LDL-cholesterol Lowering Effect of a New Dietary Supplement

Between July 2016 and April 2017 eligible patients were recruited among the outpatients attending the Obesity Center of the Endocrinology Unit 1, Cisanello Hospital, Pisa, Italy. Patients aged 18-75 years with serum LDL-C between130-180 mg/dL, not significantly modified by an appropriate dietetic regimen were considered eligible for the study. Thirty patients, all Caucasian, were screened. Ten were excluded during the screening process because they did not fulfill all the inclusion criteria (screening failure). Twenty patients were thus randomized, 10 to the A>N sequence and 10 to the N>A sequence.

Study design The study was conducted in a single center according to a controlled, randomized, open-label, cross-over design. Each patient had to assume, in a randomized sequence, both treatments (A, 1 sachet/day; N, 1 capsule/day) for 8 weeks each separated by a 4-week wash-out period. The study plan included the initial screening visit (V1), an entry visit at start of the first treatment period (V2), a visit at the end of the first treatment period (V3, 56 ±5 days after V2), a wash-out period of 4 weeks (±5 days), a visit at start of the second (crossed over) treatment period (V4), and a visit at the end of the second treatment period (V5, 56 ±5 days after V4) (Figure 1). Tot-C, LDL-C, HDL-cholesterol (HDL-C), TG, fasting blood glucose, aspartate aminotransferase (AST), alanine aminotransferase (ALT), creatinkinase (CK), gamma-glutamyl-transpeptidase (GGT), humeral blood pressure and heart rate were measured at V1, V3, V4 and V5. Blood analyses were centrally performed in the laboratory of the Endocrinology Unit using standard enzymatic techniques; LDL-C was measured directly. Clinical safety was monitored throughout the study. If required, the patient could be re-evaluated at any time during the study, aside of the visits scheduled.

Statistical Methods The minimum level of statistical significance was set to p\<0.05 two-sided, therefore 95% confidence limits (95%CIs) were calculated. All reported p-values and CIs are two-sided.

The primary efficacy variable was the LDL-C change between the start and the end of each treatment period, expressed as a percentage of the initial value. Therefore, mean and 95%CIs of changes within treatment periods (from V2 to V3 and from V4 to V5) for the experimental and the control treatment, irrespective of sequence, were calculated. The main analysis was the determination of the two-sided 95%CI of the between-treatment mean difference in the primary variable. Setting 0.10 (i.e. 10% of the initial value) as the minimum clinically relevant difference, the two treatments were considered equivalent if the two-sided 95%CI of the difference in their LDL-C change from baseline was entirely between -0.10 and +0.10. Parallel calculations were carried out on absolute, rather than relative to baseline, LDL-C changes. Tot-C and HDL-C were analysed as described above for LDL-C; for TG levels (which were approximately log-normally distributed) analogous calculations were performed on logarithmic transformations and changes were expressed as ratios. Between-treatment comparisons were expressed as A-N differences for cholesterol values and as A/N ratios for TG values. The effects on LDL-C were additionally tested in sensitivity multivariate analyses: split-plot analysis of variance for cross-over studies on final-baseline changes, and analysis of covariance on the difference between the final values adjusted for sequence and for the difference between the baseline values. Efficacy analyses had to be performed in the intention-to-treat population, i.e. all patients with at least one post-baseline control). A sensitivity analysis of the primary variable was also planned in the per-protocol population, i.e. all patients without major protocol violations. Safety results had to be reported in all patients who had assumed at least one dose of one study drug. Statistical analyses were performed by the Studio Associato Airoldi Cicogna and Ghirri, Milan, using the SAS Software version 9.4 (SAS Inc, Cary, NC).

Sample Size The sample size was calculated for the main efficacy analysis described above, i.e. the determination of the two-sided 95% CI of the between-treatment mean difference in the LDL-C change from baseline. Assuming a standard deviation (SD) of the difference no greater than 0.12, based on a previous cross-over study for the difference between monacolin K and placebo [11], it was estimated that 18 patients were required to prove the equivalence with a power of 0.80. This figure was rounded to 20 enrolled patients allowing for possible exclusions from the analysis.