Background/Purpose: Variability in response criteria for lupus nephritis (LN) clinical trials (CTs) has compromised the legitimacy and generalizability of CT results, primarily because short term response, however defined, does not necessarily equate to good long term kidney function. The aim of this study was to examine associations between short term renal parameters and long term renal outcomes to establish novel endpoints for LN CTs that predict long term kidney health.

Methods: A database of 944 patients with LN flare (baseline) and extended follow-up was established from 12 multinational clinical centers and 3 CTs. Performance of clinical variables at 12 months was assessed for predicting development of new or progressive chronic kidney disease (CKD), defined as a sustained decrease in eGFR ≥30%. Treatment was standard of care or as dictated by CT protocol. Statistical analyses included Kaplan-Meier curves for comparison of survival times using the log-rank test between categorical grouping variables and univariate Cox proportional hazards regression for continuous variables; Bonferroni correction was used for multiple comparisons. Factors significantly associated with CKD on univariate analysis (p<0.10) were included in a multivariable Cox regression analysis to build a prognostic model for time to CKD.

Results: Accounting for inclusion criteria and missing data, final analyses included 558 subjects (Table 1). 75 events of CKD occurred in 558 subjects; follow-up time ranged 22-317 months. Urine RBCs at baseline and 12 months, age, race (binary variable Black/non-Black), ISN Class, SCr at baseline and 12 months, proteinuria at 12 months and % change (%Δ) in SCr and proteinuria from baseline to 12 months were significantly associated with CKD by univariate analysis. The final multivariate model for significant predicators of CKD included the log (%Δ in proteinuria, p<0.0001), log (SCr at 12 months, p<0.0001) and race (p=0.04). The hazard ratios (HR) for CKD were 1.86 (95% CI: 1.52, 2.65) for log (%Δ proteinuria), 5.11 (3.04, 8.58) for log SCr and 1.65 (1.003, 2.70) for race. These predictors were combined into a “Hazard Index Tool” formula for developing CKD: Hazard Index_i = 0.61859*X₁ + 1.63100*X₂ + 0.49870*X₃ + 13, where X₁= log (%Δ proteinuria), X₂=log (SCr 12 months), X₃=1 if Black; 0 if non-Black.

Conclusion: Evaluation of a longitudinal cohort of patients with LN flare demonstrates that race and 12 month measures of proteinuria and SCr predict risk for development of CKD, whereas urine RBCs do not. HRs derived from multivariable analysis of these clinical variables were used to develop a Hazard Index Tool that predicts hazard of developing CKD. This tool can be used in CTs to evaluate superiority of new LN drugs in preventing future development of CKD (i.e., comparison of mean Hazard Index scores between treatment and placebo groups). Tool validation studies are planned. Table 1.Subject characteristics and predictor variables at baseline and 12 months stratified by development of CKD defined as a sustained decrease in eGFR≥30% after month 12.

  ¹Proteinuria was measured either as random urine protein/Cr ratios or 24 hour collections   Methods: Results: Conclusion:

« Back to 2016 ACR/ARHP Annual Meeting

ACR Meeting Abstracts - https://acrabstracts.org/abstract/using-a-hazard-index-tool-based-on-short-term-renal-parameters-to-predict-long-term-outcomes-in-lupus-nephritis-a-novel-way-to-assess-new-therapies/