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

概要

Introduction
Recently, it was reported that lower serum uric acid (UA) in hemodialysis patients correlated with unfavourable outcomes (Beberashvili, I. et al.,2015). In general, hyperuricemia aggravates cardiovascular disease (CVD) and chronic kidney disease (CKD) progression. In chronic kidney disease, higher serum UA levels accelerate the deterioration of kidney function (Latif, W. et al., 20119; Kanby, M. et al., 2011; Chonchol, M. et al., 2007). A recent randomized trial of allopurinol for CKD and CVD events reported reduced progression of renal disease and reduced CV risk in CKD patients treated with allopurinol (Goicoechea M, et al, 2015). Possibly because of the extra UA lowering effect of xanthine oxidoreductase (XOR) inhibitors (XORi). This study investigated the relationship of baseline serum UA with outcomes, controlling for confounding factors, and estimated whether XOR inhibition improved the all-cause mortality. Xanthine oxidoreductase (XOR) catalyzes the conversion of hypoxanthine to xanthine and xanthine to uric acid with concomitant reduction of either NAD+ or O2. XOR was discovered approximately 100 years ago and xanthine dehydrogenase (XDH) and xanthine oxidase (XO) conversion were discovered approximately 30 years ago (Bray RC et al., 1988). In vivo, XDH preferentially reacts with NAD+, whereas XO cannot, but in the pathological condition, produces the superoxide anion (O2−) and hydrogen peroxide (H2O2) (McCord JM et al., 1985; Saugstad OD et al.,1988; Hille R et al., 1995; Bender D et al., 2018; Nishino T et al., 2008). It is well established that O2− and H2O2 cause tissue damage (Biscaglia S et al., 2016). XO expression is upregulated by ATP degradation caused by certain pathological conditions such as ischemia, hypoxia, and oxidative stress, triggered by cellular stress. Under these pathological conditions, ATP is metabolized to hypoxanthine, xanthine, and uric acid via reactions catalyzed by XO (Johnson RJ et al., 2013). XO produces reactive oxygen species (ROS) such as superoxides, hydroxyl radicals, hydrogen peroxide, and peroxynitrite (Kushiyama A et al., 2016).

Methods
Study Population: This study included 2429 hemodialysis patients who were outpatients treated in 30 hemodialysis clinics in Yokohama Japan, between April 1, 2013 and March 31, 2016

Outcomes: The primary outcome of this study was all-cause mortality. The secondary outcome was CVD event (ICD-10 codes I00.0-I99.0) from the World Health Organization ICD-10.

Statistical Analysis: We fitted the Cox proportional hazard model for the mortality and CVD outcomes. Covariates were demographic and comorbidities data along with baseline serological markers.Marginal structural model : We created a marginal structural model (MSM) to estimate the treatment effect of XORi on mortality and CVD events, reasoning that prescribing decisions are influenced by serological parameters from the previous visit and past prescription patterns (Robins, J.M. et al., 2000 ). To adjust for the confounding effect of prescribing XORi, we calculated the inverse probability of treatment weighted (IPTW) estimates for XOR inhibition treatment at each monthly visit, 36 times for each patient.

Results
Cox Hazard Analysis: We performed Cox hazard analysis to estimate the correlation between baseline characteristics and mortality and CVD events. 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.

Marginal Structural Model Analysis: We performed a marginal structural model (MSM) analysis for this longitudinal repeated measure analysis to estimate the treatment effect of XORi on hemodialysis patients, according to the method of Robins and Hernan. The treatment effect of XOR inhibitors, estimated by MSM with stabilized weight for all-cause mortality, was associated with an HR of 0.24 (95% CI 0.15 - 0.38). The treatment effect of XORi on CVD events was associated with an HR of 0.92 (95% CI 0.76–1.12) .

Discussion
We estimated an HR of 0.23 for all-cause mortality in patients receiving XOR inhibitor in this 3-year follow up study using observational MSM. MSM was used in the study to analyse repeated measured data and each visit and each patient was adjusted by using inverse probability of treatment weighted (IPTW) estimates with 2429 cases and 36 visits. To estimate the effect of XOR inhibitor in hemodialysis, serum uric acid level alone should not be considered suitable as a marker of hyperuricemia. Ives et al. reported that XOR elevate reactive oxygen species (ROS) and not by UA. Macrophage secretion of interleukin (IL)-1 beta is rapidly mediated by XOR activation, while XOR inhibition impairs IL-1 beta secretion (Ives A et al., 2015). Yisireyili et al. reported that febuxostat decreased ROS production (H2O2 and 8-OHdG and lipid peroxidation) and visceral adipose tissue inflammation in stress-induced hyperuricemic mice (Yisireyili M. et al., 2017). This MSM analysis also indicated that XOR inhibition reduced new onset of uncontrolled hypertensive diseases, suggesting that XOR inhibition can act directly on the systemic vascular wall. The recommendation from our study is that a higher dose of XORi should be used in the higher serum UA level group and also in lower UA level groups with high oxidative stress due to malnutrition, severe lower leg arteriosclerosis, or aspiration pneumonia. Improved nutritional status leads to a more favorable outcome in the lower serum UA group, and we recommended XORi administration even when the UA is approximately 6.0 mg/dL. In hemodialysis patients, it has been recommended to lower UA to the same level as in CKD patients, and that a further XOR inhibitory effect has an anti-inflammatory effect rather than an effect resulting from UA lowering. This study did not estimate the effect of UA lowering, but rather evaluated the effects of XOR inhibition on mortality due to CVD events. In this repeated measured analysis, XOR inhibition was estimated to improve all-cause mortality and CVD events in end stage renal disease. The true therapeutic goal of treatment with XORi is reduction of underlying oxidative stress rather than control of UA.