An evaluation of the 2010 histopathological classification of anti-neutrophil cytoplasmic antibody (ANCA)-associated glomerulonephritis: a Bayesian network meta-analysis

Background

In 2010, a new histopathological classification of ANCA-associated glomerulonephritis (ANCA-GN) based on four categories (focal, crescentic, mixed, and sclerotic) was proposed to predict renal outcome. However, this classification was the source of much debate in different populations. We aimed to evaluate this classification scheme in terms of renal survival using Bayesian network meta-analysis.

Methods

We searched Pubmed, Embase, and Medline for articles published between 1 October 2010 and 9 August 2017 that assessed the predictive value of this classification. We pooled hazard ratios (HRs) for end-stage kidney disease (ESRD) among the four categories using a Bayesian random-effects model. Clinical factors that could potentially influence renal outcome among the different trials were explored by meta-regression.

Results

Sixteen case-controlled studies, including a total of 1945 patients with 1695 endpoint events, were included. Compared with sclerotic, the HRs for ESRD were 0.45 [95% confidence interval (CI) 0.26–0.79] for crescentic, 0.34 (0.22–0.51) for mixed, and 0.24 (0.12–0.51) for focal. Pooled results showed no obvious difference between the crescentic and the mixed sub-groups (HR 1.35; 95% CI 0.90–2.0). Baseline eGFR (P?=?0.002) and the ANCA serology (P?=?0.029) were associated with renal survival.

Conclusions

The 2010 ANCA-GN classification and the extent of ANCA serology and baseline eGFR were shown to be significant predictors of renal outcome, although there was no significant prognostic difference between crescentic and mixed. The ANCA-GN classification scheme should, therefore, be optimized by combining with other established parameters, such as tubular atrophy and the proportion of normal glomeruli.