Evaluation of the Effectiveness of Xanthine Oxidoreductase Inhibitors on Haemodialysis Patients using a Marginal Structural Model

Takeo Ishii,Masataka Taguri,Kouichi Tamura,Kunio Oyama

doi:10.1038/s41598-017-13970-4

Abstract

A lower serum uric acid (UA) level has been associated with a higher mortality rate in haemodialysis patients. We investigated the long-term confounding factors of UA and mortality, and fitted a marginal structural model (MSM) based on the causal effect of xanthine oxidoreductase inhibitors (XORi). In total, 2429 patients on regular dialysis from April 2013 to March 2016 were included, and divided into quintiles by serum UA with Kaplan Meier (KM) curves and log rank analysis. Baseline characteristics were evaluated for relationships with all-cause mortality and cardiovascular disease (CVD) using the Cox hazard model. The MSM was used to control for time-dependent confounders of the XORi treatment effect. KM curves indicated that patients in the highest UA quintile had better outcomes than those in the lowest UA quintile. UA was not correlated with all-cause mortality or CVD events in the Cox model; however, the hazard ratio (HR) for mortality was 0.96 for the baseline administration of XORi. The MSM analysis for the effect of XORi treatment on all-cause mortality revealed a HR of 0.24 (95% confidence interval: 0.15-0.38) in all cohorts. These results suggest that XORi improved all-cause mortality in end-stage renal disease, irrespective of the serum UA level.

Highlights

Several recent studies have reported that a lower serum uric acid (UA) in haemodialysis CKD patients was associated with unfavourable outcomes[9,10,11] than in nondialysis CKD patients
The present study investigated the association of baseline UA, controlling for the confounding factors mortality and cardiovascular disease (CVD) events in patients treated in our haemodialysis centres over a period of three years
Chen et al.[18] reported that patients with gout treated with UA-lowering agents for over 2 years had a lower risk of all-cause mortality (HR 0.39 for allopurinol, 0.24 for benzbromarone) in a matched cohort in Taiwan

Summary

Introduction

Several recent studies have reported that a lower serum UA in haemodialysis CKD patients was associated with unfavourable outcomes[9,10,11] than in nondialysis CKD patients. The reason for these paradoxical findings, i.e. that a lower serum UA predicted unfavourable outcomes and a higher UA may even be favourable in haemodialysis patients, is unclear. Investigations are needed to explore the relationships between UA and outcomes, and the effect of UA lowering treatments on mortality in haemodialysis patients. Hosoya et al.[14] reported the renal protective effect of topiroxostat with decreases in urine albumin excretion in patients with CKD, possibly due to the activity of XOR inhibitors (XORi) rather than the UA lowering effect. The treatment effects of allopurinol and febuxostat on mortality and CVD events were estimated using MSM analysis[15,16]

Methods

Results

Conclusion