Lingfang Xu

Objective: This study investigated that serum metabolomics, before urate‐lowering therapy (ULT) initiation, could serve as a biomarker for responsiveness to colchicine prophylaxis in patients with gout commencing treat‐to‐target ULT. Methods: We studied a multicenter prospective cohort (n = 409) initiating treat‐to‐target ULT plus colchicine prophylaxis. Untargeted metabolomics were measured at enrollment. We defined putative colchicine response or colchicine nonresponse as no flare or one or more flares, respectively, during six months after first colchicine dose. Cox regression and machine learning models defined the metabolomic signatures associated with reported flare activity on colchicine prophylaxis in the cohort. Results: In our cohort, 36.9% participants experienced at least one flare. Many inflammation‐related metabolites were differentially abundant in the recurrent flare group and correlated with recurrent flare risk. A significant association was found between a four‐metabolite risk score and gout flare activity (adjusted hazard ratio 3.21, 95% confidence interval [CI] 2.24–4.59), independent of the adjusted clinical factors. The machine learning model, which incorporated the top‐ranked metabolites selected by multivariable methods with unbiased variable selection along with traditional clinical factors, reached an area under the receiver operating characteristic of 0.771 (95% CI 0.713–0.828) and 0.724 (95% CI 0.611–0.837) for predicting flare risk in the discovery and validation cohort, respectively. Conclusion: Metabolomic profile–defined inflammation status was associated with gout flare outcomes in ULT initiators receiving colchicine prophylaxis. Identification of these novel metabolomic predictive signatures before ULT could potentially help optimize clinical decision‐making for flare prophylaxis colchicine dosing and duration in the first six months after initiation of treat‐to‐target ULT in gout. image