Effects of Intermittent Fasting on Regulation of Metabolic Homeostasis: A Systematic Review and Meta-Analysis in Health and Metabolic-Related Disorders

Intermittent fasting (IF) is an emerging dietetic intervention that has been associated with improved metabolic parameters. Nowadays, the most common IF protocols are Alternate-Day Fasting (ADF) and Time-Restricted Fasting (TRF), but in this review and meta-analysis we have also considered Religious Fasting (RF), which is similar to TRF but against the circadian rhythm. The available studies usually include the analysis of a single specific IF protocol on different metabolic outcomes. Herein, we decided to go further and to conduct a systematic review and meta-analysis on the advantages of different IF protocols for metabolic homeostasis in individuals with different metabolic status, such as with obesity, diabetes mellitus type 2 (T2D) and metabolic syndrome (MetS). Systematic searches (PubMed, Scopus, Trip Database, Web of Knowledge and Embase, published before June 2022) of original articles in peer-review scientific journals focusing on IF and body composition outcomes were performed. Sixty-four reports met the eligibility criteria for the qualitative analysis and forty-seven for the quantitative analysis. Herein, we showed that ADF protocols promoted the major beneficial effects in the improvement of dysregulated metabolic conditions in comparison with TRF and RF protocols. Furthermore, obese and MetS individuals are the most benefited with the introduction of these interventions, through the improvement of adiposity, lipid homeostasis and blood pressure. For T2D individuals, IF impact was more limited, but associated with their major metabolic dysfunctions—insulin homeostasis. Importantly, through the integrated analysis of distinct metabolic-related diseases, we showed that IF seems to differently impact metabolic homeostasis depending on an individual’s basal health status and type of metabolic disease.


Introduction
Intermittent fasting (IF) defines eating patterns in which individuals switch between extended periods of fasting and normal eating, on a recurring basis. IF comprises two main dietary regimens, one that includes fasting for entire days (1 to 4 per week), with a 60-100% energy restriction on fasting days (between 300 and 600 kcal) with ad libitum energy consumption on fed days, and another comprising daily time-restricted eating (fasting in a daily window, from 12 to 20 h). Essentially, these two regimens rely in approaches with both less frequent long fasting periods and short-term but frequent fasting periods [1]. Within these two main IF protocols, it is possible to describe four different types of fasting: (1) Alternate-Day Fasting (ADF) (where no food or energy-containing drinks are consumed ("fasting" [MeSH Major Topic]) AND (("obesity" [MeSH Terms]) OR ("overweight" [MeSH Terms]) OR ("metabolic syndrome" [MeSH Terms]) OR ("diabetes mellitus" [MeSH Terms]) OR ("insulin resistance" [MeSH Terms])) and, for Scopus, Trip Database, Web of Knowledge and Embase ("Intermittent fasting") AND ("obesity" OR "overweight" OR "metabolic syndrome" OR "diabetes" OR "insulin resistance" OR "insulin sensitivity"), from 2000 up to 2022.

Eligibility
Original articles in peer-review scientific journals focusing on IF and body composition outcomes were retrieved. Studies were included when fulfilling the following criteria: (i) adult participants, regardless of gender; (ii) data related to adiposity (weight, body mass index (BMI), waist circumference); lipid homeostasis (HDL-c, LDL-c, total cholesterol, triglycerides), insulin homeostasis (fasting glucose, fasting insulin, insulin resistance (HOMA-IR)) or blood pressure (Systolic Blood Pressure (SBP) and Diastolic Blood Pressure (DBP)); (iii) healthy individuals and/or individuals with metabolic related disorders, such as type 2 diabetes mellitus (T2D), metabolic syndrome (MetS) or obesity; and, (iv) articles presenting pre-and post-intervention measures. For the meta-analysis randomized and non-randomized controlled trials, cohort studies, case-control studies and case series were included.
To avoid significant bias for the systematic review and meta-analysis, studies were excluded when one of the following criteria was observed: (i) non-original articles (systematic reviews, protocols, reviews and book chapters); (ii) articles using animal models; (iii) articles written in languages other than English; (iv) grey literature (conference abstracts, letters to editor); (v) articles where other diets or interventions were mixed, such as IF plus Mediterranean, ketogenic, paleo or vegan diets; (vi) articles on other pathologies or conditions, eating disorders, alcohol abuse, smokers, medications (especially the ones that could affect glucose and lipid profile) and pregnant or breastfeeding women; (vii) results from groups in trials that tested both IF and other diets (such as the ketogenic diet) or that had any other intervention besides IF with or without caloric restriction; (viii) studies with individuals with metabolic related disorders, such as type 2 diabetes mellitus (T2D), metabolic syndrome (MetS) or obesity associated with other severe diseases.

