Mis-matches between adiposity and glycaemic traits: A replicated genome-wide association study for metabolic disparity

Whilst increased Body Mass Index (BMI) is correlated with a range of cardio-metabolic risk factors two sub-phenotypes of obesity have been characterised that contradict this. Metabolically obese normal-weight (MNOW) individuals present with a normal BMI but have a poor metabolic profile. In contrast, metabolically healthy obese (MHO) individuals have high BMI, but do not suffer the metabolic disturbances expected. By developing a novel phenotypic outcome acting as an indicator of metabolic disparity, we explore the biological pathways underlying these sub-phenotypes using genomewide association analyses in two independent collections.

Analyses were performed in a collection from the Avon Longitudinal Study of Parents and Children (ALSPAC, mean age 15.5yrs) and in a replication cohort of adults from the UK Adult Twin Registry (TwinsUK, mean age 50.9yrs). A compound phenotype was generated by subtracting standardised BMI from standardised fasting plasma glucose to yield a continuously distributed variable where positive values were indicative of “healthier” adiposity status relative to glucose, zero was indicative of matching adiposity/glucose status and negative values were indicative of “healthier” glucose status relative to adiposity.

Genomewide analysis of this trait in ALSPAC (n=2584) gave evidence of association at the G6PC2 locus where each additional T allele at the well-known variant rs560887 was related to a -0.25(SE 0.04, p=3.9e-10) change in standardised disparity score. This was replicated in TwinsUK (n=2599)(-0.11(SE 0.04, p=0.003)) and a meta-analysis gave a joint p-value 6.1e-11. This signal was also seen when using DXA derived fat mass in place of BMI (meta-analysis p=4.4e-10) and, as anticipated, with an inverted beta coefficient when replacing glucose status with HOMA_b.

This work illustrates how derived phenotypes can be used as indicators of disparity between adiposity and glycaemic traits. Using a standardised disparity score allows a powerful analysis of metabolic features avoiding the costly procedure of defining MONW and MHO by threshold. Results are in line with the carriage of additional T alleles at G6PC2(rs560887) being associated with improved glycaemic profile. Work here did not yield signals of variants showing independent contributions to adiposity; this likely the result of genetic architecture and the causal effect that BMI has on many downstream, causal effects.