Longitudinal biomarkers and kidney disease progression after acute kidney injury

Abstract

BACKGROUNDLongitudinal investigations of murine acute kidney injury (AKI) suggest that injury and inflammation may persist long after the initial insult. However, the evolution of these processes and their prognostic values are unknown in patients with AKI.METHODSIn a prospective cohort of 656 participants hospitalized with AKI, we measured 7 urine and 2 plasma biomarkers of kidney injury, inflammation, and tubular health at multiple time points from the diagnosis to 12 months after AKI. We used linear mixed-effect models to estimate biomarker changes over time, and we used Cox proportional hazard regressions to determine their associations with a composite outcome of chronic kidney disease (CKD) incidence and progression. We compared the gene expression kinetics of biomarkers in murine models of repair and atrophy after ischemic reperfusion injury (IRI).RESULTSAfter 4.3 years, 106 and 52 participants developed incident CKD and CKD progression, respectively. Each SD increase in the change of urine KIM-1, MCP-1, and plasma TNFR1 from baseline to 12 months was associated with 2- to 3-fold increased risk for CKD, while the increase in urine uromodulin was associated with 40% reduced risk for CKD. The trajectories of these biological processes were associated with progression to kidney atrophy in mice after IRI.CONCLUSIONSustained tissue injury and inflammation, and slower restoration of tubular health, are associated with higher risk of kidney disease progression. Further investigation into these ongoing biological processes may help researchers understand and prevent the AKI-to-CKD transition.FUNDINGNIH and NIDDK (grants U01DK082223, U01DK082185, U01DK082192, U01DK082183, R01DK098233, R01DK101507, R01DK114014, K23DK100468, R03DK111881, K01DK120783, and R01DK093771).

Keywords: Chronic kidney disease; Diagnostics; Epidemiology; Nephrology.

Figure 1. Flow chart for the inclusion of study participants in ASSESS-AKI cohort.

Figure 2. Associations between slopes of biomarkers of kidney injury, inflammation, and tubular health with the composite outcome of incident CKD and CKD progression.

(A–C) Association between biomarkers of kidney injury (A), inflammation (B), and tubular health (C) with the composite CKD outcome in 656 participants with AKI in the ASSESS-AKI cohort. Cox proportional hazard regression models were adjusted for biomarker at hospitalization, age, sex, race, Hispanic ethnicity, hypertension, diabetes, atherosclerotic disease, congestive heart failure, smoking status, baseline eGFR, albuminuria at hospitalization, urine creatinine at hospitalization, and the slope of urine creatinine from hospitalization to 12 months after discharge. Follow-up started at 12 months after hospitalization, and participants who died were censored. Median follow-up duration was 4.3 years.

Figure 3. Distribution of monthly biomarker changes and the associations between biomarker slopes using restricted cubic splines with the composite CKD outcome in 656 participants with AKI.

Distribution of biomarker slopes were visualized using kernel density plot to aid the interpretation of the confidence in hazard ratio estimates. Cox proportional hazard regression models were adjusted for biomarker at hospitalization, age, sex, race, Hispanic ethnicity, hypertension, diabetes, atherosclerotic disease, congestive heart failure, smoking status, baseline eGFR, albuminuria at hospitalization, urine creatinine at hospitalization, and the slope of urine creatinine from hospitalization to 12 months after discharge.

Figure 4. Comparison of the gene expressions of biomarkers of kidney injury, inflammation, and tubular health in mouse models of atrophy and repair after ischemic reperfusion injury.

(A) Overview of animal experiments. WT mice were subjected to 27 minutes of unilateral IRI with the contralateral kidney intact (atrophy model) or unilateral IRI with contralateral nephrectomy (repair model) followed by single-cell isolation and sequencing (scRNA-Seq and qPCR) at the indicated time points (14). (B) Expression of mRNA for biomarkers across 95,343 cells of the indicated identity (y axis) in mouse models of repair and atrophy at multiple time points after IRI (n = 2 kidneys for control; n = 2 kidneys/time point/model for days 7, 14, and 30). (C–J) qPCR analysis for Havcr1, Lcn2, Ccl2, Il18, Chi3l1, Tnfrsf1a, Tnfrsf1b, and Umod was performed on whole-kidney RNA harvested 0, 1, 7, 14, 30, and 90 days after injury; n = 10 kidneys/time point/model. *P < 0.05, **P < 0.01, ****P < 0.0001 at the indicated time point using 2-way ANOVA followed by Bonferroni post hoc test for subgroup analysis.