46
Intermittent Fasting: Weight Loss and Cardiovascular Effects

Intermittent Fasting: Weight Loss and Cardiovascular Effects · nate day fasting (ADF; fasting every other day) is organ-ized with alternating “feast days,” on which there is

  • Upload
    others

  • View
    32

  • Download
    4

Embed Size (px)

Citation preview

Page 1: Intermittent Fasting: Weight Loss and Cardiovascular Effects · nate day fasting (ADF; fasting every other day) is organ-ized with alternating “feast days,” on which there is

Intermittent Fasting: Weight Loss and Cardiovascular Effects

Page 2: Intermittent Fasting: Weight Loss and Cardiovascular Effects · nate day fasting (ADF; fasting every other day) is organ-ized with alternating “feast days,” on which there is

Moro et al. J Transl Med (2016) 14:290 DOI 10.1186/s12967-016-1044-0

RESEARCH

Effects of eight weeks of time-restricted feeding (16/8) on basal metabolism, maximal strength, body composition, inflammation, and cardiovascular risk factors in resistance-trained malesTatiana Moro1 , Grant Tinsley2 , Antonino Bianco3 , Giuseppe Marcolin1 , Quirico Francesco Pacelli1, Giuseppe Battaglia3 , Antonio Palma3 , Paulo Gentil5 , Marco Neri4 and Antonio Paoli1*

Abstract Background: Intermittent fasting (IF) is an increasingly popular dietary approach used for weight loss and overall health. While there is an increasing body of evidence demonstrating beneficial effects of IF on blood lipids and other health outcomes in the overweight and obese, limited data are available about the effect of IF in athletes. Thus, the present study sought to investigate the effects of a modified IF protocol (i.e. time-restricted feeding) during resistance training in healthy resistance-trained males.

Methods: Thirty-four resistance-trained males were randomly assigned to time-restricted feeding (TRF) or normal diet group (ND). TRF subjects consumed 100 % of their energy needs in an 8-h period of time each day, with their caloric intake divided into three meals consumed at 1 p.m., 4 p.m., and 8 p.m. The remaining 16 h per 24-h period made up the fasting period. Subjects in the ND group consumed 100 % of their energy needs divided into three meals consumed at 8 a.m., 1 p.m., and 8 p.m. Groups were matched for kilocalories consumed and macronutri-ent distribution (TRF 2826 ± 412.3 kcal/day, carbohydrates 53.2 ± 1.4 %, fat 24.7 ± 3.1 %, protein 22.1 ± 2.6 %, ND 3007 ± 444.7 kcal/day, carbohydrates 54.7 ± 2.2 %, fat 23.9 ± 3.5 %, protein 21.4 ± 1.8). Subjects were tested before and after 8 weeks of the assigned diet and standardized resistance training program. Fat mass and fat-free mass were assessed by dual-energy x-ray absorptiometry and muscle area of the thigh and arm were measured using an anthro-pometric system. Total and free testosterone, insulin-like growth factor 1, blood glucose, insulin, adiponectin, leptin, triiodothyronine, thyroid stimulating hormone, interleukin-6, interleukin-1β, tumor necrosis factor α, total cholesterol, high-density lipoprotein cholesterol, low-density lipoprotein cholesterol, and triglycerides were measured. Bench press and leg press maximal strength, resting energy expenditure, and respiratory ratio were also tested.

Results: After 8 weeks, the 2 Way ANOVA (Time * Diet interaction) showed a decrease in fat mass in TRF compared to ND (p = 0.0448), while fat-free mass, muscle area of the arm and thigh, and maximal strength were maintained in both groups. Testosterone and insulin-like growth factor 1 decreased significantly in TRF, with no changes in ND (p = 0.0476; p = 0.0397). Adiponectin increased (p = 0.0000) in TRF while total leptin decreased (p = 0.0001), although not when adjusted for fat mass. Triiodothyronine decreased in TRF, but no significant changes were detected in thyroid-stimulating hormone, total cholesterol, high-density lipoprotein, low-density lipoprotein, or triglycerides.

© 2016 The Author(s). This article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The Creative Commons Public Domain Dedication waiver (http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated.

Open Access

Journal of Translational Medicine

*Correspondence: [email protected] 1 Department of Biomedical Sciences, University of Padova, Padua, ItalyFull list of author information is available at the end of the article

Page 3: Intermittent Fasting: Weight Loss and Cardiovascular Effects · nate day fasting (ADF; fasting every other day) is organ-ized with alternating “feast days,” on which there is

Page 2 of 10Moro et al. J Transl Med (2016) 14:290

BackgroundFasting, the voluntary abstinence from food intake for a specified period of time, is a well-known practice asso-ciated with many religious and spiritual traditions. In fact, this ascetic practice is referenced in the Old Tes-tament, as well as other ancient texts such the Koran and the Mahabharata. In humans, fasting is achieved by ingesting little to no food or caloric beverages for periods that typically range from 12 h to 3 weeks. Mus-lims, for example, fast from dawn until dusk during the month of Ramadan, while Christians, Jews, Buddhists, and Hindus traditionally fast on designated days or peri-ods [1]. Fasting is distinct from caloric restriction (CR), in which daily caloric intake is chronically reduced by up to 40  %, but meal frequency is maintained [2]. In contrast to fasting, starvation is a chronic nutritional deficiency that is commonly incorrectly used as a sub-stitute for the term “fasting”. Starvation could also refer to some extreme forms of fasting, which can result in an impaired metabolic state and death. However, starva-tion typically implies chronic involuntary abstinence of food, which can lead to nutrient deficiencies and health impairment. While a prolonged period of fasting is dif-ficult to perform for the normal population, an inter-mittent fasting (IF) protocol has been shown to produce higher compliance [3]. Typically, IF is defined by a com-plete or partial restriction in energy intake (between 50 and 100  % restriction of total daily energy intake) on 1–3  days per week or a complete restriction in energy intake for a defined period during the day that extends the overnight fast. The most studied of the above form of IF is Ramadan fasting: during the holy month of Ramadan, which varies according to the lunar calendar, Muslims abstain from eating or drinking from sunrise to sunset. The effects of Ramadan have been extensively investigated, not only on health outcomes [1, 4–8], but also on exercise performance [9–16]. Moreover, in recent years a focus on other forms of IF, unrelated to religious practice, has emerged. One such form, alter-nate day fasting (ADF; fasting every other day) is organ-ized with alternating “feast days,” on which there is an “ad libitum” energy intake, and “fast days” with reduced or null energy intake.

A growing body of evidence suggests that, in general, IF could represent an useful tool for improving health in general population due to reports of improving blood lipids [17–20] and glycaemic control [3], reducing circu-lating insulin [21], decreasing blood pressure [1, 21–23], decreasing inflammatory markers [7] and reducing fat mass even during relatively short durations (8–12 weeks) [23]. These reported effects are probably mediated through changes in metabolic pathways and cellular processes such as stress resistance [24], lipolysis [3, 17, 25–27], and autophagy [28, 29]. One particular form of IF which has gained great popularity through mainstream media is the so-called time-restricted feeding (TRF). TRF allows subjects to consume ad libitum energy intake within a defined window of time (from 3–4 h to 10–12 h), which means a fasting window of 12–21  h per day is employed. A key point concerning the IF approach is that generally calorie intake is not controlled, but the feeding times are.

In sports, IF is studied mainly in relationship with Ramadan period [9–16], whilst TRF has become very popular among fitness practitioners claiming supposed effects on maintenance of muscle mass and fat loss. Very limited scientific information is available about TRF and athletes, and mixed results have been reported [22, 30, 31]. We demonstrated very recently [30] that TRF did not affect total body composition nor had negative effects on muscle cross-sectional area after 8  weeks in young previously-untrained men performing resistance training, despite a reported reduction in energy intake of ~650 kcal per fasting day in the TRF group. Thus the aim of the present study was to investigate the effects of an isoenergetic TRF protocol on body composition, ath-letic performance, and metabolic factors during resist-ance training in healthy resistance trained males. We hypothesized that the TRF protocol would lead to greater fat loss and improvements in health-related biomarkers as compared to a typical eating schedule.

MethodsSubjectsThirty-four resistance-trained males were enrolled through advertisements placed in Veneto region’s gyms.

Resting energy expenditure was unchanged, but a significant decrease in respiratory ratio was observed in the TRF group.

Conclusions: Our results suggest that an intermittent fasting program in which all calories are consumed in an 8-h window each day, in conjunction with resistance training, could improve some health-related biomarkers, decrease fat mass, and maintain muscle mass in resistance-trained males.

Keywords: Intermittent fasting, Time-restricted feeding, Resistance training, Body composition, Body builders, Fasting

Page 4: Intermittent Fasting: Weight Loss and Cardiovascular Effects · nate day fasting (ADF; fasting every other day) is organ-ized with alternating “feast days,” on which there is

Page 3 of 10Moro et al. J Transl Med (2016) 14:290

The criteria for entering the study were that subjects must have performed resistance training continuously for at least 5 years (training 3–5 days/week with at least 3  years experience in split training routines), be pres-ently engaged in regular resistance training at the time of recruitment, be life-long steroid free, and have no clinical problems that could be aggravated by the study procedures.

Fifty-three subjects responded to the advertisement, but 7 were excluded for previous use of anabolic ster-oids, and 12 declined participation after explanation of study’s protocol. Therefore, 34 subjects (age 29.21 ± 3.8; weight 84.6 ± 6.2 kg) were randomly assigned to a time-restricted feeding group (TRF; n = 17) or standard diet group (ND; n =  17) through computer-generated soft-ware. The research staff conducting outcome assessments was unaware of the assignment of the subjects (i.e. a sin-gle blind design). Anthropometric baseline character-istics of subjects are shown in Table  1. All participants read and signed an informed consent document with the description of the testing procedures approved by the ethical committee of the Department of Biomedical Sci-ences, University of Padova, and conformed to standards for the use of human subjects in research as outlined in the current Declaration of Helsinki.

DietDietary intake was measured by a validated 7-day food diary [32–34], which has been used in previous stud-ies with athletes [35], and analysed by nutritional soft-ware (Dietnext®, Caldogno, Vicenza, Italy). Subjects were instructed to maintain their habitual caloric intake, as measured during the preliminary week of the study (Table  2). During the 8-week experimental period, TRF subjects consumed 100  % of their energy needs divided into three meals consumed at 1 p.m., 4 p.m. and 8 p.m., and fasted for the remaining 16  h per 24-h period. ND group ingested their caloric intake as three meals con-sumed at 8 a.m., 1 p.m. and 8 p.m. This meal timing was chosen to create a balanced distribution of the three meals during the feeding period in the TRF protocol, while the schedule for the ND group maintained a nor-mal meal distribution (breakfast in the morning, lunch at 1 p.m. and dinner at 8 p.m.). The distribution of calories was 40, 25, and 35 % at 1 p.m., 4 p.m. and 8 p.m. respec-tively for TRF, while ND subjects consumed 25, 40 and 35 % of daily calories at 8 a.m., 1 p.m. and 8 p.m. respec-tively. The specific calorie distribution was assigned by a nutritionist and was based on the reported daily intake of each subject.

ND subjects were instructed to consume the entire breakfast meal between 8  a.m. and 9  a.m., the entire lunch meal between 1 p.m. and 2 p.m., and the entire din-ner meal between 8 p.m. and 9 p.m. TRF subjects were instructed to consume the first meal between 1  p.m. and 2 p.m., the second meal between 4 p.m. and 5 p.m., and the third meal between 8 p.m. and 9 p.m. No snacks between the meals were allowed except 20 g of whey pro-teins 30 min after each training session. Every week, sub-jects were contacted by a dietician in order to check the adherence to the diet protocol. The dietician performed a structured interview about meal timing and composition to obtain this information.

Table 1 Subject characteristics at baseline

Results presented as mean ± SD. Results are not statistically significantly different

TRF ND

Age 29.94 ± 4.07 28.47 ± 3.48

Weight (kg) 83.9 ± 12.8 85.3 ± 13

Height (cm) 178 ± 5 177 ± 4

FM (kg) 10.9 ± 3.5 11.3 ± 4.5

FFM (kg) 73.1 ± 5.7 73.9 ± 3.9

Table 2 Diet composition and  macronutrients distribution at  basal level and  during the experimental period in  both groups

Results presented as mean ± SD. No significant differences were detected between groups and within groups

TRF basal TRF exp ND basal ND exp

Total (kcal/day) 2826 ± 412.3 2735 ± 386 3007 ± 444.7 2910 ± 376.4

Carbohydrates (kcal/day) 1503.4 ± 225.95 1400.3 ± 118.8 1654 ± 222.4 1609.2 ± 201.5

Fat (kcal/day) 698 ± 178.5 683.8 ± 61.6 728.7 ± 195 647.7 ± 183.4

Protein (kcal/day) 624.5. ± 59.5 650.3. ± 62.5 637 ± 72.9 643.1 ± 69.3

% Carbohydrates 53.2 ± 1.4 51.2 ± 3.6 54.7 ± 2.2 55.3 ± 4.2

% Fat 24.7 ± 3.1 25 ± 2.8 23.9 ± 3.5 22.6 ± 3.2

% Protein 22.1 ± 2.6 23.8 ± 3.1 21.4 ± 1.8 22.1 ± 3.2

Protein (g/kgbw) 1.86 ± 0.2 1.93 ± 0.3 1.9 ± 0.3 1.89 ± 0.4

Page 5: Intermittent Fasting: Weight Loss and Cardiovascular Effects · nate day fasting (ADF; fasting every other day) is organ-ized with alternating “feast days,” on which there is

Page 4 of 10Moro et al. J Transl Med (2016) 14:290

TrainingTraining was standardized for both groups, and all sub-jects had at least 5 years of continuous resistance train-ing experience prior to the study. Training consisted of 3 weekly sessions performed on non-consecutive days for 8 weeks. All participants started the experimental proce-dures in the months of January or February 2014.

The resistance training program consisted of 3 differ-ent weekly sessions (i.e. a split routine): session A (bench press, incline dumbell fly, biceps curl), session B (mili-tary press, leg press, leg extension, leg curl), and session C (wide grip lat pulldown, reverse grip lat pulldown and tricep pressdown). The training protocol involved 3 sets of 6–8 repetitions at 85–90  % 1-RM, and repetitions were performed to failure (i.e. the inability to perform another repetition with correct execution) with 180 s of rest between sets and exercises [36]. The technique of training to muscular failure was chosen because it is one of the most common practices for body builders, and it was a familiar technique for the subjects. As expected, the muscle action velocity varied between subjects due to their different anatomical leverage. Although there was slight variation of repetition cadence for each subject, the average duration of each repetition was approximately 1.0 s for the concentric phase and 2.0 s for the eccentric phase [37].

The research team directly supervised all routines to ensure proper performance of the routine. Each week, loads were adjusted to maintain the target repetition range with an effective load. Training sessions were per-formed between 4:00 and 6:00 p.m. Subjects were not allowed to perform other exercises other than those included in the experimental protocol.

MeasurementsBody weight was measured to the nearest 0.1 kg using an electronic scale (Tanita BWB-800 Medical Scales, USA), and height to the nearest 1  cm using a wall-mounted Harpenden portable stadiometer (Holtain Ltd, UK). Body mass index (BMI) was calculated in kg/m2. Fat mass and fat-free mass were assessed by dual energy X-ray absorp-tiometry (DXA) (QDR 4500  W, Hologic Inc., Arling-ton, MA, USA). Muscle areas were calculated using the following anthropometric system. We measured limb circumferences to the nearest 0.001 m using an anthro-pometric tape at the mid-arm and mid-thigh. We also measured biceps, triceps, and thigh skinfolds to the near-est 1 mm using a Holtain caliper (Holtain Ltd, UK). All measurements were taken by the same operator (AP) before and during the study according to standard pro-cedures [38, 39]. Muscle areas were then calculated using a previously [40] validated software (Fitnext®, Caldogno, Vicenza, Italy). Cross-sectional area (CSA) measured

with Fitnext® has an r2 = 0.88 compared to magnetic res-onance and an ICC of 0.988 and 0.968 for thigh and arm, respectively [40–42].

Ventilatory measurements were made by standard open-circuit calorimetry (max Encore 29 System, Vmax, Viasys Healthcare, Inc., Yorba Linda, CA, USA) with breath-by-breath modality. The gas analysis system was used: Oxygen uptake and carbon dioxide output val-ues were measured and used to calculate resting energy expenditure (REE) and respiratory ratio (RR) using the modified Weir equation [43]. Before each measurement, the calorimeter was warmed according to the manufac-turer’s instructions and calibrated with reference gases of known composition prior to each participant.

Oxygen uptake was measured (mL/min) and also nor-malized to body weight (mL/kg/min), and the respira-tory ratio was determined. After resting for 15 min, the data were collected for 30 min, and only the last 20 min were used to calculate the respiratory gas parameters [37, 44]. All tests were performed in the morning between 6 and 8 a.m. while the subjects were supine. The room was dimly lit, quiet, and approximately 23  °C. Subjects were asked to abstain from caffeine, alcohol consumption and from vigorous physical activity for 24 h prior to the measurement.

Blood collection and analysis protocolBlood samples taken from the antecubital vein at base-line and after 8 weeks were collected in BD Vacutainers Tubes (SST™ II Advance, REF 367953). Samples were centrifuged (4000 RPM at 4  °C using centrifuge J6-MC by Beckman), and the resultant serum was aliquoted and stored at −80  °C. All samples were analysed in the same analytical session for each test using the same rea-gent lot. Before the analytical session, the serum sam-ples were thawed overnight at 4  °C and then mixed. Interleukin-6 (IL-6), tumor necrosis factor-α (TNF-α), and interleukin-1β (IL-1β) were measured using Quan-tikine HS Immunoassay Kit (R&D Systems, Minneapo-lis, MN, USA). The inter-assay coefficient of variations (CVs) were 3.5–6.2 and 3.2–6.3  % for IL-6, TNF-α and IL-1β respectively. Insulin-like growth factor 1 (IGF-1) was measured using the analyzer Liaison XL (DiaSorin S.p.A, Vercelli-Italy). This test is a sandwich immunoas-say based on a chemiluminescent revelation, and the CV for IGF-1 was between 5.6 and 9.6 %; the reference range for this test depends on age and gender. Fasting total cho-lesterol, high-density lipoprotein cholesterol (HDL-C), low-density lipoprotein cholesterol (LDL-C), and triglyc-erides (TG) were measured by an enzymatic colorimet-ric method using a Modular D2400 (Roche Diagnostics, Basel, Switzerland). LDL-C fraction was calculated from Friedewald’s formula: LDL-C = TC − HDL-C − (TG/5).

Page 6: Intermittent Fasting: Weight Loss and Cardiovascular Effects · nate day fasting (ADF; fasting every other day) is organ-ized with alternating “feast days,” on which there is

Page 5 of 10Moro et al. J Transl Med (2016) 14:290

The inter-assay CVs for total cholesterol, HDL-C, and triacylglycerol concentrations were 2.9, 1.8, and 2.4  %, respectively. Glucose was measured in triplicate by the glucose oxidase method (glucose analyzer, Beckman Instruments, Palo Alto, CA, USA), with a CV of 1.2  %. Leptin and adiponectin were measured by radioim-munoassay using commercially available kits (Leptin: Mediadiagnost; Adiponectin: DRG Diagnostic); insulin was measured with a chemiluminescent immunoassay (Siemens Immulite 2000). The assay sensitivity was 1 ng/mL, and inter- and intra-assay CVs were less than 10, 5, and 6 % for leptin, adiponectin, and insulin, respectively. Thyroid-stimulating hormone (TSH), free thyroxine (T4), and free triiodothyronine (T3) were measured by auto-mated chemiluminescence methods (ACS 180 SE; Bayer, Milan, Italy). Plasma testosterone was determined using Testosterone II (Roche Diagnostics, Indianapolis, IN, USA) performed on Modular Analytics E 170 analyzer with electrochemiluminescent detection.

Strength testsOne repetition maximum (1-RM) for the leg press and the bench press exercises was measured on separate days. Subjects executed a specific warm-up for each 1-RM test by performing 5 repetitions with a weight they could nor-mally lift 10 times. Using procedures described elsewhere [45], the weight was gradually increased until failure occurred in both of the exercises tested. The greatest load lifted was considered the 1-RM. Previously published ICCs for test–retest reliability for leg press and bench press 1-RM testing was 0.997 and 0.997, respectively, in men, with a coefficient of variation of 0.235 for LP and 0.290 for BP [46]. 1-RM was also assessed at baseline and after 4 and 8 weeks for all training exercises so that the necessary adjustments for possible strength increases could be made, thus ensuring that subjects continued to train at a relative intensity of 85–90 % of their 1-RM.

Statistical analysisResults are presented as mean ± standard deviation. The sample size was obtained assuming an interaction of a Root Mean Square Standardized Effect (RMSSE) of 0.25 with a fixed power of 80 % and an alpha risk of 5 % for the main variable. Through the Shapiro–Wilk’s  W test, we assessed the normality. An independent samples t test was used to test baseline differences between groups. The two-way repeated-measures ordinary ANOVA was performed (using time as the within-subject factor and diet as the between-subject factor) in order to assess dif-ferences between groups over the course of the study. Moreover we adopted a mixed model ANOVA with the fixed variable fat mass expressed in kg as covariate vs Time  *  Diet as random variables. All differences were

considered significant at P < 0.05. Post-hoc analyses were performed using the Bonferroni test. In order to reduce the influence of within group variability a univariate test of significance (ANCOVA) was performed. We fixed as depended variable the Δ pre-post for each group and the baseline values of the outcomes were adopted as covari-ate; IF vs ND were assumed as categorical predictors.

The analysis was performed through STATISTICA software (Vers. 8.0 for Windows, Tulsa, USA) and Prism 5 GraphPad software (Abacus Concepts GraphPad Soft-ware, San Diego, USA).

ResultsAfter 8  weeks, a significant decrease in fat mass was observed in the TRF group (−16.4 vs −2.8  % in ND group), while fat-free mass was maintained in both groups (+0.86 vs +0.64 %). The same trend was observed for arm and thigh muscle cross-sectional area. Leg press maximal strength increased significantly, but no differ-ence was present between treatments. Total testosterone and IGF-1 decreased significantly in TRF after 8  weeks while no significant differences were detected in ND. Blood glucose and insulin levels decreased significantly only in TRF subjects and conformingly a significant improvement of HOMA-IR was detected. In the TRF group, adiponectin increased, leptin decreased (but this was not significant when normalized for fat mass), and T3 decreased significantly compared to ND, without any significant changes in TSH. No significant changes were detectable for lipids (total cholesterol, HDL-c and LDL-c), except for a decrease of TG in TRF group. TNF-α and IL-1β were lower in TRF at the conclusion of the study as compared to ND. A significant decrease of respiratory ratio in TRF group was recorded (Tables 3, 4).

DiscussionFasting is a relatively well-studied metabolic state in sports and physical exercise due to studies of the “Rama-dan” period observed by Muslim athletes [12, 14]. How-ever, only a single study has reported its effect during a resistance training program aimed at achieving skeletal muscle growth [30]. Our data demonstrate that during a RT program, TRF was capable of maintaining mus-cle mass, reducing body fat, and reducing inflammation markers. However, it also reduced anabolic hormones such testosterone and IGF-1.

A key point of the TRF approach utilized in the pre-sent study is that total daily calorie intake remained the same while the frequency of meals (i.e. time between meals) was altered. This is dissimilar to many other IF regimens. There are a number of different IF protocols, most of which have the goal of reducing total energy intake. Additionally, unlike ADF and some other forms of

Page 7: Intermittent Fasting: Weight Loss and Cardiovascular Effects · nate day fasting (ADF; fasting every other day) is organ-ized with alternating “feast days,” on which there is

Page 6 of 10Moro et al. J Transl Med (2016) 14:290

IF, the regimen utilized in the present study employed the same schedule each day, consisting of 16 h fasting and 8 h feeding.

Although IF has received a great amount of attention in recent years, the majority of studies have investigated the effects of IF in overweight, obese or dyslipidemic subjects [19–21, 47–50]. However, little is known about the effects of such nutritional regimens in athletes, and more specifically, in body builders or resistance-trained individuals. The present study provides the first in-depth investigation of IF in this population of athletes. With the exception of reduced triglycerides, our results do not confirm previous research suggesting a positive effect of

IF on blood lipid profiles [17–19, 47, 49, 51, 52], how-ever, it has to be taken into account that our subjects were normolipemic athletes. The magnitude of reduction in triglycerides was also smaller than is typically seen in individuals who have elevated concentrations prior to IF.

As reported, a decrease of fat mass in individuals per-forming IF was observed. Considering that the total amount of kilocalories and the nutrient distribution were not significantly different between the two groups (Table 2), the mechanism of greater fat loss in IF group cannot simply be explained by changes in the quantity or quality of diet, but rather by the different temporal meal distribution. Many biological mechanisms have been

Table 3 Major results of experiment with statistics adopted highlighted in italics text

Results are presented as mean ± SD

n.s. not statistically significantly different (p > 0.05)

IF pre IF post ΔIF t test ND pre ND post ΔND t test 2 Way ANOVA Time * Diet

FFM (kg) 73.08 ± 3.88 73.72 ± 4.27 n.s. 73.93 ± 3.9 74.41 ± 3.59 n.s. n.s.

FM (kg) 10.90 ± 3.51 9.28 ± 2.47 0.0005 11.36 ± 4.5 11.05 ± 4.274 n.s. 0.0448

Arm muscle CSA (cm2) 48.52 ± 3.80 49.37 ± 3.66 n.s. 48.93 ± 4.05 50.17 ± 6.27 n.s. n.s.

Thigh CSA (cm2) 148 ± 34.87 153.77 ± 36.83 n.s. 150.26 ± 22.21 157.35 ± 32.56 n.s. n.s.

Bench press 1-RM (kg) 107.08 ± 18.01 110.36 ± 16.53 n.s. 109.82 ± 14.72 110.57 ± 15.11 n.s. n.s.

Leg press 1-RM (kg) 282.8 ± 30.11 290.00 ± 27.77 n.s. 298.56 ± 25.76 309 ± 68.94 n.s. n.s.

