A Tool for Translating Polygenic Scores onto the Absolute Scale Using Summary Statistics

Abstract

Background: There is growing interest in the clinical application of polygenic scores as their predictive utility increases for a range of health-related phenotypes. However, providing polygenic score predictions on the absolute scale is an important step for their safe interpretation. Currently, polygenic scores can only be converted to the absolute scale when a validation sample is available, presenting a major limitation in the interpretability and clinical utility of polygenic scores. Methods: We have developed a method to convert polygenic scores to the absolute scale for binary and normally distributed phenotypes. This method uses summary statistics, requiring only the area-under-the-ROC curve (AUC) or variance explained (R2) by the polygenic score, and the prevalence of binary phenotypes, or mean and standard deviation of normally distributed phenotypes. Polygenic scores are converted using normal distribution theory. Given the AUC/R2 of polygenic scores may be unknown, we also evaluate two methods (AVENGEME, lassosum) for estimating these values from genome-wide association study (GWAS) summary statistics alone. We validate the absolute risk conversion and AUC/R2 estimation using data for eight binary and three continuous phenotypes in the UK Biobank sample. Results: When the AUC/R2 of the polygenic score is known, the observed and estimated absolute values were highly concordant. Across binary phenotypes, the mean absolute difference between the observed and estimated proportion of cases was 5%. For continuous phenotypes, the mean absolute difference between observed and estimated means was <0.3%. Estimates of AUC/R2 from the lassosum pseudovalidation method were most similar to the observed AUC/R2 values, though estimated values deviated substantially from the observed for autoimmune disorders. Conclusion: This study enables accurate interpretation of polygenic scores using only summary statistics, providing a useful tool for educational and clinical purposes. Furthermore, we have created interactive webtools implementing the conversion to the absolute scale for binary and normally distributed phenotypes (https://opain.github.io/GenoPred/PRS_to_Abs_tool.html). Several further barriers must be addressed before clinical implementation of polygenic scores, such as ensuring target individuals are well represented by the GWAS sample.

Performance of polygenic risk scores in screening, prediction, and risk stratification

“…Because polygenic risk scores, like many non-genetic risk factors 22 , display a Gaussian distribution in the population 23 and a relatively weak log-linear association with disease risk, more cases occur among those at average than those at extreme risk 24 . This means that where there are safe, inexpensive preventative interventions (e.g., statins for CVD) there is greater public health benefit in extending rather than limiting their access 25 and this has been achieved by the progressive lowering of the risk threshold for statin treatment over time.…”

Patient and provider perspectives on polygenic risk scores: implications for clinical reporting and utilization

“…But while the gold standard for use of an absolute risk model includes incorporation of multiple risk factors from multiple longitudinal cohort studies, evidence of the populationbased prevalence of risk factors, and evidence-based guidelines tied to thresholds of risk, our results suggest that simple absolute risk models, perhaps just including age and sex, may be more appropriate than not reporting absolute risk at all. There is already a tool to convert a PRS into absolute risk solely via the incorporation of disease prevalence [38].…”

Teacher and Parent Perspectives on the Use of Genetic Information in Schools

“…Specific education‐related diagnoses have not yet been significantly predicted by polygenic scores (Shero et al, 2021; Warne, 2021). However, as these articles exemplify, this knowledge continues to be pursued and will inevitably advance (e.g., Genc et al, 2021; Pain, Gillett, Austin, Folkersen, & Lewis, 2022; Pain, Glanville, et al, 2021; von Stumm et al, 2020), indicating a strong potential for genetic information to eventually contribute to individualized education in some way. Thus, policymakers should be aware of and prepared for how such information may be received and interpreted.…”