Table 2.
Disease, authors (y) | SNVs, n | Participants, n cases/N total* | Biobank | Participants: ancestry/ethnicity, age, y | Outcome 1: HR/OR (95% CI) (included covariates controlled for)* | Outcome 2: AUC/C statistic (95% CI) (included covariates controlled for)* | Outcome 3: AUC/C-statistic (95% CI): clinical risk tool vs PRS+clinical risk tool | Outcome 4: NRI (95% CI) comparison |
---|---|---|---|---|---|---|---|---|
AF, Mars et al23 (2020) | 6 171 733 | 12 809/135 300 For model comparing clinical risk model, 229/21 030 |
FinnGenn | European, 59.2±16.6 | HR per SD, 1.62 (1.59–1.65) (collection year, genotyping array/batch and the first 10 principal components of ancestry, and stratified the models by sex) | C statistic, 0.751 (0.744–0.757) (collection year, genotyping array/batch and the first 10 principal components of ancestry, and stratified the models by sex) | CHARGE-AF, 0.725 (0.719–0.732) CHARGE-AF+PRS, 0.734 (0.728–0.741) | Using >5% risk threshold over 5 y, 10.4 (4.1–16.7) CHARGE-AF vs CHARGE-AF+PRS |
CAD, Riveros-Mckay et al63 (2021) | >3 500 000 | 4247/186 541 | UK Biobank | European, 40–69 | HR per SD, 1.90 (1.86–1.95) (age, sex, principle components of ancestry) | C statistic, 0.633 (0.625–0.641) (age, sex, principle components of ancestry) | AHA/ACC PCE, 0.76 (0.75–0.76) AHA/ACC PCE+PRS, 0.79 (0.78–0.79) | Using >7.5% threshold over 10 y, 5.9% (4.7%–7.0%) AHA/ACC PCE vs AHA/ACC PCE+PRS |
T2D, Mars et al23 (2020) | 6 437 380 | 17 519/135 300 For model comparing clinical risk model, 1346/10 561 |
FinnGenn | European, 59.2±16.6 | HR per SD, 1.74 (1.72–1.77) (collection year, genotyping array/batch and the first 10 principal components of ancestry, and stratified the models by sex) | C statistic, 0.763 (0.758–0.767) (collection year, genotyping array/batch and the first 10 principal components of ancestry, and stratified the models by sex) | ADA, 0.835 (0.831–0.839) ADA+PRS, 0.845 (0.841–0.849) | NRI using 10-y risk ≥33% risk threshold, 4.5% (3.0%–6.1%) ADA criteria |
ADA indicates American Diabetes Association; AF, atrial fibrillation; AHA/ACC, American Heart Association/American College of Cardiology; AUC, area under the receiver-operating characteristic curve; CAD, coronary artery disease; CHARGE-AF, Cohorts for Heart and Aging Research in Genomic Epidemiology model for atrial fibrillation; HR, hazard ratio; NRI, net reclassification index; OR, odds ratio; PCE, Pooled Cohort Equation; PRS, polygenic risk score; SNV, single nucleotide variant; and T2D, type 2 diabetes.
Data from validation data set/analysis.