Figure 4
From: Identifying and characterizing disease subpopulations that most benefit from polygenic risk scores
Comparative prevalence rates and mean values between top 10% or bottom 10% subgroup cases and all other cases, respectively, for selected significant differentiating features for each disease. CAD coronary artery disease (refers to myocardial infarction), BC breast cancer, SZ schizophrenia, BMI body mass index, LDL low density lipoprotein, PRS polygenic risk score. The term, “All Others” represents all other cases in the population that are not in the top 10% subgroup that most benefits from PRSs. (a) For BC three significant features that differentiate the cases in the top 10% subgroup that most benefits from PRSs for risk prediction. One polygenic risk score, BC_5k, is shown to illustrate the difference in distribution of the score between the top 10% cases and all other cases. The top 10% subgroup cases for breast cancer are significantly younger (~ 2 years on average) and slightly leaner, but still overweight on average. (b) For CAD, a selection of numerical features is shown to compare the top 10% subgroup cases that most benefit from PRSs for risk prediction to all other 10-year incident first-time myocardial infarction cases. Apolipoprotein B, LDL, triglycerides, and cholesterol, all of which are biomarkers associated with lipid metabolism, were significantly higher in the top 10% subgroup cases than in other cases, however the mean body mass index was lower, suggesting a slightly leaner phenotype despite signs of poor cardiovascular health. Still, the difference was small (Cohen’s d = − 0.16), with both groups of cases being overweight (BMI > 25 kg/m2) on average. (c) For SZ, some of the binary features with the highest effect size in differentiating the top 10% subgroup cases are shown. Although less 5% of all other 10-year SZ incident cases had a recent marital separation or divorce, this was the case for ~ 17% of the top 10% subgroup cases. Other differentiating factors included having recent financial difficulties (~ 55%) and being male (~ 65%). (d) Some of the most important binary features differentiating the bottom 10% subgroup for SZ are shown.