Study Selection
After removing duplicates, three researchers (AIS, MD and BSM) independently screened the title and abstract of every citation followed by full-text screening against the inclusion/exclusion criteria. To qualify for inclusion, the authors had to be in agreement.

Outcomes Measures
The main outcomes evaluated were the impact of IF on weight, BMI, waist circumference, HDL-c, LDL-c, total cholesterol, triglycerides, fasting glucose, fasting insulin, HOMA-IR, SBP and DBP. Although the effects of different IF protocols were not subject of analysis in this systematic review, they were briefly summarized for deeper comprehension and future reference. Different IF regimens were considered but all had to fall into one of two main groups: (1) IF with a daily window where no food or energy containing drinks were ingested or (2) whole days of the week with caloric intake greatly reduced or eliminated, as described on the characterization of the included studies.

Quality Assessment
The risk of bias was not assessed, as this review included different type of studies. The quality of the articles was carefully assessed, according to: (i) the Critical Appraisal Skills Programme (CASP) [14] checklist for RCT, CS and CCS; (ii) the Joanna Briggs [15] checklist for CRS and CSS; and (iii) the Methodological Index for Non-Randomized Studies (MINORS) [16] checklist for NRCT. Taking into consideration that it would be nearly impossible for patients and staff of the study to be blind, items of the checklist concerning this parameter were not considered for the overall assessment of quality.

Search Results
PubMed, Scopus, Embase, Trip database and Web of Science generated 2365 publications ( Figure 1). After the removal of 96 duplicates, 2269 publications were identified as potentially eligible. After screening the title and abstract and checking the full text for detailed information and data extraction, 64 publications were included in the systematic review (Table S1) with 47 [7,9, being included in the meta-analysis ( Figure 1 and Table 1).

Meta-Analysis
The meta-analysis results of the effects of IF on the regulation of metabolic homeostasis in health and disease (T2D, MetS and obesity) were based on data from 47 studies selected considering the completeness of the outcomes measured in four main categories: adiposity (weight, BMI, waist circumference), lipid homeostasis (HDL-c, LDL-c, total cholesterol, triglycerides), insulin homeostasis (fasting glucose, fasting insulin, HOMA-IR) and blood pressure (SBP and DBP). According to IF protocols, these metabolic parameters were evaluated independently in 13 ADF, 13 TRF and 21 RF studies.

Characterization of Participants Included in Each IF Protocol
Changes in body weight, BMI, waist circumference, glucose and total cholesterol of the participants included in this meta-analysis are reported in Table 2. Overall, weight loss, BMI, waist circumference and total cholesterol levels were similar among participants included in each of the IF regimes. Nevertheless, we noticed that participants from RF studies presented higher glucose levels in comparison with the participants that integrate the ADF and TRF studies. This difference could be related to the fact that most of the RF studies were performed with T2D individuals, that by itself present higher glucose levels. Concerning the adiposity measurements, the analyzed outcomes were weight, BMI and waist circumference. In general, all the tested IF protocols (ADF, TRF and RF) showed significant positive results for the analyzed outcomes, when comparing pre-and postfasting timepoints (Tables 3-5; Figures 2-4).
For ADF protocols, a total of 13 studies were included and data showed a significant reduction in the body weight (k = 12, 5. 54 Figure 2).

