Background/Purpose: Randomized controlled trials (RCTs) of new SLE treatments are hampered by the lack of effective outcome measures that are responsive to change and clinically relevant. To address this, we developed a new Multivariable Lupus Outcome Score (LuMOS) from raw data from a previous RCT and validated it using data from an independent trial.

Methods:

The new LuMOS formula was developed by analyzing raw data of two pivotal RCTs: BLISS-52 and BLISS-76, the basis for approval of belimumab for treatment of SLE. Using data from BLISS-76 as the learning dataset, we optimized discrimination between outcomes for patients treated with 10mg/kg belimumab (BM-10mg) versus placebo over the first 52 weeks of follow-up. LuMOS was developed through multivariable logistic regression analyses that combined a priori inclusion of some variables, such as change in SLEDAI ≥4 points, with forward selection of others. The final model was selected using the Akaike Information Criterion. Performance of LuMOS was assessed using an independent validation dataset from the BLISS-52 RCT. For each belimumab trial, we tested significance of the differences in the mean LuMOS in the Active Treatment versus Placebo groups. To compare discriminatory ability of LuMOS with the conventional SLE Response Index (SRI-4), used as the primary outcome measure in both BLISS trials, we relied on Effect Size (ES = Cohen’s d = mean difference divided by within-group standard deviation).

Results:

The final LuMOS model incorporated reduction in SELENA-SLEDAI ≥4 points, increase in C4 concentration, decrease in anti-dsDNA Ab titer, and changes in BILAG organ system manifestations: no new symptoms or no worsening in renal and improvements in mucocutaneous components. Decreases in prednisone doses and increases in C3 concentration, included based on clinical considerations, had very minor impacts on total LuMOS. In all analyses of BLISS-76 and BLISS-52 RCTs, mean LuMOS were significantly higher (p < 0.0001) for BM-10mg and BM-1mg treatment groups than placebo. LuMOS also found significant differences between active treatment and placebo when SRI did not, as for BM 1mg in BLISS-76. The ES were consistently and significantly much higher, and p-values systematically much lower with LuMOS compared with SRI (Table 1).

Conclusion:

The proposed evidenced-based LuMOS developed with data from BLISS-76 and validated with data from BLISS-52 exhibits superior capacity to discriminate responders from nonresponders, compared to the SRI-4. Use of LuMOS may improve the efficiency and power of analyses in future lupus trials.

« Back to 2017 ACR/ARHP Annual Meeting

ACR Meeting Abstracts - https://acrabstracts.org/abstract/development-and-validation-of-a-novel-evidence-based-lupus-multivariable-outcome-score-for-clinical-trials/