Analytical Validation for an Integrated Risk Tool for Cardiovascular Disease Combining Conventional Risk Factors and Polygenic Risk Inferred from Snp Array and Low-Pass Whole Genome Sequencing

Genetic risk factors contribute to cardiovascular disease (CVD) risk and are imperfectly captured by family history. The development of integrated risk tools (IRT) that integrate polygenic risk scores (PRS) with conventional risk factors enables improved risk assessment. However, IRTs rely on accurate estimation of the genetic component across millions of genetic variants, either typed or imputed. To support two real-world implementations of CVD IRTs, we present analytical validation data for three biosample types (whole blood, saliva and buccal swab samples) and two genetic assay technologies (a custom-designed genotyping array (845,608 markers) and low pass whole genome sequencing (lpWGS) with 1.8x average read depth, both combined with computational imputation) across 48 ethnically-diverse donors. IRT measurements (range 4%-23%) were performed on the 10-year CVD risk percentage scale, assuming a non-genetic (pooled cohort equations) risk component of 10%. We find measurement errors are below 1% across the three biosample types, so that an IRT of 4% will at most be measured as 3% or 5%. Array and lpWGS showed different bias-variance trade offs. The array technology was more reproducible, with a standard deviation on the percentage scale of 0.09 (0.072-0.129 95% CI) for blood samples compared with 0.30 (0.201-0.617 95% CI) for lpWGS. In contrast, the array technology showed somewhat larger bias when compared with high depth sequencing data (mean absolute error on the percentage scale of 0.49 (0.217-0.755 95% CI) for blood samples against 0.30 (0.008-0.597 95% CI) for lpWGS, partially reflecting the sequencing basis of the reference data. Pass rate was high for both technologies and all three biosample types. In conclusion, both genotyping array and lpWGS approaches were appropriate and performed well for the purpose of assessing the genetic component of integrated cardiovascular disease risk using either blood, saliva or buccal samples.