Understanding onset and response of two key endpoints in Systemic Lupus Erythematosus
Anna Largajolli (1), Raj Thatavarti (1), Matthew L. Zierhut (1), Garrett Nieddu (2), Tjerk JH Bueters (2), Nancy Kim (2), Martin Johnson (3)
(1) Certara USA, Inc., Princeton NJ, (2) Merck & Co., Inc., Rahway, NJ, USA (3) Merck Sharp & Dohme (UK) Limited, London, UK
Objectives:
Systemic Lupus Erythematosus (SLE) is a multiorgan autoimmune disease characterized by chronic inflammation of skin, joints, heart, lung, kidneys, circulating blood cells, and brain, which may result in permanent tissue damage from persistent disease activity [1]. Accurately measuring lupus disease activity and the changes in disease activity has proven to be a difficult task. Numerous different scoring systems have been proposed for assessing the severity of SLE and risk to vital organs. The consistent and successful application of these measures as endpoints in clinical trials remains elusive, which makes it difficult to directly compare results between therapies [2]. Understanding the onset and extent of response for different endpoints across different drugs, and how the response endpoints may correlate is crucial for designing clinical trials in SLE.
Methods:
We focused on two response endpoints curated in the Certara SLE clinical outcome database [3]: proportion of patients achieving Systemic Lupus Erythematosus Responder Index 4 (SRI-4) and/or The British Isles Lupus Assessment Group-based Composite Lupus Assessment (BICLA) response. SRI-4 response is defined as a reduction in the SLE Disease Activity Index (SLEDAI) by ≥4 points, with no worsening of the British Isles Lupus Assessment Group (BILAG) index (new grade 1A or 2B score) or deterioration from baseline ≥0.3 points in the physician global assessment [4]. BICLA response is defined as (1) at least one gradation of improvement in baseline BILAG scores in all body systems with moderate or severe disease activity at entry (e.g., all A (severe disease) scores falling to B (moderate), C (mild), or D (no activity) and all B scores falling to C or D); (2) no new BILAG A or no more than one new BILAG B scores; (3) no worsening of total SLEDAI score from baseline; (4) ≤10% deterioration in physicians global assessment; and (5) no initiation of non-protocol treatment [5, 6]. The clinical outcomes database captured the published response data as proportion of responders within each treatment arm. The two endpoints of interest were captured in multiple trials and at different time points, and this allowed us to perform a model based meta-analysis (MBMA) that characterized the time course and the response of the endpoints. The analysis focused on randomized controlled studies, most with placebo as a comparator. All timepoints and drug classes available in the trial outcome database were included. Endpoints were explored graphically based on placebo-corrected outcomes. Onset of SRI-4 and BICLA response was analyzed using nonlinear regression implemented in the generalized nonlinear least squares (gnls) routine in the R programming language (version 4.0.2). Impact of different covariates on the base model was also investigated. Different methods to impute covariate values, when missing, were explored on top of using the median value as a reference, including weighted average that accounts for the number of subjects informing each data point, random forest as implemented in the missForest R package, and predictive mean matching as implemented in the mice R package.
Results:
The total number of studies considered for the SRI-4 endpoint was 29 studies, including 86 treatment arms relative to 13 different drug classes, for a total of 441 longitudinal data points belonging to 53998 subjects. The total number of studies considered for the BICLA endpoint analysis was 16 studies, including 54 treatment arms relative to 9 different drug classes, for a total of 287 longitudinal data points belonging to 39306 subjects.
For both endpoints, the time course base model used an unstructured placebo with an exponential onset and a drug-class-specific treatment effect estimated in the logit domain. In general, the onset was found to be slower in SRI4 than BICLA. The magnitude of drug-class effects was different across the two endpoints, suggesting that a joint modeling exercise cannot be pursued.
For both endpoints, the time course base model used an unstructured placebo with an exponential onset and a drug-class-specific treatment effect estimated in the logit domain. In general, the onset was found to be slower in SRI4 (0.066, 1/week) than BICLA (0.183, 1/week). The magnitude of drug-class effects was different across the two endpoints, suggesting that a joint modeling exercise cannot be pursued (eg, interferon alpha drug class has respectively an estimated effect of 0.904 in SRI4 and 1.43 in BICLA).
Different methods to impute covariate values, when missing, were explored on top of using the median value as a reference, but all methods converged to the same final covariate model.
Conclusions:
We were able to characterize the time course of two key endpoints in SLE using an MBMA framework. These analyses can be instrumental to optimize the design of clinical trials in SLE.
References:
[1] Farounriakis, Antonis et al. “Update on the diagnosis and management of systemic lupus erythematosus,” Ann Rheum Dis 2021; 80: 14-25.
[2] Connelly, Kathryn, et al. "Clinician-reported outcome measures in lupus trials: a problem worth solving." The Lancet Rheumatology 3.8 (2021): e595-e603.
[3] Atopic Dermatitis Clinical Outcomes Database version 20221006 In, https://codex.certara.com/codex/sle/.
[4] Furie, Richard A., et al. "Novel evidence‐based systemic lupus erythematosus responder index." Arthritis Care & Research: Official Journal of the American College of Rheumatology 61.9 (2009): 1143-1151.
[5] Thanou, Aikaterini, et al. "Which outcome measures in SLE clinical trials best reflect medical judgment?." Lupus Science & Medicine 1.1 (2014): e000005.
[6] Wallace, Daniel et al. Evaluation of treatment success in systemic lupus erythematosus trials: development of the British Isles Lupus Assessment Group-based Composite Lupus Assessment Endpoint. Presented at ACR 2011: Poster 2265.