Adiponectin (μg/mL) 11.8 ± 4.3 13.9 ± 3.7 0.0001 10.8 ± 5.5 10.9 ± 4.3 n.s. 0.0000

Leptin (ng/mL) 2.1 ± 0.6 1.8 ± 0.4 0.0002 2.4 ± 0.5 2.3 ± 0.4 n.s. 0.0001

Leptin (ng/mL/kg bw) 0.21 ± 0.07 0.2 ± 0.06 n.s. 0.24 ± 0.11 0.24 ± 0.11 n.s. n.s.

IL-6 (ng/L) 1.33 ± 0.23 1.08 ± 0.22 0.0035 1.24 ± 0.38 1.19 ± 0.33 n.s. n.s.

TNF-α (ng/L) 5.58 ± 0.92 5.13 ± 0.8 0.0001 5.69 ± 0.77 5.86 ± 0.72 n.s. n.s.

IL-1β (ng/L) 0.93 ± 0.19 0.81 ± 0.07 0.0042 0.92 ± 0.12 0.94 ± 0.12 n.s. 0.0235

Testosterone total (nmol/L)

21.26 ± 6.51 16.86 ± 4.25 0.0001 18.60 ± 5.68 18.85 ± 4.57 n.s. 0.0476

IGF-1 (ng/mL) 216.94 ± 49.55 188.90 ± 31.48 0.0109 215.59 ± 56.25 218.41 ± 42,24 n.s. 0.0397

Insulin (mU/mL) 2.78 ± 0.6 1.77 ± 0.9 0.0303 2.56 ± 0.5 2.22 ± 0.4 n.s. n.s.

TSH (mUI/L) 1.28 ± 0.6 1.27 ± 0.7 n.s. 1.30 ± 0.8 1.31 ± 0.6 n.s. n.s.

T3 (ng/dL) 83.21 ± 17.23 74.32 ± 26.66 0.0001 81.12 ± 20.00 82.35 ± 25.55 n.s. n.s.

Glucose (mg/dL) 96.64 ± 5.1 85.92 ± 7.13 0.0011 95.21 ± 47.77 96.02 ± 65.32 n.s. n.s.

Total cholesterol (mg/dL)

193.45 ± 6.6 191.37 ± 11.2 n.s. 196.33 ± 9.93 197.12 ± 15.66 n.s. n.s.

Cortisol (ng/mL) 174.25 ± 56.78 186.05 ± 68.5 n.s. 191.24 ± 70.34 185.78 ± 65.89 n.s. n.s.

HDL-c (mg/dL) 54.11 ± 5.89 58.06 ± 6.11 0.0142 53.33 ± 9.67 54.12 ± 9.9 n.s. n.s.

LDL-c (mg/dL) 114.58 ± 11.33 110.26 ± 12.27 n.s. 115.58 ± 9.9 116.08 ± 11.56 n.s. n.s.

TG (mg/dL) 123.78 ± 15.12 115.23 ± 11.77 0.0052 137.10 ± 16.98 134.58 ± 15.66 n.s 0.0201

REE (kcal/day) 1880 ± 94.15 1891 ± 100.56 n.s. 1901 ± 88.76 1895 ± 93.56 n.s. n.s.

RR 0.83 ± 0.02 0.81 ± 0.01 0.0421 0.83 ± 0.03 0.83 ± 0.02 n.s. n.s.

Mixed model ANOVA with FM as covariate

Adiponectin (μg/mL) 11.8 ± 4.3 13.9 ± 3.7 10.8 ± 5.5 10.9 ± 4.3 0.0000

Leptin (ng/mL) 2.1 ± 0.6 1.8 ± 0.4 2.4 ± 0.5 2.3 ± 0.4 0.0002

Leptin (ng/mL/kg bw) 0.21 ± 0.07 0.2 ± 0.06 0.24 ± 0.11 0.24 ± 0.11 0.0135

IL-1β (ng/L) 0.93 ± 0.19 0.81 ± 0.07 0.92 ± 0.12 0.94 ± 0.12 0.0224

Page 8: Intermittent Fasting: Weight Loss and Cardiovascular Effects · nate day fasting (ADF; fasting every other day) is organ-ized with alternating “feast days,” on which there is

Page 7 of 10Moro et al. J Transl Med (2016) 14:290

advocated to explain these effects. One is the increase of adiponectin that interacts with adenosine 5′-monophos-phate-activated protein kinase (AMPK) and stimulates Peroxisome proliferator-activated receptor gamma coac-tivator 1-alpha (PGC-1α) protein expression and mito-chondrial biogenesis. Moreover, adiponectin acts in the brain to increase energy expenditure and cause weight loss [53]. It is notable that in the present study, the dif-ferences in adiponectin between groups remained even when normalized relative to body fat mass, whereas the significant decrease of leptin (that might be considered a unfavorable factor for fat loss) was no longer significant when normalized for fat mass. Other hypothesis is an enhanced thermogenic response to epinephrine [54] or an increase in REE [55] after brief periods of fasting, but our preliminary data didn’t support this point.

Interestingly, although reductions in the anabolic hor-mones testosterone and IGF-1 were observed, this did not correspond to any deleterious body composition changes or compromises of muscular strength over the duration of the study. It has been previously reported that men performing caloric restriction have lower testoster-one than those consuming non-restricted Western diets [56], however, the present experiment did not restrict calories in the IF group. In animal models, IF influences the hypothalamo-hypophysial-gonadal axis and testos-terone concentration probably through a decrease in leptin-mediated effects [57], but it must be considered that mice on a an every-other-day feeding regimen con-sume about 30–40 % less calories over time compared to free feeding animals and that in our study, no differences in leptin concentration were seen when normalized for fat mass. Also, the reduction of IGF-1 in the TRF group deserves some discussion. A previous study by Bohulel et al. [11] reported no changes in the GH/IGF-1 during Ramadan intermittent fasting. Even though it is plausible

that IF mimics caloric restriction through common path-ways (e.g. AMPK/ACC) (adenosine 5′-monophosphate-activated protein kinase/acetyl-CoA-carboxylase) [58], recent data on humans showed no influences of caloric restriction on IGF-1 [59, 60]. It is possible that the increase of adiponectin and the decrease of leptin could influence the IGF-1 concentration, even though it is unclear to what extent changes in adipokines impact cir-culating IGF-1 levels following weight loss [59].

Previous studies have reported mixed results concern-ing the ability to maintain lean body mass during IF, but the vast majority of these studies imposed calorie restric-tion and did not utilize exercise interventions [22]. In our study, the nutrient timing related to training session was different between the two groups, and this could affect the anabolic response of the subjects [61] even though these effects are still unclear [62]. However, we did not find any significant differences between groups in fat-free mass, indicating that the influence of nutrient timing may be negligible when the overall content of the diet is similar.

There is an increasing amount of data suggesting that IF could potentially be a feasible nutritional scheme to combat certain diseases. In the present study, both blood glucose and insulin concentrations decreased in the IF group. The potential of IF to modulate blood glucose and insulin concentrations has previously been discussed, but primarily in the context of overweight and obese indi-viduals [3]. The concurrent increase in adiponectin and decrease in insulin may be related to modulation of insu-lin sensitivity, as adiponectin concentrations have been positively correlated with insulin sensitivity [21, 50, 63, 64]. Moreover, related to the well-known anti-inflamma-tory effect of adiponectin, it is possible that the reduction of inflammatory markers is related to the improvement of insulin sensitivity. Inflammation plays an pivotal role

Table 4 Univariate tests of significance (ANCOVA)

The Δ pre–post for each depended variable group were considered and the baseline values of the outcomes were adopted as covariate; TRF vs ND were assumed as categorical predictors

Univariate tests of significance (ANCOVA)

Observations Dependent variables (pre–post Δ)

Categorical predictors (ΔIF vs ΔND)

Covariates (baseline values)

P values

Body weight −0.40 ± 1.76 −0.97 ± 1.58 vs 0.16 ± 1.78 84.63 ± 6.17 0.0354

FM (kg) −0.96 ± 1.72 −1.61 ± 1.53 vs −0.30 ± 1.70 11.12 ± 3.98 0.0070

Adiponectin (μg/mL) −0.45 ± 3.07 2.04 ± 1.52 vs −2.94 ± 1.97 12.83 ± 1.98 0.0000

Leptin (ng/mL) 0.09 ± 0.67 −0.36 ± 0.31 vs 0.54 ± 0.64 1.97 ± 0.52 0.0000

IL-6 (ng/L) −0.15 ± 0.27 −0.25 ± 0.30 vs −0.04 ± 0.20 1.28 ± 0.31 0.0378

TNF-α (ng/L) −0.14 ± 0.47 −0.45 ± 0.37 vs 0.17 ± 0.34 5.63 ± 0.83 0.0000

IL-1β (ng/L) −0.05 ± 0.14 −0.12 ± 0.15 vs 0.02 ± 0.09 0.92 ± 0.15 0.0000

Testosterone total (nmol/L) −2.07 ± 3.60 −4.40 ± 3.02 vs 0.25 ± 2.43 19.93 ± 6.00 0.0000

IGF-1 (ng/mL) −12.38 ± 33.04 −28.00 ± 40.11 vs 3.23 ± 11.13 216.32 ± 38.84 0.0003

Page 9: Intermittent Fasting: Weight Loss and Cardiovascular Effects · nate day fasting (ADF; fasting every other day) is organ-ized with alternating “feast days,” on which there is

Page 8 of 10Moro et al. J Transl Med (2016) 14:290

in insulin resistance development through different cytokines that influence numerous molecular pathways. For example, insulin resistance could be triggered by TNF-α via JNK and IKKβ/NF-κB (jun amino-terminal kinase/inhibitor of NF-κβ kinase) pathways, which may increase serine/threonine phosphorylation of insu-lin receptor substrate 1. Moreover IL-6 could decrease insulin sensitivity in skeletal muscle by inducing toll-like receptor-4 (TLR-4) gene expression through STAT3 (activator of transcription 3) activation. This relation-ship is potentially bidirectional as the activation of IKKβ/NF-κB signalling could, in turn, stimulate the production of TNF-α [65]. Modulation of some of these inflamma-tory markers by IF was seen in the present study: TNF-α and IL-1β were lower in the TRF group than ND at the conclusion of the study, while IL-6 appeared to decrease in the TRF group, but was not significantly different from ND. Previous information on the impact of IF on inflam-matory markers is limited, but a previous investigation by Halberg et al. [66] reported no changes in TNF-α or IL-6 after two weeks of modified IF in a small sample of healthy young men.

Although a reduction in T3 was observed in the IF group, no changes in TSH or resting energy expenditure were observed. The observed reduction in RR in the TRF group indicates a very small shift towards reliance on fatty acids for fuel at rest, although a significant statistical interaction for RR was not present. Fasting RR has been previously reported to be a predictor of substantial future weight gain in non-obese men, with individuals who have higher fasting RR being more likely to gain weight [67]. Interestingly, it was reported by Seidell et  al. [67] that although RR was related to future weight gain, RMR was not. It should be noted that individuals with the highest risk of future weight gain had fasting RR > 0.85 (as com-pared to individuals who had RR < 0.76). In the present study, the RR at the end of the study in both the TRF group and ND group do not directly fall into either of these categories (RR = 0.81 and 0.83, respectively).

Based on the present study, a modified IF protocol (i.e. TRF) could be feasible for strength athletes without negatively affecting strength and muscle mass. Interest-ingly, even though androgen concentrations were low-ered by TRF, there was no difference in muscle mass changes between groups (+0.64  kg in TRF vs +0.48  kg in ND). Caloric restriction in rodents has been reported to decrease testosterone and IGF-1 even though human data on long-term severe caloric restriction does not demonstrate a decrease in IGF-1 levels, but instead an increased serum insulin-like growth factor binding pro-tein 1 (IGFBP-1) concentration [60, 68]. However, no data are available for most forms of IF. Decrease the activity of the IGF-1 axis could be a desirable target for reducing

cancer risk [69], but it is also well known that the activa-tion of the IGF-1/AKT/mTOR (insulin-like growth fac-tor-1/protein kinase B/mammalian target of rapamycin) pathway is one of the keys for muscular growth. In addi-tion to altering IGF-1, fasting can promote autophagy [28], which is important for optimal muscle health [70]. Additionally, there is a possibility that the different eating patterns of the groups in the present study impacted the relative contributions of different hypertrophic pathways in each group.

Some limitations of the present study should be taken into account. One is the different timing of meals in rela-tionship to the training sessions that could have affected the subjects’ responses. On this point, there is not a consensus among researchers. The beneficial effects of pre-exercise essential amino acid-carbohydrate sup-plement have been suggested [61], but the same group found that ingesting 20  g of whey protein either before or 1  h after 10 sets of leg extension resulted in simi-lar rates of AA uptake [62]. Additionally, other studies have reported no benefit with pre-exercise AA feeding [71, 72]. Another limitation of the present study is that the energy and macronutrient composition of the diet was based on interview, and this approach has known weaknesses. Because of the limitations of this method, it is possible that differences in energy or nutrient intake between groups could have existed and played a role in the observed outcomes.

ConclusionsIn conclusion, our results suggest that the modified IF employed in this study: TRF with 16  h of fasting and 8 h of feeding, could be beneficial in resistance trained individuals to improve health-related biomarkers, decrease fat mass, and at least maintain muscle mass. This kind of regimen could be adopted by athletes dur-ing maintenance phases of training in which the goal is to maintain muscle mass while reducing fat mass. Additional studies are needed to confirm our results and to investigate the long-term effects of IF and peri-ods after IF cessation.

AbbreviationsIF: intermittent fasting; TRF: time-restricted feeding; ND: normal diet; ADF: alternate day fasting; IL-6: interleukin-6; TNF-α: tumor necrosis factor-α; IL-1β: interleukin-1β; IGF-1: insulin-like growth factor-1; HDL-C: high-density lipopro-tein cholesterol; LDL-C: low-density lipoprotein cholesterol; TG: triglycerides; TSH: thyroid-stimulating hormone; T4: free thyroxine; T3: free triiodothyronine; 1-RM: one repetition maximum; REE: resting energy expenditure; RR: respira-tory ratio; ACC: acetyl-CoA-carboxylase; AMPK: adenosine 5′-monophosphate-activated protein kinase; PGC-1α: Peroxisome proliferator-activated receptor gamma coactivator 1-alpha; HOMA-IR: homeostasis model assessment–insu-lin-resistance; mTOR: mammalian target of rapamycin; AKT: protein kinase B; IGFBP-1: insulin-like growth factor binding protein 1; JNK: jun amino-terminal kinase; IKKβ/NF-κB: inhibitor of NF-κβ kinase; STAT3: activator of transcription 3; TLR-4: toll-like receptor-4.

Page 10: Intermittent Fasting: Weight Loss and Cardiovascular Effects · nate day fasting (ADF; fasting every other day) is organ-ized with alternating “feast days,” on which there is

RESEARCH Open Access

Intermittent fasting combined with calorierestriction is effective for weight loss andcardio-protection in obese womenMonica C Klempel, Cynthia M Kroeger, Surabhi Bhutani, John F Trepanowski and Krista A Varady*

Abstract

Background: Intermittent fasting (IF; severe restriction 1 d/week) facilitates weight loss and improves coronaryheart disease (CHD) risk indicators. The degree to which weight loss can be enhanced if IF is combined with calorierestriction (CR) and liquid meals, remains unknown.

Objective: This study examined the effects of IF plus CR (with or without a liquid diet) on body weight, bodycomposition, and CHD risk.

Methods: Obese women (n = 54) were randomized to either the IFCR-liquid (IFCR-L) or IFCR-food based (IFCR-F)diet. The trial had two phases: 1) 2-week weight maintenance period, and 2) 8-week weight loss period.

Results: Body weight decreased more (P = 0.04) in the IFCR-L group (3.9 ± 1.4 kg) versus the IFCR-F group (2.5 ±0.6 kg). Fat mass decreased similarly (P < 0.0001) in the IFCR-L and IFCR-F groups (2.8 ± 1.2 kg and 1.9 ± 0.7 kg,respectively). Visceral fat was reduced (P < 0.001) by IFCR-L (0.7 ± 0.5 kg) and IFCR-F (0.3 ± 0.5 kg) diets. Reductionsin total and LDL cholesterol levels were greater (P = 0.04) in the IFCR-L (19 ± 10%; 20 ± 9%, respectively) versus theIFCR-F group (8 ± 3%; 7 ± 4%, respectively). LDL peak particle size increased (P < 0.01), while heart rate, glucose,insulin, and homocysteine decreased (P < 0.05), in the IFCR-L group only.

Conclusion: These findings suggest that IF combined with CR and liquid meals is an effective strategy to helpobese women lose weight and lower CHD risk.

Keywords: Intermittent fasting, Calorie restriction, Liquid diet, Body weight, Visceral fat, Cholesterol, Coronary heartdisease, Obese women

IntroductionCoronary heart disease (CHD) remains the number oneof killer of women in the United States [1]. Weight gainduring adulthood increases the risk of CHD [1]. Epi-demiological evidence suggests that a modest reductionin weight (i.e. 5% of body weight) in female subjectsreduces the incidence and progression of CHD [2]. Al-though several diet strategies exist to help individualslose weight, one regimen that has gained considerablepopularity in the past decade is intermittent fasting (IF)[3]. This diet strategy generally involves severe restric-tion (75-90% of energy needs) on 1 or 2 days per week.Results from a recent 24-week randomized clinical trial

revealed that IF can reduce body weight by 7% in obesewomen [4]. LDL cholesterol and triglyceride concentra-tions also decreased by 10% and 17%, respectively, inthese subjects [4]. Though these findings are promising,this regimen is limited in that a long duration of time(i.e. 24 weeks) is required to experience only modestreductions in body weight. One possible method ofaugmenting the rate of weight loss is to combine IFwith a daily calorie restriction (CR) protocol. In thisway, the individual would fast one day per week, andthen undergo mild CR (i.e. 20% restriction of energyneeds) on 6 days per week. This combination therapy(IF plus CR) produces greater reductions in weightand superior changes in CHD risk parameters whencompared to each intervention alone in animal models[5]. Whether the beneficial effects of IF plus CR on

* Correspondence: [email protected] of Kinesiology and Nutrition, University of Illinois at Chicago,1919 West Taylor Street, Room 506F, Chicago, IL 60612, USA

© 2012 Klempel et al.; licensee BioMed Central Ltd. This is an Open Access article distributed under the terms of the CreativeCommons Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, andreproduction in any medium, provided the original work is properly cited.

Klempel et al. Nutrition Journal 2012, 11:98http://www.nutritionj.com/content/11/1/98

Page 11: Intermittent Fasting: Weight Loss and Cardiovascular Effects · nate day fasting (ADF; fasting every other day) is organ-ized with alternating “feast days,” on which there is

weight and CHD risk can be reproduced in humansubjects has yet to be elucidated.It is well known that many obese subjects are unable

to adequately estimate portion sizes [6]. This inability toestimate portions during periods of dieting results in ex-cessive food intake, which then blunts overall weight loss[6]. In order to take the guesswork out of estimatingportion sizes, some CR protocols implement portioncontrolled liquid meals to replace 1 or 2 meals per day.When liquid meal replacements are employed, subjectstend to lose greater amounts of weight when comparedto subjects receiving diet counseling alone [7,8].Whether or not the implementation of liquid meal repla-cements during periods of IF plus CR accelerates weightloss remains unknown.Accordingly, the objective of the present study was to

examine the effects of an IF protocol combined with CRon body weight, body composition, and CHD risk factorsin obese women. Whether the addition of liquid mealreplacements to this protocol would result in greaterweight loss and more pronounced CHD risk reductionwas also assessed.

MethodsSubjectsSubjects were recruited by means of advertisementsplaced on and around the University of Illinois campus

in downtown Chicago. A total of 77 individualsresponded to the advertisements, but only 60 weredeemed eligible to participate after the preliminary ques-tionnaire, body mass index (BMI) and waist circumfer-ence assessment (Figure 1). Key inclusion criteria wereas follows: female, age 35–65 y, BMI between 30 and39.9 kg/m2, waist circumference >88 cm, weight stablefor 3 months prior to the beginning of the study (i.e.<5 kg weight loss or gain), non-diabetic, no history ofcardiovascular disease, no history of cancer, sedentary orlightly active for 3 months prior to the beginning of thestudy (i.e. <3 h/week of light-intensity exercise at 2.5–4.0 metabolic equivalents (METS)), non-smoker, notclaustrophobic, and not taking weight loss, lipid-lower-ing, or glucose-lowering medications. Peri-menopausalwomen were excluded from the study, and post-menopausal women (defined as absence of menses for2 y) were required to maintain their current hormonereplacement therapy regimen for the duration of thestudy. The experimental protocol was approved by theOffice for the Protection of Research Subjects at the Uni-versity of Illinois, Chicago, and all volunteers gavewritten informed consent to participate in the trial.

Experimental designRandomization was performed by way of a stratified ran-dom sample. The sample frame was divided into strata

Figure 1 Study flow chart. IFCR-L: Intermittent fasting calorie restriction-liquid diet; IFCR-F: Intermittent fasting calorie restriction-food based diet.

Klempel et al. Nutrition Journal 2012, 11:98 Page 2 of 9http://www.nutritionj.com/content/11/1/98

Page 12: Intermittent Fasting: Weight Loss and Cardiovascular Effects · nate day fasting (ADF; fasting every other day) is organ-ized with alternating “feast days,” on which there is

based on BMI and age. Subjects from each stratum werethen randomly assigned to either the IFCR-liquid diet(IFCR-L) group (n = 30) or IFCR-food based diet (IFCR-F)group (n = 30). The 10-week trial consisted of two dietaryphases: 1) a 2-week baseline weight maintenance period,and 2) an 8-week weight loss period.

Diet protocolBaseline weight maintenance diet (Week 1–2)Before commencing the 8-week weight loss intervention,each subject participated in a 2-week baseline weightmaintenance period. During this period, subjects wererequested to maintain a stable weight and continue eat-ing their usual diet.

Weight loss diets (Week 3–10)Following the baseline period, subjects participated inthe IFCR-L or IFCR-F protocol for 8 weeks. Energyrequirements were measured using the Mifflin equation[9]. IFCR-L subjects (n = 30) consumed a calorie-restricted liquid diet for the first 6 days of the week, andthen underwent a fast on the last day of the week (waterconsumption + 120 kcal of juice powder only, for 24 h).The liquid diet (during the CR period) consisted of a li-quid meal replacement for breakfast (240 kcal) and a li-quid meal replacement for lunch (240 kcal). All liquidmeal replacements were provided to the subjects inpowder-form in premeasured packets (Isalean Shake,Isagenix LLC, Chandler, AZ). At dinnertime, IFCR-Lsubjects consumed a 400–600 kcal meal. Food was notprovided to the subjects for the dinnertime meal. In-stead, subjects met with a Registered Dietician weekly tolearn how to make healthy eating choices that are incompliance with the National Cholesterol EducationProgram Therapeutic Lifestyle Changes (TLC) diet (i.e.<35% of kcal as fat; 50-60% of kcal as carbohydrates;<200 mg/d of dietary cholesterol; and 20–30 g/d offiber). In following this regimen, each subject was calorierestricted by 30% of their baseline needs. IFCR-F sub-jects (n = 30) consumed a calorie-restricted food-baseddiet for the first 6 days of the week, and then underwenta fast on the last day of the week (water consumption +120 kcal of juice powder only, for 24 h). IFCR-F subjectsate 3 meals per day in accordance with the TLC dietguidelines. Food was not provided to the subjects. In-stead, subjects met with a Registered Dietician weekly tolearn how to make healthy eating choices that were incompliance with the TLC diet. Subjects were instructedto eat approximately 240 kcal for breakfast, 240 kcal forlunch, and 400–600 kcal for dinner. In following thisregimen, each IFCR-F subject was calorie restricted by30% of their baseline needs.

AnalysesAdherence with dietsA 7-d food record was used to assess adherence to thediets. Subjects completed the food records at week 3and 10. The Registered Dietician provided 15 min of in-struction to each participant on how to complete thefood records. These instructions included verbal infor-mation and detailed reference guides on how to estimateportion sizes and record food items in sufficient detail toobtain an accurate estimate of dietary intake. All dietaryinformation from the food records was entered into thefood analysis program, Nutritionist Pro (Axxya Systems,Stafford, TX) by a single trained operator to alleviateinter-investigator bias. In addition, IFCR-L subjects wereprovided with a checklist each day to monitor: 1) adher-ence to the liquid meal protocol, and 2) adherence tothe fast day regimen. IFCR-F subjects were also given achecklist to monitor their adherence to the fast day regi-men. If an IFCR-L or IFCR-F subject consumed morethat 100 kcal on the fast day (above the 120 kcal of juicepowder allotted to the subject), the subject was consid-ered non-adherent with the fast day protocol.

Maintenance of physical activity habitsChanges in daily energy expenditure associated withalterations in physical activity habits were quantified bythe use of a pattern recognition monitor (Sense WearMini (SWM), Bodymedia, Pittsburgh, PA). Subjects worethe lightweight monitor on their upper arm for 7 d atweek 3 and 10 of the trial. The SWM is a wireless multi-sensor activity monitor that integrates motion data froma triaxial accelerometer along with several other physio-logical sensors (heat flux, skin temperature and galvanicskin response). The data was analyzed using BodymediaSoftware V.7.0, algorithm V.2.2.4 (Bodymedia, Pitts-burgh, PA).

