PERFORMANCE OF THE DIFFERENT PREDICTION TOOLS FOR ANCA-ASSOCIATED GLOMERULONEPHRITIS: A RETROSPECTIVE COHORT STUDY

Introduction:

In the last decades several prediction tools have been proposed to predict outcomes of ANCA-associated glomerulonephritis (ANCA-GN). The recently presented updated ANCA Kidney Risk Score (AKRiS) demonstrated capability to optimize prognostication. The aim of our study was to compare the diagnostic performance of AKRiS with the previous version of the same tool (ANCA Renal Risk Score, ARRS) and the histopathologic classification by Berden et al (2010) in the real-world setting.

Methods:

In our single center retrospective cohort study, we enrolled adult (≥18 years) patients with biopsy-proven ANCA-GN who were followed at Tareev Clinic of Internal Diseases between 2010 and 2024. Histopathologic classes, ARRS and AKRiS risk groups were established in accordance with the definitions provided in the original full-text publications. The primary end-point was end-stage kidney disease (ESKD) that was defined as the need for kidney replacement therapy for ≥12 weeks. Time to ESKD was measured from the date of kidney biopsy. Kidney function was measured in serum creatinine and eGFR using the 2021 CKD-EPI equation. Cox proportional hazards, Kaplan–Meier curves, Harrell’s C-statistic, and receiver operating characteristics were used to compare the characteristics of the prognostication tools. The study was approved by the Sechenov University local ethics committee (#10-22, 19.05.2022).

Results:

We enrolled 84 patients, 30 (36%) males and 54 (64%) females. Median age at AAV onset was 49.5 (35; 57) years. Seventy-four (88%) patients were ANCA-positive. Seventeen (20%) patients were diagnosed with renal-limited AAV. Median Birmingham vasculitis activity score (BVAS v.3) at biopsy was 14.5 (12; 18). Median follow-up after kidney biopsy was 16.5 (8.0; 48.7) months. According to the Berden et al classification 15 (18%) patients had focal, 17 (20%) had crescentic, 34 (40%) had mixed and 18 (21%) had sclerotic class. According to the ARRS 20 (24%) patients were classified as low-risk, 36 (43%) as moderate-risk, and 28 (33.3%) as high-risk. They were retrospectively re-evaluated using AKRiS as follows: 30 (36%) - low-risk, 35 (42%) – moderate-risk, 12 (14%) – high-risk, and 7 (8%) – very high risk. Nineteen (23%) patients developed end-stage kidney disease. All three prognostication models showed statistically significant differences in kidney survival (Fig. 1), however histopathologic classification demonstrated a significantly lower predictive ability (Harrel’s C=0.651, 95% CI 0.514-0.788) compared to ARRS (C=0.839, 95% CI 0.785-0.894, p=0.012) and AKRiS (C=0.861, 95% CI 0.787-0.936, p=0.019).

Conclusions:

In our single center cohort, all three prognostication tools showed their ability to predict kidney outcomes in patients with ANCA-GN, therefore they all can be used to personalize treatment. However, ARRS and its updated version AKRiS demonstrated a significantly better prognostication ability compared to the histopathologic classification by Berden et al.

I have no potential conflict of interest to disclose.

I did not use generative AI and AI-assisted technologies in the writing process.