Study population

Patients who met the 2013 American College of Cardiology/American Heart Association criteria for receiving lipid-lowering therapy were eligible for this study. They consisted of patients who had a prior history of atherosclerotic cardiovascular disease, those who had diabetes mellitus, or high cardiovascular risk. The patients were statin-naïve or had discontinued any lipid-lowering agent at least for 3 months before the enrollment unless they had previous cardiovascular disease. Patients were excluded if they were pregnant or breast feeding, had a history of acute cardio- or cerebro-vascular disease within 3 months before the study, uncontrolled hypertension or diabetes mellitus, thyroid dysfunction, serum transaminase >2 times the upper limit of normal, serum creatinine >1.5 mg/dL, an acute or chronic infection or inflammation, or a history of cancer or adverse events associated with test drugs including myopathy. All the patients provided written informed consent.

Study protocol

The present study was a sub-study of an 8-week, randomized, open-label interventional study that was approved by Yonsei University Health System, Severance Hospital, Institutional Review Board (4-2013-0281). This study was actually a sub-study of a main trial as mentioned in the methods section, and we applied the clinical trial protocol that was used in the main study. Since our trial followed the protocol, which was revised after the study began, we included the flowchart of revised version of the protocol (Additional file 1: Figure S1). At the initial screening visit, the patients were interviewed to obtain their medical histories and then underwent laboratory assessments. Those who met the criteria for lipid-lowering therapy were subsequently randomized in a 1:1 ratio into two treatment groups for 8 weeks: atorvastatin 20 mg (Lipitor, Pfizer, New York, NY, USA) or atorvastatin/ezetimibe 5 mg/10 mg (Lipitor, Pfizer and Ezetrol, Merck & Co., Whitehouse Station, NJ, USA). These two regimens were selected on our previous studies that showed these two regimens reduced LDL-C levels similarly [13, 14]. Thirty-two patients were initially screened but 3 of them did not complete the study: 2 due to refuse to follow-up and 1 due to protocol violation. Other eight patients were excluded due to insufficient blood sampling. Because of high censoring rate in the atorvastatin monotherapy group, enrollment with unequal allocation was performed to avoid decreasing the power of the comparison. HDL function and proteins were finally analyzed in 21 patients (11 in the atorvastatin monotherapy group and 10 in the combination group; Additional file 1: Figure S1).

Blood sampling and isolation of HDL

Blood samples were collected from the patients on enrollment and after the 8-week drug treatment. The patients were instructed to fast and avoid alcohol beverages or smoking for at least 12 h before the collection of the samples, which were analysed within 4 h. All the analyses were performed by a local laboratory, certified by the Korean Society of Laboratory Medicine. The lipid levels were measured using an auto-analyzer.

HDL was isolated by ultracentrifugation described below. Briefly, 2 mL of serum sample was transferred into a 12 mL-ultracentrifuge tube (Polyallomer, Beckman Coulter Korea Ltd, Seoul, Korea) and then 0.12 g potassium bromide (KBr) and 0.045 g sucrose were added to be dissolved. Then, 2 mL of solution B (1 mL distilled water, sodium chloride [NaCl] 0.012 g, and KBr 0.135 g), 4 mL of solution A (distilled water 1 mL plus NaCl 0.012 g plus KBr 0.318 g), and 4 mL of distilled water were sequentially added. Ultracentrifugation was conducted using a Beckman Coulter XL-100 K Table Top Ultracentrifuge with a Beckman fixed-angle rotor (SW41Ti) for 18 h at 35,000 rpm. Then, the supernatant, contained the very low-density lipoprotein and LDL was removed and HDL was aspirated. The isolated HDL was subsequently desalted and concentrated with an Amicon 3 k Ultracentrifugal Filter device (Merck Millipore Korea, Seoul, Korea) at 3000 rpm at 4 °C.

In vitro tests of HDL function

The cholesterol efflux assay was performed using a previously described method [15]. Briefly, the J774 cells were plated and radiolabeled with 2 μCi of ³H-choelsterol/mL for 24 h. For the upregulation of adenosine triphosphate (ATP)-binding cassette transporter subfamily member A1 (ABCA1), the cells were incubated with medium containing 0.2% bovine serum albumin (BSA) and 0.3 mM cyclic adenosine monophosphate (cAMP) for 2 h. Then, the medium was changed to a medium containing 0.2% BSA and HDL for 4 h. The experiment was conducted by treatment the cells with acyl-coenzyme A:cholesterol acyltransferase inhibitor 2 μg/mL. The cholesterol efflux proportion was calculated using the following formula: Cholesterol efflux capacity (%) = [³H-cholesterol (μCi) in medium containing HDL/(³H-cholesterol {μCi} in medium containing HDL + μCi of ³H-cholesterol {μCi} in cells)] x 100. The values were adjusted based on the efflux capacity of the pooled serum run in each plate. Each sample was run in duplicate.