Body weight and body composition assessmentBody weight measurements were taken to the nearest0.5 kg at the beginning of each week in light clothingand without shoes using a balance beam scale (Health-OMeter; Sunbeam Products, Boca Raton, FL). Heightwas assessed using a wall-mounted stadiometer to thenearest 0.1 cm. BMI was assessed as kg/m2. Fat massand fat free mass were assessed by dual energy X-ray ab-sorptiometry (DXA) at weeks 1, 3 and 10 (QDR 4500 W,Hologic Inc. Arlington, MA). Waist circumference wasmeasured by a flexible tape to the nearest 0.1 cm, mid-way between the lower costal margin and super iliaccrest during a period of expiration. Abdominal visceraladipose tissue and subcutaneous adipose tissue volumeswere quantified using magnetic resonance imaging(MRI) at week 3 and 10 [10]. Images were acquiredusing a 1.5-T superconducting magnet (Siemens, Iselin,

Klempel et al. Nutrition Journal 2012, 11:98 Page 3 of 9http://www.nutritionj.com/content/11/1/98

Page 13: Intermittent Fasting: Weight Loss and Cardiovascular Effects · nate day fasting (ADF; fasting every other day) is organ-ized with alternating “feast days,” on which there is

NJ), and were obtained every 1 cm from the 9th thoracicvertebra (T9) to the first sacral vertebra (S1). Imagelocations were defined relative to a common anatomicallandmark, the L4–L5 intervertebral space. To facilitatethe comparison of individual image data, we limited ouranalyses to the set of images ranging from 20 cm aboveL4–L5 (+20 cm) to 3 cm below L4–L5 (−3 cm), for atotal of 24 images per subject. Segmentation of the axialimages into visceral adipose tissue and subcutaneousadipose tissue areas (cm2) was performed using Hippo-Fat image analysis software [11]. Visceral adipose tissueand subcutaneous adipose tissue areas were summedacross all 24 images to obtain visceral adipose tissue andsubcutaneous adipose tissue volumes, and then thesevolumes were multiplied by 0.916 g/cm3, the density ofadipose tissue, to obtain total visceral adipose tissuemass (kg) and total subcutaneous adipose tissue mass(kg) [10].

Blood collection protocolTwelve-hour fasting blood samples were collected be-tween 6.00 am and 9.00 am at weeks 1, 3 and 10. Thesubjects were instructed to avoid exercise, alcohol, andcoffee for 24 h before each visit. Blood was centrifugedfor 10 min at 520 × g at 4°C to separate plasma fromred blood cells and was stored at −80°C until analyzed.

Plasma lipid and LDL particle size determinationPlasma total cholesterol, direct LDL cholesterol, HDLcholesterol, and triglyceride concentrations were mea-sured in duplicate by enzymatic kits (Biovision Inc,Mountainview, CA). The interassay coefficient of varia-tions (CV) for total cholesterol, LDL cholesterol, HDLcholesterol, and triglycerides were 2.4%, 3.7%, 4.0%, and3.5%, respectively. LDL particle size was measured bylinear polyacrylamide gel electrophoresis (QuantimetrixLipoprint System, Redondo Beach, CA) [12]. High-resolution 3% polyacrylamide gel tubes were used forelectrophoresis. Briefly, 25 μL of sample was mixed with200 μL of liquid loading gel containing Sudan black, andadded to the gel tubes. After photopolymerization atroom temperature for 30 min, samples were electro-phoresed for 1 h (3 mA/gel tube). Lipoware computersoftware (Quantimetrix, Redondo Beach, CA) was thenused to divide LDL into small (<255 Å), medium (255–260 Å), and large (>260 Å) particles [12].

Coronary heart disease risk indicator assessmentBlood pressure and heart rate were measured in tripli-cate each week using a digital automatic blood pressure/heart rate monitor (Omron HEM 705 LP, Kyoto, Japan)with the subject in a seated position after a 10-min rest.Fasting plasma glucose was measured in duplicate atweek 1, 3, and 10 with a glucose hexokinase reagent kit

(A-gent glucose test, Abbott, South Pasadena, CA; inter-assay CV: 2.8%). Insulin, C-reactive protein (CRP),homocysteine, adiponectin, and leptin were assessed induplicate at week 1, 3, and 10 by ELISA (R&D Systems,Minneapolis, MN; inter-assay CVs: 3.0%, 3.4%, 3.9%,4.7%, and 4.2%, respectively).

StatisticsResults are presented as mean ± SEM. An independentsamples t-test was used to test baseline differences be-tween groups. Repeated-measures ANOVA was per-formed (taking time as the within-subject factor and dietas the between-subject factor) to assess differencesbetween groups over the course of the study. Post-hocanalyses were performed using the Tukey test. Differ-ences were considered significant at P < 0.05. All datawas analyzed using SPSS software (version 20.0, SPSSInc, Chicago, IL).

ResultsSubject dropout and baseline characteristicsOf the 30 subjects that were randomized to the IFCR-Lgroup, 1 subject dropped out due to scheduling conflictsand 1 subject dropped out because they could not ad-here to the diet (Figure 1). Thus 28 subjects completedthe IFCR-L protocol. Of the 30 subjects that were rando-mized to the IFCR-F group, 2 dropped out due to sched-uling conflicts and 2 others dropped out because theycould not adhere to the diet. Baseline characteristics ofthe IFCR-L and IFCR-F groups are reported in Table 1.There were no differences between groups for age, eth-nicity, menopausal status, waist circumference, BMI,plasma lipids, fasting glucose, or insulin.

Adherence to diets and physical activity maintenanceDuring the weight loss period (weeks 3–10), adherenceto the liquid meal protocol was 92 ± 3% in the IFCR-Lgroup over the course of the 8 weeks. IFCR-F subjectsachieved their energy restriction goal on 80 ± 2% ofthe CR days during each week. As for fast day compliance,there were no differences between groups (P = 0.91) (96 ±4% and 98 ± 3% compliance for the IFCR-F and IFCR-Lgroups, respectively). The degree of CR achieved duringthe weight loss period by the IFCR-L group (29 ± 3%)was greater (P < 0.05) than that achieved by the IFCR-F(22 ± 4%). The macronutrient composition of the IFCR-Ldiet (28 ± 2% kcal from fat, 20 ± 1% kcal from protein,52 ± 3% kcal from carbs) did not differ from that of theIFCR-F diet (31 ± 2% kcal from fat, 19 ± 2% kcal fromprotein, 50% ± 3 kcal from carbs) during the 8-weekintervention. Alterations in physical activity habits werequantified by the use of a pattern recognition monitor(i.e. an accelerometer). There were no differences in activityenergy expenditure between baseline and post-treatment in

Klempel et al. Nutrition Journal 2012, 11:98 Page 4 of 9http://www.nutritionj.com/content/11/1/98

Page 14: Intermittent Fasting: Weight Loss and Cardiovascular Effects · nate day fasting (ADF; fasting every other day) is organ-ized with alternating “feast days,” on which there is

the IFCR-L group (week 3: 249 ± 28 kcal/d, week 10: 283 ±27 kcal/d, P = 0.24), and the IFCR-F group (week 3: 246 ±37 kcal/d, week 10: 258 ± 43 kcal/d, P = 0.63).

Weight loss and body compositionChanges in body weight and body composition arereported in Figure 2. During the weight maintenanceperiod, body weight remained stable in both the IFCR-Lgroup (week 1: 95 ± 3, week 3: 95 ± 3 kg) and IFCR-F

group (week 1: 94 ± 3, week 3: 94 ± 3 kg). During theweight loss period, body weight decreased (P < 0.0001)by 3.9 ± 1.4 kg (4.1 ± 1.5%) in the IFCR-L group and by2.5 ± 0.6 kg (2.6 ± 0.4%) in the IFCR-F group. Thus, atweek 10, body weight of the IFCR-L and IFCR-F groupwas 91 ± 3 kg and 91 ± 2 kg, respectively. The IFCR-Lgroup lost more body weight compared to the IFCR-Fgroup (P = 0.04). BMI decreased (P < 0.0001) by 1.3 ± 0.5and 0.8 ± 0.5 kg/m2, respectively, in the IFCR-L andIFCR-F groups. Fat mass decreased (P < 0.0001) in theIFCR-L and IFCR-F groups by 2.8 ± 1.2 kg and 1.9 ± 0.7 kg,respectively. Fat free mass remained unchanged through-out the course of the trial in both groups. Visceral fatwas reduced (P < 0.001) in the IFCR-L (0.7 ± 0.5 kg) andIFCR-F (0.3 ± 0.5 kg) groups after 8 weeks of treatment.Abdominal subcutaneous fat was not affected by eitherintervention. There were no differences between groupsfor fat mass, fat free mass, visceral adipose tissue, or ab-dominal subcutaneous adipose tissue at any time point.Figure 3 depicts the change in visceral and subcutaneousadipose tissue in one subject in the IFCR-L group beforeand after the intervention.

Plasma lipids and LDL particle sizePlasma lipids did not change during the baseline periodin either the IFCR-L or IFCR-F group. Changes inplasma lipid concentrations during the weight lossperiod are displayed in Figure 4. Total cholesterol con-centrations decreased (P = 0.04) to a greater extent inthe IFCR-L group (19 ± 10%) compared to the IFCR-Fgroup (8 ± 3%). LDL cholesterol concentrations werealso reduced (P = 0.03) to a greater degree by the IFCR-Ldiet (20 ± 9%) versus the IFCR-F diet (7 ± 4%). HDLcholesterol was not affected by either intervention. Trigly-cerides decreased (P < 0.0001) in the IFCR-L group only(17 ± 9%). Changes in LDL particle size characteristics

Table 1 Subject characteristics at baseline 1

Characteristic IFCR-L IFCR-F

n 28 26

Age (y) 47 ± 2 48 ± 2

Ethnicity

African American 16 18

Asian 3 2

Caucasian 4 2

Hispanic 5 4

Pre-menopausal women 16 14

Post-menopausal women 12 12

Body weight (kg) 95 ± 3 94 ± 3

Height (cm) 165 ± 2 164 ± 2

Waist circumference (cm) 103 ± 1 102 ± 3

Body mass index (kg/m2) 35 ± 1 35 ± 1

Total cholesterol (mg/dl) 185 ± 8 188 ± 7

LDL cholesterol (mg/dl) 110 ± 7 115 ± 6

HDL cholesterol (mg/dl) 57 ± 4 55 ± 3

Triglycerides (mg/dl) 71 ± 7 81 ± 8

Glucose (mg/dl) 120 ± 2 120 ± 3

Insulin (uIU/ml) 14 ± 1 15 ± 21 Values reported as mean ± SEM. IFCR-L: Intermittent fasting calorierestriction-liquid diet; IFCR-F: Intermittent fasting calorie restriction-food baseddiet. No differences between groups for any parameter (Independent samplest-test).

Figure 2 Body weight and body composition changes during the weight loss period. IFCR-L: Intermittent fasting calorie restriction-liquiddiet (n = 28); IFCR-F: Intermittent fasting calorie restriction-food based diet (n = 26); BW: Body weight; FM: Fat mass; FFM: Fat free mass, VAT:Visceral adipose tissue; SAT: Abdominal subcutaneous adipose tissue. A. Changes in body weight, fat mass, and fat free mass in the IFCR-L andIFCR-F groups. IFCR-L group lost more body weight (P = 0.04) compared to the IFCR-F group (Repeated-measures ANOVA). *Week 10 absolutevalues significantly different from baseline (week 3) absolute values, P < 0.0001 (Repeated-measures ANOVA). B. Changes in visceral adipose tissueand abdominal subcutaneous adipose tissue in the IFCR-L and IFCR-F groups. *Week 10 absolute values significantly different from baseline (week3) absolute values, P < 0.001 (Repeated-measures ANOVA).

Klempel et al. Nutrition Journal 2012, 11:98 Page 5 of 9http://www.nutritionj.com/content/11/1/98

Page 15: Intermittent Fasting: Weight Loss and Cardiovascular Effects · nate day fasting (ADF; fasting every other day) is organ-ized with alternating “feast days,” on which there is

are reported in Table 2. LDL peak particle size increased(P < 0.01) by 2 ± 1 Å in the IFCR-L group only. The pro-portion of large and medium particles was augmented(P < 0.05) by the IFCR-L diet, but not the IFCR-F diet.The proportion of small particles was reduced (P < 0.05)in the both the IFCR-L (week 3: 37 ± 1%, week 10: 28 ±2%) and IFCR-F groups (week 3: 39 ± 1%, week 10: 36 ±1%). However, a greater reduction (P < 0.05) in the pro-portion of small particles was noted in the IFCR-L group.

Coronary heart disease risk indicatorsThere were no changes in CHD risk parameters duringthe weight maintenance period. Changes in CHD risk in-dices during the weight loss period are displayed inTable 3. Blood pressure was not altered by either theIFCR-L or IFCR-F intervention. Heart rate decreased(P < 0.05) in the IFCR-L group only. Glucose and insulinconcentrations decreased (P < 0.05) by the IFCR-L diet,but were not affected by the IFCR-F diet. CRP remainedunchanged in both intervention groups. Homocysteine

concentrations decreased (P < 0.01) in the IFCR-L grouponly. Adiponectin and leptin levels were reduced (P < 0.01)by both the IFCR-L and IFCR-F diets.

DiscussionOur findings show, for the first time, that the combin-ation of IF plus CR is an effective means of reducingbody weight, fat mass, and visceral fat mass in obesewomen. This novel regimen also decreased key indica-tors of CHD risk, such as LDL cholesterol, triglycerides,and the proportion of small LDL particles. When liquidmeal replacements were incorporated into the IFCRregimen, greater reductions in body weight and indica-tors of heart disease risk were noted.Studies of IF in human populations are very limited.

To our knowledge, only two studies have examined theeffect of IF on body weight [4,13]. In a trial conductedby Williams et al. [13], obese subjects consumed a very-low calorie diet (VLCD; <500 kcal/d) 1 day per week,and ate ad libitum every other day of the week. After20 weeks of treatment, body weight decreased by 9%(9 kg) from baseline [13]. Similar decreases in bodyweight were also observed in a recent trial by Harvieet al. [4]. In this study, obese women underwent 2 daysof VLCD (600 kcal/d) and ate ad libitum on every otherday of the week, for 24 weeks [4]. Body weight wasreduced by 7% (6 kg), fat mass decreased by 13% (4 kg),and waist circumference (an indirect indicator of visceralfat) decreased by 6% (6 cm) [4]. In the present study, weobserved modest weight loss in both the IFCR-L andIFCR-F groups after 8 weeks of treatment. We also

Figure 3 Change in visceral and subcutaneous adipose tissuearea after the IFCR-L intervention. A. Abdominal visceral fat(58 cm2) and subcutaneous fat (245 cm2) before the intermittentfasting calorie restriction-liquid diet (IFCR-L) at the L3-L4 vertebrae(week 3). B. Abdominal visceral fat (17 cm2) and subcutaneous fat(173 cm2) after the IFCR-L regimen at the L3-L4 vertebrae (week 10).

Figure 4 Plasma lipid changes during the weight loss period.IFCR-L: Intermittent fasting calorie restriction-liquid diet (n = 28);IFCR-F: Intermittent fasting calorie restriction-food based diet(n = 26); TC: Total cholesterol; LDL: Low density lipoproteincholesterol; HDL: High density lipoprotein cholesterol; TG:Triglycerides. *Week 10 values significantly different from baseline(week 3) values, P < 0.01 (Repeated-measures ANOVA). IFCR-L groupexperienced significantly greater reductions in total cholesterol andLDL cholesterol concentrations when compared to the IFCR-F group(P < 0.05) (Repeated-measures ANOVA).

Klempel et al. Nutrition Journal 2012, 11:98 Page 6 of 9http://www.nutritionj.com/content/11/1/98

Page 16: Intermittent Fasting: Weight Loss and Cardiovascular Effects · nate day fasting (ADF; fasting every other day) is organ-ized with alternating “feast days,” on which there is

observed that the addition of liquid meals to the proto-col resulted in greater weight loss (IFCR-L group: 4.1%weight loss versus IFCR-F group: 2.6% weight loss).Reductions in fat mass and visceral fat mass were alsodemonstrated at the end of the trial, but did not differbetween groups. As for fat free mass, no significantchanges were noted in either intervention group. This issurprising as a previous study that implemented energyrestricted liquid meals observed small but significantdecreases in fat free mass (i.e. 3% reductions from base-line) after 8 weeks of treatment [14]. The greater weightloss by the IFCR-L intervention is most likely due to bet-ter dietary adherence. Analysis of food intake revealed agreater degree of energy restriction in the IFCR-L group(29%) compared to the IFCR-F group (22%) over thecourse of the trial. This greater overall restriction in theIFCR-L group, and hence better adherence, is not sur-prising as these subjects were given portion-controlledliquid meals for breakfast and lunch everyday. Providingsuch meals has been shown to boost initial weight lossduring dietary restriction protocols as it takes the guess-work out of having to estimate calories from varyingfoods [7]. Although these liquid meals are effective for

helping with initial weight loss, lasting weight loss andweight maintenance requires extensive dietary counsel-ing to instill healthy behaviors that can be employedlong-term [8,15]. In view of this, both groups met with aRegistered Dietician weekly to incorporate TLC dietaryguidelines into their daily lives and to help make the leapfrom the liquid diet to a wholesome food-based regimenafter the study was over.Beneficial modulations in key lipid risk factors were

also observed by both diets. For instance, total and LDLcholesterol decreased in the IFCR-L group (19% and20%, respectively) and IFCR-F group (8% and 7%, re-spectively), with greater changes noted for IFCR-L. Incontrast, only the IFCR-L group experienced reductionsin triglycerides (17% from baseline). The proportion ofsmall LDL particles was also decreased by both theIFCR-L group (week 3: 37 ± 1%, week 10: 28 ± 2%) andIFCR-F group (week 3: 39 ± 1%, week 10: 36 ± 1%).However, increases in LDL peak particle size and theproportion of large LDL particles, were only noted inthe IFCR-L group. Taken together, these results suggestthat IF combined with CR is an effective means of im-proving lipid profile in a short-term (8 week)

Table 2 LDL particle size changes during the weight loss period 1

IFCR-L IFCR-F

Week 3 Week 10 Change 2 Week 3 Week 10 Change 2

LDL peak size (Å) 260 ± 1 262 ± 1 3 2 ± 1 4 261 ± 1 261 ± 1 0 ± 1

Proportion large particles (%) 34 ± 1 39 ± 1 3 5 ± 2 35 ± 1 37 ± 1 2 ± 1

Proportion medium particles (%) 29 ± 2 33 ± 2 3 4 ± 4 26 ± 2 27 ± 2 1 ± 2

Proportion small particles (%) 37 ± 1 28 ± 2 3 −9 ± 4 4 39 ± 1 36 ± 1 3 −3 ± 11 Values reported as mean ± SEM. IFCR-L: Intermittent fasting calorie restriction-liquid diet (n = 28); IFCR-F: Intermittent fasting calorie restriction-food based diet(n = 26). Small LDL particles (<255 Å), medium LDL particles (255–260 Å), and large LDL particles (>260 Å).2 Change expressed as the difference between week 3 and week 10 absolute values.3 Week 3 values significantly (P < 0.05) different from week 10 values within group (Repeated-measures ANOVA).4 Absolute change significantly different (P < 0.05) between the IFCR-L and IFCR-F group (Repeated-measures ANOVA).

Table 3 Coronary heart disease risk parameter changes during the weight loss period 1

IFCR-L IFCR-F

Week 3 Week 10 Change 2 Week 3 Week 10 Change 2

Systolic BP (mm Hg) 120 ± 3 118 ± 2 −2 ± 6 116 ± 4 114 ± 2 −2 ± 4

Diastolic BP (mm Hg) 83 ± 3 79 ± 3 −4 ± 4 80 ± 3 80 ± 2 0 ± 3

Heart rate (bpm) 73 ± 3 70 ± 4 3 −3 ± 4 4 78 ± 2 81 ± 2 3 ± 2

Glucose (mg/dl) 120 ± 2 116 ± 2 3 −4 ± 3 120 ± 3 117 ± 3 −3 ± 4

Insulin (uIU/ml) 14 ± 1 11 ± 1 3 −3 ± 3 15 ± 2 13 ± 2 −2 ± 2

C-Reactive protein (mg/dl) 0.4 ± 0.1 0.4 ± 0.1 0 ± 0.1 0.6 ± 0.2 0.4 ± 0.1 −0.2 ± 0.2

Homocysteine (ng/dl) 10 ± 1 8 ± 1 3 −2 ± 1 4 10 ± 1 10 ± 1 0 ± 1

Adiponectin (ng/ml) 7893 ± 1002 5462 ± 576 3 −2431 ± 802 8442 ± 1201 5931 ± 964 3 −2511 ± 771

Leptin (ng/ml) 37 ± 3 27 ± 3 3 −10 ± 2 38 ± 2 29 ± 2 3 −9 ± 21 Values reported as mean ± SEM. IFCR-L: Intermittent fasting calorie restriction-liquid diet (n = 28); IFCR-F: Intermittent fasting calorie restriction-food based diet(n = 26); BP: Blood pressure.2 Change expressed as the difference between week 3 and week 10 absolute values.3 Week 3 values significantly (P < 0.05) different from week 10 values within group (Repeated-measures ANOVA).4 Absolute change significantly different (P < 0.05) between the IFCR-L and IFCR-F group (Repeated-measures ANOVA).

Klempel et al. Nutrition Journal 2012, 11:98 Page 7 of 9http://www.nutritionj.com/content/11/1/98

Page 17: Intermittent Fasting: Weight Loss and Cardiovascular Effects · nate day fasting (ADF; fasting every other day) is organ-ized with alternating “feast days,” on which there is

intervention. We also show that adding a liquid dietcomponent may enhance this lipid-lowering effect. Thegreater decreases in plasma lipids by the IFCR-L diet ismost likely due to the greater weight loss noted in thisgroup. For every kg of body weight loss, LDL cholesterolis estimated to decrease by 2 mg/dl [16]. Since theIFCR-L group lost 1.4 kg more body weight than theIFCR-F group, this may explain why the reductions inLDL cholesterol by the liquid diet intervention weremore pronounced. The decreases in lipids demonstratedin the present study are similar to what has beenreported in previous trials of IF [4,13]. Williams et al.[13] noted a 10% and 52% lowering of LDL cholesteroland triglycerides, respectively, after 20 weeks of treat-ment. In accordance with these findings, Harvie et al. [4]observed a 10% decrease in LDL cholesterol and a 17%reduction in triglycerides. No trial to date has examinedthe effect of IF on LDL particle size, thus there is nodata for which to compare our findings.Modulations in other CHD risk parameters were also

more pronounced in the IFCR-L group compared to theIFCR-F group. For instance, fasting plasma glucose andinsulin were only decreased by the liquid intervention,suggesting that this diet therapy may benefit glycemiccontrol. Heart rate and homocysteine concentrationswere reduced solely in the IFCR-L group. Leptin concen-trations, on the other hand, were lowered by both diets.The decreases in leptin are most likely mediated by thereductions in fat mass and visceral fat mass observed inboth groups [17]. Leptin may be involved in atheroscler-otic plaque formation through its effect on cholesterolbiosynthesis in monocytes [18]. Thus, these reductionsin leptin by IFCR may play a systemic anti-atherogenicrole [19]. Adiponectin levels were also decreased by bothdiet interventions. This finding is not surprising, as adi-ponectin levels have been shown to decrease during thefirst 8–12 weeks of CR, and then increase only once a10% weight loss has been achieved [20]. Since thepresent trial only ran for 8 weeks, and since weight losswas <5%, this may explain why adiponectin concentra-tions were lowered from baseline. Blood pressure andCRP also remained unchanged throughout the course ofthe trial in both groups. Accumulating evidence suggestthat a 5 and 10% reduction in body weight is required todecrease blood pressure and CRP, respectively [21,22].This degree of weight loss was not attained, which mayexplain why these two variables were unaffected by ei-ther treatment.This study is limited in that it did not tease apart the

effect of IF and CR on body weight and CHD risk. Thus,the independent contributions of the IF diet versus theCR diet on these various parameters, are not known. Inorder to answer these key questions, a future studyshould be performed that compares the effect of IF

combined with CR, to that of IF alone, and CR alone.An additional limitation of the study was that it did notcarefully control for food intake by providing food-basedmeals to the intervention groups (i.e. dinner meal forthe IFCR-L subjects, and 3 meals/d for the IFCR-F sub-jects). If meals were provided, a more precise measure-ment of energy restriction and dietary adherence couldhave been obtained. The study is also limited in that itemployed HippoFat software to quantify visceral fatmass from MRI images. This software is limited in thatit underestimates visceral fat and overestimates subcuta-neous fat, particularly in larger individuals [11]. Anotherlimitation of the study is that we employed food recordsto estimate overall calorie restriction in each group. It iswell known that obese individuals underreport food in-take by 20-40% when completing food records [23]. Fu-ture studies in this field should therefore implementmore robust measures of energy assessment such as thedoubly-labeled water technique [24]. The last disadvan-tage of the study is that only female subjects wereemployed, and as such, the applicability of these findingsto males remains uncertain.In summary, these findings suggest that IFCR may be

effective for reducing body weight, visceral fat mass, andCHD risk in obese women. We also report that incorp-orating liquid meal replacements into an IFCR regimenmay facilitate greater weight loss and lipid-lowering.From a clinical standpoint, we would recommend thisdiet to individuals who wish to boost the weight lossthey see with IF, by adding a daily CR regimen. Thiscombination may also help reduce the boredom typicallyassociated with attempting only one dietary plan. Al-though these short-term findings are promising, thelong-term effects of this novel diet strategy still requireconfirmation in a large-scale human trial.

Competing interestsKrista Varady has a consulting relationship with the sponsor of the research,lsagenix, LLC.

Authors’ contributionsMCK conducted the clinical trial, analyzed the data, and assisted with thepreparation of the manuscript. CMK assisted with the conduction of theclinical trial and performed the laboratory analyses. SB and JFT performedthe laboratory analyses, and assisted with data analyses. KAV designed theexperiment and wrote the manuscript. All authors read and approved thefinal manuscript.

Funding sourceThis study was funded by Isagenix LLC., Chandler, AZ

Received: 14 August 2012 Accepted: 20 November 2012Published: 21 November 2012

References1. Mercuro G, Deidda M, Bina A, Manconi E, Rosano GM: Gender-specific

aspects in primary and secondary prevention of cardiovascular disease.Curr Pharm Des 2011, 17(11):1082–1089.

2. Lindstrom J, Uusitupa M: Lifestyle intervention, diabetes, andcardiovascular disease. Lancet 2008, 371(9626):1731–1733.

