Evidence synthesis in neuromyelitis optica spectrum disorders: modeling a clinical trial based on published data

Abstract

Relevance. Optic neuromyelitis spectrum disorders are a severe orphan nosology for which the search for new highly effective therapeutic options is a priority area of medical care. At the same time, due to the nature of this group of diseases, conducting new clinical trials (CT) is significantly complicated not only by the long and complex recruitment of patients and the unethical nature of using a parallel control group without treatment, but also by conducting comparisons with a historical control group or indirect comparisons, for which planning the sample size in the CT for a new therapy option is a separate methodological task. Objective. The aim of this study was to estimate the power for the analysis of the primary endpoint in the single-arm BCD-132-6/AQUARELLE trial, which was planned to be performed using an adjusted indirect comparison (AIC) of the study drug divosilimab with published data on the placebo group in the randomized N-MOmentum trial. Materials and methods. Using the data of the studies identified using a systematic search, a meta-regression model was estimated for the dependence of the average annual relapse rate (ARR) on therapy (anti-CD20 or placebo), the assessment horizon, and baseline patient characteristics. According to its results, for the expected BCD-132-6/AQUARELLE population, a 6-month ARR of 0.116 (95 % CI: 0.030–0.212) was predicted for anti-CD20 therapy. Next, a stepwise effective sample size (ESS) simulation was performed in the divosilimab group with power assessment. Results. The expected ratio of the ARR of divosilimab and placebo at 6 months was 0.117 (90 % CI: 0.045–0.214) with-out weighting and 0.156 (90% CI: 0.050–0.315) after weighting (ESS). The minimum EES required to achieve a power of 90 % when testing the hypothesis of superiority in the ARR at 6 months for divosilimab therapy compared with placebo with a one-sided type I error probability of 0.05 is 35 people. Conclusion. The obtained results can be used to support the statistical inference of the BCD-132-6/AQUARELLE study.

Keywords

• meta-regression
• neuromyelitis optica
• adjusted indirect comparison
• divosilimab
• bcd-132-6/ aquarelle
• statistical power
• simulation modeling
• effective sample size