Effect of Epidemic Intermittent Fasting on Metabolic Syndrome: A Systematic Review and Meta-analysis of Randomized Controlled Trials

Background and aims: Intermittent fasting (IF) has gained attention as a promising diet for weight loss and dysmetabolic diseases management. This systematic review aimed to investigate the effects of IF on metabolic syndrome (MetS). Methods: A systematic literature search was carried out using three electronic databases, namely PubMed, Embase, and the Cochrane Library, until October 2020. Randomized controlled trials that compared the IF intervention with a control group diet were included. Effect sizes were expressed as weighted mean difference (WMD) using a xed-effects model and 95% condence intervals (CI). Results: Forty-six studies were included. Compared to the ones within control groups, participants exposed to the IF intervention reduced their body weight (WMD, -1.78 kg; 95% CI, -2.21 to -1.35; p < 0.05), waist circumference (WMD, -1.19 cm; 95% CI, -1.8 to -0.57; p < 0.05), fat mass (WMD, -1.26 kg; 95% CI, -1.57 to -0.95; p < 0.05), body mass index (WMD, -0.58 kg/m 2 ; 95% CI, -0.8 to -0.37; p < 0.05), systolic blood pressure (WMD, -2.14 mmHg; 95% CI: -3.54 to -0.73; p < 0.05), diastolic blood pressure (WMD: -1.38 mmHg, 95% CI, -2.35 to -0.41, p < 0.05), fasting blood glucose (WMD, -0.96 mg/dL; 95% CI, -1.89 to -0.03; p < 0.05), fasting insulin (WMD, -0.8 μU/mL; 95% CI, -1.15 to -0.44; p < 0.05), insulin resistance (WMD, -0.21; 95% CI, -0.36 to -0.05; p < 0.05), total cholesterol (WMD, -3.75 mg/dL; 95% CI, -6.64 to -0.85; p < 0.05), triglycerides (WMD, -7.54 mg/dL; 95% CI, -11.45 to -3.63; p < 0.05). No effects were observed for low-density lipoprotein cholesterol, high-density lipoprotein cholesterol, or glycosylated hemoglobin. Conclusions: This meta-analysis supports IF’s role in the improvement of MetS, compared to a control group diet. Further research on IF interventions should take into account long-term and well-designed administration to draw denitive conclusions. Two investigators independently extracted the relevant data from the eligible studies using predesigned forms. Data included study, publication year, country, study design, inclusion and exclusion criteria, total number of participants, participant details, study duration, intervention details and control groups, baseline patient characteristics (mean age, sex), body composition, BP, glycemic control, and lipid panel. Disagreements were resolved by consensus. When necessary, we emailed the corresponding author to acquire study details.


Introduction
Metabolic syndrome (MetS) is an emerging health issue throughout the world. Clinical diagnostic criteria for MetS [1] include increased waist circumference (WC) [2], increased triglycerides (TG), changes in lipoprotein, increased blood pressure (BP) [3], and increased fasting blood glucose (FBG) levels [4,5]. Moreover, studies have demonstrated that MetS doubled the risk for atherosclerotic cardiovascular diseases (CVDs) [6] and increased the risk for type 2 diabetes ve-fold [7]. Considering that calorie restriction and exercise are effective management strategies for MetS, intermittent fasting (IF) can also be used as an important treatment [7].
IF has gradually come into focus in our daily lives [8]. At present, a large number of studies have shown that IF is bene cial in the treatment of metabolic diseases, and because of its simple use, it is easy to accept. Dietary restrictions [9] through IF have been shown to improve metabolic disease risk indicators.
Further, IF reportedly plays a considerable role in regulating cardiovascular risk indicators [10], insulin resistance (HOMA-IR), and circulating blood glucose levels [11,12]. There are different types of IF that act as an energy-limiting diet for a speci c period, including alternate-day fasting (ADF), alternate-modi edday fasting (AMDF), periodic fasting (PF), time-restricted feeding (TRF), and religious fasting (Table 1). Intermittent energy restriction (IER) is an alternative of IF in this study, and the control group is often a continuous energy restriction (CER). Table 1 Types of intermittent fasting and description

Type of Fast Description
Alternate-day fasting (ADF) A circular diet that requires fasting for a day (consumption of no calories) and then eating freely for a day [73].
Alternatemodi ed-day fasting (AMDF) A circular feeding pattern that requires fasting (consumption of 20-25% of energy needs) for a day, and then eating freely for a day; the popular 5:2 diet includes a discontinuous strict energy limit of 2 days a week and 5 other days of random eating [14,73].

Time-restricted feeding (TRF)
Complete fast (no calories) for at least 12 hours a day, and eating freely the rest of the time; the 16:8 fasting pattern currently prevails [14,106,107].
Periodic fasting (PF) sample size, a variety of IF types, and multiple effect indicators to determine the effectiveness of IF interventions in improving health outcomes and modi able risk factors for people with MetS.

Methods
This study used a systematic review and a meta-analysis of PRISMA's preferred reporting items as a guide for reporting research results [20,21].

