Estimating Time to ESRD in Children With CKD

Abstract

Rationale & objective: The KDIGO (Kidney Disease: Improving Global Outcomes) guideline for chronic kidney disease (CKD) presented an international classification system that ranks patients' risk for CKD progression. Few data for children informed guideline development.

Study design: Observational cohort study.

Settings & participants: Children aged 1 to 18 years enrolled in the North American Chronic Kidney Disease in Children (CKiD) cohort study and the European Effect of Strict Blood Pressure Control and ACE Inhibition on the Progression of CRF in Pediatric Patients (ESCAPE) trial.

Predictor: Level of estimated glomerular filtration rate (eGFR) and proteinuria (urine protein-creatinine ratio [UPCR]) at study entry.

Outcome: A composite event of renal replacement therapy, 50% reduction in eGFR, or eGFR<15mL/min/1.73m². eGFR was estimated using the CKiD-derived "bedside" equation.

Analytical approach: Accelerated failure time models of the composite outcome using a conventional generalized gamma distribution. Likelihood ratio statistics of nested models were used to amalgamate levels of similar risk.

Results: Among 1,232 children, median age was 12 (IQR, 8-15) years, median eGFR was 47 (IQR, 33-62) mL/min/1.73m², 60% were males, and 13% had UPCRs>2.0mg/mg at study entry. 6 ordered stages with varying combinations of eGFR categories (60-89, 45-59, 30-44, and 15-29mL/min/1.73m²) and UPCR categories (<0.5, 0.5-2.0, and >2.0mg/mg) described the risk continuum. Median times to event ranged from longer than 10 years for eGFRs of 45 to 90mL/min/1.73m² and UPCRs<0.5mg/mg to 0.8 years for eGFRs of 15 to 30mL/min/1.73m² and UPCRs>2mg/mg. Children with glomerular disease were estimated to have a 43% shorter time to event than children with nonglomerular disease. Cross-validation demonstrated risk patterns that were consistent across the 10 subsample validation models.

Limitations: Observational study, used cross-validation rather than external validation.

Conclusions: CKD staged by level of eGFR and proteinuria characterizes the timeline of progression and can guide management strategies in children.

Keywords: Pediatric; children; chronic kidney disease (CKD); disease progression; disease staging; end-stage renal disease (ESRD); estimated glomerular filtration rate (eGFR); proteinuria; risk pattern; urinary protein-creatinine ratio (UPCR).

Figure 1

Distribution of persons, person-years, CKD diagnosis, and events by categories of baseline glomerular filtration rate (GFR) and urine protein-creatinine ratio (UPCR), N=1169. Each cell in the table above describes: the number of subjects, overall and distributed by cohort affiliation; prevalence of glomerular CKD; number of events; number of person-years; and empirical incidence rate (per 100 person-years) of the composite outcome event (50% GFR decline, renal replacement therapy, or GFR < 15 mL/min/1.73m²) with 95% confidence interval. Cell coloring defines the final six GFR/proteinuria risk stages ordered from best prognosis to worst prognosis as follows: A (dark green), B (light green), C (gold), D (tan), E (salmon), F (red). Incidence rates (IR) expressed as events per 100 person years. Cells with <15 subjects do not have an incidence rate calculated (n/a).

Figure 2

Amalgamated number of subjects, events, person-years and empirical incidence rates in the final six GFR/proteinuria risk stages, A through F, overall, by CKD diagnosis and cohort affiliation. This table shows the preservation of risk order for the final risk stages within diagnosis and cohort substrata. Areas under the Receiver-Operator Curve (AUC) quantify the ability of the six risk stages to discriminate risk of progression in the diagnosis and cohort substrata. Abbreviations: IR, incidence rate expressed in events per 100 person years (PY). ^a – The area under the ROC curve is calculated at 4 years of follow up.

Figure 3

Parametric and non-parametric survival curves for the six risk stages (A-F) modeling time from study enrollment (baseline) to composite clinical event (50% GFR decline, renal replacement therapy, or GFR less than 15 mL/min/1.73m²), stratified by CKD diagnosis (N=868 non-glomerular and N=285 glomerular CKD children. Parametric survival curves are generated from an accelerated failure time model using a conventional generalized gamma distribution with seven beta indicator parameters: six risk stages (A-F) and glomerular CKD.

Figure 4

Expected times – 10^th, 25^th and 50^th (percentile) - from GFR/proteinuria risk characterization (study baseline) to clinical composite event (50% GFR decline, renal replacement therapy, or GFR less than 15 mL/min/1.73m²), for six GFR/proteinuria risk stages, by CKD diagnosis. Estimated event times are generated from an accelerated failure time model using a conventional generalized gamma distribution with seven beta indicator parameters: six risk stages (A-F) and glomerular CKD. Time to event is from baseline (time at which risk level was defined). Event is defined as RRT, GFR<15mL/min/173m2 or 50% GFR decline.