LAUSR

The prognosis of lupus nephritis (LN) has become progressively more favorable thanks to the introduction of cyclophosphamide and mycophenolate as the mainstay of induction of remission treatment regimens. However, 10-15% of patients still progress to end-stage renal disease (ESRD). Early predictors of ESRD, i.e. in the first six months between kidney biopsy and the completion of induction, are currently limited to few histological and clinical features: ≥ 25% interstitial fibrosis and tubular atrophy (IFTA), fibrinoid necrosis, fibrous crescents, and thrombotic microangiopathy (TMA) [Rijnink EC et al CJASN 2017; Song D Arthritis Res Ther 2013]; lack of decrease in proteinuria < 0.5 g/24-h at 3 and 6 months from kidney biopsy [Tamirou F Ann Rheum Dis 2016], baseline GFR ≤ 90 ml/min/1.73 m2, lack of decrease in urinary protein-to-creatinine ratio (UPCR) < 1 and anti-dsDNA positivity at the end of induction [Dall’Era M Lupus Sci Med 2015]. In this study we sought to identify further clinical and histological predictors of ESRD in LN. Adult patients diagnosed with LN between 1995 and 2018 in two centers (NIAMS, Bethesda, Maryland, USA, and Nefrologia, AOU di Parma, Italy) were retrospectively identified. Patients with available serum C3 and C4 levels at the time of biopsy and 6 months thereafter, and a follow-up of at least 6 months, were included. Baseline and follow-up data (until March 2019) including age, sex, ethnicity, clinical, histological and laboratory findings were collected. Histology slides were reviewed by an experienced renal pathologist and biopsies re-scored using the ISN/RPS classification and NIH activity and chronicity indices. Distinct histological features were assessed individually (e.g. TMA). Persistent C3 hypocomplementemia was defined as decreased serum C3 levels at the time of biopsy and after 6 months (i.e. after the completion of induction), with concurrent normal serum C4 levels at 6 months. Early renal recovery was defined as either an increase in eGFR above 60 in those with a baseline eGFR < 60 ml/min/1.73 m2, or a 50% decrease in proteinuria in those with a baseline eGFR ≥ 60 ml/min/1.73 m2 and ≥ 0.5 g/24-h or g/g UPCR at the time of biopsy. Variables were tested for their predictive power of death-censored ESRD in univariate and multivariate Cox regression models. 74 patients (NIAMS n = 36; Parma n = 38) met our criteria. Median follow-up duration was 64 months (range 6-230). On univariate analysis, the following parameters predicted ESRD: Hispanic ethnicity; age at biopsy; persistent C3 hypocomplementemia; normalization of both C3 and C4; renal recovery after induction; NIH activity index; presence of TMA; ≥ 25% IFTA. Multivariate Cox regression models for ESRD were created considering statistically significant variables (p < 0.05). In a model including Hispanic ethnicity, age at biopsy, and persistent C3 hypocomplementemia, the latter predicted ESRD with an HR of 5.22 (95% CI [1.33, 20.58] p = 0.018) when adjusting for renal recovery after induction. Upon including histological features in the model, persistent C3 hypocomplementemia, TMA, and the NIH activity index lost significance, while ≥ 25% IFTA predicted ESRD with an HR of 27.26 (95% CI [2.12, 350.54], p = 0.011). In patients with LN, ≥ 25% IFTA at baseline biopsy is a predictor of ESRD, allowing for early risk stratification with the potential of informing treatment strategies. Where percent IFTA is unavailable or its assessment unreliable (e.g. inadequate biopsy specimen for tubulointerstitial assessment), persistent C3 hypocomplementemia represents a reliable and reproducible early predictor of ESRD, irrespective of early renal recovery after induction.