Performance of the Anca Renal and Kidney Risk Scores in Predicting Renal Outcome in A New Zealand Cohort

There is emerging evidence for the ANCA Renal Risk Score (ARRS) and ANCA Kidney Risk Score (AKRiS) as simple and effective measures to predict risk of end-stage kidney disease (ESKD) in glomerulonephritis (GN) secondary to ANCA-associated vasculitis (AAV). We sought to externally validate these scores on an Aotearoa New Zealand (AoNZ) cohort.

Using a biopsy database, cases of pauci-immune GN in the three metropolitan Auckland districts between 1 January 2008 and31 December 2018 were retrospectively identified and histopathological data extracted. Hospital electronic records were screened to obtain relevant clinical information. Cox proportional hazards model was used to estimate survival distributions and perform receiver operating characteristic curve analysis.

One hundred twenty-one patients had sufficient data for risk score calculation. The cohort was predominantly male (60%) and of European ethnicity (71%). Māori and Pacific Peoples comprised 12% of the cohort. The vast majority (82%) of patients received cyclophosphamide induction. Ethnicity was associated with AKRiS score (p = 0.015) with Pacific Peoples being over-represented in the high risk AKRiS category. Both ARRS and AKRiS were significantly associated with ESKD (p < 0.001), with ordinal increases in probability of ESKD with risk category (apart from very high risk AKRiS group). Harrell's C-index for ARRS was 0.78 (95% CI: 0.705-0.862), and for AKRiS was 0.83 (95% CI: 0.753-0.896), without a significant increase with inclusion of ethnicity in the models. Brier scores were 0.131 for ARRS and 0.119 for AKRiS.

In a cohort of AoNZ AAV GN patients, both ARRS and AKRiS performed well in predicting the risk of ESKD.