Background Recent lupus nephritis (LN) trials have used various primary endpoints, but it is not yet known which endpoint is most sensitive to detecting differences between groups. We recently compared the performance of various definitions of complete response (CR) in a post-hoc analysis of data from a double-blind, controlled Phase II/III study of abatacept (ABA) in patients (pts) with LN who had not met the primary efficacy endpoint. Several CR definitions from other major LN trials suggested benefit for the treatment groups compared with the control.
Objectives The present analysis was designed to determine whether CR is the most discriminatory outcome measure or whether alternative measures may be more discriminatory between groups in this data set.
Methods In this study, 298 pts were randomized 1:1:1 to: ABA 30 mg/kg on Days 1, 15, 28, 57, and then 10 mg/kg through Week 52 (30/10); ABA 10 mg/kg on Days 1, 15, 28, and then every 28 days (10/10); or placebo. All pts received mycophenolate mofetil and glucocorticoids (GC). In our previous analyses, the response criteria from the LUNAR trial1 discriminated most effectively between groups. Therefore, we used LUNAR criteria to compare the following outcome measures: (i) CR at Weeks 24 and 52; (ii) CR + partial response (PR); (iii) major clinical response (MCR), which included pts who met CR criteria plus pts who were nephrotic at baseline but achieved a urine protein:creatinine ratio (UPCR) <1.0; (iv) ≥75% reduction in UPCR; (v) ≥25% increase in estimated glomerular filtration rate (eGFR); and (vi) frequency of treatment failure (defined as persistent nephrotic levels of proteinuria; failure of UPCR to improve by at least 25%; eGFR below normal and reduced by ≥25% relative to baseline; failure to taper GC to ≤10 mg/day; or withdrawal due to worsening LN, infection, or drug toxicity).
Results MCR and CR discriminated best and were comparable (group sizes of 50 pts would have been sufficient to demonstrate a statistically significant difference based on the observed response rates). The other outcome measures were distinctly less effective at discriminating treatment from control. Their order, from most to least discriminating, and the approximate number of pts that would have been required to show statistical significance, was: CR at Week 24 (∼90/group), treatment failure (∼90/group), UPCR reduced by ≥75% at Week 52 (∼90/group), total response at Week 52 (CR+PR) (∼100/group), total response at Week 24 (∼250/group), and eGFR increased by ≥25% at Week 52 (>500/group).
Conclusions The choice among possible outcome measures dramatically influences the likelihood of demonstrating a significant difference between groups. For this data set, MCR and CR at Week 52 showed the greatest difference between treatment and control. It remains to be determined which, if any, of the measures tested correlates with long-term outcome in future studies of abatacept or other agents.
Rovin B, et al. Arthritis Rheum 2012 Jan 9 [epub ahead of print].
Disclosure of Interest D. Wofsy Consultant for: Bristol-Myers Squibb, Genentech and Biogen IDEC Inc, Merck Serono, Teva Pharmaceuticals, S. Meadows–Shropshire Employee of: Bristol-Myers Squibb, J. Hillson Employee of: Bristol-Myers Squibb, B. Diamond Grant/Research support from: Bristol-Myers Squibb