Klempel et al. Nutrition Journal 2012, 11:98 Page 8 of 9http://www.nutritionj.com/content/11/1/98

Page 18: Intermittent Fasting: Weight Loss and Cardiovascular Effects · nate day fasting (ADF; fasting every other day) is organ-ized with alternating “feast days,” on which there is

RESEARCH Open Access

Alternate day fasting for weight loss in normalweight and overweight subjects: a randomizedcontrolled trialKrista A Varady*, Surabhi Bhutani, Monica C Klempel, Cynthia M Kroeger, John F Trepanowski, Jacob M Haus,Kristin K Hoddy and Yolian Calvo

Abstract

Background:Alternate day fasting (ADF; ad libitum “feed day”, alternated with 25% energy intake “fast day”), iseffective for weight loss and cardio-protection in obese individuals. Whether these effects occur in normal weightand overweight individuals remains unknown. This study examined the effect of ADF on body weight and coronaryheart disease risk in non-obese subjects.

Methods:Thirty-two subjects (BMI 20–29.9 kg/m2) were randomized to either an ADF group or a control group for12 weeks.

Results:Body weight decreased (P < 0.001) by 5.2 ± 0.9 kg (6.5 ± 1.0%) in the ADF group, relative to the controlgroup, by week 12. Fat mass was reduced (P < 0.001) by 3.6 ± 0.7 kg, and fat free mass did not change, versus controls.Triacylglycerol concentrations decreased (20 ± 8%, P < 0.05) and LDL particle size increased (4 ± 1 Å, P < 0.01) in theADF group relative to controls. CRP decreased (13 ± 17%, P < 0.05) in the ADF group relative to controls at week 12.Plasma adiponectin increased (6 ± 10%, P < 0.01) while leptin decreased (40 ± 7%, P < 0.05) in the ADF group versuscontrols by the end of the study. LDL cholesterol, HDL cholesterol, homocysteine and resistin concentrations remainedunchanged after 12 weeks of treatment.

Conclusion:These findings suggest that ADF is effective for weight loss and cardio-protection in normal weightand overweight adults, though further research implementing larger sample sizes is required before solid conclusioncan be reached.

Keywords:Alternate day fasting, Calorie restriction, Weight loss, Cholesterol, Blood pressure, Adipokines, Coronary heartdisease, Non-obese humans

IntroductionIntermittent fasting regimens, particularly alternateday fasting (ADF) protocols, have gained considerablepopularity in the past decade. Alternate day fastinginvolves a “fast day” where individuals consume 25%of energy needs, alternated with a “feed day” wheresubjects eat ad libitum [1]. Only a handful of studieshave been performed to test the effects of ADF onbody weight and coronary heart disease (CHD) riskreduction, and almost all of these studies have beenundertaken in obese populations (BMI 30–39.9 kg/m2)[2-4]. Results from these initial trials indicate that

ADF is effective for weight loss (5-6% reductions inbody weight) and visceral fat mass loss (5–7 cm reductionsin waist circumference) in 8–12 weeks of treatment [2-4].These reports also suggest that ADF may aid in theretention of lean mass in obese individuals [2-4]. Inaddition to these favorable body composition changes,improvements in CHD risk have also been noted. Forinstance, decreases in LDL cholesterol concentrations(20-25%), triacylglycerol concentrations (15-30%), andincreases in LDL particle size are often observed withshort-term ADF (8–12 weeks) [2-4]. Beneficial changesin blood pressure and adipokine profile (i.e. increasesin adiponectin, and decreases in leptin and resistin) havealso been reported [2-4]. Taken together, this preliminary

* Correspondence: [email protected] of Kinesiology and Nutrition, University of Illinois at Chicago,1919 West Taylor Street, Room 506 F, Chicago, IL 60612, USA

© 2013 Varady et al.; licensee BioMed Central Ltd. This is an open access article distributed under the terms of the CreativeCommons Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, andreproduction in any medium, provided the original work is properly cited.

Varady et al. Nutrition Journal 2013, 12:146http://www.nutritionj.com/content/12/1/146

Page 19: Intermittent Fasting: Weight Loss and Cardiovascular Effects · nate day fasting (ADF; fasting every other day) is organ-ized with alternating “feast days,” on which there is

work suggests that ADF may be effective for weightloss and CHD risk reduction in obese adults.An important question that remains unresolved is

whether the favorable effects of ADF can also be observedin normal weight and overweight populations. Only twohuman studies [5,6] have tested the effect of ADF onbody weight and CHD risk in non-obese subjects. In astudy by Heilbronn et al. [5], normal weight men andwomen (BMI 23 kg/m2) participated in an ADF regimenfor 3 weeks. Body weight decreased by 2% from baseline,while triacylglycerol concentrations decreased only inmen [5]. Contrary to these findings, Halberg et al. [6]demonstrated no change in body weight after 2 weeksof ADF in overweight men (BMI 26 kg/m2). Whilethese trials [5,6] lay some groundwork, they are limitedby their short durations (2–3 weeks) and their lack of acontrol group. As such, a longer-term trial (12 weeks)that employs a control group is well warranted.Accordingly, the present study examined the effect of

ADF on body weight, body composition, and CHD riskparameters in both normal weight and overweight adultsin a 12-week randomized controlled feeding trial. Wehypothesized that ADF would reduce body weight andCHD risk in normal weight and overweight participants,when compared to controls.

Subjects and methodsSubjectsSubjects were recruited from the Chicago area by meansof advertisements placed around the University of Illinois,Chicago campus. A total of 107 individuals expressedinterest in the study, but only 32 were recruited to partici-pate after screening via a preliminary questionnaire andBMI assessment (Figure 1). Inclusion criteria were asfollows: BMI between 20 and 29.9 kg/m2; age between35 and 65 years; pre-menopausal or post-menopausal(absence of menses for more than 2 years); lightly active(< 3 h/week of light intensity exercise at 2.5 to 4.0metabolic equivalents (METs) for 3 months prior to thestudy); weight stable for 3 months prior to the beginningof the study (< 4 kg weight loss or weight gain); non-diabetic; no history of cardiovascular disease; non-smoker;and not taking weight loss, lipid- or glucose-loweringmedications. The experimental protocol was approvedby the University of Illinois, Chicago, Office for theProtection of Research Subjects, and all research partici-pants gave their written informed consent to participatein the trial. The research protocol was in compliancewith the Helsinki Declaration.

Study designExperimental designA 12-week, randomized, controlled, parallel-arm feedingtrial was implemented as a means of testing the study

objectives. Subjects were randomized by KAV by way ofa stratified random sample. Subjects were first dividedinto strata based on sex (M/F), age (35–50 y/51-65 y),and BMI (20–24.9 kg/m2/ 25–29.9 kg/m2), and thensubjects from each stratum were randomized 1:1 intoeither the ADF or control group (Figure 1).

Diet protocolDuring the dietary intervention period, ADF subjectsconsumed 25% of their baseline energy needs on the fastday (24 h), and then ate ad libitum on each alternatingfeed day (24 h). Energy needs for each subject weredetermined by the Mifflin equation [7]. The feed andfast days began at midnight each day, and all fast daymeals were consumed between 12.00 pm and 2.00 pmto ensure that each subject was undergoing the sameduration of fasting. ADF subjects were provided withmeals on each fast day (ranging from 400–600 kcal),and ate ad libitum at home on the feed day. All ADFfast day meals were prepared in the metabolic kitchenof the Human Nutrition Research Center (HNRU) atthe University of Illinois, Chicago. Fast day meals wereprovided as a 3-day rotating menu, and were formulatedbased on the American Heart Association (AHA) guide-lines (30% kcal from fat, 15% kcal from protein, 55% kcalfrom carbohydrate) [8]. All meals were consumed outsideof the research center. ADF subjects were permitted toconsume energy-free beverages, tea, coffee, and sugar-free gum, and were encouraged to drink plenty ofwater. Control subjects were permitted to eat ad libitum

Figure 1 Study flow chart. ADF: Alternate day fasting.

Varady et al. Nutrition Journal 2013, 12:146 Page 2 of 8http://www.nutritionj.com/content/12/1/146

Page 20: Intermittent Fasting: Weight Loss and Cardiovascular Effects · nate day fasting (ADF; fasting every other day) is organ-ized with alternating “feast days,” on which there is

every day, and were not provided with meals from theresearch center.

Blood collection protocolTwelve-hour fasting blood samples were collectedbetween 6.00 am and 9.00 am at baseline (week 1) andpost-treatment (week 12). Participants were instructedto avoid exercise, alcohol, and coffee for 24 h beforeeach visit. Blood was centrifuged for 15 min at 520 × gand 4°C to separate plasma from RBCs, and was storedat −80°C until analysed.

AnalysesEnergy intake on feed and fast daysDuring the 12-week diet intervention, subjects in the ADFgroup were instructed to eat only the foods providedon each fast day. To assess energy intake on the fastdays, ADF subjects were asked to report any extra fooditems consumed (i.e. those not provided) using an“Extra food log”. Additionally, subjects were instructedto return any leftover food items to the HNRU forweighing. To assess energy intake on the feed days,ADF and control subjects were asked to complete a3-day food record on 2 feed days during the week, andon 1 feed day during the weekend, at week 1 and 12.At baseline, the Research Dietician provided 15 min ofinstruction to all participants on how to completethe food records. These instructions included verbalinformation and detailed reference guides on howto estimate portion sizes and record food items insufficient detail to obtain an accurate estimate ofdietary intake. All dietary information from the foodlogs/records was entered into the food analysis program,Nutritionist Pro (version 5, Axxya Systems, Stafford,TX) to assess energy intake.

Hunger, satisfaction, and fullnessA validated visual analog scale (VAS) was used to measurehunger, fullness, and satisfaction with the ADF diet [9].The scale was completed on 3 fast days (before bedtime)at week 1 and 12. In brief, the VAS consisted of 100-mmlines, and subjects were asked to make a vertical markacross the line corresponding to their feelings from 0(not at all) to 100 (extremely) for hunger, satisfaction,or fullness. Quantification was performed by measuring thedistance from the left end of the line to the vertical mark.

Weight loss and body compositionBody weight was assessed to the nearest 0.25 kg at thebeginning of every week without shoes and in lightclothing using a balance beam scale at the HNRU(HealthOMeter, Sunbeam Products, Boca Raton, FL).BMI was assessed as kg/m2. Body composition (fatmass and fat free mass) was measured using dual x-ray

absorptiometry (DXA) (Hologic QDR 4500 W, HologicInc., Waltham, MA).

Lipid coronary heart disease risk factorsPlasma total cholesterol, HDL-cholesterol, and triacyl-glycerol concentrations were measured in duplicate usingenzymatic kits (Biovision Inc., Moutainview, CA) at week1 and 12. The concentration of LDL-cholesterol wascalculated using the Friedewald, Levy and Fredricksonequation. LDL particle size was measured by linear poly-acrylamide gel electrophoresis (Quantimetrix LipoprintSystem, Redondo Beach, CA, USA) at week 1 and 12[10,11]. High-resolution 3% polyacrylamide gel tubeswere used for electrophoresis. Briefly, 25 μL of samplewas mixed with 200 μL of liquid loading gel containingSudan black, and added to the gel tubes. After photo-polymerization at room temperature for 30 min, sampleswere electrophoresed for 1 h (3 mA/gel tube). Lipowarecomputer software (Quantimetrix, Redondo Beach, CA,USA) was then used to divide LDL into small (<255 Å),medium (255–260 Å), and large (>260 Å) particles, andto assess mean LDL particle size [10]. The intra-assaycoefficients of variation (CV) for total cholesterol, HDLcholesterol, triacylglycerol, and LDL particle size were3.6%, 4.8%, 2.5%, and 4.1%, respectively.

Non-lipid coronary heart disease risk factorsAll measurements were taken at week 1 and 12. Bloodpressure was measured in triplicate with the subject ina seated position after a 10-min rest. C-reactive protein(CRP) was measured in duplicate using Immulite 1000High Sensitivity CRP kits (Diagnostic Products Corpor-ation, Los Angeles, CA). Plasma homocysteine measure-ments were carried out in duplicate using HPLC withfluorometric detection. Adiponectin, leptin and resistinwere measure by ELISA (R&D Systems, Minneapolis,MN). The intra-assay coefficients of variation (CV) forCRP, homocysteine, adiponectin, leptin, and resistinwere 5.0%, 4.3%, 3.3%, 3.0%, and 4.7%, respectively.

StatisticsResults are presented as means ± standard error ofthe mean (SEM). Tests for normality were included inthe model. No variables were found to be not normal.Differences between groups at baseline were tested byindependent samples t-test. Within-group changes fromweek 1 to 12 were tested by a paired t-test. Between-group differences were tested by an independent sam-ples t-test. Sample size was calculated assuming a 10%change in LDL-cholesterol concentrations in the ADFgroup, with a power of 80% and an alpha risk of 5%.P-values of < 0.05 were considered significant. Data wereanalyzed by using SPSS software (version 21.0 for MacOS X; SPSS Inc., Chicago, IL).

Varady et al. Nutrition Journal 2013, 12:146 Page 3 of 8http://www.nutritionj.com/content/12/1/146

Page 21: Intermittent Fasting: Weight Loss and Cardiovascular Effects · nate day fasting (ADF; fasting every other day) is organ-ized with alternating “feast days,” on which there is

ResultsSubject baseline characteristics and dropoutsThirty-two subjects commenced the study, with 30 com-pleting the entire 12-week trial (Figure 1). After loss dueto dropouts, the remaining subjects in each interventiongroup were as follows: ADF (n = 15) and control (n = 15).Baseline characteristics of the subjects who completed thetrial are presented in Table 1. There were no significantdifferences at the beginning of the study between groupsfor age, sex, ethnicity, body weight, body composition,height or BMI.

Energy intake, hunger, satisfaction and fullnessEnergy intake, hunger, satisfaction, and fullness are reportedin Table 2. At baseline, there were no differences betweenthe ADF and control groups for feed day energy intake.From week 1 to 12 of the study, energy intake remainedconstant on both feed and fast days in the ADF group.Adherence to the fast day protocol was high in theADF group (98 ± 5%). Hunger levels were moderate asbaseline, and did not change by week 12 in eithergroup. Satisfaction and fullness increased (P < 0.01)from baseline to post-treatment in the ADF group, withno change in the control group.

Weight loss and body compositionChanges in body weight and body composition aredisplayed in Figure 2. Body weight decreased (P < 0.001)by 5.2 ± 0.9 kg (6.5 ± 1.0%) in the ADF group, relativeto the control group at week 12. Fat mass was reduced(P < 0.001) by 3.6 ± 0.7 kg, and fat free mass did notchange, versus controls.

Lipid coronary heart disease risk factorsChanges in plasma lipids and LDL particle size arereported in Table 3. Total cholesterol concentrationsdecreased (P < 0.01) in the ADF group when post-treatment values were compared to baseline. However,changes in total cholesterol levels were not significantlydifferent from controls at week 12. LDL cholesterolconcentrations were reduced (P = 0.01) within the ADFgroup, but no significant between-group differenceswere noted. HDL cholesterol concentrations remainedunchanged throughout the trial. Triacylglycerol concen-trations decreased (P = 0.01) in the ADF group relativeto controls at week 12. Non-HDL cholesterol levelswere reduced (P < 0.01) within the ADF group, but nosignificant between-group differences were observed.LDL particle size increased (P < 0.01) in the ADF grouprelative to controls by the end of the study.

Non-lipid coronary heart disease risk factorsChanges in blood pressure, homocysteine, CRP, andadipokines are shown in Table 4. Systolic and diastolicblood pressure decreased (P < 0.05) within the ADFgroup, but no significant between-group differenceswere noted. CRP decreased (P = 0.01) in the ADF grouprelative to controls at week 12. Plasma adiponectinincreased (P < 0.01) while leptin decreased (P = 0.03) inthe ADF group versus controls by the end of the study.Plasma homocysteine and resistin concentrationsremained unchanged after 12 weeks of treatment.

DiscussionThis study shows, for the first time, that ADF is aneffective strategy for moderate weight loss (6%) in nor-mal weight and overweight subjects. This diet strategymay also have cardio-protective effects in non-obesesubjects, by way of lowering triacylglycerols, CRP andleptin, while increasing LDL particle size and adiponectinconcentrations.The primary goal of this study was to determine if

non-obese individuals could benefit from ADF in termsof weight loss. Previous ADF studies implementingnon-obese subjects report inconsistent findings [5,6].While one study demonstrated decreases in body weightof 2% from baseline after 3 weeks of ADF [5], anotherstudy showed no effect after 2 weeks of diet [6]. Thelimited amount of weight loss reported previously isundoubtedly a factor of the short trial durations imp-lemented [5,6]. Thus, we wanted to determine if thedegree of weight loss could be amplified if the trialduration was extended to 12 weeks. We show here thatnormal weight and overweight subjects can indeedbenefit from ADF, as body weight was reduced by 6%(5 kg) by the end of the trial. This degree of weight lossin non-obese participants is similar to what has been

Table 1 Subject characteristics at baselineADF-ALL Control P-value1

n 15 15

Age (y) 47 ± 3 48 ± 2 0.18

Sex (M/F) 5 / 10 3 / 12 0.44

Ethnicity (n)

African American 5 8 0.30

Caucasian 8 6

Hispanic 2 1

Other 0 0

Body weight (kg) 77 ± 3 77 ± 3 0.79

Fat mass (kg) 26 ± 2 27 ± 1 0.13

Fat free mass (kg) 51 ± 3 50 ± 3 0.78

Height (cm) 171 ± 3 170 ± 2 0.71

BMI (kg/m2) 26 ± 1 26 ± 1 0.75

Values reported as mean ± SEM. ADF: Alternate day fasting, BMI: Body massindex, F: Female, M: Male.1P-value between groups at baseline: Independent samples t-test.

Varady et al. Nutrition Journal 2013, 12:146 Page 4 of 8http://www.nutritionj.com/content/12/1/146

Page 22: Intermittent Fasting: Weight Loss and Cardiovascular Effects · nate day fasting (ADF; fasting every other day) is organ-ized with alternating “feast days,” on which there is

reported for obese individuals undergoing ADF [2-4].For instance, Bhutani et al. [4] demonstrated 5% (5 kg)weight loss after 12 weeks of ADF in obese men andwomen. In line with these findings, Klempel et al. [3]and Varady et al. [2] report 5-6% (5–6 kg) weight lossafter 8 weeks of treatment in obese subjects. Thus, ADFmay produce a mean rate of weight loss of approximately0.5 kg/week, independent of the starting weight or BMIclass of the subject. Fat free mass was also retainedafter 12 weeks of ADF in non-obese individuals. Thisfinding is similar to what has been reported in previousshort-term studies of ADF [2-4]. As such, the beneficialpreservation of fat free mass observed in obese individuals[2-4] may be replicated in non-obese subjects participatingin ADF protocols.

Our findings also indicate that normal weight andoverweight subjects have no problem adhering to thefast day protocol for 12 weeks. Dietary adherence wasvery high at baseline (98%), and did not wane over thecourse of the study. It should be noted, however, thatone normal weight subject dropped out of the trial dueto an inability to adhere to the diet. Notwithstanding,our dropout rate was still less than 10%, which is similarto the dropout rate of studies performed in obese individ-uals [2-4]. Complementary to previous reports [12,13],there was very little or no hyperphagic response on thefeed day in response to the lack of food on the fast day.This lack of hyperphagia allowed for overall energyrestriction to remain high throughout the study, andundoubtedly contributed to the sizeable degree of weightloss observed here. As for eating behaviors, perceivedhunger was moderate at baseline and did not changeby week 12. This is contrary to findings in obese par-ticipants, which consistently show declines in hungerafter 8–12 weeks of ADF [11,12]. Dietary satisfactionand feelings of fullness, on the other hand, increasedfrom baseline to post-treatment. These increases insatisfaction and fullness have also been noted in obesesubjects [11,12], and may play a role in long-termadherence to the diet.The cardio-protective effects of ADF were also exam-

ined. Reductions in triacylglycerol concentrations (20%)were noted after 12 weeks of ADF. LDL particle sizealso increased post-treatment (4 Å from baseline).These changes in lipid risk factors are in line with whathas been reported for obese ADF subjects [14,15]. Intwo recent ADF studies, triacylglycerols decreased by15% and LDL particle size increased by 2–3 Å after8 weeks of treatment in obese men and women [14,15].Thus, ADF may improve plasma lipids to the sameextent in non-obese subjects as it does in obese

Table 2 Energy intake, hunger, satisfaction and fullness during the 12-week studyIntervention Week 11 Week 12 P-value2 P-value3 Change4 P-value5

Feed day energy intake (kcal/d) ADF 1874 ± 136 1856 ± 229 0.93 0.92 −18 ± 186 0.50

Control 1873 ± 243 1790 ± 286 0.71 −82 ± 75

Fast day energy intake (kcal/d) ADF 482 ± 19 489 ± 20 0.74 7 ± 5

Hunger (mm) ADF 5 ± 1 4 ± 1 0.44 0.56 −1 ± 1 0.38

Control 5 ± 1 5 ± 1 0.46 0 ± 1

Satisfaction (mm) ADF 4 ± 1 7 ± 1 <0.01 0.81 3 ± 1 0.22

Control 6 ± 1 7 ± 1 0.34 1 ± 1

Fullness (mm) ADF 2 ± 1 4 ± 1 <0.01 0.78 2 ± 1 0.01

Control 6 ± 1 6 ± 1 0.57 0 ± 1

Values reported as mean ± SEM. ADF: Alternate day fasting.1Baseline values were not significantly different between intervention groups for any parameter: Independent samples t-test.2P-value between week 1 and week 12: Paired t-test.3P-value between groups at week 12: Independent samples t-test.4Absolute change between week 1 and week 12 values.5P-value between groups for absolute change: Independent samples t-test.

Figure 2 Body weight and body composition changes at week12. Values reported as mean ± SEM. ADF: Alternate day fasting.*Body weight and fat mass significantly different (P < 0.001) from thecontrol group at week 12 (Independent samples t-test). No differencebetween groups for fat free mass at week 12 (Independentsamples t-test).

Varady et al. Nutrition Journal 2013, 12:146 Page 5 of 8http://www.nutritionj.com/content/12/1/146

Page 23: Intermittent Fasting: Weight Loss and Cardiovascular Effects · nate day fasting (ADF; fasting every other day) is organ-ized with alternating “feast days,” on which there is

subjects. Additional vascular benefits, including decreasesin circulating leptin and CRP concentrations, in con-junction with increases in adiponectin, were also notedin non-obese subjects undergoing ADF. As for HDLcholesterol, homocysteine, and resistin concentrations,no effect was observed. This lack of effect is not surprisingas these CHD risk parameters are generally only improvedwith >10% weight loss [16-18].

It will be of interest in future studies to determinehow alterations in macronutrient intake on the fast daymay affect weight loss and cardiovascular outcomes. Forinstance, it has been well established that Mediterranean[19] and certain low-carbohydrate diets [20] help tomaintain a healthy body weight and reduce CHDrisk. Whether further reductions in body weight andCHD risk would occur if ADF were combined with

Table 3 Lipid coronary heart disease risk factor changes during the 12-week studyIntervention Week 11 Week 12 P-value2 P-value3 Change4 P-value5

Total cholesterol (mg/dl) ADF 201 ± 9 175 ± 12 <0.01 0.12 −26 ± 6 0.50

Control 211 ± 11 202 ± 9 0.54 −9 ± 5

LDL cholesterol (mg/dl) ADF 118 ± 9 99 ± 9 0.01 0.82 −18 ± 6 0.29

Control 128 ± 10 119 ± 6 0.08 −9 ± 4

HDL cholesterol (mg/dl) ADF 56 ± 3 54 ± 4 0.49 0.77 −2 ± 3 0.51

Control 57 ± 2 58 ± 4 0.83 1 ± 2

Triacylglycols (mg/dl) ADF 109 ± 13 87 ± 9 0.06 0.01 −22 ± 11 0.22

Control 108 ± 18 118 ± 19 0.34 10 ± 7

Non-HDL cholesterol (mg/dl) ADF 149 ± 11 124 ± 12 <0.01 0.54 −25 ± 5 0.96

Control 153 ± 12 144 ± 10 0.79 −9 ± 5

LDL particle size (Å) ADF 254 ± 1 258 ± 2 <0.01 <0.01 4 ± 1 0.13

Control 252 ± 2 250 ± 3 0.16 −2 ± 1

Values reported as mean ± SEM. ADF: Alternate day fasting.1Baseline values were not significantly different between intervention groups for any parameter: Independent samples t-test.2P-value between week 1 and week 12: Paired t-test.3P-value between groups at week 12: Independent samples t-test.4Absolute change between week 1 and week 12 values.5P-value between groups for absolute change: Independent samples t-test.

Table 4 Non-lipid coronary heart disease risk factor changes during the 12-week studyIntervention Week 11 Week 12 P-value2 P-value3 Change4 P-value5

Systolic BP (mm Hg) ADF 124 ± 4 117 ± 4 0.02 0.85 −7 ± 2 0.51

Control 119 ± 3 120 ± 4 0.67 1 ± 3

Diastolic BP (mm Hg) ADF 78 ± 3 72 ± 2 0.03 0.05 −6 ± 2 0.17

Control 82 ± 4 84 ± 5 0.28 2 ± 6

Homocysteine (umol/dl) ADF 9 ± 1 9 ± 1 0.37 0.50 0 ± 1 0.21

Control 9 ± 1 9 ± 1 0.86 0 ± 1

C-reactive protein (mg/L) ADF 2 ± 1 1 ± 1 0.29 0.01 −1 ± 1 0.01

Control 1 ± 1 1 ± 1 0.78 0 ± 1

Adiponectin (ng/ml) ADF 10728 ± 1251 11401 ± 1197 0.58 0.15 672 ± 1191 <0.01

Control 11350 ± 1369 10509 ± 1316 0.26 −842 ± 623

Leptin (ng/ml) ADF 25 ± 4 15 ± 3 0.04 0.55 −10 ± 3 0.03

Control 22 ± 7 18 ± 6 0.07 −4 ± 3

Resistin (ng/ml) ADF 18 ± 3 15 ± 4 0.15 0.57 −3 ± 3 0.20

Control 23 ± 4 21 ± 4 0.26 −2 ± 2

Values reported as mean ± SEM. ADF: Alternate day fasting.1Baseline values were not significantly different between intervention groups for any parameter: Independent samples t-test.2P-value between week 1 and week 12: Paired t-test.3P-value between groups at week 12: Independent samples t-test.4Absolute change between week 1 and week 12 values.5P-value between groups for absolute change: Independent samples t-test.

