Participants were recruited through flyers and newspaper advertisements in Seoul, South Korea. Subjects were first screened using detailed and general medical questionnaires and trained nurse interviewed each subject for enrollment. Subjects were selected if they were non-smokers aged 20–39. Other inclusion criteria included a body mass index (BMI) of 17–30 kg/m² and a habitual diet close to that of a typical Korean diet. Subjects were excluded if they had any previous hypertension, arteriosclerosis, or metabolic diseases, had experienced any significant weight change in the last six months, were chronically ill or taking medications, consumed nuts frequently (more than twice per week), had erratic exercise habits, were smokers, or consumed alcohol more than twice per week. Women who had an irregular menstrual cycle, took birth control pills, or were pregnant or lactating were also excluded. This study was conducted according to the guidelines laid down in the Declaration of Helsinki, and all procedures involving human subjects were approved by the Ethics Committee of Sookmyung Women's University, Seoul, Korea (SMWU-1407-BR-006). Written informed consent was obtained from all study subjects.

This study was a randomized, 3-arms parallel, controlled 16-week trial. After being screened, eligible participants were simply randomized without any restriction into one of the three study arms using a computerized random number generator by the primary investigator. (1) In the pre-meal group (PM), participants were instructed to consume 56 g of almonds per day as a preload when having regular meals. They were directed to prepare the table first and consume nearly 1/3 amount of almond before each meal. (2) In the snack group (SN), participants were instructed to consume 56 g of almonds between meals as snacks. A snack was defined as an eating event that occurred between participants' regular meals, specifically two hours before and after meals. All the participants in the PM and SN groups were provided daily portions of packaged almonds. (3) In the control group (CL), participants were provided with high-carbohydrate control food items (66 g of commercial cookies; 320 kcal, 18 g fat, 3.9 g protein, 1.2 g total dietary fiber and 0.79 mg α-tocopherol) that had a similar number of calories as 56 g of almonds (independent analyses indicates that they contain 340 kcal, 30 g fat, 12 g protein, 6.2 g total dietary fiber, and 14 mg α-tocopherol). All of the subjects were able to determine the frequency of ingestion per day as desired, if they keep the guidelines for their groups. Participants were instructed to maintain their habitual dietary intake and usual level of physical activity and avoid consuming any additional nuts or nut products throughout the study. Participants, care providers, and analysts for dietary record were not blinded because of obvious differences between treatments, whereas personnel who were in charge of assessing outcomes and data management were blinded.

Compliance was measured by self-report. A consumption log sheet was provided for participants to keep a record of their almond or cookies consumption. On every Monday, subjects reported to care providers, trained registered dietitians using text message how many packages they did not consume during previous week. Good compliance was defined as > 80% adherence to the treatment. 80% compliance was defined as intake of almond or cookies over 90 times during 16 weeks (112 days). The three-day diet records, including two consecutive weekdays and one weekend day, were done once before the trial and twice during the trial. A registered dietitian provided detailed instructions of how to fill out a diet record and report their diet by taking pictures and sending these via text messages. All dietary records were analyzed to provide an estimation of the daily average energy and nutrient intake using CAN-Pro 4.0 software (Korean Nutrition Society, Seoul, Korea).

All measurements were conducted with participants going barefoot and wearing light clothing for the baseline (week 0) and main-trial days (weeks 8 and 16). Body composition including body fat percentages was assessed through multi-frequency whole-body bioimpedance measurement using InBody 620 (Biospace Co., Ltd, Seoul, Korea). The results included body composition analysis (total body water, protein, and body fat mass), muscle-fat analysis (body weight, skeletal muscle mass, and fat free mass), obesity analysis (body fat percentage, BMI (calculated as weight (kg) divided by height in meters (m) squared)).

For measuring blood lipid profiles, after a 12-hour fast, blood samples were taken at the baseline (week 0) and at weeks 8 and 16 by standard venipuncture. Blood was centrifuged at 3,000 rpm for 15 min at 4℃. Aliquots of serum were prepared and stored at −80℃ for following biochemical measurements. The serum total cholesterol and triglycerides levels were measured by the enzymatic-colorimetric method using a Cobas 8000 c702 chemistry analyzer (Roche Diagnostics; Mannheim, Germany). HDL cholesterol and LDL cholesterol levels were determined via homogeneous enzymatic colorimetry. Non-HDL cholesterol levels were calculated from the total cholesterol and HDL cholesterol using the following formula: Non-HDL cholesterol (mg/dL) = total cholesterol - HDL cholesterol [13].

