Slope of Kidney Function and Its Association with Longitudinal Mortality and Cardiovascular Disease among Individuals with CKD

Abstract

Background: Slopes of eGFR have been associated with increased risks of death and cardiovascular events in a U-shaped fashion. Poor outcomes in individuals with rising eGFR are potentially attributable to sarcopenia, hemodilution, and other indicators of clinical deterioration.

Methods: To investigate the association between eGFR slopes and risks of death or cardiovascular events, accounting for multiple confounders, we studied 2738 individuals with moderate to severe CKD participating in the multicenter Chronic Renal Insufficiency Cohort (CRIC) Study. We used linear, mixed-effects models to estimate slopes with up to four annual eGFR assessments, and Cox proportional hazards models to investigate the association between slopes and the risks of death and cardiovascular events.

Results: Slopes of eGFR had a bell-shaped distribution (mean [SD], -1.5 [-2] ml/min per 1.73 m² per year). Declines of eGFR that were steeper than the average decline associated with progressively increasing risks of death (hazard ratio [HR], 1.23; 95% confidence interval [95% CI], 1.09 to 1.39; for a slope 1 SD below the average) and cardiovascular events (HR, 1.19; 95% CI, 1.03 to 1.38). Rises of eGFR or declines lower than the average decline were not associated with the risk of death or cardiovascular events.

Conclusions: In a cohort of individuals with moderate to severe CKD, we observed steep declines of eGFR were associated with progressively increasing risks of death and cardiovascular events; however, we found no increased risks associated with eGFR improvement. These findings support the potential value of eGFR slopes in clinical assessment of adults with CKD.

Keywords: cardiovascular events; chronic kidney disease; epidemiology and outcomes; glomerular filtration rate; mortality risk; risk factors.

Graphical abstract

Figure 1.

Cohort definition. (A) Total Number of participants after exclusions. (B) The primary exposure was defined within the first 3 years from baseline, while the time-to outcomes started at the end of this interval.

Figure 2.

Association between slopes of eGFR and the relative risks of death and CVD events. (A) Unadjusted association between slopes of eGFR and the risk of death (model 1). (B) Multivariable-adjusted association between slopes of eGFR and the risk of death (model 8). (C) Unadjusted association between slopes of eGFR and the risk of cardiovascular events (model 1). (D) Multivariable-adjusted association between slopes of eGFR and the risk of cardiovascular events (model 8). Model 8 includes adjustments for age, sex, race, educational attainment, smoking status, history of CVD, diabetes, SBP, urinary protein, eGFR, FFM, body mass index, waist circumference, ejection fraction, serum sodium, HDL cholesterol, high-sensitivity Troponin-T, N-terminal probrain natriuretic peptide, reported use of angiotensin-converting enzyme inhibitors/angiotensin receptor blockers and diuretics, hemoglobin, serum bicarbonate, albumin, phosphate, fibroblast growth factor 23, intact parathyroid hormone, high-sensitivity C-reactive protein, urinary neutrophil gelatinase–associated lipocalin, number of hospitalizations, and the number of days spent in the hospital in the 3 years between baseline and the index date. The solid black lines represent HRs. Boundaries of the 95% CIs for the HRs are shown by the solid gray lines. Reference risk for a slope of eGFR of −1.5 ml/min per 1.73 m² per year (mean slope of eGFR for the study population) is depicted by dashed lines. Reported P values are the results of chi-squared tests, checking for the overall association between the slope of eGFR and the risk of all-cause mortality. Slopes of eGFR (CRIC Study equation) were defined using annual assessments of eGFR over a 3-year period, using linear, mixed-effects models. Analyses were performed in multiple imputed datasets. The estimated HRs for the slopes of −5.5 (2 SD below the average), −3.5 (1 SD below the average), 0.5 (1 SD above the average), and 2.0 (2 SD above the average) are presented for each model in the tables below the graphs. Ref, reference.

Figure 3.

Association between slopes of eGFR and the relative risks of death and CVD events according to significant subgroups. (A) Association between slopes of eGFR and the risk of death among subjects who presented a rate of change of (log) urinary protein over time <0.05 g/24 h per year during the period in which slopes were defined. (B) Association between slopes of eGFR and the risk of death among subjects who presented a rate of change of (log) urinary protein over time ≥0.05 g/24 h per year during the period in which slopes were defined. (C) Association between slopes of eGFR and the risk of CVD events among non-Hispanic White subjects. (D) Association between slopes of eGFR and the risk of CVD events among non-Hispanic Black and Hispanic subjects. Models were adjusted for age, sex, race, educational attainment, smoking status, history of CVD, diabetes, SBP, urinary protein, eGFR, FFM, body mass index, waist circumference, ejection fraction, serum sodium, HDL cholesterol, high-sensitivity Troponin-T, N-terminal probrain natriuretic peptide, reported use of angiotensin-converting enzyme inhibitors/angiotensin receptor blockers and diuretics, hemoglobin, serum bicarbonate, albumin, phosphate, fibroblast growth factor 23, intact parathyroid hormone, high-sensitivity C-reactive protein, urinary neutrophil gelatinase–associated lipocalin, number of hospitalizations, and the number of days spent in the hospital in the 3 years between baseline and the index date. HRs are depicted by the solid black lines. Boundaries of the 95% CIs for the HRs are shown by the solid gray lines. Reference risk for a slope of eGFR of −1.5 ml/min per 1.73 m² per year (mean slope of eGFR for the study population) is depicted by dashed lines. Reported P values are the results of chi-squared tests, checking for the overall association between the slope of eGFR and the risk of all-cause mortality. Slopes of eGFR (CRIC Study equation) were defined using annual assessments of eGFR over a 3-year period, using linear, mixed-effects models. Analyses were performed in multiple imputed datasets. The estimated HRs for the slopes of −5.5 (2 SD below the average), −3.5 (1 SD below the average), 0.5 (1 SD above the average), and 2.0 (2 SD above the average) are presented for each model in the tables below the graphs. Subgroup analyses were performed on the basis of significant interactions on the association between slopes of eGFR and the risks of death and CVD events. Ref, reference.