability to South Asian populations, largely due to differences in genetic architecture and GWAS training data. While Bayesian methods offer improved predictive accuracy, their utility remains constrained without ancestry-specific GWAS and reference panels. Integrating genetic and lifestyle risk factors provides a more robust framework for CHD risk prediction in South Asians, highlighting the need for more inclusive genomic research to support equitable clinical implementation",
  "authors": [
    {
      "affiliations": [
        "Imperial College London, London, United Kingdom"
      ],
      "name": "Orion O’Connor"
    },
    {
      "affiliations": [
        "Imperial College London, London, United Kingdom"
      ],
      "name": "Madison Lee"
    },
    {
      "affiliations": [
        "Imperial College London, London, United Kingdom"
      ],
      "name": "Anusha Tahir"
    },
    {
      "affiliations": [
        "Imperial College London, London, United Kingdom"
      ],
      "name": "Harshita Bihani"
    }
  ],
  "title": "136 Efficacy of polygenic scores in predicting CHD in South Asian individuals",
  "uid": "95aada44-b4dd-58a6-bf1e-9a7b2f935742"
}