Effect of Oral Fingolimod Treatment on Brain Volume Loss in Patients with Relapsing–Remitting Multiple Sclerosis Estimated Using Bayesian Methodology

Background: In patients with multiple sclerosis (MS), loss of brain volume (BV) occurs at an increased rate. Bayesian methodology, which uses prior evidence to analyze current data, was applied to provide new estimates of treatment effects of fingolimod on BV loss in FREEDOMS II (N=1083) with probabilistic interpretation; estimates may be enhanced by using prior information from FREEDOMS (N=1272).

Objectives: To estimate loss of BV in patients with relapsing–remitting MS treated with fingolimod in FREEDOMS II using Bayesian power prior methodology.

Methods: Structural Image Evaluation, using Normalization, of Atrophy (SIENA) was used to assess normalized BV at baseline and percentage BV change (PBVC) from baseline to months 6, 12 and 24, and during months 6–12 and 12–24. Bayesian power prior methodology was used to estimate PBVC for fingolimod 0.5 mg versus placebo in FREEDOMS II, with data from FREEDOMS as the informative prior and appropriate prior distributions of model parameters.

Results: From baseline to month 24, Bayesian analysis with zero weight given to FREEDOMS data estimated a treatment difference in PBVC of 0.42% in fingolimod-treated patients (Bayesian credible interval [BCI]: 0.19% to 0.67%) relative to placebo (mean PBVC −1.28%, BCI −1.46% to −1.11%). Observed PBVC at month 24 in FREEDOMS was: fingolimod 0.5mg, −0.84%; placebo, −1.31%. Applying equal weight to the likelihoods of prior and current study data gave an estimated treatment difference in PBVC of 0.35% (BCI: 0.22% to 0.48%) in fingolimod-treated patients relative to placebo (mean PBVC, −0.93%; BCI: −1.03% to −0.84%). Power priors with weights from 0.1–0.9 were explored in increments of 0.1.

Conclusions: Estimates of BV loss obtained using Bayesian power prior methodology were consistent with previously reported results in phase 3 fingolimod studies. Using prior clinical data and Bayesian methodology to analyze new data may improve estimates of treatment effects and their interpretability.