Lipid Homeostasis
To study the impact of the different types of IF on lipid homeostasis, the outcomes analyzed were HDL-c, LDL-c, total cholesterol and triglyceride. ADF, TRF and RF showed distinct impact on the analyzed lipid homeostasis outcomes (Tables 3-5; Figures 5-7). Starting with the ADF protocols, data collected revealed that this intervention promotes statistically significant positive effects on HDL-c (k = 11,  Figure 5).
Regarding HDL-c, LDL-c and total cholesterol parameters, only people with obesity presented statistically significant positive effects (k = 8,  Figure 5).

Figure 2.
Forest plot of the data from random effects meta-analysis shown as mean difference with 95% confidence intervals on adiposity outcomes, weight, body mass index (BMI) and waist circumference, for the studies that presented data concerning these parameters. Data presented are related with alternate-day fasting (ADF) and the metabolic conditions: healthy or disease condition, obesity, T2D or MetS. For each study, the square represents the mean difference between baseline and fasting conditions, with the horizontal line intersecting it as the lower and upper limits of the 95% confidence interval [9,22,23,25,29,32,34,43,47,54,55,61]. Forest plot of the data from random effects meta-analysis shown as mean difference with 95% confidence intervals on adiposity outcomes, weight, body mass index (BMI) and waist circumference, for the studies that presented data concerning these parameters. Data presented are related with alternate-day fasting (ADF) and the metabolic conditions: healthy or disease condition, obesity, T2D or MetS. For each study, the square represents the mean difference between baseline and fasting conditions, with the horizontal line intersecting it as the lower and upper limits of the 95% confidence interval [9,22,23,25,29,32,34,43,47,54,55,61]. Figure 3. Forest plot of the data from random effects meta-analysis shown as mean difference with 95% confidence intervals on adiposity outcomes, weight, body mass index (BMI) and waist circumference, for the studies that presented data concerning these parameters. Data presented are related with time-restricted fasting (TRF) and the metabolic conditions: healthy or disease condition, obesity, T2D or MetS. For each study, the square represents the mean difference between baseline and fasting conditions, with the horizontal line intersecting it as the lower and upper limits of the 95% confidence interval [7,19,20,24,26,30,35,38,41,46,51,56,58]. . Forest plot of the data from random effects meta-analysis shown as mean difference with 95% confidence intervals on adiposity outcomes, weight, body mass index (BMI) and waist circumference, for the studies that presented data concerning these parameters. Data presented are related with time-restricted fasting (TRF) and the metabolic conditions: healthy or disease condition, obesity, T2D or MetS. For each study, the square represents the mean difference between baseline and fasting conditions, with the horizontal line intersecting it as the lower and upper limits of the 95% confidence interval [7,19,20,24,26,30,35,38,41,46,51,56,58].

Figure 4.
Forest plot of the data from random effects meta-analysis shown as mean difference with 95% confidence intervals on adiposity outcomes, weight, body mass index (BMI) and waist circumference, for the studies that presented data concerning these parameters. Data presented are related with and religious fasting (RF) and the metabolic conditions: healthy or disease condition, obesity, T2D or MetS. For each study, the square represents the mean difference between baseline and fasting conditions, with the horizontal line intersecting it as the lower and upper limits of the 95% confidence interval [17,18,21,27,28,31,36,37,39,40,42,44,45,[48][49][50]52,53,57,59,60]. . Forest plot of the data from random effects meta-analysis shown as mean difference with 95% confidence intervals on adiposity outcomes, weight, body mass index (BMI) and waist circumference, for the studies that presented data concerning these parameters. Data presented are related with and religious fasting (RF) and the metabolic conditions: healthy or disease condition, obesity, T2D or MetS. For each study, the square represents the mean difference between baseline and fasting conditions, with the horizontal line intersecting it as the lower and upper limits of the 95% confidence interval [17,18,21,27,28,31,36,37,39,40,42,44,45,[48][49][50]52,53,57,59,60].   Figure 3). With exception between the HDL-c studies, the heterogeneity between the studies compiling the data for LDL-c, total cholesterol and triglycerides was high (Table 4; Figure 6).