Waist and hip circumference measurements were measured by trained technical staff using a measuring tape (SECA-201, SECA Ltd., Hamburg, Germany). Waist circumferences were obtained at the mid-point between the lowest rib and the iliac crest to the nearest 0.1 cm after inhalation and exhalation. Hip circumferences were rounded to the nearest 0.1 cm at the widest part between the waist and knees. Weight while fasting was measured and rounded to the nearest 0.1 kg, and height was rounded to the nearest 0.1 cm using a stadiometer (BSM330, Biospace Co., Ltd, Seoul, Korea). Blood pressure was measured on the right arm using an up-load blood pressure monitor (BPBIO320S, Biospace Co., Ltd, Seoul, Korea) with participants in a comfortably seated position after at least a five-minute rest.

Serum Interleukin-6 (IL-6) was measured using an enzyme immunometric assay kit (ADI-900-033, Enzo Life Science, Farmingdale, NY, USA). Serum malondialchehyche (MDA) and oxidized LDL-C (cholesterol) were measured using commercially available kits, a human MDA ELISA kit (MBS724277, MyBioSource, San Diego, CA , USA) and oxidized LDL ELISA kit (10-1143-01, Mercodia AB, Uppsala, Sweden) with an Epoch microplate spectrophotometer (BIOTEK, Inc., Winooski, VT, USA). The coefficients of variation (CV) for all measurements were less than 4%.

The sample size was calculated based on the sample size estimation for clinical trials using OpenEpi version 3.01. Percentages of the subjects who may show the primary outcomes in the almond groups and snack group were set at 25 and 5%, respectively. In total, 45 subjects were required in each group to provide over 80% power. The significance of the differences between the baseline (week 0) and main-trial periods (weeks 8 and 16) with different interventions was assessed using the repeated measures analysis of variance (RMANOVA) with Duncan multi-comparison, and the model assumptions were checked. Differences in baseline and Δ (changes from the baseline at weeks 8 or 16) among the three groups were detected using one-way ANOVA. All data were expressed as means with standard deviations unless otherwise specified. All statistical analyses were performed using of SAS (version 9.3; SAS Institute, Cary, NC).

In total, 169 (n = 58 for PM; 55 for SN; and 56 for CL) out of the 227 participants completed the study between June 2014 and June 2015. Fifty-eight participants dropped out due to schedule conflict, poor compliance, nausea and stomach pain, traveling, and pregnancy (Fig. 1). The basic characteristics of the participants are presented in Table 1. Age, BMI and blood lipid profiles at baseline indicated that the subjects were healthy and young. The results of the dietary analyses (Table 2) showed that at weeks 8 and 16, the carbohydrate intake of the PM and SN groups was lower than that of the CL group, while the intake of vegetable fat, vegetable protein, fiber, monounsaturated fatty acid (MUFA) and polyunsaturated fatty acid (PUFA) of the PM and SN groups was significantly higher than that of the CL group.

Changes from baseline in body composition including total body protein, body water, body fat mass, skeletal muscle mass, and visceral fat levels are presented in Table 3(A). Body weight changes from the baseline did not differ significantly for each time point among the three groups, whereas significant intervention effect was observed for changes of body fat mass (P = 0.025, RMANOVA) as results with one-way ANOVA revealed that body fat mass of PM group was decreased compared to CL at both weeks 8 and 16. In addition, body fat percentages in both almond groups were significantly reduced compared to the control group at week 16 using one-way ANOVA and significant interaction effect (P = 0.032, RMANOVA) was also observed. The visceral fat level significantly decreased in the PM group compared to the control group at both 8 and 16 weeks. In summary, consuming almond as a preload before meals reduced body fat mass, body fat percentages, and visceral obesity without reducing body weight.

Blood biochemistries including lipid profiles, inflammation and anti-oxidants markers were presented in Table 3(B). At baseline, CL group showed lower levels of triglycerides, LDL, and non-HDL cholesterol when compared to the PM and SN groups. Total blood cholesterol levels were significantly reduced in SN group compared to CL at week 8, not at week 16 due to the great reduction of cholesterol levels in the CL group. SN group showed significantly greater reductions of LDL cholesterol levels at week 16 than the CL and PM groups using one-way ANOVA. These results suggest that consuming almonds as a daily snack may reduce total cholesterol and LDL cholesterol levels. However, changes in blood HDL cholesterol levels were not observed throughout the trial. Due to the great reduction in the total cholesterol for the almond groups, a significant decrease in non-HDL cholesterol was observed in the SN group at both weeks 8 and 16 compared with the CL. Changes in the blood triglycerides levels did not significantly differ among the three groups based on the RMANOVA. There were significant intervention effects in changes of total cholesterol (P = 0.043), LDL cholesterol (P = 0.011), and non-HDL cholesterol (P = 0.013). Most of the blood chemistries also showed significant time effects as well, whereas there was no interaction between time and intervention. None of the changes in the oxidative and inflammation indicators were not significantly different among groups.