This was a randomized, double-blind, active-controlled, 2-arm, multicenter study conducted at 60 sites (11 in the US; 7 in Malaysia; 6 each in Hungary, Poland, and Spain; 5 in Romania; 4 each in Costa Rica, Guatemala, Latvia, and Peru; 2 in Estonia; and 1 in Israel) from November 2008 to September 2010. The protocol (Protocol 134) and amendments were reviewed and approved by an institutional review board. The study was conducted in conformance with applicable country or local requirements regarding ethical committee review, informed consent, and other statutes or regulations regarding the protection of the rights and welfare of human subjects participating in biomedical research. All subjects signed informed consent prior to any study procedures being performed.
Eligible subjects were adults 18-79 years of age at high risk for CHD with primary hypercholesterolemia. For 6 weeks prior to screening, subjects eligible for the run-in period were currently taking either atorvastatin 20 mg or another lipid-lowering therapy with potency ≤ atorvastatin 20 mg, or were naïve to lipid-lowering therapy (naïve was defined as not being treated with a statin and/or ezetimibe). During the 5-week run-in period, subjects received open label atorvastatin 20 mg and lifestyle, diet counseling, and treatment compliance recommendations. All subjects must have been at high risk as defined by the National Cholesterol Education Program Adult Treatment Panel (NCEP ATP) III, which included subjects with CHD, having a CHD risk equivalent medical condition, and those who had 2+ risk factors that confer a 10-year risk for CHD > 20% as determined by the Framingham calculation [3
]. Laboratory entry criteria included LDL-C ≥ 100 mg/dL (2.59 mmol/L) and ≤ 160 mg/dL (4.14 mmol/L) at baseline (Visit 2, which is a treated baseline following the active treatment run-in period), triglyceride level ≤ 4.0 mmol/L (≤ 350 mg/dL), liver transaminases (alanine aminotransferase [ALT] and aspartate aminotransferase [AST]) ≤ 2.0 × upper limit of normal (ULN) with no active liver disease, and creatine kinase (CK) levels ≤ 3x ULN. Subjects were excluded from participating if they were taking simvastatin 80 mg, atorvastatin 40 or 80 mg, or rosuvastatin 10, 20, or 40 mg, or were taking other prescription and/or over-the-counter-drugs with the potential for significant lipid effects (other than study drug) or with potential drug interactions with the statins. Females who were pregnant or lactating were also excluded.
Blinding & randomization
The Clinical Biostatistics department of the study sponsor generated the randomized allocation schedule for study treatment assignment. At the end of the run-in period, eligible subjects were randomized in a 1:1 ratio to atorvastatin 40 mg or ezetimibe/simvastatin 10/40 mg. To achieve balance across treatment groups, subjects were stratified by their baseline LDL-C values at randomization (taken at Visit 3): Stratum 1: ≥ 2.59 mmol/L (100 mg/dL) and < 3.36 mmol/L (130 mg/dL) and Stratum 2: ≥ 3.36 mmol/L (130 mg/dL) and ≤ 4.14 mmol/L (160 mg/dL). Subjects received open label run-in bottles containing atorvastatin 20 mg. Treatment was supplied in blinded kits, each containing 2 bottles labeled A and B. Subjects were instructed to take one tablet daily from each of the bottles provided. The final database was not unblinded until medical/scientific review was performed, protocol violators were identified, and data had been declared final and complete.
LDL-C was calculated using the Friedewald method when triglycerides were < 4.52 mmol/L (400 mg/dL). The beta quantification ultracentrifugation method was used if triglycerides reached ≥ 4.52 mmol/L (400 mg/dL). Total cholesterol (TC), high-density lipoprotein cholesterol (HDL-C), hs-CRP, apolipoprotein (Apo) A-I, Apo B, and triglyceride fasting plasma levels were determined by a central laboratory. Non-HDL-C, LDL-C/HDL-C ratio, TC/HDL-C ratio, non-HDL-C/HDL-C ratio, and Apo B/Apo A-I ratio were calculated using the laboratory measurements.