Figure 5.
Forest plot of the data from random effects meta-analysis shown as mean difference with 95% confidence intervals on lipid homeostasis outcomes, HDL-c, LDL-c, total cholesterol, triglycerides, for the studies that presented data concerning these parameters. Data are presented according to alternate-day fasting (ADF) and the metabolic conditions: healthy or disease condition, obesity, T2D or MetS. For each study, the square represents the mean difference between baseline and fasting conditions, with the horizontal line intersecting it as the lower and upper limits of the 95% confidence interval [9,22,23,25,29,[32][33][34]43,47,54,55].

Figure 5.
Forest plot of the data from random effects meta-analysis shown as mean difference with 95% confidence intervals on lipid homeostasis outcomes, HDL-c, LDL-c, total cholesterol, triglycerides, for the studies that presented data concerning these parameters. Data are presented according to alternate-day fasting (ADF) and the metabolic conditions: healthy or disease condition, obesity, T2D or MetS. For each study, the square represents the mean difference between baseline and fasting conditions, with the horizontal line intersecting it as the lower and upper limits of the 95% confidence interval [9,22,23,25,29,[32][33][34]43,47,54,55].

Figure 6.
Forest plot of the data from random effects meta-analysis shown as mean difference with 95% confidence intervals on lipid homeostasis outcomes, HDL-c, LDL-c, total cholesterol, triglycerides, for the studies that presented data concerning these parameters. Data are presented according to time-restricted fasting (TRF) and the metabolic conditions: healthy or disease condition, obesity, T2D or MetS. For each study, the square represents the mean difference between baseline and fasting conditions, with the horizontal line intersecting it as the lower and upper limits of the 95% confidence interval [7,24,30,38,41,46,51,56,58].

Figure 6.
Forest plot of the data from random effects meta-analysis shown as mean difference with 95% confidence intervals on lipid homeostasis outcomes, HDL-c, LDL-c, total cholesterol, triglycerides, for the studies that presented data concerning these parameters. Data are presented according to time-restricted fasting (TRF) and the metabolic conditions: healthy or disease condition, obesity, T2D or MetS. For each study, the square represents the mean difference between baseline and fasting conditions, with the horizontal line intersecting it as the lower and upper limits of the 95% confidence interval [7,24,30,38,41,46,51,56,58].

Figure 7.
Forest plot of the data from random effects meta-analysis shown as mean difference with 95% confidence intervals on lipid homeostasis outcomes, HDL-c, LDL-c, total cholesterol, triglycerides, for the studies that presented data concerning these parameters. Data are presented according to religious fasting (RF) and the metabolic conditions: healthy or disease condition, obesity, T2D or MetS. For each study, the square represents the mean difference between baseline and fasting conditions, with the horizontal line intersecting it as the lower and upper limits of the 95% confidence interval [17,18,21,28,36,37,39,40,45,[48][49][50]57].

Figure 7.
Forest plot of the data from random effects meta-analysis shown as mean difference with 95% confidence intervals on lipid homeostasis outcomes, HDL-c, LDL-c, total cholesterol, triglycerides, for the studies that presented data concerning these parameters. Data are presented according to religious fasting (RF) and the metabolic conditions: healthy or disease condition, obesity, T2D or MetS. For each study, the square represents the mean difference between baseline and fasting conditions, with the horizontal line intersecting it as the lower and upper limits of the 95% confidence interval [17,18,21,28,36,37,39,40,45,[48][49][50]57].    Table 5). Heterogeneity between the studies for LDL-c and total cholesterol was high (Table 4) and it was moderate between the studies presenting data for LDL-c and triglycerides levels (Table 5; Figure 7).

