Background There is controversy about what outcome measure in a lupus nephritis (LN) trial can be relied upon to predict longterm preservation of renal function.
Objectives This analysis was undertaken to determine which of the outcome measures commonly employed in LN trials, used alone or in combination, correlate with longterm preservation of renal function and which measure has the greatest positive and negative predictive value.
Methods We analyzed data from 81 of the 90 participants in the Euro-Lupus Nephritis Trial (ELNT) who have been followed for 5-12 yrs (median 10 yrs). Longterm data were not available for the other 9 subjects. Renal parameters were analyzed at 3, 6, and 12 mos after initiation of therapy, and those results were then correlated with the final serum creatinine (creat) measurement 5-12 years later. For the purpose of this analysis, the complete response (CR) targets were defined as: (i) proteinuria (prot) ≤500 mg/24 hrs; (ii) creat no worse than 0.2 mg/dL above baseline; and (iii) ≤5 RBCs/hpf on urinalysis. Partial response (PR) required a 50% improvement in proteinuria and creat no worse than 0.2 mg/dL above baseline, but did not require absence of RBCs. We tested three definitions for longterm preservation of renal function at the final visit: (i) creat ≤1.0 mg/dL; (ii) creat ≤1.3 mg/dL; and (iii) creat no worse than 0.2 mg/dL above the value at entry into ELNT.
Results 89% of subjects who met prot criteria for CR at 12 mos had creat<1.0 mg/dL 5-12 years later, compared to 61% of subjects who met prot criteria for PR (p<0.05) and only 16% of subjects who did not meet the prot response criteria (either CR or PR) at 12 mos (p<0.0001). Similar results were obtained for the two other definitions of a good longterm outcome. In all of the analyses performed, improvement in prot alone had the best positive and negative predictive value (Table). Inclusion of a creat component in the definition of CR slightly reduced the value of the outcome measure as a predictor of longterm outcome, and inclusion of an RBC component in the definition of CR substantially undermined the predictive value of the outcome measure. This pattern applied across all three definitions of “good longterm outcome”. The predictive value of a response (CR or PR) was better at 12 mos than at 3 or 6 mos (Table 1), and CR was consistently a better predictor of outcome than PR.
Conclusions Data from the ELNT support the conclusion that inclusion of an RBC component substantially weakens the positive predictive value of an early proteinuria drop in predicting longterm preservation of renal function. The clinical status at 3 or 6 mos does not correlate as well with longterm outcome as the status at 12 mos. Taken together, these data suggest that the primary outcome measure in LN trials should be based on improvement in proteinuria alone.
Disclosure of Interest None declared