Varady et al. Nutrition Journal 2013, 12:146 Page 6 of 8http://www.nutritionj.com/content/12/1/146

Page 24: Intermittent Fasting: Weight Loss and Cardiovascular Effects · nate day fasting (ADF; fasting every other day) is organ-ized with alternating “feast days,” on which there is

Mediterranean or low-carbohydrate diets, undoubtedlywarrants investigation.A couple of adverse events were reported during the

study. Two subjects experienced mild headaches duringweek 1 of the trial, which may or may not be related todietary treatment. One other subject reported constipa-tion during week 1 and 2 of the trial. The subject wasadvised to consume more fruits and vegetables on feeddays, and the constipation subsided by week 3 of thedietary intervention period.This study has several limitations. First and foremost,

it must be acknowledged that this pilot study was origin-ally designed to compare the effects of ADF in normalweight versus overweight individuals on body weightand CHD risk. Due to a low recruitment rate, we wereonly able to recruit n = 8 subjects into the normal weightgroup and n = 8 subjects into the overweight group. Inview of this, we decided to combine the normal weightand overweight groups into one group to increasesample size. This post hoc change should be takeninto consideration when interpreting the findings ofthis paper. Secondly, physical activity was not assessedthroughout the trial, thus the degree of weight lossassociated with increased energy expenditure fromexercise is not known. Thirdly, the sample size of eachgroup was small (n = 15). Thus, this study may not beadequately powered to detect changes in certain CHD riskparameters (e.g. resistin). Fourthly, this study employedfood records to assess dietary intake/adherence. It iswell known that overweight subject underreport foodintake by ~30% [21,22]. Thus, our findings for the hy-perphagic response on the feed day may be inaccurate.In summary, these preliminary findings suggest that

ADF is a viable weight loss strategy for normal weightand overweight individuals wishing to lose a moderateamount of weight (5–6 kg) within a relatively shortperiod of time (12 weeks). This diet may also help lowerCHD risk in non-obese individuals, though furtherinvestigation is warranted to confirm these effects. Itshould also be noted that the purpose of this paper is toreport pilot feasibility findings. It is our hope that thispreliminary data will be utilized to design larger-scalelonger-term trials with similar objectives, in normalweight and overweight participants undergoing ADF.

Competing interestThe authors have no conflicts of interest to report.

Authors’ contributionsKAV designed the experiment, analyzed the data, and wrote themanuscript. SB, MCK, CMK, and JFT assisted with the conduction of theclinical trial and performed the laboratory analyses. JMH assisted withthe data analyses and the preparation of the manuscript. KKH and YCassisted with the laboratory analyses. All authors read and approved thefinal manuscript.

Funding sourceDepartmental grant from Kinesiology and Nutrition at the University ofIllinois, Chicago.

Received: 3 July 2013 Accepted: 4 November 2013Published: 12 November 2013

References1. Varady KA, Hellerstein MK: Alternate-day fasting and chronic disease

prevention: a review of human and animal trials. Am J Clin Nutr 2007,86:7–13.

2. Varady KA, Bhutani S, Church EC, Klempel MC: Short-term modifiedalternate-day fasting: a novel dietary strategy for weight loss and cardio-protection in obese adults. Am J Clin Nutr 2009, 90:1138–1143.

3. Klempel MC, Kroeger CM, Varady KA: Alternate day fasting (ADF) with ahigh-fat diet produces similar weight loss and cardio-protection as ADFwith a low-fat diet. Metabolism 2013, 62:137–143.

4. Bhutani S, Klempel MC, Kroeger CM, Trepanowski JF, Varady KA: Alternateday fasting and endurance exercise combine to reduce body weightand favorably alter plasma lipids in obese humans. Obesity (Silver Spring)2013. Epub ahead of print.

5. Heilbronn LK, Smith SR, Martin CK, Anton SD, Ravussin E: Alternate-dayfasting in nonobese subjects: effects on body weight, body composition,and energy metabolism. Am J Clin Nutr 2005, 81:69–73.

6. Halberg N, Henriksen M, Söderhamn N, Stallknecht B, Ploug T, Schjerling P,Dela F: Effect of intermittent fasting and refeeding on insulin action inhealthy men. J Appl Physiol 2005, 99:2128–2136.

7. Mifflin MD, St Jeor ST, Hill LA, Scott BJ, Daugherty SA, Koh YO: A newpredictive equation for resting energy expenditure in healthyindividuals. Am J Clin Nutr 1990, 51:241–247.

8. Gidding SS, Lichtenstein AH, Faith MS, Karpyn A, Mennella JA, Popkin B,Rowe J, Van Horn L, Whitsel L: Implementing American heart associationpediatric and adult nutrition guidelines: a scientific statement from theAmerican heart association nutrition committee of the council onnutrition, physical activity and metabolism, council on cardiovasculardisease in the young, council on arteriosclerosis, thrombosis andvascular biology, council on cardiovascular nursing, council onepidemiology and prevention, and council for high blood pressureresearch. Circulation 2009, 119:1161–1175.

9. Flint A, Raben A, Blundell JE, Astrup A: Reproducibility, power and validityof visual analogue scales in assessment of appetite sensations in singletest meal studies. Int J Obes Relat Metab Disord 2000, 24:38–48.

10. Chung M, Lichtenstein AH, Ip S, Lau J, Balk EM: Comparability of methodsfor LDL subfraction determination: a systematic review. Atherosclerosis2009, 205:342–348.

11. Varady KA, Lamarche B: Lipoprint adequately estimates LDL sizedistribution, but not absolute size, versus polyacrylamide gradient gelelectrophoresis. Lipids 2011, 46:1163–1167.

12. Klempel MC, Bhutani S, Fitzgibbon M, Freels S, Varady KA: Dietary andphysical activity adaptations to alternate day modified fasting:implications for optimal weight loss. Nutr J 2010, 9:35.

13. Bhutani S, Klempel MC, Kroeger CM, Aggour E, Calvo Y, Trepanowski JF,Hoddy KK, Varady KA: Effect of exercising while fasting on eatingbehaviors and food intake. J Int Soc Sports Nutr 2013, 10(1):50.

14. Varady KA, Bhutani S, Klempel MC, Lamarche B: Improvements in LDLparticle size and distribution by short-term alternate day modifiedfasting in obese adults. Br J Nutr 2011, 105:580–583.

15. Klempel MC, Kroeger CM, Varady KA: Alternate day fasting increases LDLparticle size independently of dietary fat content in obese humans.Eur J Clin Nutr 2013. Epub ahead of print.

16. Dattilo AM, Kris-Etherton PM: Effects of weight reduction on blood lipidsand lipoproteins: a meta-analysis. Am J Clin Nutr 1992, 56:320–328.

17. Ryan AS, Nicklas BJ: Reductions in plasma cytokine levels with weight lossimprove insulin sensitivity in overweight and obese postmenopausalwomen. Diabetes Care 2004, 27:1699–1705.

18. Jae SY, Fernhall B, Heffernan KS, et al: Effects of lifestyle modifications onC-reactive protein: contribution of weight loss and improved aerobiccapacity. Metabolism 2006, 55:825–831.

19. Kris-Etherton P, Eckel RH, Howard BV, St Jeor S, Bazzarre TL: Nutritioncommittee population science committee and clinical sciencecommittee of the American heart association. AHA science advisory:

Varady et al. Nutrition Journal 2013, 12:146 Page 7 of 8http://www.nutritionj.com/content/12/1/146

Page 25: Intermittent Fasting: Weight Loss and Cardiovascular Effects · nate day fasting (ADF; fasting every other day) is organ-ized with alternating “feast days,” on which there is

RESEARCH Open Access

Improvement in coronary heart disease riskfactors during an intermittent fasting/calorierestriction regimen: Relationship toadipokine modulationsCynthia M Kroeger1, Monica C Klempel1, Surabhi Bhutani1, John F Trepanowski1, Christine C Tangney2

and Krista A Varady1*

Abstract

Background:The ability of an intermittent fasting (IF)-calorie restriction (CR) regimen (with or without liquid meals)to modulate adipokines in a way that is protective against coronary heart disease (CHD) has yet to be tested.

Objective:Accordingly, we examined the effects of an IFCR diet on adipokine profile, body composition, andmarkers of CHD risk in obese women.

Methods:Subjects (n = 54) were randomized to either the IFCR-liquid (IFCR-L) or IFCR-food based (IFCR-F) diet for10 weeks.

Results:Greater decreases in body weight and waist circumference were noted in the IFCR-L group (4 ± 1 kg; 6 ± 1 cm)versus the IFCR-F group (3 ± 1 kg; 4 ± 1 cm). Similar reductions (P < 0.0001) in fat mass were demonstrated in theIFCR-L (3 ± 1 kg) and IFCR-F group (2 ± 1 kg). Reductions in total and LDL cholesterol levels were greater (P = 0.04)in the IFCR-L (19 ± 10%; 20 ± 9%, respectively) versus the IFCR-F group (8 ± 3%; 7 ± 4%, respectively). LDL peakparticle size increased (P < 0.01) in the IFCR-L group only. The proportion of small LDL particles decreased (P < 0.01)in both groups. Adipokines, such as leptin, interleukin-6 (IL-6), tumor necrosis factor-alpha (TNF-alpha), andinsulin-like growth factor-1 (IGF-1) decreased (P < 0.05), in the IFCR-L group only.

Conclusion:These findings suggest that IFCR with a liquid diet favorably modulates visceral fat and adipokines in away that may confer protection against CHD.

Keywords:Intermittent fasting, Calorie restriction, Liquid diet, Body weight, Visceral fat, Cholesterol, Coronary heartdisease, Obese women

IntroductionIntermittent fasting (IF) is a novel weight loss regimenthat has been steadily growing in popularity over thepast decade [1]. This diet strategy generally involves se-vere restriction (75-90% of energy needs) on 1–2 daysper week. Though clinical trial evidence is still limited[2,3], preliminary findings indicate that IF may be effect-ive for weight loss and coronary heart disease (CHD)risk reduction. For instance, two recent trials of IF

demonstrate decreases in body weight of 7-9% andreductions in LDL cholesterol of 10% after 20–24 weeksof treatment [2,3]. While these data are encouraging,this diet therapy is limited in that a long duration oftime, i.e. 24 weeks, is required to experience modestreductions in weight. One possible way to boost the rateof weight loss would be to combine IF with daily calorierestriction (CR). In following this protocol, the individ-ual would fast one day per week, and then undergo mildCR, i.e. 20% restriction of energy needs, on 6 days perweek. The incorporation of portion-controlled liquidmeals may also enhance weight loss as it helps indivi-duals to stay within the confines of their prescribed

* Correspondence: [email protected] of Kinesiology and Nutrition, University of Illinois at Chicago,Chicago, IL, USAFull list of author information is available at the end of the article

© 2012 Kroeger et al.; licensee BioMed Central Ltd. This is an Open Access article distributed under the terms of the CreativeCommons Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, andreproduction in any medium, provided the original work is properly cited.

Kroeger et al. Nutrition & Metabolism 2012, 9:98http://www.nutritionandmetabolism.com/content/9/1/98

Page 26: Intermittent Fasting: Weight Loss and Cardiovascular Effects · nate day fasting (ADF; fasting every other day) is organ-ized with alternating “feast days,” on which there is

energy goals [4,5]. The effect of IF combined with CR(with or without liquid meals) on body weight and CHDrisk has yet to be tested.Although the mechanisms remain unclear, the lipid-

lowering actions of dietary restriction protocols maybe mediated, in part, by modulations in adipokines, i.e.fat-cell derived hormones and cytokines [6]. Leptin isan adipokine that plays a role in CHD development byincreasing platelet aggregation [7], and stimulating theproliferation and migration of endothelial cells [8].Interleukin-6 (IL-6) and tumor necrosis factor-alpha(TNF-alpha) are pro-inflammatory mediators releasedby adipose tissue that are strong independent predic-tors of CHD [9]. C-reactive protein (CRP) is producedby adipose tissue in response to a rise in IL-6 [10].CRP may play a role in atherogenesis by binding tooxidized LDL and promoting foam cell formation [11].Isoprostanes are another group of compounds releasedby adipose tissue [12]. Isoprostanes act as markers ofoxidative stress, and have been shown to accumulatein atherosclerotic lesions of carotid arteries derivedfrom CHD patients [12]. Insulin-like growth hormone-1 (IGF-1), another adipose tissue-derived protein, mayplay a role in the development of CHD by stimulatingthe proliferation of vascular smooth muscle cells [13].Circulating concentrations of these hormones are dic-tated by regional fat distribution [14]. Excess visceraladiposity, as determined by an increased waist circum-ference, is related to an increased incidence of dyslipi-demia [14]. Viscerally obese women (defined as a waistcircumference >88 cm) have higher circulating levelsof each of the above-mentioned adipokines, relative tosubcutaneously obese women [15]. The ability of IFCRto reduce visceral fat mass, and in turn, improve circulatingadipokine profile, remains unknown.Accordingly, the objective of the present study was to

examine the effect of IFCR with a liquid-diet or foodbased diet on body weight and lipid risk factors for CHDin obese women, and to evaluate how changes in adipo-kines are related to these modulations in vascular diseaserisk.

MethodsSubjectsObese women were recruited from the Chicago area bymeans of advertisements placed on and around theUniversity of Illinois campus. Seventy-seven individualsresponded to the advertisements, and 60 were deemedeligible to participate after the preliminary question-naire, body mass index (BMI) and waist circumferenceassessment. Key inclusion criteria were as follows: female,age 35–65 y, BMI between 30 and 39.9 kg/m2, waist cir-cumference >88 cm, weight stable for 3 months prior tothe beginning of the study, i.e. <5 kg weight loss or gain,

non-diabetic, no history of cardiovascular disease, no his-tory of cancer, sedentary or lightly active for 3 monthsprior to the beginning of the study, i.e. <3 h/week oflight-intensity exercise at 2.5–4.0 metabolic equivalents(METS), non-smoker, and not taking weight loss,lipid-lowering, or glucose-lowering medications. Peri-menopausal women were excluded from the study, andpostmenopausal women (defined as absence of mensesfor 2 y) were required to maintain their current hormonereplacement therapy regimen for the duration of thestudy. The experimental protocol was approved by theOffice for the Protection of Research Subjects at theUniversity of Illinois, Chicago, and all volunteers gavewritten informed consent to participate in the trial.

Diet interventionsSubjects were randomized by way of a stratified randomsample, based on BMI and age, into either the IFCR-liquid diet (IFCR-L) group (n = 30) or IFCR-food baseddiet (IFCR-F) group (n = 30). A random number tablewas used to randomize the subjects from each strata intothe intervention groups. The 10-week trial consisted oftwo dietary phases: 1) a 2-week baseline weight mainten-ance period, and 2) an 8-week weight loss period.

Baseline weight maintenance diet (Week 1–2)Each subject participated in a 2-week baseline weightmaintenance period before commencing the 8-weekweight loss intervention (Figure 1). During this period,subjects were requested to continue eating their usualdiet and to maintain a stable body weight.

Weight loss diets (Week 3–10)After the baseline period, subjects partook in either theIFCR-L or IFCR-F intervention for 8 weeks (Figure 1).The Mifflin equation was used to measure energyrequirements [16]. IFCR-L subjects (n = 30) consumed acalorie-restricted liquid diet for the first 6 days of theweek, and then underwent a fast on the last day of theweek (water consumption + 120 kcal of juice powderonly, for 24 h). The liquid diet consisted of a liquid mealreplacement for breakfast (240 kcal) and a liquid mealreplacement for lunch (240 kcal). All liquid meal repla-cements were provided to the subjects in powder-formin premeasured envelopes (Isalean Shake, Isagenix LLC,Chandler, AZ). For the dinnertime meal, IFCR-L subjectsconsumed a 400–600 kcal meal. Food was not providedto the subjects for the dinner meal. Instead, subjects metwith a Registered Dietician weekly to learn how to makehealthy eating choices that are in compliance with theNational Cholesterol Education Program TherapeuticLifestyle Changes (TLC) diet (i.e. <35% of kcal as fat;50-60% of kcal as carbohydrates; <200 mg/d of dietarycholesterol; and 20–30 g/d of fiber). In following this

Kroeger et al. Nutrition & Metabolism 2012, 9:98 Page 2 of 8http://www.nutritionandmetabolism.com/content/9/1/98

Page 27: Intermittent Fasting: Weight Loss and Cardiovascular Effects · nate day fasting (ADF; fasting every other day) is organ-ized with alternating “feast days,” on which there is

7 d intervention, IFCR-L subjects were energy restrictedby 30% of their baseline needs. IFCR-F subjects (n = 30)consumed a calorie-restricted food-based diet for thefirst 6 days of the week, and then underwent a fast on thelast day of the week (water consumption + 120 kcal ofjuice powder only, for 24 h). IFCR-F subjects ate 3 mealsper day in accordance with the TLC diet guidelines. Foodwas not provided to the subjects. Instead, subjects metwith a Registered Dietician weekly to learn how to makehealthy eating choices by implementing the TLC guide-lines. Subjects were instructed to eat approximately 240kcal for breakfast, 240 kcal for lunch, and 400–600 kcalfor dinner. In following this 7 d intervention, IFCR-Fsubjects were energy restricted by 30% of their baselineneeds.

AnalysesDietary intake and physical activity assessmentA multiple-pass, telephone-administered, 24-h recall wasused to assess dietary intake. The recalls were performedat weeks 1, 3 and 10 by a trained Registered Dietician.Dietary intake data were analyzed using Nutrition DataSystem (NDS) software (version 2012; University of Min-nesota, Minneapolis, MN). Furthermore, IFCR-L subjectswere provided with a checklist each day to monitor: 1)adherence to the liquid meal protocol, and 2) adherenceto the fast day regimen. IFCR-F subjects were also givena checklist to monitor their adherence to the fast dayregimen. Alterations in energy expenditure associatedwith changes in physical activity were measured by theuse of a pattern recognition monitor (Sense Wear Mini(SWM), Bodymedia, Pittsburgh, PA). Subjects wore thelightweight monitor on their upper arm for 7 d atweek 3 and 10 of the trial. The data was analyzedusing Bodymedia Software V.7.0, algorithm V.2.2.4(Bodymedia, Pittsburgh, PA).

Hunger, satisfaction, and fullness assessmentSubjects completed a validated visual analog scale (VAS)on each fast day approximately 5 min before going tobed (reported bedtime ranged from 8.00 pm to 1.20 am).In brief, the VAS consisted of 100-mm lines, and sub-jects were asked to make a vertical mark across the linecorresponding to their feelings from 0 (not at all) to 100(extremely) for hunger, satisfaction with diet, or fullness.

The VAS was collected at the weigh-in each week andreviewed for completeness. Quantification was per-formed by measuring the distance from the left end ofthe line to the vertical mark.

Body weight and body composition assessmentBody weight measurements were taken to the nearest0.5 kg at the beginning of every week in light clothingand without shoes using a balance beam scale (Health-OMeter; Sunbeam Products, Boca Raton, FL). Heightwas assessed using a wall-mounted stadiometer to thenearest 0.1 cm. BMI was assessed as kg/m2. Fat massand fat free mass were assessed by dual energy X-rayabsorptiometry (DXA) at weeks 1, 3 and 10 (QDR4500 W, Hologic Inc. Arlington, MA). Waist circum-ference was measured by a flexible tape to the nearest0.1 cm, midway between the lower costal margin andsuper iliac crest during a period of expiration.

Plasma lipids and adipokine assessmentFasting blood samples were collected between 6.00 amand 9.00 am at weeks 1, 3 and 10 after a 12-h fast. Bloodwas centrifuged for 10 min at 520 × g at 4°C to separateplasma from red blood cells and was stored at −80°Cuntil analyzed. Plasma total cholesterol, direct LDL chol-esterol, HDL cholesterol, and triglyceride concentrationswere measured in duplicate by enzymatic kits (BiovisionInc, Mountainview, CA). LDL particle size was measuredby linear polyacrylamide gel electrophoresis (Quantime-trix Lipoprint System, Redondo Beach, CA), as previouslydescribed [17]. Leptin, IL-6, TNF-alpha, CRP, 8-isopros-tane, and IGF-1 were assessed in duplicate at week 1, 3,and 10 by ELISA (R&D Systems, Minneapolis, MN).

StatisticsResults are presented as mean ± SEM. Sample size wascalculated as n = 30 subjects per group, assuming a 5%decrease in body weight in the IFCR-L group, with apower of 80% and an α risk of 5%. An independent sam-ples t-test was used to test baseline differences betweengroups. Repeated-measures ANOVA was performed,taking time as the within-subject factor and diet as thebetween-subject factor, to assess differences betweengroups over the course of the study. Post-hoc analyseswere performed using the Tukey test. Differences were

Figure 1 Experimental design.

Kroeger et al. Nutrition & Metabolism 2012, 9:98 Page 3 of 8http://www.nutritionandmetabolism.com/content/9/1/98

Page 28: Intermittent Fasting: Weight Loss and Cardiovascular Effects · nate day fasting (ADF; fasting every other day) is organ-ized with alternating “feast days,” on which there is

considered significant at P < 0.05. All data was analyzedusing SPSS software (version 20.0, SPSS Inc, Chicago, IL).

ResultsSubject dropout and baseline characteristicsSixty subjects (IFCR-L n = 30, IFCR-F n = 30) commencedthe study. Two subjects dropped out of the IFCR-L groupdue to scheduling conflicts (n = 1) and problems adheringto the diet (n = 1). Four subjects dropped out of the IFCR-F group because of scheduling conflicts (n = 2) and inabil-ity to adhere to the diet protocol (n = 2). Thus, a total of28 and 26 subjects completed the IFCR-L and IFCR-Finterventions, respectively. There were no differences be-tween groups for age, ethnicity, or BMI (Table 1).

Dietary intake and physical activityDiet and physical activity data are displayed in Table 2.Adherence to the fast day protocol was similar betweengroups (P = 0.91) (IFCR-l: 96 ± 4% compliance; IFCR-F:98 ± 3% compliance). Compliance with the liquid mealprotocol was 92 ± 3% in the IFCR-L group over thecourse of the 8 weeks. Energy intake decreased (P < 0.05)in both the IFCR-L and IFCR-F groups between week 3and 10. There were no changes in fat, protein, carbohy-drate, cholesterol, or fiber intake from the beginning tothe end of the study in either group. Activity expenditureand steps/d remained stable over the course of the trial inboth intervention groups. Hours of sleep per night alsodid not change during the 8-week weight loss period ineither the IFCR-L or IFCR-F group.

Hunger, satisfaction, and fullness assessmentHunger scores were low, and did not differ over thecourse of the trial in the IFCR-L (week 3: 27 ± 8 mm,week 10: 28 ± 7 mm) or IFCR-F group (week 3: 46 ± 7mm, week 10: 39 ± 7 mm). Satisfaction with the dietsremained elevated from the beginning to the end of the

study in the IFCR-L (week 3: 72 ± 7 mm, week 10: 78 ±6 mm) and IFCR-F group (week 3: 55 ± 9 mm, week 10:66 ± 6 mm). Fullness scores were moderate and stableduring the trial in the IFCR-L (week 3: 66 ± 7 mm, week10: 82 ± 6 mm) or IFCR-F group (week 3: 58 ± 7 mm,week 10: 64 ± 6 mm). No differences were noted be-tween groups for hunger, satisfaction or fullness at eithertime point.

Body weight and body compositionChanges in body weight and body composition arereported in Table 3. Body weight remained stable in boththe IFCR-L group (week 1: 95 ± 3, week 3: 95 ± 3 kg) andIFCR-F group (week 1: 94 ± 3, week 3: 94 ± 3 kg) duringthe weight maintenance period. Body weight decreased toa greater extent (P < 0.05) in the IFCR-L group (4 ± 1 kg)versus the IFCR-F group (3 ± 1 kg) during the weight lossperiod. Similar decreases (P < 0.0001) in fat mass wereobserved in the IFCR-L (3 ± 1 kg) and IFCR-F (2 ± 1 kg)groups after 8 weeks of treatment. Fat free mass remainedunchanged throughout the course of the trial in bothgroups. Greater decreases (P < 0.05) in waist circumfer-ence were demonstrated in the IFCR-L (6 ± 1 cm) whencompared to the IFCR-F group (4 ± 1 cm). BMI decreased(P < 0.0001) by 1 ± 1 and 1 ± 1 kg/m2, respectively, in theIFCR-L and IFCR-F groups.

Plasma lipids and adipokinesChanges in plasma lipid concentrations and LDL particlesize are displayed in Figure 2. Total and LDL cholesterolconcentrations were reduced (P < 0.05) to a greaterextent in the IFCR-L group (19 ± 10%; 20 ± 9%, respect-ively) compared to the IFCR-F group (8 ± 3%; 7 ± 4%,respectively). HDL cholesterol was not changed by eitherdiet. Triglycerides decreased (P < 0.0001) in the IFCR-Lgroup only (17 ± 9%). LDL peak particle size increased(P < 0.01) in the IFCR-L group only, while the propor-tion of small particles was reduced (P < 0.05) in the boththe IFCR-L and IFCR-F groups. Changes in adipokinesare reported in Table 4. Leptin decreased (P < 0.05) to asimilar extent in the IFCR-L (−9 ± 2 ng/ml) and theIFCR-F group (−8 ± 2 ng/ml). IL-6, TNF-alpha, andIGF-1 concentrations were reduced (P < 0.05) in theIFCR-L group only. Significant associations betweenplasma lipids and adipose tissue parameters were notedin the IFCR-L group only. For instance, decreases inLDL cholesterol were related to reductions in waist circum-ference (r = 0.26, P = 0.04), and leptin (r = 0.37, P = 0.04).Reductions in triglycerides were also related todecreased leptin (r = 0.29, P = 0.04) and TNF-alpha(r = 0.33, P = 0.03). Furthermore, decreased waist cir-cumference was related to lower circulating leptinlevels (r = 0.45, P = 0.03).