Clinical evaluation included physical examinations, vital signs (blood pressure, weight, height, pulse rate), and laboratory safety tests. Clinical safety evaluations, including serious adverse events (AEs) and discontinuations due to clinical and/or laboratory AEs, were performed at Visits 2, 3, 4, and 5. Laboratory safety test evaluations were performed at Visits 1, 3, and 5 during the study.
Approximately 240 patients were expected to be randomized to ensure that 220 evaluable subjects would be included in the analysis of the full analysis set (FAS) population (110 per treatment group). With this enrollment, the study was assumed to have 95% power to detect a treatment difference of 10% in percent change from baseline between ezetimibe/simvastatin 10/40 mg and atorvastatin 40 mg assuming a standard deviation (SD) of 20% (at significance level = 0.05, 2-sided).
Primary and secondary efficacy data were analyzed based on the FAS population, which included all randomized subjects who had a baseline measurement and received at least 1 dose of study medication. For the primary and secondary efficacy endpoints involving percent change from baseline (except triglycerides and hs-CRP), between-group differences were analyzed using a constrained longitudinal data analysis (LDA) model, including both baseline and post-baseline percent change from baseline measurements as the response variable, and with terms for treatment, time, the interaction of time-by-treatment, stratum, and the interaction of time-by-stratum with a restriction of the same baseline mean across treatment groups (due to randomization). Due to the non-normal distribution associated with percent changes from baseline in triglycerides and hs-CRP, data were log-transformed. The percentage of subjects reaching LDL-C < 1.81 mmol/L (70 mg/dL), < 2.00 mmol/L (77 mg/dL) and < 2.59 mmol/L (100 mg/dL), respectively, was analyzed using a logistic regression model with terms for treatment and stratum.
There were 1 primary and 3 secondary efficacy hypotheses. Since there was only 1 primary efficacy endpoint and only 1 primary treatment comparison, no adjustment for multiplicity was applied for the primary hypothesis and the treatment comparison was performed at significance level 0.05. Analysis of the secondary variables related to secondary hypotheses was performed if the primary variable was significant at level 0.05. A step-down ordered testing procedure, which keeps the overall Type I error at level 0.05 if each test is performed at level 0.05, was performed for the key secondary variables. Specifically, the primary comparison of ezetimibe/simvastatin (10/40 mg) versus atorvastatin 40 mg in percent change from baseline in LDL-C was performed first. The tests of the secondary hypotheses were performed in the following order if the primary endpoint was significant at level 0.05: (1) percentage of subjects reaching LDL-C goal of < 1.81 mmol/L (70 mg/dL), (2) percentage of subjects reaching LDL-C goal of < 2.00 mmol/L (77 mg/dL), (3) percentage of subjects reaching LDL-C goal of < 2.59 mmol/L (100 mg/dL). The ordered testing procedure was to stop at any step where statistical significance was not achieved.
The All Patients as Treated (APaT) population, which consisted of all randomized subjects who received at least one dose of study treatment, was used for the analysis of safety data. Inferential testing provided statistical significance levels for between-group comparisons on the following safety parameters: gastrointestinal-related AEs, gallbladder-related AEs, allergic reaction or rash AEs, hepatitis-related AEs, elevations in ALT/AST ≥ 3× ULN, elevations in CK ≥ 10× ULN, elevations in CK ≥ 10× ULN with muscle symptoms and elevations in CK ≥ 10× ULN with drug-related muscle symptoms. Ninety-five percent confidence intervals (CIs) for treatment differences in proportions using Miettinen and Nurminen were provided for safety parameters including: 1 or more AEs, drug-related AEs, serious AEs, discontinuations due to an AE, AEs and predefined limits of change in laboratory safety parameters with incidence occurring in at least 4 subjects within at least 1 of the treatment groups. Descriptive statistics were used for all other AEs and predefined limits of change