BackgroundThe ABCG2 Q141K (rs2231142) and rs10011796 variants associate with hyperuricaemia (HU). The effect size of ABCG2 rs2231142 on urate is ∼60% that of SLC2A9, yet the effect size on gout is greater. We tested the hypothesis that ABCG2 plays a role in the progression from HU to gout by testing for association of ABCG2 rs2231142 and rs10011796 with gout using HU controls.MethodsWe analysed 1,699 European gout cases and 14,350 normourciemic (NU) and HU controls, and 912 New Zealand (NZ) Polynesian (divided into Eastern and Western Polynesian) gout cases and 696 controls. Association testing was performed using logistic and linear regression with multivariate adjusting for confounding variables.ResultsIn Europeans and Polynesians, the ABCG2 141K (T) allele was associated with gout using HU controls (OR=1.85, P=3.8E-21 and ORmeta =1.85, P=1.3E-03, respectively). There was evidence for an effect of 141K in determining HU in European (OR=1.56, P=1.7E-18) but not in Polynesian (ORmeta=1.49, P=0.057). For SLC2A9 rs11942223, the T allele associated with gout in the presence of HU in European (OR=1.37, P=4.7E-06), however significantly weaker than ABCG2 rs2231142 141K (PHet=0.0023). In Western Polynesian and European, there was epistatic interaction between ABCG2 rs2231142 and the genetically-independent rs10011796. Combining the presence of the 141K allele with the rs10011796 CC-genotype increased gout risk, in the presence of HU, 21.5-fold in Western Polynesian (P=0.009) and 2.6-fold in European (P=9.9E-06). The 141K allele positively associated with flare frequency in Polynesian (Pmeta=2.5E-03).ConclusionThese data are consistent with a role for ABCG2 141K in gout in the presence of established HU.