Table 1 Subject characteristics at baseline1

Characteristic IFCR-L IFCR-F

n 28 26

Age (y) 47 ± 2 48 ± 2

Ethnicity

African American 16 18

Asian 3 2

Caucasian 4 2

Hispanic 5 4

Body weight (kg) 95 ± 3 94 ± 3

Height (cm) 165 ± 2 164 ± 2

Body mass index (kg/m2) 35 ± 1 35 ± 11 Values reported as mean ± SEM. IFCR-L: Intermittent fasting calorierestriction-liquid diet; IFCR-F: Intermittent fasting calorie restriction-food baseddiet. No differences between groups for any parameter (Independentsamples t-test).

Kroeger et al. Nutrition & Metabolism 2012, 9:98 Page 4 of 8http://www.nutritionandmetabolism.com/content/9/1/98

Page 29: Intermittent Fasting: Weight Loss and Cardiovascular Effects · nate day fasting (ADF; fasting every other day) is organ-ized with alternating “feast days,” on which there is

DiscussionThis study is the first to show that IFCR with liquidmeals can produce potent decreases in CHD risk, andthat these effects are mediated in part by improvementsin adipokines. More specifically, we show here thatIFCR-L is an effective diet therapy to modulate lipidindicators of CHD risk, i.e. reduce LDL cholesterol, tri-glycerides, and the proportion of small LDL particles,while increasing LDL peak particle size. Our findingsalso demonstrate that these favorable changes in lipidswere related to reduced waist circumference (visceral fatmass) and reductions in pro-atherogenic adipokines,such as leptin and TNF-alpha.Although both of the interventions produced favorable

changes in lipids, superior modulations were shown inthe IFCR-L group when compared to the IFCR-F group.The reductions in plasma lipids by IFCR-L (LDL-

cholesterol: 19%, triglycerides: 20%) are similar to whathas been reported in previous trials of IF [2,3]. For in-stance, in a trial by Williams et al. [3], obese subjectsconsumed a very-low calorie diet (VLCD; <500 kcal/d) 1day per week, and ate ad libitum every other day of theweek. After 20 weeks of treatment, LDL cholesteroland triglyceride concentrations decreased by 10% and52%, respectively [3]. In the trial by Harvie et al. [2],obese women underwent 2 days of VLCD (600 kcal/d)and ate ad libitum on every other day of the week, for24 weeks. Post-treatment LDL cholesterol and trigly-ceride levels were reduced by 10% and 17% from base-line [2]. The mechanism by which IFCR modulatescirculating lipid concentrations is not clear. Nonethe-less, recent evidence from CR studies indicate that theoxidation of circulating free fatty acids (FFA) is increasedduring periods of weight loss, while FFA synthesis is

Table 2 Dietary intake and physical activity during the weight loss period1

IFCR-L IFCR-F

Diet variables Week 3 Week 10 Change 2 Week 3 Week 10 Change 2

Energy (kcal) 1708 ± 135 1255 ± 102 3 −453 ± 210 1694 ± 180 1444 ± 132 3 −250 ± 146

Total fat (g) 54 ± 6 51 ± 11 −3 ± 14 62 ± 9 47 ± 5 −15 ± 11

Saturated fat (g) 20 ± 2 18 ± 3 −2 ± 4 22 ± 4 18 ± 2 −4 ± 4

Monounsaturated fat (g) 22 ± 2 17 ± 3 −5 ± 5 26 ± 4 17 ± 1 −9 ± 4

Polyunsaturated fat (g) 12 ± 2 16 ± 5 4 ± 6 14 ± 2 12 ± 2 −2 ± 3

Protein (g) 75 ± 5 84 ± 12 9 ± 14 67 ± 6 65 ± 9 −2 ± 4

Carbohydrates (g) 226 ± 20 215 ± 49 −11 ± 60 210 ± 23 174 ± 20 −36 ± 20

Cholesterol (mg) 215 ± 27 196 ± 36 −19 ± 51 224 ± 43 169 ± 28 −55 ± 53

Fiber 17 ± 1 23 ± 7 6 ± 8 20 ± 2 16 ± 2 −4 ± 2

Physical activity variables Week 3 Week 10 Change 2 Week 3 Week 10 Change 2

Activity expenditure (kcal/d) 249 ± 28 283 ± 27 34 ± 76 246 ± 37 258 ± 43 12 ± 25

Steps (steps/d) 6975 ± 526 7375 ± 426 400 ± 261 5876 ± 621 6405 ± 599 529 ± 610

Sleep (h/d) 6.4 ± 0.4 6.0 ± 0.2 −0.4 ± 1.0 6.2 ± 0.5 5.5 ± 0.4 0.7 ± 0.51 Values reported as mean ± SEM. IFCR-L: IFCR-L: Intermittent fasting calorie restriction-liquid diet (n = 28); IFCR-F: Intermittent fasting calorie restriction-foodbased diet (n = 26).2 Change expressed as the difference between week 3 and week 10 absolute values.3 Week 3 values significantly (P < 0.05) different from week 10 values within group (Repeated-measures ANOVA).No significant difference between groups for any nutrient except energy.

Table 3 Body weight and body composition changes during the weight loss period1

IFCR-L IFCR-F

Week 3 Week 10 Change 2 Week 3 Week 10 Change 2

Body weight (kg) 95 ± 3 91 ± 3 3 −4 ± 1 4 94 ± 2 91 ± 2 3 −3 ± 1

Fat mass (kg) 45 ± 1 42 ± 1 3 −3 ± 1 45 ± 1 43 ± 1 3 −2 ± 1

Fat free mass (kg) 50 ± 1 49 ± 2 −1 ± 1 49 ± 1 48 ± 1 −1 ± 1

Waist circumference (cm) 103 ± 1 97 ± 1 3 −6 ± 1 4 102 ± 3 98 ± 3 3 −4 ± 1

Body mass index (kg/m2) 35 ± 1 34 ± 1 3 −1 ± 1 35 ± 1 34 ± 1 3 −1 ± 11 Values reported as mean ± SEM. IFCR-L: IFCR-L: Intermittent fasting calorie restriction-liquid diet (n = 28); IFCR-F: Intermittent fasting calorie restriction-foodbased diet (n = 26).2 Change expressed as the difference between week 3 and week 10 absolute values.3 Week 3 values significantly (P < 0.05) different from week 10 values within group (Repeated-measures ANOVA).4 Absolute change significantly different (P < 0.05) between the IFCR-L and IFCR-F group (Repeated-measures ANOVA).

Kroeger et al. Nutrition & Metabolism 2012, 9:98 Page 5 of 8http://www.nutritionandmetabolism.com/content/9/1/98

Page 30: Intermittent Fasting: Weight Loss and Cardiovascular Effects · nate day fasting (ADF; fasting every other day) is organ-ized with alternating “feast days,” on which there is

decreased [18]. Lower availability of precursor FFAresults in a reduction in hepatic synthesis and secretionof very-low density lipoprotein (VLDL) into plasma.Lower secretion of VLDL contributes to reduced plasmaconcentrations of LDL, since VLDL is quickly convertedto LDL in the circulation [19]. Although this mechanismhas only been demonstrated for CR, it possible that IFCRmay alter circulating lipids in a similar fashion.The greater improvement in lipid profile by IFCR-L is

most likely due to the more pronounced reductions inbody weight and visceral fat mass observed. After 8weeks of treatment, body weight and waist circumfer-ence, an indirect indicator of visceral fat, decreased to agreater extent in the IFCR-L (4 kg and 6 cm, respect-ively) when compared to the IFCR-F group (3 kg and 4cm, respectively). The greater weight loss by the IFCR-Lgroup is not surprising as these individuals had a larger

daily calorie deficit relative to the IFCR-F group (453kcal/d, 250 kcal/d, respectively). These decreases in waistcircumference in the IFCR-L group were related toreductions in LDL cholesterol concentrations. An accu-mulation of adipose tissue in visceral depots may con-tribute to the development of dyslipidemia in severalways. For instance, lipolysis of fat tissue in visceral adi-pocytes is higher than that of subcutaneous adipocytes.This can lead to an augmented efflux of FFA from vis-ceral depots [20]. These FFAs released from visceral fatare then collected by the portal vein and reach the liverat much higher concentrations than they do the systemiccirculation [20]. In the liver, these FFA trigger the aug-mented production of triglycerides, and the elevated se-cretion of VLDL [20]. The increased secretion of VLDLthen results in higher plasma levels of LDL [20]. In viewof this, it is conceivable that the decrease in visceral fat

Figure 2 Changes in lipid indicators of coronary heart disease risk during the weight loss period. Values reported as mean ±SEM changebetween week 3 and 10. IFCR-L: Intermittent fasting calorie restriction-liquid diet (n = 28); IFCR-F. Intermittent fasting calorie restriction-foodbased diet (n = 26). A. Total cholesterol. B. LDL cholesterol. C. HDL cholesterol. D. Triglycerides. E. LDL peak particle size. F. Proportion of smallLDL particles. *Week 3 values significantly (P < 0.05) different from week 10 values within E group (Repeated-measures ANOVA). # Significantlydifferent (P < 0.05) between the IFCR-L and IFCR-F group (Repeated-measures ANOVA).

Table 4 Changes in adipokines during the weight loss period1

IFCR-L IFCR-F

Week 3 Week 10 Change 2 Week 3 Week 10 Change 2

Leptin (ng/ml) 36.5 ± 2.5 27.1 ± 3.1 3 −9.4 ± 1.6 37.6 ± 2.5 29.4 ± 2.4 3 −8.2 ± 1.7

IL-6 (pg/ml) 3.1 ± 0.3 2.5 ± 0.3 3 −0.6 ± 0.3 3.1 ± 0.4 3.3 ± 0.4 0.2 ± 0.4

TNF-alpha (pg/ml) 1.6 ± 0.3 1.2 ± 0.3 3 −0.4 ± 0.1 1.2 ± 0.3 1.1 ± 0.2 −0.1 ± 0.1

C-Reactive protein (mg/dl) 0.4 ± 0.1 0.4 ± 0.1 0 ± 0.1 0.6 ± 0.2 0.4 ± 0.1 −0.2 ± 0.2

8-isoprostane (pmol/mg) 1.8 ± 0.1 1.9 ± 0.1 0.1 ± 0.1 1.8 ± 0.1 1.9 ± 0.1 0.1 ± 0.1

IGF-1 (ng/ml) 67.7 ± 3.8 60.8 ± 4.2 3 −6.9 ± 2.3 69.6 ± 3.8 67.6 ± 5.4 −2.0 ± 4.91 Values reported as mean ± SEM. IFCR-L: Intermittent fasting calorie restriction-liquid diet (n = 28); IFCR-F: Intermittent fasting calorie restriction-food based diet(n = 26). IL-6: Interleukin-6, TNF-alpha: Tumor necrosis factor-alpha, IGF-1: Insulin-like growth factor-1.2 Change expressed as the difference between week 3 and week 10 absolute values.3 Week 3 values significantly (P < 0.05) different from week 10 values within group (Repeated-measures ANOVA).

Kroeger et al. Nutrition & Metabolism 2012, 9:98 Page 6 of 8http://www.nutritionandmetabolism.com/content/9/1/98

Page 31: Intermittent Fasting: Weight Loss and Cardiovascular Effects · nate day fasting (ADF; fasting every other day) is organ-ized with alternating “feast days,” on which there is

by IFCR played a role in the reduced LDL cholesterolconcentrations shown here.Decreases in pro-atherogenic adipokines, such as leptin

(26%), IL-6 (19%), TNF-alpha (25%), and IGF-1 (10%)were observed by the IFCR-L group. No changes in adi-pokines were noted in the IFCR-F group, however, whichis most likely due to insufficient weight loss to achievechanges in these parameters [21]. The decreases in leptinnoted here are similar to those observed by Harvie et al.[2]. After 24 weeks of IF, obese women experiencedpotent reductions in leptin of 40% [2]. The greater reduc-tions in leptin in this previous trial are most likely due tothe greater weight loss achieved with the longer trial dur-ation [22]. In the IFCR-L group, lower plasma leptin wasrelated to decreased triglyceride levels. In vitro studiesdemonstrate that leptin is a potent stimulator of lipolysisand fatty acid oxidation in adipocytes and other cell types[23]. Accordingly, leptin is also a regulator of circulatingtriacylglycerol concentrations [23]. Reduced leptin levelswere also related to lower LDL cholesterol levels post-treatment, though the mechanism by which leptin maybe involved in reducing LDL cholesterol concentrationsis not clear. We also observed a positive association be-tween visceral fat mass and leptin levels. Thus, it is pos-sible that the decrease in visceral fat by the IFCR-Lintervention stimulated a reduction in leptin, which inturn contributed to the lipid profile improvementsdemonstrated here. Reductions in TNF-alpha werealso correlated to decreases in plasma triglycerides.Evidence in rodent models indicates that TNF-alphainduces hyperlipidemia by increasing hepatic triglycer-ide production [24]. Thus, a reduction in circulatingTNF-alpha by IFCR-L may be related to the reduc-tions in triglycerides observed.This study is limited in that it did not carefully control

for food intake by providing food-based meals to the inter-vention groups, i.e. dinner meal for the IFCR-L group, and3 meals/d for the IFCR-F group. Providing all the mealsduring the study would allow for a more precise assess-ment of dietary adherence. Another disadvantage is thatonly female subjects were employed, and as such, the ap-plicability of these findings to males remains uncertain.Taken together, our results suggest that IFCR with liquid

meals is an effective diet therapy to reduce body weight,visceral fat mass, and lipid indicators of CHD risk. Ourfindings also demonstrate that the beneficial modulationsin vascular disease risk by IFCR may be mediated, in part,by reductions in visceral fat mass and pro-atherogenic adi-pokines. This study is an important first step to under-standing the underlying mechanisms that mediate thecardio-protective effects of this novel diet regimen.

Competing interestsKAV has a consulting relationship with the sponsor of the research, lsagenix,LLC.

Authors’ contributionsCMK and MCK conducted the clinical trial, analyzed the data, and assistedwith the preparation of the manuscript. SB and JFT performed the laboratoryanalyses, and assisted with data analyses. CCT performed the analysis ofdietary intake data. KAV designed the experiment and wrote the manuscript.All authors read and approved the final manuscript.

Funding sourceThis study was funded by Isagenix LLC., Chandler, AZ.

Author details1Department of Kinesiology and Nutrition, University of Illinois at Chicago,Chicago, IL, USA. 2Department of Clinical Nutrition, Rush University, Chicago,IL, USA.

Received: 13 August 2012 Accepted: 5 October 2012Published: 31 October 2012

References1. Mattson MP, Wan R: Beneficial effects of intermittent fasting and caloric

restriction on the cardiovascular and cerebrovascular systems. J NutrBiochem 2005, 16(3):129–137.

2. Harvie MN, Pegington M, Mattson MP, Frystyk J, Dillon B, Evans G, et al: Theeffects of intermittent or continuous energy restriction on weight lossand metabolic disease risk markers: a randomized trial in youngoverweight women. Int J Obes (Lond) 2011, 35(5):714–727.

3. Williams KV, Mullen ML, Kelley DE, Wing RR: The effect of short periods ofcaloric restriction on weight loss and glycemic control in type 2diabetes. Diabetes Care 1998, 21(1):2–8.

4. Heymsfield SB, van Mierlo CA, van der Knaap HC, Heo M, Frier HI: Weightmanagement using a meal replacement strategy: meta and poolinganalysis from six studies. Int J Obes Relat Metab Disord 2003, 27(5):537–549.

5. Tsai AG, Wadden TA: The evolution of very-low-calorie diets: an updateand meta-analysis. Obesity (Silver Spring) 2006, 14(8):1283–1293.

6. Northcott JM, Yeganeh A, Taylor CG, Zahradka P, Wigle JT: Adipokines andthe cardiovascular system: mechanisms mediating health and disease.Can J Physiol Pharmacol 2012, 90(8):1029–1059.

7. Konstantinides S, Schafer K, Koschnick S, Loskutoff DJ: Leptin-dependentplatelet aggregation and arterial thrombosis suggests a mechanism foratherothrombotic disease in obesity. J Clin Invest 2001, 108(10):1533–1540.

8. Goetze S, Bungenstock A, Czupalla C, Eilers F, Stawowy P, Kintscher U, et al:Leptin induces endothelial cell migration through Akt, which is inhibitedby PPARgamma-ligands. Hypertension 2002, 40(5):748–754.

9. Tuomisto K, Jousilahti P, Sundvall J, Pajunen P, Salomaa V: C-reactiveprotein, interleukin-6 and tumor necrosis factor alpha as predictors ofincident coronary and cardiovascular events and total mortality. Apopulation-based, prospective study. Thromb Haemost 2006,95(3):511–518.

10. Madjid M, Willerson JT: Inflammatory markers in coronary heart disease.Br Med Bull 2011, 100:23–38.

11. Singh U, Dasu MR, Yancey PG, Afify A, Devaraj S, Jialal I: Human C-reactiveprotein promotes oxidized low density lipoprotein uptake and matrixmetalloproteinase-9 release in Wistar rats. J Lipid Res 2008,49(5):1015–1023.

12. Basarici I, Altekin RE, Demir I, Yilmaz H: Associations ofisoprostanes-related oxidative stress with surrogate subclinical indicesand angiographic measures of atherosclerosis. Coron Artery Dis 2007,18(8):615–620.

13. Shanahan CM, Weissberg PL: Smooth muscle cell phenotypes inatherosclerotic lesions. Curr Opin Lipidol 1999, 10(6):507–513.

14. Matsuzawa Y, Funahashi T, Nakamura T: The concept of metabolicsyndrome: contribution of visceral fat accumulation and its molecularmechanism. J Atheroscler Thromb 2011, 18(8):629–639.

15. Fontana L, Eagon JC, Trujillo ME, Scherer PE, Klein S: Visceral fat adipokinesecretion is associated with systemic inflammation in obese humans.Diabetes 2007, 56(4):1010–1013.

16. Mifflin MD, St Jeor ST, Hill LA, Scott BJ, Daugherty SA, Koh YO: A newpredictive equation for resting energy expenditure in healthyindividuals. Am J Clin Nutr 1990, 51(2):241–247.

17. Hoefner DM, Hodel SD, O'Brien JF, Branum EL, Sun D, Meissner I, et al:Development of a rapid, quantitative method for LDL subfractionation

Kroeger et al. Nutrition & Metabolism 2012, 9:98 Page 7 of 8http://www.nutritionandmetabolism.com/content/9/1/98

Page 32: Intermittent Fasting: Weight Loss and Cardiovascular Effects · nate day fasting (ADF; fasting every other day) is organ-ized with alternating “feast days,” on which there is

Cioffi et al. J Transl Med (2018) 16:371 https://doi.org/10.1186/s12967-018-1748-4

REVIEW

Intermittent versus continuous energy restriction on weight loss and cardiometabolic outcomes: a systematic review and meta-analysis of randomized controlled trialsIolanda Cioffi1, Andrea Evangelista2, Valentina Ponzo3, Giovannino Ciccone2, Laura Soldati4, Lidia Santarpia1, Franco Contaldo1, Fabrizio Pasanisi1, Ezio Ghigo3 and Simona Bo3*

Abstract Background: This systematic review and meta-analysis summarized the most recent evidence on the efficacy of intermittent energy restriction (IER) versus continuous energy restriction on weight-loss, body composition, blood pressure and other cardiometabolic risk factors.

Methods: Randomized controlled trials were systematically searched from MEDLINE, Cochrane Library, TRIP data-bases, EMBASE and CINAHL until May 2018. Effect sizes were expressed as weighted mean difference (WMD) and 95% confidence intervals (CI).

Results: Eleven trials were included (duration range 8–24 weeks). All selected intermittent regimens provided ≤ 25% of daily energy needs on “fast” days but differed for type of regimen (5:2 or other regimens) and/or dietary instructions given on the “feed” days (ad libitum energy versus balanced energy consumption). The intermittent approach deter-mined a comparable weight-loss (WMD: − 0.61 kg; 95% CI − 1.70 to 0.47; p = 0.87) or percent weight loss (WMD: − 0.38%, − 1.16 to 0.40; p = 0.34) when compared to the continuous approach. A slight reduction in fasting insulin concentrations was evident with IER regimens (WMD = − 0.89 µU/mL; − 1.56 to − 0.22; p = 0.009), but the clinical relevance of this result is uncertain. No between-arms differences in the other variables were found.

Conclusions: Both intermittent and continuous energy restriction achieved a comparable effect in promoting weight-loss and metabolic improvements. Long-term trials are needed to draw definitive conclusions.

Keywords: Continuous energy restriction, Intermittent energy restriction, Fasting glucose, Triglycerides, Weight loss

© The Author(s) 2018. This article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creat iveco mmons .org/licen ses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The Creative Commons Public Domain Dedication waiver (http://creat iveco mmons .org/publi cdoma in/zero/1.0/) applies to the data made available in this article, unless otherwise stated.

BackgroundIn the last decade, much interest has been focused on dietary strategies that manipulate energy intake uncon-ventionally, known as intermittent fasting or intermit-tent energy restriction (IER) [1–4]. This dietary approach has gained greater attention and popularity as a way for

losing weight alternative to the conventional weight-loss diets, characterized by continuous (non-intermittent) energy restriction (CER). The two most popular forms of IER are: the 5:2 diet characterized by two consecu-tive or non-consecutive “fast” days and the alternate-day energy restriction, commonly called alternate-day fast-ing, alternate-day modified fasting, or every-other-day fasting, consisting of a ‘‘fast” day alternated with a ‘‘feed” day [5]. Commonly, during “fast” days, the energy intake is severely restricted, ranging from complete abstinence

Open Access

Journal of Translational Medicine

*Correspondence: [email protected] 3 Department of Medical Sciences, University of Turin, c.so AM Dogliotti 14, 10126 Turin, ItalyFull list of author information is available at the end of the article

Page 33: Intermittent Fasting: Weight Loss and Cardiovascular Effects · nate day fasting (ADF; fasting every other day) is organ-ized with alternating “feast days,” on which there is

Page 2 of 15Cioffi et al. J Transl Med (2018) 16:371

from foods to a daily maximum intake roughly corre-sponding to 75% energy restriction. Therefore, the term “fast” often does not involve a true complete abstinence from caloric intake. The term IER will be used to describe all intermittent energy-restricted/fasting regimens.

The time-restricted feeding [2, 6–9] and the very-low-calorie or energy diets [2, 3] are other types of dietary interventions which were often included in previous sys-tematic reviews and meta-analyses on IER. Indeed, in the former, individuals are allowed to eat within a specific range of time, thus, every day there is a period without food intake, varying from 12 to 21  h [10–12] (i.e. the Muslim Ramadan). On the other hand, there is no daily intermittency in a very-low-calorie-diet, although the overall energy intake may be similar to those of the IER regimens [13].

To the best of our knowledge, an overall evaluation of the impact of IER on multiple metabolic variables, on percent body fat changes, and on the effects of balanced versus ad libitum “feed” days, as well as on the benefits of the different “fasting” regimens is at present lacking.

The primary objective of this systematic review and meta-analysis was to update the efficacy of IER on weight loss, limiting the analyses to regimens which actually included a weekly intermittent energy restriction, i.e. from 1 up to 6 “fast days” per week. Furthermore, the impact of IER on fat mass (FM), fat free mass (FFM), arterial blood pressure (BP) and other cardiometabolic risk factors was assessed. The effects of IER according to the specific type of nutritional regimen on all these out-comes were evaluated too.

Materials and methodsWe followed the Preferred Reporting Items for System-atic Reviews and Meta-Analyses (PRISMA) guidelines in the reporting of this study [14].

Search strategyThe following electronic databases were queried using a combination of search terms until the 31th of May 2018: PubMed (National Library of Medicine), the TRIP database, the Cochrane Library, EMBASE, and Cumu-lative Index to Nursing and Allied Health Literature (CINAHL). The construction of the search strategy was performed using database specific subject headings and keywords. Both medical subject headings (MeSH) and free text search terms were employed. Restrictions to human studies were placed.

The search terms included combinations of “inter-mittent fasting” or “alternate day fasting” or “intermit-tent energy restriction” or “periodic fasting”, and weight loss, weight gain, obesity, weight, fat mass, blood pres-sure, blood glucose, insulin, insulin-resistance, insulin

sensitivity, glycated hemoglobin A1c (HbA1c), type 2 diabetes mellitus (T2DM), cholesterol, and triglycerides (free-term and MESH as possible) (Additional file  1). These search strategies were implemented by hand searching the references of all the included studies and systematic reviews on the field.

Study selectionWe included studies with the following characteristics: (1) randomized controlled trials (RCTs); (2) a detailed description of the IER regimen; (3) 75% of energy restriction on “fast” days, with a maximum cut-off of 500/660  kcal/day for females/males, respectively; (4) weekly intermittency of energy restriction (from 1 up to 6 “fast” days per week); (5) trial duration > 4 weeks; (6) con-taining as comparator a group on a CER regimen and (7) including changes in body weight or percent body weight as one of the study’s outcome.

We excluded studies with the following characteris-tics: (i) uncontrolled trials or study design other than RCTs; (ii) studies not including body weight as an out-come and/or lacking sufficient information on weight change; (iii) including time restricted feeding interven-tion; (iv) reporting very-low-calorie or fasting regimens for > 6  days consecutive/week; and (v) providing > 500–660 kcal/day or not reporting the amount of calorie pre-scribed on “fast” days.

In trials with multiple interventional arms (i.e. exercise arm, intervention arm with specific diets), the IER and the CER arms were considered, while other arms were not analyzed, since out of the scope of this review.

Two authors (IC, SB) separately screened abstracts for their inclusion or exclusion; retrieving full text arti-cles from potentially relevant abstracts. Any discrepancy about inclusion was resolved by discussing with a third author (AE).

OutcomesThe primary outcome of the review was evaluating changes in body weight or in percent body weight. Sec-ondary outcomes were: changes in body mass index (BMI), waist circumference, FM, FFM, arterial BP, and the blood values of fasting glucose and insulin, insu-lin resistance, insulin sensitivity, HbA1c, total choles-terol, HDL- and LDL-cholesterol, and triglycerides. The changes of these outcomes according to the specific type of IER regimen were also evaluated.

