Deep genome-wide profiling has increasingly become a standard measurement in established cardiovascular population cohorts and in emerging precision medicine biobanks. In this issue of JAMA Cardiology, Lyall et al1 harness genetic and clinical information from the UK Biobank to demonstrate a causal association between elevated body mass index (BMI) and increased risks for hypertension, type 2 diabetes, and coronary heart disease (CHD). Mendelian randomization (MR) analysis requires triangulation of 3 associations: gene variants and BMI, BMI and cardiometabolic disease, and BMI gene variants and cardiometabolic disease. The MR approach takes advantage of natural randomization at birth to germline genotypes that are robustly associated with BMI and are selected as BMI proxies (“instrumental variables”). Because the instrumental variable is an unconfounded marker of BMI over the lifecourse, MR protects against confounding by reverse causality. For the instrumental variable, a genetic risk score was used of 97 genetic variants reported in the largest genome-wide association study for BMI.2 Using MR analysis, the authors report significant “causal” associations of BMI with hypertension (odds ratio, 1.64; 95% CI, 1.48-1.83), type 2 diabetes (odds ratio, 2.53; 95% CI, 2.04-3.13), and CHD (odds ratio, 1.35; 95% CI, 1.09-1.69) but not stroke, even after use of a conservative (MR-Egger) approach.
Prior MR studies have supported (eg, for low-density lipoprotein cholesterol or lipoprotein[a]) or refuted (eg, for C-reactive protein or high-density lipoprotein cholesterol) evidence of risk factor causality with CHD. Strengths and limitations of the MR approach have been well described.3 A unique feature and strength of the current analysis is its conduct within a single mega-cohort, instead of meta-analysis of multiple different cohorts. Among the study limitations was the conduct of a cross-sectional rather than prospective association analysis of BMI with cardiovascular disease events.
In well-established cohorts, such as Framingham, BMI is clearly associated with both traditional risk factors and with CHD, but the association of BMI with CHD is nonsignificant after adjustment for other major modifiable risk factors. Of note, multivariable office-based cardiovascular disease risk prediction functions that incorporate BMI instead of lipid measures perform nearly as well in risk prediction.4 From a preventive cardiology perspective, the MR study of Lyall et al1 refocuses attention on, and strengthens the body of literature for, a “causal” connection of BMI with increasing blood pressure, diabetes, and CHD. Nevertheless, further research is needed to expand our understanding of mechanism from molecular genetic and/or interventional clinical studies. We invite reports of clinically relevant cardiovascular disease findings using well-designed genomic studies in large, well-phenotyped biobank cohorts as well as consortia of meta-analyzed cohorts, and we encourage studies that further our understanding of both causality and mechanisms of cardiovascular disease and health.
Corresponding Author: Christopher J. O’Donnell, MD, MPH, Boston Veterans Affairs Healthcare System, 1400 VFW Parkway, Boston, MA 02132 (email@example.com).
Conflict of Interest Disclosures: The author has completed and submitted the ICMJE Form for Disclosure of Potential Conflicts of Interest and none were reported.
O’Donnell CJ. Harnessing Genomic Biobanks to Understand Obesity in Cardiometabolic DiseaseProspects and Pitfalls. JAMA Cardiol. Published online July 05, 2017. doi:10.1001/jamacardio.2016.5805