Figure 8.
Forest plot of the data from random effects meta-analysis shown as mean difference with 95% confidence intervals on insulin homeostasis outcomes, fasting glucose, fasting insulin, insulin resistance (HOMA-IR), for the studies that presented data concerning these parameters. Data are presented according to alternate-day fasting (ADF) and the metabolic conditions: healthy or disease condition, obesity, T2D or MetS. For each study, the square represents the mean difference between baseline and fasting conditions, with the horizontal line intersecting it as the lower and upper limits of the 95% confidence interval [9,22,23,25,29,32,34,43,47,54,61]. Figure 8. Forest plot of the data from random effects meta-analysis shown as mean difference with 95% confidence intervals on insulin homeostasis outcomes, fasting glucose, fasting insulin, insulin resistance (HOMA-IR), for the studies that presented data concerning these parameters. Data are presented according to alternate-day fasting (ADF) and the metabolic conditions: healthy or disease condition, obesity, T2D or MetS. For each study, the square represents the mean difference between baseline and fasting conditions, with the horizontal line intersecting it as the lower and upper limits of the 95% confidence interval [9,22,23,25,29,32,34,43,47,54,61]. Figure 9. Forest plot of the data from random effects meta-analysis shown as mean difference with 95% confidence intervals on insulin homeostasis outcomes, fasting glucose, fasting insulin, insulin resistance (HOMA-IR), for the studies that presented data concerning these parameters. Data are presented according to time-restricted fasting (TRF) and the metabolic conditions: healthy or disease condition, obesity, T2D or MetS. For each study, the square represents the mean difference between baseline and fasting conditions, with the horizontal line intersecting it as the lower and upper limits of the 95% confidence interval [7,19,20,24,30,35,46,51,56,58]. Figure 9. Forest plot of the data from random effects meta-analysis shown as mean difference with 95% confidence intervals on insulin homeostasis outcomes, fasting glucose, fasting insulin, insulin resistance (HOMA-IR), for the studies that presented data concerning these parameters. Data are presented according to time-restricted fasting (TRF) and the metabolic conditions: healthy or disease condition, obesity, T2D or MetS. For each study, the square represents the mean difference between baseline and fasting conditions, with the horizontal line intersecting it as the lower and upper limits of the 95% confidence interval [7,19,20,24,30,35,46,51,56,58]. Figure 10. Forest plot of the data from random effects meta-analysis shown as mean difference with 95% confidence intervals on insulin homeostasis outcomes, fasting glucose, fasting insulin, insulin resistance (HOMA-IR), for the studies that presented data concerning these parameters. Data are presented according to religious fasting (RF) and the metabolic conditions: healthy or disease condition, obesity, T2D or MetS. For each study, the square represents the mean difference between baseline and fasting conditions, with the horizontal line intersecting it as the lower and upper limits of the 95% confidence interval [17,18,21,27,28,37,39,40,44,45,[48][49][50]53]. Figure 10. Forest plot of the data from random effects meta-analysis shown as mean difference with 95% confidence intervals on insulin homeostasis outcomes, fasting glucose, fasting insulin, insulin resistance (HOMA-IR), for the studies that presented data concerning these parameters. Data are presented according to religious fasting (RF) and the metabolic conditions: healthy or disease condition, obesity, T2D or MetS. For each study, the square represents the mean difference between baseline and fasting conditions, with the horizontal line intersecting it as the lower and upper limits of the 95% confidence interval [17,18,21,27,28,37,39,40,44,45,[48][49][50]53].

Blood Pressure
To study the impact of different types of IF protocols on blood pressure, the outcomes systolic blood pressure (SBP) and diastolic blood pressure (DBP) were analyzed. Relatively to blood pressure outcomes, ADF, TRF and RF showed positive impact (Tables 3-5 and Figures 11-13).
This meta-analysis demonstrates high heterogeneity in the results, particularly in studies of individuals with obesity and MetS (Tables 3-5). Therefore, a sensitive analysis was performed by two different strategies. In the first, the sensitive analysis integrated all the studies (Table S3), and in the second analysis the impact of IF, without the RF studies was evaluated (Table S4). The sensitivity analysis did not decrease heterogeneity neither when all studies are integrated nor when RF studies were removed. Therefore, data suggest that heterogeneity is most probably related with differences in study design, populations, interventions or outcomes across the included studies. Figure 11. Forest plot of the data from random effects meta-analysis shown as mean difference with 95% confidence intervals on blood pressure outcomes systolic blood pressure (SBP) and diastolic blood pressure (DBP), for the studies that presented data concerning these parameters. Data are presented according to alternate-day fasting (ADF) and the metabolic conditions: healthy or disease condition, obesity, T2D or MetS. For each study, the square represents the mean difference between baseline and fasting conditions, with the horizontal line intersecting it as the lower and upper limits of the 95% confidence interval [9,25,29,32,33,43,47,54].

