Brain atrophy and associations with long-term disability and cognitive function in participants with relapsing multiple sclerosis treated with ozanimod: Results from phase 3 and open-label extension trials

Abstract

Background: In phase 3 trials, ozanimod reduced brain atrophy and improved cognitive processing speed compared with interferon β-1a (IFN) in participants with relapsing multiple sclerosis (RMS).

Objectives: To assess long-term brain volume changes and associations with clinical/cognitive outcomes during an open-label extension ([OLE] DAYBREAK [NCT02576717]).

Methods: Completers of phase 3 "parent" trials were eligible to receive ozanimod 0.92 mg in DAYBREAK. Whole brain, thalamic, and cortical gray matter volumes (WBV, TV, and CGMV, respectively) were analyzed annually.

Results: Participants receiving continuous ozanimod had sustained, low rates of WBV loss through OLE month (M)60 (annualized least-squares mean percent change from parent baseline: RADIANCE, -0.27; SUNBEAM, -0.35). Compared with participants switched from IFN, these participants had lower reductions in WBV (parent baseline through OLE M48 [RADIANCE] and OLE M60 [SUNBEAM]). Larger baseline brain volumes were associated with numerically better Symbol Digit Modalities Test scores and lower 3-month confirmed disability progression (CDP) incidence. Annualized TV atrophy ⩽1.0% was associated with lower 3-month CDP.

Conclusion: This study confirms the sustained efficacy of ozanimod in reducing brain atrophy rates for up to 7 years. Brain volume preservation was associated with faster cognitive processing speed and slower physical disability progression.

Keywords: MRI; Multiple sclerosis; Symbol Digit Modalities Test; confirmed disability progression; sphingosine 1-phosphate receptor modulators; thalamus.

Figure 1.

LS mean percentage change and annualized percentage change in whole brain volume (a), thalamic volume (b), and cortical gray matter volume (c) loss over time relative to parent-trial baseline. BL: baseline; CGMV: cortical gray matter volume; IFN: interferon; LS: least squares; ns: not statistically significant (p ⩾ 0.05); OLE: open-label extension (DAYBREAK); SE: standard error; TV: thalamic volume; WBV: whole brain volume. p-values presented are nominal, as treatment comparisons in this post hoc, exploratory analysis were not subject to multiplicity adjustment. LS mean percentage change from parent-trial baseline was estimated using a mixed model for repeated measures. The model included percentage change or atrophy rate from parent-trial baseline in brain volume as the dependent variable and stratification factors (region [Eastern Europe vs. rest of the world], baseline Expanded Disability Status Scale category [⩽3.5 vs. >3.5]), age at parent-trial baseline, treatment, time point, and the interaction between treatment and time point as fixed effects, parent-trial baseline brain volume as a continuous covariate, and subject as a random effect. For OLE visits, treatment duration during the parent trial was added as a fixed effect in addition to the effects above. *p < 0.05; ^†p < 0.01; ^‡p < 0.001; ^§p < 0.0001, versus IFN β-1a.

Figure 2.

SDMT mean scores (a) and categories of SDMT score change in participants treated with IFN β-1a (b) and ozanimod 0.92 mg (c) by TV baseline tertile in SUNBEAM. CI: confidence interval; IFN: interferon; LS: least squares; M: month; OLE: open-label extension; PT: parent trial; SDMT: Symbol Digit Modalities Test; SE: standard error; TV: thalamic volume. ^aBased on t-distribution. ^bOLE month 12 visit included data collected after OLE first dose date up to OLE month 12. ^cBased on Wald 95% CI.

Figure 3.

Proportions of participants with 3-month CDP^a by baseline thalamic volume tertiles (a) and by thalamic atrophy (⩽1.0%) rates at the end of parent trial^b (b). CI: confidence interval; CDP: confirmed disability progression; HR: hazard ratio; IFN: interferon; OLE: open-label extension; TV: thalamic volume. ^aParticipants could have a 3-month CDP at any time during the parent trial or OLE. ^bFor SUNBEAM, this was at least 12 months; for RADIANCE, this was 24 months. The 95% CIs were calculated using the Clopper–Pearson method.

Figure 4.

Kaplan-Meier curves of 3-month CDP^a by baseline thalamic volume tertile (low vs. high) in participants treated with IFN β-1a or ozanimod 0.92 mg in SUNBEAM (a) and RADIANCE (b). CDP: confirmed disability progression; IFN: interferon; M: month; OLE: open-label extension. ^aParticipants could have a 3-month CDP at any time during the parent trial or OLE.

Figure 5.

Predictors of 3-month CDP in participants treated with IFN β-1a or ozanimod 0.92 mg in SUNBEAM (a) and RADIANCE (b). CDP: confirmed disability progression; CGMV: cortical gray matter volume; EDSS: Expanded Disability Status Scale; GdE: gadolinium-enhancing; HR: hazard ratio; IFN: interferon; MS: multiple sclerosis; OLE: open-label extension; PASAT: Paced Auditory Serial Addition Test; SDMT: Symbol Digit Modalities Test; TV: thalamic volume; WBV: whole brain volume. ^aFor Z-scores at parent baseline, the predictor was normalized to mean 0 and unit variance. ^bAt parent baseline. ^cHazard ratios were based on the Cox proportional hazards model on 3-month CDP. Participants could have 3-month CDP at any time during the parent trial or OLE. A separate model was produced for each independent predictor as a factor, and nominal p-value is from Wald’s chi-square test. Other predictors that were tested but are not reported here are age,^b sex,^b body mass index,^b weight,^b years since MS symptom onset, pre-existing cardiac disease flag, pre-existing hepatic disease flag, prior MS treatment flag, number of GdE lesions group (>0 vs. 0),^b number of GdE lesions at parent month 12 group (>0 vs. 0), EDSS score group (>2 vs. ⩽2),^b log transformations of T2, unenhancing T1 lesion volume^b (µL), and GdE lesion volume^b (µL); T2 lesion counts, unenhancing T1 lesion counts, unenhancing T1 lesion counts,^b GdE lesions counts,^b and WBV,^b TV,^b and CGMV^b (cm³).