Renal Survival in Children with Glomerulonephritis with Crescents: A Pediatric Nephrology Research Consortium Cohort Study

Abstract

There is no evidence-based definition for diagnosing crescentic glomerulonephritis. The prognostic implications of crescentic lesions on kidney biopsy have not been quantified. Our objective was to determine risk factors for end-stage kidney disease (ESKD) in patients with glomerulonephritis and crescents on kidney biopsy. A query of the Pediatric Nephrology Research Consortium's Pediatric Glomerulonephritis with Crescents registry identified 305 patients from 15 centers. A retrospective cohort study was performed with ESKD as the primary outcome. Median age at biopsy was 11 years (range 1-21). The percentage of crescents was 3-100% (median 20%). Etiologies included IgA nephropathy (23%), lupus (21%), IgA vasculitis (19%) and ANCA-associated GN (13%), post-infectious GN (5%), and anti-glomerular basement membrane disease (3%). The prevalence of ESKD was 12% at one year and 16% at last follow-up (median = 3 years, range 1-11). Median time to ESKD was 100 days. Risk factors for ESKD included %crescents, presence of fibrous crescents, estimated GFR, and hypertension at biopsy. For each 1% increase in %crescents, there was a 3% decrease in log odds of 1-year renal survival (p = 0.003) and a 2% decrease in log odds of renal survival at last follow-up (p < 0.001). These findings provide an evidence base for enrollment criteria for crescentic glomerulonephritis in future clinical trials.

Keywords: cellular crescents; fibrous crescents; glomerulonephritis; immunosuppression; targeted biologic therapies.

Figure 1

Eligibility and inclusion criteria for the registry and this study.

Figure 2

Percent crescents in children with glomerulonephritis and crescents, stratified by etiology. Median percentage of glomerular crescents are shown for all 305 children in the study. Boxes and whiskers represent IQR with 5th and 95th percentiles. Etiologies: anti-glomerular basement membrane disease (anti-GBM), membranous GN (MGN), anti-neutrophil cytoplasmic antibody associated nephritis (ANCA), immune complex GN not otherwise specified (ICGN-NOS), dense deposit disease (DDD), C3 glomerulopathy (C3GN), post-infectious GN (PIGN), lupus nephritis (LN), Ig-predominant membranoproliferative GN (MPGN-Ig), IgA nephropathy (IgAN), IgA renal vasculitis (IgAV).

Figure 3

Predicting 1-year renal survival in children with glomerulonephritis with crescents. Receiver operating characteristic curve describing association of percentage of glomerular crescents at biopsy (black line, area under the curve = 0.82) and composite model (gray line, area under the curve = 0.93) with ESKD at one year post biopsy. The AUC for the composite model is significantly higher (p-value = 0.001). Under the composite model, a predicted probability of 0.14 is the optimal cutoff. When only percent crescents are used to predict end-stage kidney disease (ESKD) at 1-year, 43% is the optimal cutoff.

Figure 4

Predicting long-term renal survival in children with glomerulonephritis with crescents. Kaplan Meier survival analysis demonstrates that children with glomerulonephritis and ≥43% crescents (n = 72) have decreased survival than children with <43% crescents (n = 233), at a median of 3 years follow up (hazard ratio of 6.77, 95% CI: 3.76–12.2). The numbers in each subset remaining at risk after each year of follow-up are indicated.