Figure 12.
Forest plot of the data from random effects meta-analysis shown as mean difference with 95% confidence intervals on blood pressure outcomes systolic blood pressure (SBP) and diastolic blood pressure (DBP), for the studies that presented data concerning these parameters. Data are presented according to time-restricted fasting (TRF) and the metabolic conditions: healthy or disease condition, obesity, T2D or MetS. For each study, the square represents the mean difference between baseline and fasting conditions, with the horizontal line intersecting it as the lower and upper limits of the 95% confidence interval [7,19,20,24,26,30,41,46,51,56,58]. Figure 11. Forest plot of the data from random effects meta-analysis shown as mean difference with 95% confidence intervals on blood pressure outcomes systolic blood pressure (SBP) and diastolic blood pressure (DBP), for the studies that presented data concerning these parameters. Data are presented according to alternate-day fasting (ADF) and the metabolic conditions: healthy or disease condition, obesity, T2D or MetS. For each study, the square represents the mean difference between baseline and fasting conditions, with the horizontal line intersecting it as the lower and upper limits of the 95% confidence interval [9,25,29,32,33,43,47,54].
J. Clin. Med. 2023, 12, x FOR PEER REVIEW 33 of 42 Figure 11. Forest plot of the data from random effects meta-analysis shown as mean difference with 95% confidence intervals on blood pressure outcomes systolic blood pressure (SBP) and diastolic blood pressure (DBP), for the studies that presented data concerning these parameters. Data are presented according to alternate-day fasting (ADF) and the metabolic conditions: healthy or disease condition, obesity, T2D or MetS. For each study, the square represents the mean difference between baseline and fasting conditions, with the horizontal line intersecting it as the lower and upper limits of the 95% confidence interval [9,25,29,32,33,43,47,54].

Figure 12.
Forest plot of the data from random effects meta-analysis shown as mean difference with 95% confidence intervals on blood pressure outcomes systolic blood pressure (SBP) and diastolic blood pressure (DBP), for the studies that presented data concerning these parameters. Data are presented according to time-restricted fasting (TRF) and the metabolic conditions: healthy or disease condition, obesity, T2D or MetS. For each study, the square represents the mean difference between baseline and fasting conditions, with the horizontal line intersecting it as the lower and upper limits of the 95% confidence interval [7,19,20,24,26,30,41,46,51,56,58]. Forest plot of the data from random effects meta-analysis shown as mean difference with 95% confidence intervals on blood pressure outcomes systolic blood pressure (SBP) and diastolic blood pressure (DBP), for the studies that presented data concerning these parameters. Data are presented according to time-restricted fasting (TRF) and the metabolic conditions: healthy or disease condition, obesity, T2D or MetS. For each study, the square represents the mean difference between baseline and fasting conditions, with the horizontal line intersecting it as the lower and upper limits of the 95% confidence interval [7,19,20,24,26,30,41,46,51,56,58]. Figure 13. Forest plot of the data from random effects meta-analysis shown as mean difference with 95% confidence intervals on blood pressure outcomes systolic blood pressure (SBP) and diastolic blood pressure (DBP), for the studies that presented data concerning these parameters. Data are presented according to religious fasting (RF) and the metabolic conditions: healthy or disease condition, obesity, T2D or MetS. For each study, the square represents the mean difference between baseline and fasting conditions, with the horizontal line intersecting it as the lower and upper limits of the 95% confidence interval [18,21,27,44,49,52,57].

