#4404 kidney function trajectory before and after arteriovenous access creation in patients with predialysis ckd

Abstract Background and Aims Timely arteriovenous (AV) access creation in view of starting hemodialysis is challenged by non-linear kidney function decline and the prospect of competing mortality. In addition, some studies have shown slower CKD progression following AV access creation in patients not on dialysis. While pathophysiological mechanisms, such as ischemic preconditioning and improved kidney perfusion, have been put forward to explain the apparent influence of AV access creation on estimated glomerular filtration rate (eGFR) trajectory, it cannot be ruled out that this finding resulted from an artefact induced by the use of models assuming linear (eGFR) decline before and after AV access creation. Our aim was to describe the kinetics of eGFR decline around the period of AV access creation and to identify different trajectory profiles using models relaxing this hypothesis. Method From 2013 through 2016, the CKD-REIN cohort included 3033 patients with CKD stages 3 to 5 from 40 nationally representative outpatient nephrology clinics in France. Participants were followed for 5 years or until initiation of kidney replacement therapy (KRT), death, or loss to follow-up, whichever came first. This study focused on patients who underwent their first AV access creation during follow-up. Linear mixed models with restricted cubic spline functions (two internal knots, one at AV access creation date, the other, one year before) were used to model a potential non-linear eGFR trajectory over time, based on routine labs. Random effects for the intercept and the spline function components allowed us to deal with individual variations in eGFR trajectory. Instantaneous rates of eGFR decline around AV access creation were then extracted. In addition, we performed latent class mixed models (LCMM) to identify distinct eGFR trajectories. Results During a median follow-up of 5.0 years (interquartile range [IQR], 4.6 to 5.2), 415 (14% of the total population) patients underwent a first AV access creation (32% women, 51% with diabetes). The median age at AV access creation was 69 years (IQR, 61 to 76), and the median eGFR, 13 ml/min/1.73 m² (IQR, 11 to 16). The median numbers of eGFR measurements before and after creation were 12 (IQR, 8 to 19) and 3 (IQR, 2 to 6) respectively. The average eGFR decline in the year before and after AV access creation, assuming constant slopes in each period, was 5.2 ml/min/1.73m² (95% confidence interval [CI], 4.8 to 5.5) and 3.4 ml/min/1.73 m² (95% CI, 3.1 to 3.7), respectively, with a mean difference of −1.8 ml/min/1.73m2 (95% CI, −1.4 to −2.1). Analysis of instantaneous rates showed that the slowdown of eGFR decline began 8.3 months on average (95% CI, 7.8 to 8.6) before AV access creation. The LCMM identified two profiles of eGFR trajectories which mostly differed in the rate of eGFR decline (Figure). In both trajectories, the mean time to the slowdown of eGFR decline preceded time of AV access creation, by 9.1 and 7.2 months in the fastest and slowest eGFR decline trajectories, respectively. Conclusion In nondialysis patients, slowdown of kidney function decline appears to occur several months before AV access creation. Our findings do not support a causal biological effect of AV access creation on CKD progression, but favor alternative hypotheses including optimal management before AV access creation, greater inaccuracy in eGFR estimation in advanced CKD due to muscle mass loss, or simply regression to the mean.