Abstract
Background and Aims
ANCA-associated vasculitis is the most common cause of rapidly progressive glomerulonephritis (GN) and can lead to ESRD or death, even after aggressive immunosuppressive therapies (Jones et al, N Engl J Med, 363(2010), 211-220). The histological assessment of renal tissue can help in predicting the outcome and a 4-tiered classification based on glomerular lesions has been proposed to stratify these cases (Berden et al, J Am Soc Nephrol, 21(2010), 1628-1636). Subsequent studies, however, failed to confirm its predictive value (Tanna et al, Nephrol Dial Transplant, 30(2015), 1185-1192), stressing the need of new prognostic markers of the disease. The present study investigates whether additional histological features can improve the performance of the current classification in predicting the outcome of these patients.
Method
A retrospective series of biopsy-proven paucimmune crescentic GN has been collected between January 2012 and June 2018 from two Italian centers (San Gerardo Hospital, Monza and Fondazione Policlinico Universitario A. Gemelli IRCCS, Roma). After the exclusion of patients with negative/unavailable serum ANCA testing, lacking clinical data, less than 12 months of follow-up, less than 5 glomeruli per biopsy and coexisting glomerular diseases, a final cohort of 52 cases was selected. Demographic and clinical data at the time of the diagnosis are reported in Figure, panel A. Renal biopsies have been reviewed by two experts in renal pathology and scored for multiple glomerular, tubulointerstitial and vascular lesions, as well as classified following the 4-tiered schema. The outcome of interest was time to need for RRT or death, whatever occurred first. Cox proportional hazards regression models were constructed with time to composite event, loss to follow-up or censoring (06/30/2019). Time at risk started at the date of renal biopsy. Histologic predictors were collapsed into binary variables (0-1 = low; 2-3 = high) and tested in univariate models, with the association expressed as hazard ratios (HRs) and 95% confidence intervals (CI); those showing a significant association with the outcome ( Figure, panel B) were included in a multivariable model together with a variable representing the Berden class. The prognostic performance of models including only the Berden class or the Berden class plus additional predictors was assessed using Harrell’s c-statistic.
Results
During a follow-up of 1,828 person-months, 13 composite events developed (8 deaths, 5 RRT), corresponding to an incidence rate of 7.1 (95% CI 4.2, 12.2) per 1,000 person-months. Of the tested predictors, Bowman’s capsule rupture (BCR, Figure, panel C) was significantly associated with the outcome (p = 0.023). Compared with a model including only Berden class (c = 0.67), the addition of this parameter significantly improved the prognostic performance (c = 0.76). In the multivariable model including Berden class, BCR remained significantly associated with the outcome (HR 3.61, 95% CI 1.15, 11.34; p = 0.028).
Conclusion
The present study demonstrated an improved performance of Berden classification in predicting patients’ outcome after the implementation of BCR. This can allow a better stratification of these cases, leading to tailored therapeutic approaches. Further investigations on a larger prospective cohort are required to confirm these results.
Corticosteroid-responsive nephrotic syndrome in children with myelodysplastic syndromes