Discussion
The main purpose of this systematic review and meta-analysis was to summarize scientific evidence on the impact of IF protocols on metabolic-related outcomes, both on healthy and metabolic-related disease conditions, namely obesity, T2D and MetS. To our knowledge, this is the first meta-analysis that presents comprehensive results on the effects of different IF protocols, Alternate-Day Fasting (ADF), Time-Restricted Fasting (TRF) and Religious Fasting (RF), on specific parameters both in health and in metabolic-related disorders. Despite the amount of data already available and the fact that many studies point to beneficial effects of IF protocols on several metabolic parameters, data are conflicting. While reasons are still elusive, we hypothesized that different basal metabolic conditions such as those herein analyzed, as well as the multitude of IF protocols could contribute for distinct results and conclusions.
Herein, we have considered not only the effectiveness of different IF protocols but also how they impact in individuals with different metabolic status. For our analysis we selected three main common metabolic disorders: obesity, T2D and MetS. Regarding each of the metabolic parameters, for the adiposity outcomes, it was possible to observe that the different IF approaches resulted in the reduction in weight, BMI and waist circumference. These data suggest that independently of the IF intervention protocol, and most probably the reasons behind these observations are most likely related with the inability of individuals to fully compensate, during non-fasting periods, the calorie deficit associated with IF protocols [76][77][78]. Furthermore, fasting periods might also diminish the hunger that these individuals usually feel [77,[79][80][81][82]. For the remaining outcomes, namely lipid, insulin and blood pressure homeostasis, ADF was the type of IF that showed the most beneficial effects in comparison with TRF and RF. The ADF major effectiveness might be related with higher fasting time of this intervention and a greater overall caloric restriction. These alterations can elicit an augmented autophagy, improved insulin sensitivity and reduce inflammation, possibly by modulating the gut microbiota and the release of inflammatory cytokines [3,83,84]. Furthermore, the increment of fasting periods Figure 13. Forest plot of the data from random effects meta-analysis shown as mean difference with 95% confidence intervals on blood pressure outcomes systolic blood pressure (SBP) and diastolic blood pressure (DBP), for the studies that presented data concerning these parameters. Data are presented according to religious fasting (RF) and the metabolic conditions: healthy or disease condition, obesity, T2D or MetS. For each study, the square represents the mean difference between baseline and fasting conditions, with the horizontal line intersecting it as the lower and upper limits of the 95% confidence interval [18,21,27,44,49,52,57].

