UCSF

Reverse translational research takes a bedside to bench approach, using sophisticated basic research to explain the biological mechanisms behind observed clinical data. Here, we used this approach to characterize the inter-individual differences in the efficacy and pharmacokinetics of a commonly prescribed drug, allopurinol. Allopurinol is the first-line treatment for chronic gout, a debilitating arthritis caused by high serum uric acid levels. Not only is the prevalence of gout growing rapidly, but high serum uric acid and gout have been shown to be important risk factors for the development of comorbidities such as renal and cardiovascular disease. Thus, allopurinol is increasingly being used in preventing these deadly diseases. However, response to allopurinol is variable, and many gout patients fail to achieve healthy serum uric acid levels, even while adherent. Our reverse translational approach begins with the largest genome-wide association study on allopurinol response to date, performed in a multi-ethnic cohort of patients taking allopurinol for the treatment of gout. Consistent with previous studies, we observed that the Q141K variant in ABCG2 (rs2231142), which encodes the efflux pump BCRP, associated with worse response to allopurinol. However, for the first time this association reached genome-wide level significance (p=8.06 x 10-11). Additionally, we identified a novel association with a variant in GREM2 (rs1934341, p=3.2 x 10-6). In vitro studies led to the identification of oxypurinol, the active metabolite of allopurinol, as an inhibitor of the uric acid transporter GLUT9, suggesting that oxypurinol may modulate uric acid disposition. These results suggest that allopurinol may have additional hypouricemic effects beyond xanthine oxidase inhibition, a mechanism previously unknown. In order to further explore the role between BCRP Q141K and allopurinol response, we performed a clinical study in which we characterized the relationship between BCRP Q141K and allopurinol pharmacokinetics and pharmacodynamics. We observed a significant difference in oxypurinol half-life in subjects harboring the Q141K variant (19.1 ± 0.5 vs 23.3 ± 0.2, p=0.02), but no differences in any other pharmacokinetic parameters. These results suggest that the association between response and BCRP Q141K may be dependent on tissue levels of the drug, rather than systemic levels. These findings have important implications in the understanding of allopurinol’s mechanism of action and clinical use, and more broadly, for the complex relationships between BCRP Q141K and the pharmacokinetics of its substrates.