The endothelial NO production was assayed as described previously [16, 17]. Briefly, the human umbilical vein endothelial cells were purchased from Lonza (Basel, Switzerland), grown until the cells had reached 90% confluency, and then incubated with serum-free medium overnight. After being treated with 50 μg/mL of HDL, cells were washed and lysed in 5 mM Tris. After centrifuging the cell lysates, supernatants were transferred to Amicon 10 kDa cut-off filter tubes (Merck Millipore Korea) and further centrifuged. Then the flow-through was collected and nitrite level was measured using a kit according to manufacturer’s instruction (Cayman Chemical, Ann Arbor, MI, USA).

The VCAM-1 level was measured by western blotting [18, 19]. Briefly, the human umbilical vein endothelial cells were grown and VCAM-1 expression was induced by 5 ng/mL of tumor necrosis factor-α in serum-free media overnight. Then, the cells were treated with 50 μg/mL of HDL for 4 h, washed, and lysed in radioimmunoprecipitation assay buffer supplemented with protease inhibitor cocktail tablet (Roche Applied Science, Penzberg, Germany). After that, the total protein concentration of cell lysate supernatant was determined, then 7 μg of protein was loaded, and subsequently separated by running the 10% sodium dodecyl sulphate-polyacrylamide gel electrophoresis (SDS-PAGE). The proteins were transferred from the gel to a nitrocellulose membrane and incubated with anti-VCAM-1 (Abcam, Cambridge, MA, USA) and mouse anti-β-actin antibodies (Santa Cruz Biotechnology, Inc., Santa Cruz, CA, USA). Protein bands were visualized using the SuperSignal West Pico Chemiluminescent substrate (ThermoFisher Scientific, Waltham, MA, USA), and the band intensity was quantified using the ImageJ software (National Institute of Health, Bethesda, MD, USA). The VCAM-1 expression was normalized to the intensity of β–actin, and the levels in cells treated with each HDL sample were presented as percentages of untreated cells.

The generation of intracellular ROS was determined using dichlorodihydrofluorescein diacetate (CM-H2DCFDA, ThermoFisher Scientific) [20, 21]. After treating J774 cells with 100 μg/mL of HDL for 24 h, they were stained with 5 μM CM-H2DCFDA in PBS for 24 min at 37 °C, incubated with or without 100 M hydroperoxide for 20 min, and then ROS generation was detected using a flow cytometer. The mean fluorescence intensity was measured in 10,000 cells using the fluorescein isothiocyanate channel.

Measurement of selected HDL proteins

From preliminary proteomic analysis of HDL samples based on the protocol we previously reported [22], we selected five HDL-related proteins that were abundantly and reproducibly detected: apoA1, apoA2, apoC1, poC2, and apoC3. The proteins were measured and quantified as follows. Briefly, the same amount (from 0.5 to 10 μg) of the HDL samples were separated by 15% SDS-PAGE and transferred onto polyvinylidene difluoride membranes, which were blocked against nonspecific binding, and then incubated with primary antibodies against apoA1, apoA2 (Santa Cruz Biotechnology), apoC1, apoC2 (Abcam), and apoC3 (Academy Bio-Medical Company Inc., Houston, TX, USA). Then, the membranes were further incubated with appropriate horseradish peroxidase-conjugated bovine anti-mouse or goat anti-rabbit secondary antibodies (Santa Cruz). The signal was detected using chemiluminescence with ECL reagent (GE Healthcare, Piscataway, NJ, USA) and the band intensities were quantified using the ImageJ software.

Statistical analysis

The clinical and laboratory variables were compared by the Student’s t-test or chi-square test. A paired t-test was used to compare the parameters before and after drug treatment. For the variables showing a skewed distribution, the Wilcoxon signed-rank test for the median was used. The Spearman correlation analysis was used to evaluate the association between the HDL functional parameters and HDL-related protein levels as well as the changes of HDL function and those of HDL protein levels. All the analyses used two-tailed tests with a significance level of 0.05. The statistics for the social sciences version 17.0 software (SPSS Inc, Chicago, IL, USA) was used for the analyses. This study is registered with ClinicalTrials.gov, number NCT02942602.

Clinical characteristics and laboratory values

Changes in HDL function after drug treatment

The baseline median cholesterol efflux capacities were similar between the atorvastatin and combination groups (13.1% and 16.4%, respectively, p = 0.32) (Fig 1a). NO and ROS production, as well as VCAM-1 expression, did not differ between the two groups (Figs 1c, e, and g). After the 8-week drug treatment, the cholesterol efflux capacity significantly increased in the atorvastatin, but not in the combination group (Fig 1a). However, the percentage change in the capacities did not differ between the groups (35.6% and 34.6%, respectively, p = 0.60, Fig 1b). The NO production did not change significantly after treatment in the monotherapy and combination groups (Fig 1c), and there were no inter-group difference in the percentage changes (9.5% and -5.2%, respectively, p = 0.21, Fig 1d). Furthermore, both groups exhibited no changes in the VCAM-1 expressions and ROS production after drug treatment (Figs 1e and g). The percentage changes in the two functional parameters were similar between the monotherapy and combination groups (-5.5% and 2.4%, VCAM-1 expression, p = 0.25; 3.0% and -7.2%, ROS production, p = 0.43, respectively, Figs 1f and h).

Relationship between changes in HDL function and HDL-related proteins