Genomic risk prediction of obesity and related disorders: body mass index vs. waist-to-hip ratio (Personalized/Predictive Medicine and Pharmacogenomics)

Introduction: Waist-to-hip ratio (WHR) has been suggested to be a better predictor of obesity and related disorders than body mass index (BMI). However, it is unclear which of the two are most appropriate for genomic risk stratification. The aim of the study was to test whether predicted genomic values (PGV) for WHR are better than those of BMI in classifying outcomes for obesity and related disorders within a Croatian (N=2,159) and into a UK (N=805) population sample. Materials and Methods: PGV were estimated in the genomic best linear unbiased prediction (GBLUP) and Bayes C framework. The discriminative power of BMI and WHR PGV in classifying outcomes for general (BMI ≥30 kg/m2) and abdominal (WHR >1.0 in men and >0.85 in women) obesity and related disorders (chronic obstructive pulmonary disease, hypertension, peripheral vascular disease and metabolic syndrome) was assessed by the area under the receiver operating characteristic curves (AUC) and bootstrap-derived confidence limits. Results: Performance of GBLUP prediction was similar to that of Bayes C in both populations; suggesting that the genetic architecture of BMI and WHR approximates the infinitesimal model. BMI classified genomic risk of obesity and related disorders as well as or better than WHR. All AUC reported in this study ranged from 0.51 to 0.81; indicating low to moderate discriminatory value. Conclusions: Inclusion of PGV in combination with the traditional risk factors (age, age2 and sex) in most cases augmented the AUC; indicating that genomic information can be used to supplement traditional risk factors in prediction models.