Use of Single Arm Studies/Disconnected Observational Studies in Comparative Effectiveness of Treatments for Fabry Disease: A Network Meta-Analysis Using a 3-Level Hierarchical Model

OBJECTIVES: With limited head-to-head data or disconnected studies in evidence network, conventional methods of network meta-analysis (NMA) are not suitable. We conducted NMA of available treatments for Fabry disease (FD) using randomized controlled trials (RCTs), single-arm, and observational studies.

METHODS: Studies evaluating enzyme replacement therapies (ERTs), migalastat and other available treatments for FD were identified from an existing review. Studies investigating at least one intervention as a part of the RCT network were considered for aggregate data-based matching. As the covariates data were not consistently reported, matching was restricted to age and gender. The complete network post-inclusion of the matched observational/single-arm studies was utilized. A three-level hierarchical model was used by incorporating data from RCTs and observational studies while also accounting for heterogeneity between the study designs. The outcomes of interest included severe renal events and estimated glomerular filtration rate (eGFR).

RESULTS: A total of eight RCTs, and eight matched studies contributed to the NMA. In the random effects NMA, agalsidase alfa 0.2 mg/kg EOW showed a numerical advantage (lower odds of occurrence of renal events) versus placebo (odds ratio (OR), 95% credible interval (CrI): 0.93 [95% CrI: 0.05, 17.16]) and undefined ERT (OR: 0.71 [95% CrI: 0.35, 1.47]). For eGFR, numerically better results were observed with agalsidase alfa 0.2 mg/kg EOW versus placebo (mean difference [MD]: 0.75 [95% CrI: -1.46, 2.78]), undefined ERT (MD: 0.56 [95% CrI: -2.36, 3.52]), agalsidase beta 0.2 mg/kg EOW (MD: 0.09 [95% CrI: -1.51, 1.73]) and migalastat 150 mg (MD: 0.66 [95% CrI: -1.99, 3.35]). No significant difference was observed across the treatments for both outcomes.

CONCLUSIONS: NMA results showed no differences in efficacy using both RCTs and observational studies in FD. Several limitations of the NMA should be acknowledged including the differences in inclusion criteria, endpoint definition, disease phenotype, and study designs.