Data collection and extractionFrom each included study, the following informa-tion were extracted (1) first author name and year of publication; (2) study design; (3) inclusion criteria of

Page 34: Intermittent Fasting: Weight Loss and Cardiovascular Effects · nate day fasting (ADF; fasting every other day) is organ-ized with alternating “feast days,” on which there is

Page 3 of 15Cioffi et al. J Transl Med (2018) 16:371

participants; (4) trial duration; (5) number of subjects enrolled in each arm; (6) type of dietary intervention; (7) age, gender, BMI of participants; (8) body composi-tion (FM and FFM); (9) systolic (SBP) and diastolic blood pressure (DBP); (10) blood concentrations of fasting glu-cose, HbA1c, insulin, total cholesterol, HDL-cholesterol, LDL-cholesterol, and triglycerides; (11) Homeostasis Model Assessment-Insulin Resistance (HOMA-IR) and insulin-sensitivity index (Si).

Risk of bias assessmentAll studies were independently assessed by two authors (IC, SB) using the “Risk of bias” tool developed by the Cochrane Collaboration for RCTs [15]. The items used for the assessment of each study were the following: ade-quacy of sequence generation, allocation concealment, blinding, addressing of dropouts (incomplete outcome data), selective outcome reporting, and other potential sources of bias. A judgment of “L” indicated low risk of bias, “H” indicated high risk of bias, and “unclear” indi-cated an unclear/unknown risk of bias. The possible disagreements were resolved by consensus, or with con-sultation with a third author (AE).

Data synthesisData synthesis was performed only for the outcomes which were reported by > 3 trials.

The pooled effect sizes were expressed as weighted mean differences (WMD) and 95% confidence interval (CI) between IER and CER arms of the mean outcome values measured at the end of follow-up.

The mean difference of changes from baseline was esti-mated for each study on the basis of reported baseline and follow-up measurements. If the standard deviation for change from baseline was not reported, we imputed missing values assuming a within-patient correlation from baseline to follow-up measurements of 0.8 as sug-gested in the Cochrane handbook [16]. When between-arms mean differences on change from baseline were already estimated [17], those data were included. For the relative weight change from baseline, the non-reported standard deviations were imputed using the mean stand-ard deviation of the available studies.

Random-effects models were applied to provide a sum-mary estimate.

Inter-study heterogeneity was assessed using Cochrane Q statistic and quantified by I2 test [18].

Subgroup analyses for all outcomes were performed based on the different dietary regimen of the “feed” days (balanced vs. ad  libitum food intake) and the effects of the different regimens of “fasting” (5:2 vs. the other

regimens). Weighting of studies was done using generic inverse variance method.

In order to evaluate the influence of each study on the overall effect size, sensitivity analysis was conducted using the one-study remove (leave-one-out) approach.

Potential publication bias was explored using visual inspection funnel plot asymmetry and Egger’s weighted regression tests.

Meta-analyses were performed by using the Stata Metan package (Stata Statistical Software, Release 13; StataCorp LP, College Station, TX); meta-regressions and Egger’s weighted regression tests for publication bias were performed using the metafor package (version 1.9-7) for R (version 3.1.2, R Foundation for Statistical Com-puting, Vienna, Austria).

ResultsIncluded studiesThe initial literature search identified 8577 records. After removing duplicates, 6943 records were screened, and, after excluding articles not meeting the inclusion crite-ria, 94 records were assessed for eligibility. After further analysis and quality assessment, a total of 11 studies were selected for the systematic review and meta-analysis (Fig.  1). All studies identified were RCTs, reporting an IER arm and a CER arm comparison; the corresponding details are shown in Table 1. Data relative to participants involved in exercise-only arms [19] or in high-protein dietary intervention [20] were not considered, because not pertinent to the aims of the study.

Characteristics of the studiesThe total number of subjects included in the present analysis was 630 at enrolment. During the course of the trials, 102 patients dropped out. Drop-out rates ranged from about 2% [21] to 38% for IER arms [22] and from 0% [23] to 50% [22] for CER. The number of participants analyzed at the end of the RCTs was 528.

There was a greater number of women among partici-pants, with the exception of 3 studies with a balanced number between men and women [21, 22, 24] and 1 enrolling only men [23]. Participants were individuals with overweight/obesity; in 2 RCTs patients with T2DM were selected [23, 25], and in 1 RCT patients with mul-tiple dysmetabolic conditions were enrolled [21]. In all RCTs except for 2 [23, 25], participants with a stable weight before the beginning of the study, without his-tory of bariatric surgery, and without drugs impacting on weight or the other study outcomes, were studied.

Trials were performed in UK [20, 22, 26], in USA [17, 19, 25, 27], in Australia [23, 24], and Norway [21, 28]. The duration of the studies ranged from 8 weeks [27] to 24 weeks [17, 21, 23, 26].

Page 35: Intermittent Fasting: Weight Loss and Cardiovascular Effects · nate day fasting (ADF; fasting every other day) is organ-ized with alternating “feast days,” on which there is

Page 4 of 15Cioffi et al. J Transl Med (2018) 16:371

Dietary interventionFour studies prescribed alternating “fast” and “feed” days [17, 19, 27, 28]. Six studies used 2 “fast” days and 5 “feed” days per week (5:2 diet) [21–24, 26]. In 1 RCT, 5 consecutives “fast” days were prescribed before a 1 “fast” day/week regimen per 15 weeks in the IER arm, while the other arm (5 “fast” days every 5 weeks) was not consid-ered, since no intermittence within the same week was present [25]. On “fast” days, diets provided a maximum of 660 kcal/day. In 2 studies, participants were instructed to consume their meals between 12:00 p.m. and 2:00 p.m. on “fast” days to ensure that subjects underwent the same duration of fasting [17, 19]. In 4 studies, meals of “fast” days were partially [17, 25] or totally supplied [19, 27]. In 1 study, a commercially available very-low energy for-mula-based food was assigned in the “fast” days [22].

On “feed” days, 6 studies prescribed healthy and bal-anced eating pattern, according to the energy require-ments [17, 20, 22, 25, 26, 28], 4 allowed for ad  libitum food intake based on the participants’ usual eating

[19, 21–24] and 1 provided a diet based on the energy requirements but allowing the access to 5–7 optional food modules (200  kcal each) [27]. In the comparator arms, energy was restricted by approximately 25% of the daily energy requirements in all studies (CER arms).

Dietary compliance and energy intake assessmentSix studies specifically assessed the compliance to the diet and the overall energy intake in both arms by filling 7-day food records at different time points [17, 20–22, 26, 28]. In 1 study, dieticians evaluated adherence by using patients’ self-recorded dietary diaries and diet histories taken during their dietetic appointments [23]. Either similar adherence between IER and CER [20, 21, 23, 26, 28], a lower [17] or a higher [22] adherence in the IER arms were reported. Adherence to the recommenda-tions in the IER arms ranged from 64% [26] to 93% [22] at the end of the RCTs, but data were difficult to compare because of their incompleteness and the different meth-ods employed to evaluate the compliance.

Published studies iden!fied through database search and addi!onal records iden!fied through other

sources (list of references)(n=8577)

Records screened(n=6943)

Full-text ar!cles assessedfor eligibility

(n=94)

Record not mee!ng the inclusion criteria

(n=6849)

Studies included in quan!ta!ve synthesis

(meta-analysis) (n = 11)

Non original ar!cles or duplicates(n=1634)

Full-text ar!cles excluded, with reasons(n =83)

IER criteria not met (n = 40)Absence of CER arm (n=16)

Interven!on < 4 weeks (n=3)Short treatment dura!on (n=3)

No RCT (n=7)Review (n=14)

Iden

!fica!o

nIn

clude

dSc

reen

ing

Elig

ibili

ty

Fig. 1 Flow of the study

Page 36: Intermittent Fasting: Weight Loss and Cardiovascular Effects · nate day fasting (ADF; fasting every other day) is organ-ized with alternating “feast days,” on which there is

Page 5 of 15Cioffi et al. J Transl Med (2018) 16:371

Tabl

e 1

Char

acte

rist

ics o

f the

 incl

uded

stud

ies

Auth

or(y

ear)

[ref

]

Stud

y de

sign

Part

icip

ants

Tria

l du

ratio

nN

Stud

y gr

oups

Age

(yea

rs)

Mal

es

(n)

BMI (

kg/

m2 )

Wai

st-c

(c

m)

FM FFM

(k

g)

SBP

DBP

(m

mH

g)

Fast

ing

gluc

ose

(mg/

dL)

Hb1

Ac

(%)

Fast

ing

insu

lin

(µU

/mL)

HO

MA-

IR

(mm

ol/L

 * µ

U/

mL)

Tota

l ch

ol

(mg/

dL)

HD

L-c

(mg/

dL)

LDL-

c (m

g/dL

)TG

(mg/

dL)

Anto

ni(2

018)

[22]

RCT

2 pa

ralle

l ar

ms

BMI >

25

kg/m

2

Age

18–6

5 ye

ars

No

com

orbi

dity

8–10

wee

ks15

aIE

R (2

day

s/w

eek)

= 2

5% o

f th

e en

ergy

nee

d on

2 c

onse

cutiv

e fa

st d

ays a

nd

ener

gy a

ccor

ding

to

nee

ds o

n fe

ed

days

42 ±

47

30 ±

131

± 2

123 ±

379

± 2

11 ±

116

2 ±

12

42 ±

410

0 ±

12

97 ±

9

102 ±

358

± 3

74 ±

3N

D1.

6 ±

0.2

12a

CER:

600

kca

l les

s th

an e

stim

ated

ne

eds

48 ±

36

31 ±

134

± 3

115 ±

379

± 4

9 ±

116

2 ±

12

39 ±

410

4 ±

880

± 9

102 ±

256

± 3

75 ±

3N

D1.

3 ±

0.1

Cart

er(2

016)

[24]

RCT

2 pa

ralle

l ar

ms

T2D

MBM

I ≥ 2

7 kg

/m2

Age ≥

18

year

sN

o co

mor

bidi

ty

12 w

eeks

31IE

R (2

day

s/w

eek)

= 4

00–

600

kcal

/day

on

fast

day

s and

no

rest

rictio

n on

feed

da

ys

61 ±

816

35 ±

538

± 9

134 ±

17

ND

ND

ND

ND

ND

ND

ND

55 ±

11

84 ±

10

7 ±

1

32CE

R =

120

0–15

00 k

cal/d

ay

ever

y da

y

62 ±

914

36 ±

540

± 1

113

8 ±

15

ND

ND

ND

ND

ND

ND

ND

54 ±

990

± 1

18 ±

1

Cate

nacc

i(2

016)

[27]

RCT

2 pa

ralle

l ar

ms

BMI ≥

30

kg/m

2

Age

18–5

5 ye

ars

Non

-sm

oker

sN

o di

abet

es, C

V di

seas

es, m

ajor

co

mor

bidi

ty

8 w

eeks

15IE

R (3

day

s/w

eek)

= 0

kca

l/da

y on

fast

day

s al

tern

ate

and

ener

gy a

ccor

ding

to

nee

ds, b

ut w

ith

the

chan

ce to

ask

fo

r mor

e fo

od, o

n fe

ed d

ays

40 ±

10

336

± 4

38 ±

8N

D88

± 7

13 ±

617

0 ±

33

38 ±

810

0 ±

31

143 ±

56

ND

53 ±

9N

DN

D

14CE

R =

400

kca

l les

s th

an n

eeds

43 ±

83

40 ±

649

± 1

0N

D92

± 8

19 ±

617

1 ±

36

39 ±

710

4 ±

29

140 ±

43

ND

61 ±

12

ND

ND

Conl

ey(2

018)

[23]

RCT

2 pa

ralle

l ar

ms

Men

with

BM

I ≥ 3

0 kg

/m2

Age

55–7

5 ye

ars

No

diab

etes

No

high

alc

ohol

in

take

or m

ajor

co

mor

bidi

ty

24 w

eeks

12IE

R (2

day

s/w

eek)

= 6

00 k

cal/

day

on 2

non

-co

nsec

utiv

e fa

st

days

and

ene

rgy

ad li

bitu

m o

n fe

ed

days

68 ±

312

33 ±

2N

D14

2 ±

14

108 ±

27

ND

151 ±

35

45 ±

12

77 ±

31

168 ±

53

114 ±

584

± 1

0N

D

12CE

R =

500

kca

l les

s th

an n

eeds

67 ±

412

36 ±

4N

D15

0 ±

18

110 ±

31

ND

166 ±

39

46 ±

12

98 ±

35

212 ±

150

123 ±

10

88 ±

14

ND

Page 37: Intermittent Fasting: Weight Loss and Cardiovascular Effects · nate day fasting (ADF; fasting every other day) is organ-ized with alternating “feast days,” on which there is

Page 6 of 15Cioffi et al. J Transl Med (2018) 16:371

Tabl

e 1

(con

tinue

d)

Auth

or(y

ear)

[ref

]

Stud

y de

sign

Part

icip

ants

Tria

l du

ratio

nN

Stud

y gr

oups

Age

(yea

rs)

Mal

es

(n)

BMI (

kg/

m2 )

Wai

st-c

(c

m)

FM FFM

(k

g)

SBP

DBP

(m

mH

g)

Fast

ing

gluc

ose

(mg/

dL)

Hb1

Ac

(%)

Fast

ing

insu

lin

(µU

/mL)

HO

MA-

IR

(mm

ol/L

 * µ

U/

mL)

Tota

l ch

ol

(mg/

dL)

HD

L-c

(mg/

dL)

LDL-

c (m

g/dL

)TG

(mg/

dL)

Cout

inho

(201

7)[2

8]

RCT

2 pa

ralle

l ar

ms

BMI =

30–

40 k

g/m

2

Age

18–6

5 ye

ars

Inac

tive,

no

men

o-pa

use

or m

ajor

co

mor

bidi

ty

12 w

eeks

18IE

R (3

day

s/w

eek)

= 5

50–

660

kcal

/day

on

fast

day

s and

en

ergy

acc

ordi

ng

to n

eeds

on

feed

da

ys

39 ±

11

436

± 3

47 ±

8N

DN

DN

DN

DN

DN

DN

D

ND

60 ±

12

17CE

R =

33%

of

ener

gy re

stric

tion

ever

y da

y

39 ±

92

35 ±

443

± 8

ND

ND

ND

ND

ND

ND

ND

ND

55 ±

9

Har

vie

(201

1)[2

6]

RCT

2 pa

ralle

l ar

ms

BMI =

24–

40 k

g/m

2

Prem

enop

ausa

l w

omen

Age

30–4

5 ye

ars

Non

-sm

oker

s, no

t us

ing

OC,

no

maj

or c

omor

-bi

dity

24 w

eeks

53IE

R (2

day

s/w

eek)

= 2

5% o

f th

e en

ergy

nee

d on

2 c

onse

cutiv

e fa

st d

ays a

nd

ener

gy a

ccor

ding

to

nee

ds o

n fe

ed

days

40 ±

40

31 ±

534

(31–

36)

115

(111

–11

9)87

(85–

88)

7 (6

–8)

197

(189

–19

7)58

(54–

58)

119

(112

–12

7)10

6 (88–

124)

102 (9

8–10

5)48

(46–

49)

77 (7

4–79

)N

D1.

5 (1

.3–1

.8)

54CE

R =

25%

ene

rgy

rest

rictio

n ev

ery

day

40 ±

40

31 ±

535

(32–

39)

117

(113

–12

0)87

(83–

88)

7 (6

–9)

200

(193

–20

8)62

(54–

66)

119

(108

–12

7)11

5 (97–

124)

102 (9

9–10

6)49

(48–

50)

75 (7

2–78

)N

D1.

6 (1

.3–1

.8)

Har

vie

(201

3)[2

0]

RCT

3 pa

ralle

l ar

ms

Wom

en w

ith

BMI =

24–

45 k

g/m

2 or b

ody

fat >

30%

of B

WAg

e 30

–45

year

sN

o m

ajor

com

or-

bidi

ty

12 w

eeks

37IE

R (2

day

s/w

eek)

= 2

5% o

f th

e en

ergy

nee

d on

2 c

onse

cutiv

e fa

st d

ays a

nd

ener

gy a

ccor

ding

to

nee

ds o

n fe

ed

days

46 ±

80

30 ±

431

(28–

34)

115

(111

–12

5)88

(85–

90)

6 (5

–8)

204

(191

–22

9)50

(49–

59)

128

(116

–13

9)08

8 (75–

102)

101 (9

7–10

4)48

.5

(46–

50)

ND

51.

6 (1

.3–1

.9)

38IE

R as

abo

ve p

lus

unlim

ited

prot

eins

an

d fa

ts (n

on-S

FA)

on fa

st d

ays

49 ±

70

31 ±

634

(30–

37)

130

(115

–13

8)90

(86–

92)

7 (6

–9)

221

(206

–23

7)55

(51–

59)

144

(130

–15

9)95

(81–

109)

104 (9

9–10

9)49

(47–

51)

ND

61.

9 (1

.5–2

.2)

40CE

R =

dai

ly 2

5%

rest

rictio

n ev

ery

day

48 ±

80

32 ±

636

(32–

39)

124

(116

–13

1)90

(86–

92)

7 (6

–9)

205

(193

–21

8)51

(48–

55)

129

(118

–13

9)97

(83–

111)

106

(102

–11

0)50

(48–

52)

ND

61.

8 (1

.5–2

.2)

Page 38: Intermittent Fasting: Weight Loss and Cardiovascular Effects · nate day fasting (ADF; fasting every other day) is organ-ized with alternating “feast days,” on which there is

Page 7 of 15Cioffi et al. J Transl Med (2018) 16:371

Auth

or(y

ear)

[ref

]

Stud

y de

sign

Part

icip

ants

Tria

l du

ratio

nN

Stud

y gr

oups

Age

(yea

rs)

Mal

es

(n)

BMI (

kg/

m2 )

Wai

st-c

(c

m)

FM FFM

(k

g)

SBP

DBP

(m

mH

g)

Fast

ing

gluc

ose

(mg/

dL)

Hb1

Ac

(%)

Fast

ing

insu

lin

(µU

/mL)

HO

MA-

IR

(mm

ol/L

 * µ

U/

mL)

Tota

l ch

ol

(mg/

dL)

HD

L-c

(mg/

dL)

LDL-

c (m

g/dL

)TG

(mg/

dL)

Sund

for

(201

8)[2

1]

RCT

2 pa

ralle

l ar

ms

BMI =

30–

45 k

g/m

2

Age

21–7

0 ye

ars

Wai

st ≥

84/

90 c

m

(mal

e/fe

mal

e)

plus

ano

ther

co

mpo

nent

of

the

met

abol

ic

synd

rom

eN

o co

mor

bidi

ty o

r al

coho

l/dru

gab

use

24 w

eeks

54IE

R (2

day

s/w

eek)

= 4

00–

600

kcal

/day

on

2 no

n-co

nsec

utiv

e fa

st d

ays a

nd

ener

gy a

s usu

al

on fe

ed d

ays

50 ±

10

2835

± 4

ND

129 ±

13

104 ±

22

ND

192 ±

35

47 ±

13

126 ±

32

162 ±

73

116 ±

10

88 ±

86 ±

1

58CE

R =

400

–600

kca

l le

ss th

an n

eeds

48 ±

12

2835

± 4

ND

128 ±

13

103 ±

13

ND

197 ±

34

45 ±

10

133 ±

32

137 ±

60

116 ±

10

86 ±

96 ±

1

Trep

anow

ski

(201

7)[1

7]

RCT

3 pa

ralle

l ar

ms

BMI =

25.

0–39

.9 k

g/m

2

Age

18–6

5 ye

ars

Non

-sm

oker

s, in

activ

eN

o m

enop

ause

, di

abet

es, C

V di

seas

es

24 w

eeks

34IE

R (a

ltern

ate

d/w

eek)

= 2

5% o

f th

e en

ergy

nee

d on

fast

day

s and

12

5% o

f ene

rgy

need

s on

feed

da

ys

44 ±

10

434

± 4

38 ±

712

4 ±

12

90 ±

12

16 ±

14

188 ±

35

57 ±

14

111 ±

13

101 ±

59

ND

55 ±

983

± 9

ND

ND

35CE

R =

dai

ly 2

5%

rest

rictio

n ev

ery

day

43 ±

12

635

± 4

40 ±

712

2 ±

17

92 ±

18

20 ±

18

184 ±

35

53 ±

11

112 ±

31

97 ±

27

ND

58 ±

12

80 ±

11

ND

ND

31C =

no

diet

ary

inte

rven

tion

44 ±

11

434

± 4

36 ±

10

121 ±

16

87 ±

816

± 9

190 ±

30

59 ±

13

112 ±

31

98 ±

43

ND

53 ±

10

81 ±

11

ND

ND

Vara

dy(2

011)

[19]

RCT

4 pa

ral-

lel

arm

s

BMI =

25–

39.9

kg/

m2

Age

35–6

5 ye

ars

Non

-sm

oker

s, in

activ

eN

o di

abet

es, C

V di

seas

es

12 w

eeks

15IE

R (3

day

s/w

eek)

= 2

5% o

f th

e en

ergy

nee

d on

fast

day

s and

ad

libi

tum

on

feed

day

s

47 ±

23

32 ±

2N

DN

DN

DN

DN

DN

D51

± 3

141 ±

9N

D

15CE

R =

25%

ene

rgy

rest

rictio

n ev

ery

day

47 ±

32

32 ±

2N

DN

DN

DN

DN

DN

D60

± 6

137 ±

9N

D

15M

oder

ate

exer

cise

pr

ogra

m, n

o di

et

inte

rven

tion

46 ±

32

33 ±

1N

DN

DN

DN

DN

DN

D51

± 4

122 ±

9N

D

15C =

no

diet

ary/

exer

cise

inte

r-ve

ntio

n

46 ±

32

32 ±

2N

DN

DN

DN

DN

DN

D57

± 3

136 ±

10

ND

Tabl

e 1

(con

tinue

d)

Page 39: Intermittent Fasting: Weight Loss and Cardiovascular Effects · nate day fasting (ADF; fasting every other day) is organ-ized with alternating “feast days,” on which there is

Page 8 of 15Cioffi et al. J Transl Med (2018) 16:371

Tabl

e 1

(con

tinue

d)

Auth

or(y

ear)

[ref

]

Stud

y de

sign

Part

icip

ants

Tria

l du

ratio

nN

Stud

y gr

oups

Age

(yea

rs)

Mal

es

(n)

BMI (

kg/

m2 )

Wai

st-c

(c

m)

FM FFM

(k

g)

SBP

DBP

(m

mH

g)

Fast

ing

gluc

ose

(mg/

dL)

Hb1

Ac

(%)

Fast

ing

insu

lin

(µU

/mL)

HO

MA-

IR

(mm

ol/L

 * µ

U/

mL)

Tota

l ch

ol

(mg/

dL)

HD

L-c

(mg/

dL)

LDL-

c (m

g/dL

)TG

(mg/

dL)

Will

iam

s(1

998)

[25]

RCT,

3 pa

ral-

lel

Arm

s

T2D

MAg

e 30

–70

year

sBW

> 2

0% id

eal

Not

cur

rent

ly o

n in

sulin

, no

liver

, re

nal, h

eart

di

seas

es

20 w

eeks

18IE

R (1

day

/w

eek)

= 4

00–

600

kcal

/day

on

fast

day

and

15

00–1

800

kcal

/da

y on

feed

day

s

51 ±

89

35 ±

5N

DN

D17

7 ±

56

20 ±

11

215 ±

37

41 ±

813

2 ±

35

197 ±

83

ND

8 ±

2N

D

18IE

R (5

day

s/w

eek)

= 4

00–

600

kcal

/day

on

fast

day

eve

ry

5 w

eeks

and

15

00–1

800

kcal

/da

y on

feed

day

s

50 ±

97

37 ±

5N

DN

D18

2 ±

58

17 ±

720

8 ±

39

42 ±

713

1 ±

29

203 ±

239

ND

8 ±

2N

D

18CE

R =

150

0–18

00 k

cal/d

ay

ever

y da

y

54 ±

77

35 ±

5N

DN

D18

4 ±

61

22 ±

921

8 ±

42

46 ±

11

127 ±

48

167 ±

89

ND

8 ±

2N

D

BMI b

ody

mas

s ind

ex, B

W b

ody

wei

ght,

Chol

cho

lest

erol

, CV

card

iova

scul

ar, C

ER c

ontin

uous

ene

rgy

rest

rictio

n, D

BP d

iast

olic

blo

od p

ress

ure,

FFM

fat f

ree

mas

s, FM

fat m

ass,

IER

inte

rmitt

ent e

nerg

y re

stric

tion,

HD

L-c

high

de

nsity

lipo

prot

ein-

chol

este

rol,

LDL-

c lo

w d

ensit

y lip

opro

tein

-cho

lest

erol

, ND

no

data

, OC

oral

con

trac

eptiv

es, R

CT ra

ndom

ized

con

trol

led

tria

l, SF

A sa

tura

ted

fatt

y ac

ids,

SBP

syst

olic

blo

od p

ress

ure,

TG

trig

lyce

rides

, T2D

M

type

2 d

iabe

tes m

ellit

us, W

aist

-c w

aist

circ

umfe

renc

e, W

k w

eek

a Not

ava

ilabl

e th

e ba

selin

e da

ta o

f all

the

rand

omiz

ed p

atie

nts;

the

stud

y re

port

ed th

e ba

selin

e ch

arac

teris

tics o

f the

stud

y co

mpl

eter

s onl

y

Page 40: Intermittent Fasting: Weight Loss and Cardiovascular Effects · nate day fasting (ADF; fasting every other day) is organ-ized with alternating “feast days,” on which there is

Page 9 of 15Cioffi et al. J Transl Med (2018) 16:371

Risk of bias assessmentSome of the analyzed trials were characterized by the lack of information about the randomization procedures (Additional file 2). If blinding of participants was not fea-sible owing to the nature of the interventions, data about blinding of the personnel performing the laboratory or statistical analyses were always unknown, except for 1 study [20]. Dropouts were higher in the IER arms [17, 26, 28] or in the CER arms [20, 22, 24, 25], thus introducing a possible selection bias between-arms, but intention-to treat analyses were performed by all studies, except for 1 RCT [22], where data of the completers only have been reported. Finally, most trials appeared to be free of selec-tive outcome reporting and of other sources of bias, apart from 1, where body weight at baseline was not reported [19].