Data source and search strategy
Articles were identi ed by searching through three electronic databases, i.e., PubMed, Embase, and the Cochrane Library until October 2020. Two reviewers (C.L. and X.L.) independently evaluated articles' eligibility, and the inconsistencies shall be made by the corresponding author (F.Y.). Solve it. The search strategy is described in detail in Supplementary Table 1.

Inclusion and Exclusion Criteria
The articles had the following characteristics: (1)

Quality assessment and publication bias
The researchers used Cochrane Collaboration's bias risk tool to evaluate the quality of the methodology included in the studies. According to the criteria of the Cochrane handbook for systematic reviews, the bias risk of each item is classi ed as low, high, or unclear [22].

Data statistical and analysis
Effect estimates were expressed as weighted mean differences (WMD) with a 95% con dence interval (CI). Inter-study heterogeneity was tested using the Higgins I 2 statistic, and I 2 >50% indicated signi cant statistical heterogeneity. The heterogeneity of the study and measurement of effect estimates were determined using the mean and standard deviation (SD) of the differences before and after IF intervention. Publication bias was evaluated using funnel plots; formal testing was conducted with Egger's test [23], and a sensitivity analysis was also performed. We used STATA 16 (StataCorp LLC, College Station, TX, USA) for the statistical analyses.
In order to determine the in uence of IF on various effect indicators, it is necessary to change the mean value before and after intervention as well as the SD of the changes. Therefore, we used a method outlined in the Cochrane handbook [22,24] to determine the SD of changes between time points. SD change =√[SD baseline 2 + SD nal 2 − (2 × R × SD baseline × SD nal )]. In addition, we performed some conversion of data units through international calculation formulas to ensure that results are clinically signi cant.

Characteristics of included studies
The PRISMA statement ow diagram is shown in Fig. 1 [25]. A total of 9087 studies were part of the initial database search (PubMed: 6128, EMBASE: 40, Cochrane Library: 2919), after the removal of 1540 duplicate studies. After ltering the titles and abstracts to exclude irrelevant articles, we found 106 studies that met the topic of interest. The full texts of the 106 records were reviewed. Of them, 60 records were excluded for the following reasons: data not available (n= 15), literature review, letter, or case report (n= 34), unrelated to relevant predictive factors (n= 2), related to protocol (n = 2), and meta-analyses (n = 7).
Finally, 46 studies from the database searches were included in the meta-analysis [15,. A total of 2681 participants were randomized in the IF intervention group (n = 1423) and the control group (n = 1258). The characteristics of the eligible trials are summarized in Table 2. All the results calculated using Stata are shown in Table3.  Risk of bias and quality assessment of studies Fig. 2 summarizes the risk of bias for RCTs. Twenty-three studies (50%) had a low risk of selection bias. This was because of the intervention type, as no RCT adequately performed blinding of the participants (blinding of dietary interventions is impossible); however, 18 studies (39%) were judged as having a low risk of bias for outcome assessment blinding, 38 studies (82%) were judged as having a low risk of bias for incomplete outcome data, 40 studies (86%) were judged as having a low risk of bias for selective reporting, and 43 studies (93%) showed a low risk of other biases. Overall, 16 studies (35%) were rated with a high risk of bias due to random sequence generation, allocation concealment, outcome assessment blinding, incomplete outcome data, and selective reporting.

Effect of IF on body composition
Body composition was operationalized in weight, WC, FM, and BMI. Forty-ve studies, with 2225 participants (case = 1136, control = 1089), showed a consistent effect of IF on weight (Fig. 3a) Fig. 4a).

Effect of IF on glycemic control
A meta-analysis of the effect of IF on glycemic control was performed including the relevant studies. A cumulative meta-analysis of 34 studies with 1863 participants (case = 947, control = 916) evaluated changes in FBG during IF (Fig. 5a). The WMD was -0.96 mg/dL (95% CI: -1.89 to -0.03, p < 0.05, a xedeffects model), which indicates signi cant FBG reduction. We observed a moderate effect heterogeneity (I 2 = 44.4%, p = 0.003). Regarding funnel plot symmetry and Egger's test, p = 0.502 (Fig. 6a).

Sensitivity analysis
In order to determine the impact of each individual study on the effect index, we used a sensitivity analysis in our meta-analysis. Finally, we did not observe the signi cant effects of any individual study (Fig.11-14).

