Statistical Analysis
Primary analyses were performed on data for the intent-to-treat (ITT) population, defined as all subjects who were randomized and to whom the study drug was administered. These analyses were repeated in the Per Protocol (PP) population
to confirm the results, defined as ITT subjects who had no major protocol deviations.
The first co-primary efficacy endpoint was the success rate based on IGA score [percent of subjects who achieved “clear†or “almost clear†ratings on the IGA scale at week 12 (ITT-LOCF)], analyzed by the Cochran-Mantel-Haenszel (CMH) test stratified by analysis site, using the general association statistic.The second co-primary efficacy endpoint was the absolute change in inflammatory lesion counts from baseline to week 12 (ITT-LOCF), analyzed by analysis of covariance (ANCOVA). Missing data at week 12 in the ITT population were imputed by the Last Observation Carried Forward (LOCF) approach. Also, sensitivity analyses were conducted to impute missing data in order to assess the robustness of the primary efficacy results.
The secondary efficacy endpoint was percent change in inflammatory
lesion counts from baseline at week 12 (ITT-LOCF), analyzed by the Mann-Whitney test using the Cochran-Mantel-Haenszel (CMH) procedure stratified by analysis center, with row mean score difference statistic using ridit score transformation.
The endpoint of the subject’s assessment of rosacea improvement was analyzed using the CMH test stratified by analysis center, with row mean score difference statistic using ridit score transformation. The QoL questionnaires were analyzed using the Wilcoxon rank sum test, and other variables were descriptively analyzed. High mean scores from the QoL questionnaires indicated a low quality of life.
RESULTS
Subject Disposition and Baseline Characteristics
A total of 683 subjects with moderate to severe PPR were randomized in Study 1 (IVM 1%: 451, vehicle: 232), and 688 subjects in Study 2 (IVM 1%: 459, vehicle: 229) (Figure 1). In Studies 1 and 2, the vast majority of subjects completed the study (91.4% and 92.6%, respectively). The treatment groups were similar at baseline in terms of demographics and baseline disease characteristics, with about 31-33 inflammatory lesions on average and the majority having moderate rosacea (Table 2). Most subjects were female (68.2% and 66.7% in Studies 1 and 2, respectively) and Caucasian/white (96.2% and 95.3%), with a mean age of 50.4 and 50.2 years, respectively. Additionally, treatment groups were comparable regarding rates/reasons for early study discontinuation (Figure 1).
Efficacy
The proportion of subjects achieving IGA success (“clear†or “almost clearâ€) at week 12 for Studies 1 and 2 were 38.4% and
40.1% for IVM 1% compared to 11.6% and 18.8% for vehicle (both P<.001; Figure 2). A significant difference between treatment arms in both studies was observed by week 4 (10.9% and 11.8% vs 5.6% and 5.7%, respectively; both P<.05).
For inflammatory lesion counts, the mean difference between IVM 1% and vehicle from baseline to week 12 was -8.13 lesions for Study 1 and -8.22 for Study 2 (both P<.001 vs vehicle), with a 95% CI of [-10.12, -6.13] and [-10.18, -6.25], respectively. Median reduction from baseline in inflammatory lesion counts for both studies was 76.0% and 75.0%, respectively, vs 50.0% for both vehicle groups (P<.001), with significant difference observed by week 2 (Figure 3).
Safety
The incidence of AEs was comparable between Studies 1 and 2 (40.5% and 36.5% for IVM 1% vs 39.4% and 36.5% for vehicle, respectively). Fewer subjects in IVM 1% groups tended to report related AEs than in vehicle groups (4.2% and 2.6% vs 7.8% and 6.5%, respectively), as well as for related dermatologic AEs (3.5% and 1.5% vs 6.9% and 5.7%) and related AEs leading to discontinuation (1.3% and 0.2%, vs 1.7% for both vehicle groups). A similarly low proportion of subjects reported serious AEs for IVM 1% and vehicle groups (0.7% and 1.5% vs 0.4% and 1.7%). There were no related serious AEs. The most common related AE in Study 1 was sensation of skin burning: 8 (1.8%) in IVM 1% subjects vs 6 (2.6%) for vehicle. For Study 2, the most common related AEs for IVM 1% were pruritus and dry skin [3 subjects each (0.7%)] compared to 0 and 2 subjects (0.9%) for vehicle, respectively. In addition, laboratory tests did not demonstrate clinically significant abnormalities.
At baseline before treatment application, a large proportion of subjects presented with local cutaneous symptoms consistent