Meta-analysisAll the outcomes of interest of this systematic review are reported in Additional file 3. Data synthesis was per-formed for the outcomes reported by > 3 trials, therefore data relative to Si values were not pooled.

Weight lossAll RCTs reported weight loss in the IER arms during the intervention, ranging from 5.2% [19] of initial weight to 12.9% [28], while in the CER arms, changes ranged from 4.3% [20] to 12.1% [28] (Additional file  3). Pooled data from random-effect analysis did not show a significant effect of IER on weight loss (WMD: −  0.61  kg, 95% CI − 1.70 to 0.47; p = 0.27) (Fig. 2). The estimated effect on body weight did not change in the leave-one-out sensitiv-ity analysis (data not shown).

Subgroup analyses based on the type of regimen (5:2 vs. other regimens) as well as on the dietary character-istics of the “feed” days of the IER interventions (ad libi-tum vs. balanced food intake) showed consistent results, as reported in Additional file 4. Analyses were repeated after the exclusion of the trial prescribing 5 consecutives “fast” days and then 1 “fast” day/week per 15 weeks [25], and the results did not change (WMD: − 0.36 kg, 95% CI − 1.48 to 0.77; p = 0.54). Finally, the RCT reporting the percent relative variations of the endpoints only [19] was included in the analyses, and the estimated effect size of weight change did not show any between-arms difference (WMD: − 0.08, 95% CI − 0.23 to 0.07; p = 0.29).

Similarly, the percent weight loss was similar in both arms (WMD: − 0.38%, 95% CI − 1.16 to 0.40; p = 0.34) and the results did not differ either in the subgroup anal-yses (Additional file 5) or in the leave-one-out sensitivity analysis.

Other anthropometric measuresSeven out of the 11 included RCTs reported changes in FM and FFM [17, 20, 22, 24, 26–28]. FM was measured by different methods: body impedance analysis (BIA) [20, 22]; dual X-ray absorptiometry (DXA) [17, 24, 27]; impedance [26]; air displacement plethysmography [28]. Pooled results showed no difference between-arms in FM (WMD: − 0.23 kg, 95% CI − 1.23 to 0.77; p = 0.66) as well as in FFM (WMD: − 0.22 kg, 95% CI − 1.01 to 0.56; p = 0.58), as shown in Additional file  6. Those results were consistent both at subgroup analyses and at sensi-tivity analyses. Five RCTs assessed waist circumference [20–23, 26] without showing any differences between arms (WMD: − 0.17 cm; 95% CI − 1.74 to 1.39; p = 0.83).

Cardiometabolic biomarkersPooled data obtained from glucose, HbA1c, insulin and HOMA-IR are presented in Fig.  3a–d respectively. Changes in fasting glucose and HbA1c values were reported respectively in 7 [17, 20–23, 26, 27] and 4 [21, 24–26] trials. Random-effect analysis showed no dif-ference either on glucose (WMD: −  0.49  mg/dL, 95% CI − 1.98 to 0.99; p = 0.51) or HbA1c (WMD: − 0.02%, 95% CI − 0.10 to 0.06; p = 0.62) changes in the IER when compared to CER arms with consistent results in sub-group/sensitivity analyses.

On the contrary, fasting insulin values were signifi-cantly reduced with IER (WMD = −  0.89 µU/mL; 95% CI − 1.56 to − 0.22; p = 0.009; I2 = 0%) and the estimated effect appeared robust in the leave-one-out sensitivity analysis (data not shown). Moreover, subgroup analy-ses showed that the 5:2 regimens were associated with increased insulin reductions (WMD: − 0.99 µU/mL; 95% CI − 1.67 to − 0.30; p = 0.005; I2 = 0) (Additional file 7). All the RCTs evaluating fasting insulin values included a balanced energy regimen for the “feed” days. HOMA-IR values were reduced, though not significantly, in the IER regimens (WMD = − 0.15 mmol/L × µU/mL; 95% CI − 0.33 to 0.02; p = 0.09).

Only 1 RCT evaluated insulin sensitivity (Si) by a fre-quently sampled intravenous glucose tolerance [21], without between-arms differences.

Pooled data obtained from 8 RCTs [17, 20–23, 25–27] did not show any significant effect of IER on triglyc-eride concentrations (WMD: −  3.11  mg/dL, 95% CI −  9.76 to 3.54; p = 0.36) (Fig.  4a). However, subgroup analyses showed a slightly significant triglyceride reduc-tion in the IER arms employing other fasting regi-mens (WMD = − 14.4 mg/dL 95% CI − 28.6 to − 0.23; p = 0.046; I2 = 0%). Characteristics of the “feed” days were not associated with differences in triglyceride changes (Additional file 8). HDL-cholesterol levels increased after IER regimens, albeit not significantly (WMD = 1.72 mg/

Page 41: Intermittent Fasting: Weight Loss and Cardiovascular Effects · nate day fasting (ADF; fasting every other day) is organ-ized with alternating “feast days,” on which there is

Page 10 of 15Cioffi et al. J Transl Med (2018) 16:371

dL 95% CI −  0.20 to 3.63; p = 0.07) (Fig.  4c). Subgroup analysis revealed a significant HDL-cholesterol increase with a balanced diet on “feed” days (WMD = 2.88  mg/dL 95% CI 0.66 to 5.09; p = 0.011; I2 = 0%) compared with ad  libitum eating (Additional file  9). No between-arm differences were found for total cholesterol and LDL-cholesterol (Fig. 4b, d). Finally, changes in both SBP and DBP did not significantly differ between arms (Additional file 10).

Publication biasWe used the Egger’s test for funnel plot asymmetry to detect a potential publication bias on reporting results on weight change. Test result (p = 0.15) did not suggest any asymmetry in the funnel plot (Additional file 11).

SafetyNo major adverse events were reported. Only 1 patient from the IER arm of the RCT supplying 0  kcal during “fast” days developed gallbladder dyskinesia and under-went cholecystectomy after completing the study, but this event was reported to be unrelated to the interven-tion [27]. Minor physical or psychological adverse effects, such as lack of energy, headaches, feeling cold, constipa-tion, bad breath, lack of concentration, bad temper, were reported in a minority of participants from the IER arms (< 20%) in a few studies [20, 21, 23, 26]. On the other hand, hunger was reported in the first weeks by about half of participants to a 5:2 regimen in 1 trial, but this symptom improved over time [23].

DiscussionAn intermittent regimen of energy restriction (at least 1  day/week) determined a loss in body weight and per-cent body weight similar to continuous (non-intermit-tent) energy restriction. Interestingly, a slight reduction in fasting insulin concentrations was evident with IER regimens employing 2  days/week “fast”, but the clinical relevance of this result is uncertain.

Effects of IER on weight loss and fat massMost systematic reviews and meta-analyses demon-strated that IER regimens achieved comparable weight loss as CER regimens [4, 5, 9], reporting an overall weight loss ranging from 4 to 8% [2, 3, 7, 9], and a difference of − 4.14 kg to + 0.08 kg versus the comparator arms [4, 5, 29]. Our results are in accordance, even if the trials previ-ously included differed from ours, since we have included only RCTs with a at least 1 day/week and no more than 6 day/week of “fasting”, and with an extremely low energy supply during the “fast” days. This latter choice derived from the idea of studying conditions simulating as much as possible a condition of fasting, whose benefits, proven

by animal studies, seem to depend on the shift in metab-olism from glucose utilization and fat synthesis/storage towards reduced insulin secretion and fat mobilization/oxidation [30, 31].

There is no clear definition of IER, and intermittent regimens providing up to 800  kcal [5, 9], with ≥ 7 “fast” days [4, 6, 9, 29], including time-restricted feeding [2, 6–8, 32], with unlimited energy restriction as a compara-tor group [2, 3, 5–7], or not randomized controlled trials [2] have been included within previous reviews. We have taken care to define precise inclusion criteria to limit var-iability and increase the comparability among trials, and we have obtained a low heterogeneity.

It could be hypothesized that the very low caloric intake on “fast” days determined an overall lower caloric intake in the IER arms, which were therefore difficult to be compared with the CER arms. In the only RCT where water and calorie-free beverages were allowed in the “fast” days, a significant between-arms difference in energy intake was evident [27]; in two studies a between-arms difference of 300–400 kcal was observed [22, 23] while most RCTs reported a negligi-ble between-arms difference (~ 100 kcal) [17, 20, 21, 25, 26]. Consistently, our sensitivity and subgroup analyses did not find significant between-arms differences.

Furthermore, the percent weight loss was highly overlapping, and no apparent superiority of a dietary regimen was evident. Indeed, participants of the IER arms from all RCTs lost ≥ 5% of their initial weights, thus confirming the clinical usefulness of this approach at least in the short term, i.e. within 24 weeks.

Previous reviews reported a FM loss ranging from 4 to 7% [3] to 11–16% [2] in the IER arms, and the only meta-analysis evaluating this outcome reported a dif-ferential loss of 1.38  kg with respect to comparator arms [5]. We failed to find significant between-arms difference for this outcome, suggesting that such a regi-men could be a valid, but not superior alternative to CER.

Intriguingly, participants to the IER regimens usually did not consume as much food in the “feed” days as to compensate for the caloric restriction of the “fast” days, thus suggesting that IER could reduce food intake even in the “feed” days, without compensatory overeating [6, 31]. This finding was not confirmed by all studies [28, 33, 34]. Furthermore, adverse events were sometimes higher with the IER regimens [20, 21, 26], and the partici-pants reported stronger feelings of hunger [21, 23]. The compliance and adherence to the intervention diets was heterogeneous among trials, the attrition rate was often higher in the IER arms [17, 22, 24, 26, 31, 35], and the percentage of participants planning to continue with the dietary regimen beyond 6 months was lower in the IER

Page 42: Intermittent Fasting: Weight Loss and Cardiovascular Effects · nate day fasting (ADF; fasting every other day) is organ-ized with alternating “feast days,” on which there is

Page 11 of 15Cioffi et al. J Transl Med (2018) 16:371

arms [26]. Overall, these data do not support the fact that IER is easier and more acceptable than CER to everyone. Moreover, the reduction in resting energy expenditure, i.e. the compensatory metabolic response which reduces the degree of weight loss, has been reported to be either reduced (favoring weight loss) [27, 36] or increased (attenuating weight loss) [22, 28] with IER regimens. Indeed, some studies suggest that IER evokes the same adaptive response as CER [6, 37].

The hypothesized benefits of IER, extensively studied in animal models, included the use of fats during severe energy restriction with preferential reduction of adipose mass, the stimulation of browning in white adipose tis-sue, increased insulin sensitivity, lowering of leptin and increased human growth hormone, ghrelin and adi-ponectin circulating levels, reduced inflammation and oxidative stress [30]. The trigger of adaptive cell response leading to enhanced ability to cope with stress, improved autophagy by sirtuin-1 activity stimulation, modification of apoptosis, increase of vascular endothelial growth fac-tor expression in white adipose tissue, the action on the metabolism via Forkhead Box A genes, and reduction of advance glycation end-products might be all possible metabolic pathways explaining the beneficial effects of IER [7, 30, 38, 39]. In mice, IER determined metabolic improvements and weight loss as a consequence of a shift

in the gut microbiota composition, leading to an increase in the production of acetate and lactate and to the selec-tive upregulation of monocarboxylate transporter in beige adipose cells which stimulate beige fat thermogen-esis [40]. At present, many of these adaptive mechanisms have been demonstrated in animal experimental models but not in humans, thus more research is still needed.

Effects of IER on cardiometabolic markersIER regimens were associated with lower circulating insulin values; a significant reduction was evident for the 5:2 “fasting” regimen only. Indeed, two RCTs, both employing this regimen, determined the difference [20, 26]. Our data are in line with the results of a previous meta-analysis reporting a significantly higher reduction in fasting insulin (−  0.67 µU/mL) in the IER arms [5]. The difference we found (− 0.89 µU/mL) was statistically significant, but not clinically relevant, above all consider-ing the fact that participants to the included RCTs were overweight/obese and therefore probably insulin-resist-ant individuals.

Our data synthesis on glucose, HOMA-IR, HbA1c showed no between-arms difference. We did not include patients with T2DM from 2 RCTS in the pooled analy-sis on fasting glucose, since most participants were on hypoglycemic drugs and their glycemic values would be

Fig. 2 Meta-analysis of the effects of intermittent energy restriction versus continuous energy restriction on weight loss. MD (mean difference) indicates the mean difference on change from baseline of the IER vs. the CER arms. The plotted points are the mean differences and the horizontal error bars represent the 95% confidence intervals. The grey areas are proportional to the weight of each study in the random-effects meta-analysis. The vertical dashed line represents the pooled point estimate of the mean difference. The solid black line indicates the null hypothesis (MD = 0)

Page 43: Intermittent Fasting: Weight Loss and Cardiovascular Effects · nate day fasting (ADF; fasting every other day) is organ-ized with alternating “feast days,” on which there is

Page 12 of 15Cioffi et al. J Transl Med (2018) 16:371

certainly influenced by the treatment [24, 25]. Highly contrasting human studies are available about the ben-efits of IER on glucose metabolism and insulin sensitivity [3, 6, 31], contrarily to animal studies strongly suggesting a benefit in T2DM prevention [1, 31]. The improvements in glucose homeostasis might be therefore comparable to those obtained by continuous energy restrictions.

Our meta-analysis did not show significant between-arms difference in lipid values and arterial blood pres-sure, with the exception of a small difference in subgroup analyses on triglyceride concentrations (− 14 mg/dL) and HDL-cholesterol (+ 2.88 mg/dL), not meaningful from a clinical point of view. Most studies showed reduction in triglyceride levels between 15 and 42% in the IER arms [31, 41], and the only available meta-analysis reported a between-arms not significant difference of 2.65  mg/dL [5]. Reduction in total cholesterol, LDL-cholesterol in the IER arms ranged respectively between 6–25%, 7–32%, with small effects on HDL-cholesterol [1, 31], and between-arms differences resulted not significant [5]. Intriguingly, a few studies reported that IER regimens determined an increase in LDL particle size [19, 42], and

reduced post-prandial hypertriglyceridemia [22], thus potentially conferring cardio-protection, since the lower the LDL size, the higher the oxidizability and the suscep-tibility to arterial penetration, and higher post-prandial hyperlipemia is a marker of atherosclerosis progression. Furthermore, fasting can act on many enzymes impli-cated in lipid and lipoprotein metabolism [27]. How-ever, all these reports need confirmation in larger human RCTs.

Similarly, data on arterial BP were controversial, with the majority of human studies reporting no differences between IER and CER regimens [1, 5, 31, 41]. Indeed, most of the published studies and RCTs included normo-tensive subjects at baseline, making it difficult to identify differences between-arms.

Therefore, unlike the very promising data on animals, evidence is not sufficiently robust to suggest the superi-ority of intermittent vs. continuous caloric restriction regimens on the main cardiovascular factors in humans.

Fig. 3 Meta-analysis of the effects of intermittent energy restriction versus continuous energy restriction on fasting glucose (a), HbA1c (b), insulin (c) and HOMA-IR (d) values. MD (mean difference) indicates the mean difference on change from baseline of the IER vs. the CER arms. The plotted points are the mean differences and the horizontal error bars represent the 95% confidence intervals. The grey areas are proportional to the weight of each study in the random-effects meta-analysis. The vertical dashed line represents the pooled point estimate of the mean difference. The solid black line indicates the null hypothesis (MD = 0)

Page 44: Intermittent Fasting: Weight Loss and Cardiovascular Effects · nate day fasting (ADF; fasting every other day) is organ-ized with alternating “feast days,” on which there is

Page 13 of 15Cioffi et al. J Transl Med (2018) 16:371

Clinical implicationsWeight loss maintenance should be an integral compo-nent of the management of obesity, owing to the weight regain usually occurring with time. The 2 RCTs includ-ing longer follow-ups (24 months) did not find between-arms differences in weight loss maintenance [17, 27]. Studies with longer follow-ups, evaluating the long-term sustainability, adherence to, and safety of IER regi-mens are needed. Furthermore, no RCT evaluated hard endpoints, such as cardiovascular outcomes or T2DM incidence. Two observational cohort studies found that fasting was associated with a lower prevalence of coro-nary artery diseases or diabetes diagnosis but are limited by a lack of a comprehensive dietary history and many potential bias [43, 44]. It could be hypothesized that IER regimens should be proposed in clinical practice, since it is possible that some individuals find easier to reduce their energy intakes for 1 or more days per week, rather than every day. It is well known that a single diet fit not all, and in the choice of the individual’s tailored regimen,

IER strategies should be considered by health care pro-fessionals. In this way, data on the feasibility of these regi-mens in “real life” would be obtained.

Strengths and limitationsThis is, to our knowledge, the largest and updated meta-analysis on the effects of IER on weight loss and multiple metabolic outcomes, setting strict inclusion criteria to increase comparability among studies.

The high variability among the RCTs in the feeding protocols, the limited follow-up, the small sample sizes, the high drop-out rates potentially leading to selection bias, the limited reporting of adverse events and blind-ing of investigators about arm allocation, or other meth-odological problems are all limitations to be considered. Finally, most studies were performed by the same authors and the majority of subjects included were adult healthy women, thus limiting the generalizability of the results.

Fig. 4 Meta-analysis of the effects of intermittent energy restriction versus continuous energy restriction on triglycerides (a), total cholesterol (b), HDL-cholesterol (c) and LDL-cholesterol (d) values. MD (mean difference) indicates the mean difference on change from baseline of the IER vs. the CER arms. The plotted points are the mean differences and the horizontal error bars represent the 95% confidence intervals. The grey areas are proportional to the weight of each study in the random-effects meta-analysis. The vertical dashed line represents the pooled point estimate of the mean difference. The solid black line indicates the null hypothesis (MD = 0)

Page 45: Intermittent Fasting: Weight Loss and Cardiovascular Effects · nate day fasting (ADF; fasting every other day) is organ-ized with alternating “feast days,” on which there is

Page 14 of 15Cioffi et al. J Transl Med (2018) 16:371

ConclusionIn overweight/obese adults, IER is as effective as CER for promoting weight loss and metabolic improvements in the short term. Further long-term investigations are needed to draw definitive conclusions.

Additional files

Additional file 1. Electronic search strategy.

Additional file 2. Risk of bias assessment in the trials included in the systematic review.

Additional file 3. Changes in outcomes at the end of the trials.

Additional file 4. Subgroup analysis of weight loss based on the type of regimen (a) and on dietary characteristics of the “feed” days (b).

Additional file 5. Percent weight loss (a) and subgroup analysis of per-cent weight loss based on the type of regimen (b) and dietary characteris-tics of the “feed” days (c).

Additional file 6. Meta-analysis of the effects of intermittent energy restriction versus continuous energy restriction on body composition (a) and waist circumference (b).

Additional file 7. Subgroup analysis of fasting insulin based on the type of regimen.

Additional file 8. Subgroup analysis of triglycerides based on the type of regimen (a) and dietary characteristics of the “feed” days (b).

Additional file 9. Subgroup analysis of HDL-cholesterol based on the type of regimen (a) and dietary characteristics of the “feed” days (b).

Additional file 10. Meta-analysis of the effects of intermittent energy restriction versus continuous energy restriction on systolic blood pressure (SBP) (a) and diastolic blood pressure (DBP) (b).

Additional file 11. Funnel plot for publication bias detection on weight loss changes.

AbbreviationsBIA: body impedance analysis; BMI: body mass index; BP: blood pressure; CER: continuous energy restriction; CI: confidence interval; CINAHL: Cumulative Index to Nursing and Allied Health Literature; DBP: diastolic blood pressure; DXA: dual X-ray absorptiometry; FFM: fat free mass; FM: fat mass; HbA1c: glycated hemoglobin A1c; HOMA-IR: Homeostasis Model Assessment-Insulin Resistance; IER: intermittent energy restriction; MeSH: medical subject head-ings; RCTs: randomized controlled trials; Si: insulin-sensitivity index; SBP: systolic blood pressure; T2DM: type 2 diabetes mellitus; WMD: weight mean difference.

Authors’ contributionsIC participated in the conception and design of the study, data collection and revision, interpretation of the findings of the study, manuscript writing and revision. AE participated in the data analysis, interpretation of the findings, manuscript writing and revision. VP, GC participated in the data analysis, interpretation of the findings, and manuscript revision. LSo, LSa, FP, FC, EG participated in the interpretation of the findings, and manuscript revision. SB participated in the conception and design of the study, data collection and revision, manuscript writing and revision. All authors read and approved the final manuscript.

Author details1 Interuniversity Center for Obesity and Eating Disorders, Department of Medi-cine and Surgery, Federico II University Hospital, Pansini, 5, Naples 80131, Italy. 2 Unit of Clinical Epidemiology, CPO, “Città della Salute e della Scienza” Hospital of Turin, Turin, Italy. 3 Department of Medical Sciences, University of Turin, c.so AM Dogliotti 14, 10126 Turin, Italy. 4 Department of Health Sciences, University of Milan, Milan, Italy.

AcknowledgementsNot applicable.

Competing interestsThe authors declare that they have no competing interests.

Availability of data and materialsAll data analyzed during this study are included in this published article.

Consent for publicationNot applicable.

Ethics approval and consent to participateNot applicable.

FundingNot applicable.

Publisher’s NoteSpringer Nature remains neutral with regard to jurisdictional claims in pub-lished maps and institutional affiliations.

Received: 31 October 2018 Accepted: 14 December 2018

References 1. Varady KA, Hellerstein MK. Alternate-day fasting and chronic disease pre-

vention: a review of human and animal trials. Am J Clin Nutr. 2007;86:7. 2. Varady KA. Intermittent versus daily calorie restriction: which diet regi-

men is more effective for weight loss? Obes Rev. 2011;12:593. 3. Barnosky AR, Hoddy KK, Unterman TG, Varady KA. Intermittent fasting vs

daily calorie restriction for type 2 diabetes prevention: a review of human findings. Transl Res. 2014;164:302.

4. Headland M, Clifton PM, Carter S, Keogh JB. Weight-loss outcomes: a systematic review and meta-analysis of intermittent energy restriction trials lasting a minimum of 6 months. Nutrients. 2016;8:8.

5. Harris L, Hamilton S, Azevedo LB, Olajide J, De Brún C, Waller G, et al. Inter-mittent fasting interventions for treatment of overweight and obesity in adults: a systematic review and meta-analysis. JBI Database Syst Rev Implement Rep. 2018;16:507.

6. Seimon RV, Roekenes JA, Zibellini J, Zhu B, Gibson AA, Hills AP, et al. Do intermittent diets provide physiological benefits over continuous diets for weight loss? A systematic review of clinical trials. Mol Cell Endocrinol. 2015;418:153.

7. Horne BD, Muhlestein JB, Anderson JL. Health effects of intermittent fast-ing: hormesis or harm? A systematic review. Am J Clin Nutr. 2015;102:464.

8. Varady KA. Impact of intermittent fasting on glucose homeostasis. Curr Opin Clin Nutr Metab Care. 2016;19:300.

9. Davis CS, Clarke RE, Coulter SN, et al. Intermittent energy restriction and weight loss: a systematic review. Eur J Clin Nutr. 2016;70:292.

10. Rothschild J, Hoddy KK, Jambazian P, Varady KA. Time-restricted feeding and risk of metabolic disease: a review of human and animal studies. Nutr Rev. 2014;72:308.

11. Patterson RE, Laughlin GA, LaCroix AZ, Hartman SJ, Natarajan L, Senger CM, et al. Intermittent fasting and human metabolic health. J Acad Nutr Diet. 2015;115:1203.

12. Hutchison AT, Heilbronn LK. Metabolic impacts of altering meal fre-quency and timing—does when we eat matter? Biochimie. 2016;124:187.

13. Lantz H, Peltonen M, Agren L, Torgerson JS. Intermittent versus on-demand use of a very low calorie diet: a randomized 2-year clinical trial. J Intern Med. 2003;253:463.

14. Moher D, Liberati A, Tetzlaff J, Group P. Preferred reporting items for systematic reviews and meta-analyses: the PRISMA statement. PLoS Med. 2009;6:e1000097. https ://doi.org/10.1371/journ al.pmed.10000 97.

15. Higgins JP, Altman DG, Gotzsche PC, Jüni P, Moher D, Oxman AD, et al. The Cochrane Collaboration’s tool for assessing risk of bias in randomised trials. BMJ. 2011;343:5928.

Page 46: Intermittent Fasting: Weight Loss and Cardiovascular Effects · nate day fasting (ADF; fasting every other day) is organ-ized with alternating “feast days,” on which there is

"This course was developed from the open access article: Effects of eight weeks of time-restricted feeding

(16/8) on basal metabolism, maximal strength, body composition, inflammation, and cardiovascular risk

factors in resistance-trained males Moro et al. J Transl Med (2016) 14:290 (DOI 10.1186/s12967-016-1044-0),

used under the Creative Commons Attribution License."

"This course was developed from the open access article: Klempel et al.: Intermittent fasting combined with

calorie restriction is effective for weight loss and cardio-protection in obese women. Nutrition Journal 2012

11:98. (DOI 10.1186/1475-2891-11-98), used under the Creative Commons Attribution License."

"This course was developed from the open access article: Varady et al.: Alternate day fasting for weight loss in

normal weight and overweight subjects: a randomized controlled trial. Nutrition Journal 2013 12:146.

(DOI 10.1186/1475-2891-12-146), used under the Creative Commons Attribution License."

"This course was developed from the open access article: Kroeger et al.: Improvement in coronary heart

disease risk factors during an intermittent fasting/calorie restriction regimen: Relationship to adipokine

modulations. Nutrition & Metabolism 2012 9:98. (DOI 10.1186/1743-7075-9-98), used under the Creative

Commons Attribution License."

"This course was developed from the open access article: Intermittent versus continuous energy restriction on

weight loss and cardiometabolic outcomes: a systematic review and meta-analysis of randomized controlled

Trials, Cioffi et al. J Transl Med (2018) 16:371 (DOI /10.1186/s12967-018-1748-4), used under the Creative

Commons Attribution License."