FC 116WEIGHT CHANGE AS A PREDICTOR OF MORTALITY IN CHRONIC HEMODIALYSIS PATIENTS

Abstract

Abstract Background and Aims Serial decrease in body weight can be a prognostic marker reflecting Protein Energy Wasting (PEW) condition in haemodialysis patients. However, the impact of longitudinal body weight changes on mortality have rarely been evaluated. Therefore, we applied landmark analysis and time-scale mixed regression analyses to clarify the impact of long-term changes in body weight on all-cause mortality. Method This study is a post-hoc analysis of the participants in the Olmesartan Clinical Trial in Okinawan Patients Under OKIDS (OCTOPUS) conducted between June 2006 and June 2011. We additionally followed up with the participants until 31 July, 2018. During the OCTOPUS trial, the body weight of surviving participants was repeatedly measured every 6 months. The primary aim of the study was to clarify the association of serial change in body weights and all-cause mortality. The participants were categorized into 4 groups based on weight change slope during the first 2 years (bottom to highest group correspond with the first quartile to the fourth quartile group). We conducted a landmark analysis to evaluate the influence of weight change slope on the subsequent survival with Cox-proportional hazard regression. We also conducted time-scale multilevel regression analyses to draw marginal plots of body weight with a backward time scale to describe the difference in trajectories of body weight between deceased and surviving patients. Results Out of 461 patients in the cohort, 404 participants were analysed in the landmark analysis. Of the participants, 168 (41.6%) were deceased, and 236 participants (58.4%) survived. Although baseline body weights and indices about body components such as a creatinine index were similar among the 4 groups, multivariate Cox-proportional hazard regression showed that the higher rate of body weight loss was associated with higher mortality. Hazard ratios were 2.02 (1.28–3.20), 1.77 (1.10–2.85), 1.00 (Reference), and 1.11 (0.67–1.83) for first quartile, second quartile, third quartile (Reference), and fourth quartile group, respectively (P for trend was <0.001). The association was also described using Kaplan-Meier plots, with the P-value of general Log-rank test being 0.015 (Fig 1a). Marginal plots drawn with a multivariate adjusted time-scale multilevel regression model, which included all 461 participants, with a backward time-scale, showed that body weight constantly decreased in deceased patients, while body weight was longitudinally stable in surviving patients (Fig 1b). These trends of weight change have been soundly observed in sensitivity analyses. Conclusion In chronic haemodialysis patients, the higher rate of body weight loss, such as a decrease in body weight by 1 kg (1.7%) per 6 months, was significantly associated with higher mortality. Persistent loss of body weight is a significant predictor of death among chronic dialysis patients.