Fig. 2.

Gd-enhancing T1 lesion counts and ratios over 12 months in patient subgroups defined by demographic factors and baseline disease characteristics (intent-to-treat population). Gd-enhancing T1 lesion counts were estimated using a negative binomial regression model, log-link, adjusted for treatment for the overall result, and adjusted for treatment subgroup and treatment by subgroup variable interaction for the subgroup analyses. n number of observations included in the analysis (patients with non-missing Gd-enhancing T1 lesion count assessments at month 12; scans obtained <30 days after the last use of steroids to treat MS were excluded from this analysis). ^aPatients were categorized according to whether they were treatment-naïve (had received no form of medication for MS before the study) or had previously received treatment for MS with any medication at any time before study enrollment. ^bp value for the treatment contrast within the subgroup. ^cp value for the treatment by subgroup interaction, which evaluates heterogeneity of the treatment effect (see “Methods”). EDSS expanded disability status scale, Gd gadolinium, IF interferon, MS multiple sclerosis