Discussion
In this study, 46 RCTs were systematically reviewed to evaluate the effects of IF on MetS. The pooled analysis showed that IF had signi cantly reduced body composition (weight, WC, FM, and BMI), BP (SBP, DBP), lipid panel (TC, TG), and improved glycemic control by reducing FBG, Fins, and HOMA-IR; however, it did not affect the HbA1c level and lipid pro le (LDL-C and HDL-C).
Overall, in terms of body composition, there was a signi cant positive correlation between BMI and weight loss during IF (i.e., the higher the starting BMI, the greater the weight loss during the fasting period). This suggests that IF may be more effective for people with a higher BMI. The results for the effect of IF on body composition were similar to those obtained in a previous meta-analysis, by [72], which involved 11 trials that found that TRF was effective in promoting weight loss and reducing FBG compared to not limiting meal times approaches. In addition, IER was more effective in reducing weight than a regular control diet. Moreover, it was also more effective in reducing FM level than CER [73]. In a meta-analysis on religious fasting [74], it was found that overweight participants had a greater reduction in weight and percentage of fat than normal people. A recent meta-analysis of RCTs showed that ADF effectively lowered body composition and TC in overweight adults within 6 months compared to the control group [75]. However, in another meta-analysis of 12 RCTs, researchers con rmed that lean mass was relatively conserved in the IF group and no signi cant weight reduction was identi ed [34]. In addition, a recent study by Lowe, depleted, and the body begins to absorb fatty acids from fat cells to replace glucose for combustion, which improves insulin sensitivity [59,77].
Previous studies have shown that IF is not only bene cial in reducing the production of free radicals or weight loss; it also has several health bene ts [7,[78][79][80]. IF can cause an evolutionarily conservative adaptive cellular response, improve blood glucose regulation, enhance anti-stress ability, and inhibit in ammation between and within organs. During fasting, cells activate pathways that enhance the body's defense against oxidation and metabolic stress as Human beings have gradually formed a 24-hour circadian rhythm during evolution [85]. The master clock is mainly produced by the suprachiasmatic nucleus of the hypothalamus, while the peripheral oscillators are found in the esophagus, liver, pancreas, spleen, skin, and thymus. There is an important relationship between the feeding signal and peripheral clock rhythm. Thus, energy consumption outside the normal eating phase (i.e., late-night eating in humans) may disrupt the balance of some peripheral clocks [86]. Meanwhile, irregular mealtimes may cause a shift or an internal desynchronization of the peripheral clock, which may lead to its decoupling, followed by a series of unhealthy consequences such as MetS [87]. Daily rhythms also exist in glucose homeostasis, and the decline of insulin sensitivity and glucose oxidation at night is higher than the decline experienced in the morning [88]. People who eat late lunches are less likely to lose weight because glucose tolerance and insulin function decline at night [89]. This nding is critical because studies have suggested that mealtimes may change the central biological clock [90]. These data highlight that appropriate mealtimes play a key role in health. TRE can improve metabolic dysfunction and weight loss by adjusting circadian rhythms in obese individuals [91,92].
Some studies have found that the mammalian TOR pathway activated by diets alters the stability of the biological clock [93]. In contrast, fasting activates the AMP-dependent protein kinase pathway to degrade the cryptochrome [94]. Moreover, nicotinamide adenine dinucleotide and sirtuins uctuate with the cell's energy state, affecting circadian rhythms [95][96][97]. Therefore, the feeding/fasting cycle enhances the oscillation of circadian activators and repressors, thereby regulating rhythmic tissue-speci c transcriptomes [98,99], and ultimately translating to a healthier phenotype. Meanwhile, researchers have found that the gut microbiota is associated with circadian rhythms and dietary habits [100]. In fact, feeding alters the inherent daily rhythm of the intestinal microbes, and both food content and feeding time play a role in the process [100][101][102]. Defense against oxidative and metabolic stress as well as clearance of damaged molecules also enhance and provide greater diversity of intestinal ora during fasting [101,103]. In addition, an earlier study suggested that IF promotes browning of white fat and reduces obesity by shaping the intestinal ora [104]. Another study showed that TRF reduced the number of several obese microbes and increased the proportion of hypothetical obesity protective bacteria [101]. TRF is associated with periodic microorganism uctuations and improves the intestinal microenvironment [105], resulting a total amount of intestinal bacteria and Firmicutes increased during the awake/eating phase. Further, the bacteroids, proteobacteria, and microbiota increase during the sleep/fasting phase. There is also evidence of diurnal variations in microbial metabolites, which in turn affect host circadian rhythms and metabolism.
Even though, many experiments have shown that IF is bene cial to human health and suitable for a wide range of metabolic diseases, this dietary pattern is rarely used in practice. The main reason for this is the three-meals-a-day habit in our daily lives. Second, when switching to an IF program, some people feel hungry, irritable, and lose concentration. Finally, doctors are required to prescribe speci c training for IF interventions, which requires a standardized use of IF.
There are some limitations to this meta-analysis. As this is the case with some studies, some of the included ones have a small sample size, and there are several studies with a high risk of bias. Second, the number of long-term studies conducted is very limited, and larger long-term trials with a longer duration are needed to understand the effects of IF on weight loss and long-term weight management. Moreover, different types of IF have different characteristics in various metabolic diseases, and we did not analyze each of them individually. Finally, although IF has a variety of components, a comparison with other types of IF could not be conducted due to the lack of RCT research on religious fasting and lack of data on other kinds of fasting.

Conclusions
This systematic review has demonstrated that IF may improve body composition (weight, WC, FM, and BMI) and moderate BP, TC, TG, and blood glucose, but there may be no difference regarding the LDL-C, HDL-C, and HbA1c levels; components of MetS are also risk factors for the development of diabetes and CVDs. Therefore, high-quality and long-term RCTs are needed to provide data on the persistence of the effect and to strengthen the certainty of the evidence. Risk-of-bias assessment of the studies included in the meta-analysis.