Quantifying genetic, lifestyle and socioeconomic mediation of familial cardiometabolic risk
S Demeester, A Misra, S Haidermota, W Hornsby, R Bhattacharya, A Fahed, P Natarajan, A P PatelAbstract
Background
Family history of cardiometabolic (CM) disease reflects both shared genetic and nongenetic factors. Understanding contributions to this risk factor could help inform cascade screening strategies.
Purpose
To quantify and compare familial risk mediated by shared polygenic risk score (PRS), and shared lifestyle and socioeconomic factors (LSFs: smoking, diet, physical activity, deprivation, income and education) for CM traits.
Methods
We included unique pairs of kinship-verified first-degree relatives [0.177-0.354] in UK Biobank (UKBB). We examined the association of trait-specific, UKBB-independent PRSs with three CM diseases: coronary artery disease (CAD), type 2 diabetes (T2DM), and atrial fibrillation (Afib), and three continuous traits: body mass index (BMI), systolic blood pressure (SBP), and lipoprotein(a) (Lp(a)). Shared high PRS was defined as both index patient and relative being in the top 20% of the disease-specific PRS distribution. Shared LSFs were defined as concordant status of smoking, healthy diet score (≥4/5), physical activity (≥150/75 min a week moderate/vigorous activity), deprivation (Townsend deprivation index, top 20%), low income (<£18,000), or a college/university degree. Mediation analysis using structural equation modeling quantified familial risk mediated by shared PRS versus shared LSFs. Models were adjusted for age, sex and the first ten principal components of genetic ancestry.
Results
Among 26,796 index patients (41.2% male, mean age 57.0 (SD 8.5) years) and 25,540 relatives, median concordance for high PRS was 44.2% (IQR 43.9%–44.8%) and 26.8% (IQR 20.0%–36.8%) for LSFs. High PRS in the index patients was significantly associated with all CM traits, including CAD (OR 3.23; 95%CI 2.83-3.69). LSFs varied more across outcomes, although socioeconomic factors showed the most frequent associations across traits. Specifically for CAD: smoking (OR 1.69; 1.46-1.95), deprivation (OR 1.36; 1.21-1.54), low income (OR 1.53; 1.36-1.72) and having a degree (OR 0.66; 0.58-0.75) reached significance, while physical activity (OR 0.88; 0.75-1.02) and healthy diet (OR 0.87; 0.74-1.02) did not. Mediation via shared high PRS was significantly greater than via shared LSFs for Lp(a) (47.1%; 95%CI [44.0-50.2] vs. 0.1% [0.0-0.2], P<0.001), CAD (17.7% [9.7-25.8] vs. 5.4% [2.1-8.6], P=0.006) and SBP (6.6% [5.1-8.1] vs. 2.0% [0.9-3.0], P<0.001). In contrast, shared LSFs mediated larger proportions than shared PRS for BMI (5.3% [4.4-6.1] vs. 0.8% [0.5-1.2], P<0.001) and T2DM (4.9% [2.7-7.2] vs. 1.1% [0.2-2.0], P=0.003). For Afib, the proportion mediated by combined LSFs did not differ significantly from that of shared PRS (P=0.194).
Conclusions
Familial clustering of CM traits is largely driven by PRS for Lp(a), SBP and CAD, while LSFs dominate for T2DM and BMI. These findings support using family history to guide PRS cascade screening of first-degree relatives of affected individuals for select CM traits.