Discussion
The main purpose of this systematic review and meta-analysis was to summarize scientific evidence on the impact of IF protocols on metabolic-related outcomes, both on healthy and metabolic-related disease conditions, namely obesity, T2D and MetS. To our knowledge, this is the first meta-analysis that presents comprehensive results on the effects of different IF protocols, Alternate-Day Fasting (ADF), Time-Restricted Fasting (TRF) and Religious Fasting (RF), on specific parameters both in health and in metabolic-related disorders. Despite the amount of data already available and the fact that many studies point to beneficial effects of IF protocols on several metabolic parameters, data are conflicting. While reasons are still elusive, we hypothesized that different basal metabolic conditions such as those herein analyzed, as well as the multitude of IF protocols could contribute for distinct results and conclusions.
Herein, we have considered not only the effectiveness of different IF protocols but also how they impact in individuals with different metabolic status. For our analysis we selected three main common metabolic disorders: obesity, T2D and MetS. Regarding each of the metabolic parameters, for the adiposity outcomes, it was possible to observe that the different IF approaches resulted in the reduction in weight, BMI and waist circumference. These data suggest that independently of the IF intervention protocol, and most probably the reasons behind these observations are most likely related with the inability of individuals to fully compensate, during non-fasting periods, the calorie deficit associated with IF protocols [76][77][78]. Furthermore, fasting periods might also diminish the hunger that these individuals usually feel [77,[79][80][81][82]. For the remaining outcomes, namely lipid, insulin and blood pressure homeostasis, ADF was the type of IF that showed the most beneficial effects in comparison with TRF and RF. The ADF major effectiveness might be related with higher fasting time of this intervention and a greater overall caloric restriction. These alterations can elicit an augmented autophagy, improved insulin sensitivity and reduce inflammation, possibly by modulating the gut microbiota and the release of inflam-matory cytokines [3,83,84]. Furthermore, the increment of fasting periods contributes to a decreased in energy intake [85], to a depletion of liver glycogen and a metabolic switch from lipid/cholesterol synthesis and fat storage to the utilization of fat as a substrate [86]. Therefore, IF protocols' effectiveness might be depend on different factors such as the duration and frequency of fasting periods, the amount and type of food consumed during feeding periods, and individual variation in genetic and metabolic factors [3,22,87].
Although RF have a similar fasting time, when compared to TRF, this last appears to present more pronounced beneficial effects. The differences between TRF and RF are most probably associated with the alterations promoted by RF in the circadian rhythm. It has been postulated that TRF, due to the limited eating windows, allow for the synchronization of the circadian system, consequently optimizing the metabolic function [88]. Indeed, it was already described that both TRF and ADF have beneficial effects aligned with the circadian rhythm, which have consequently benefits on glucose regulation, beta cell responsiveness, body composition and weight, reduction in oxidative stress and metabolic switch [3,83,84,89]. RF disturbs the circadian system, which, as previously described, can predispose to several dysfunctions, such as the impairment of glucose tolerance, reduction in sensitivity to insulin and increased arterial blood pressure [3,83,84,89].
Our analysis shows that a significant weight loss, reduction in BMI and waist circumference is particularly observed in obese and MetS individuals independently of IF approach. MetS constitutes a group with a small number of studies that did not provide enough information to allow a categorization of the participants in either people with obesity or T2D. However, one has to keep in mind that MetS is a combination of several metabolic disfunctions, including obesity, (one of the metabolic disorders herein analyzed) insulin resistance, hypertriglyceridemia, hypercholesterolemia, hypertension and reduced high-density lipoprotein (HDL)-cholesterol concentrations [90]. Therefore, the promoted metabolic shift and the reduction in fat mass, induced by the IF protocols, could underlie the beneficial effects observed in the lipid profile of obese and MetS individuals, in regards to the increment of HDL-c and the reduction in the total cholesterol and triglycerides, as already reported in the literature (reviewed in [91]). T2D individuals benefit by the implementation of IF protocols through the reduction in fasting glucose and insulin. T2D is triggered by a combination of two essential factors, a defect in insulin secretion and/or the inability of insulin-sensitive tissues to respond appropriately to insulin [90]. It is proposed that these insulin defects are linked to increased adiposity and subsequent chronic inflammation, leading to the development of insulin resistance in tissues [90]. Therefore, we hypothesized that reduction in caloric intake and the consequent metabolic changes, implied by IF, underly the establishment of a better insulin homeostasis. Furthermore, it is also described that IF protocols might incite a prolonged decrease in insulin production and increased levels of AMPK, which can result in the improvement of insulin sensitivity and glucose homeostasis [92]. In relation to healthy individuals, data herein presented have no power to infer the role of IF in the modulation of the metabolic parameters analyzed, under healthy circumstances. Cohorts of healthy individuals are reduced and essentially subjected to RF, which presented the worst IF results in health promotion.
This work faced several limitations including great heterogeneity between studies, lack of blinding and variable quality of result reporting. The high heterogeneity observed in the global analysis can be explained by the strategy of grouping participants with different metabolic status, the diversity of IF interventions, variation in population age, gender ratio and geographical localization. In addition, individual weight loss and behavior change, such as increased physical activity or improved dietary habits, can be confounding factors when assessing the benefits of IF. Weight loss, a common outcome of IF, can, independently of IF intervention, lead to improvements in metabolic markers, such as insulin sensitivity and lipid profiles, making it difficult to determine which factor is responsible for the enhancement of metabolic health [93,94]. Therefore, it is important to control for these factors when designing and conducting studies on the effects of IF. However, the observed heterogeneity, by other side, can be seen as a major find, suggesting that IF should be personalized. Other aspect is that the literature search ranged from 2000 to June 2022. Although the concept of IF and its variations have been around for centuries, and various cultures and religions have practiced fasting for spiritual, health or cultural reasons, the scientific understanding and interest in IF gained more attention and recognition in the scientific community in the 21st century. By the year 2000, the term IF was widely used in the scientific literature to describe a variety of different fasting protocols, including ADF, TRF and RF. Furthermore, it is only in the last few decades that IF protocols have become more standardized. Therefore, for a more consistent classification, we decided to select studies published after the year 2000, keeping in mind that this is one of the study limitations.
Another point to consider in this study is that the commonly used measures of IF, such as changes in body weight, blood glucose levels or lipid profiles, may not be perfect proxies to estimate the metabolic effects of IF. Indeed, these measures can be easily influenced by other factors, for example body weight, changes in water weight, muscle mass and measurements under fasting or feeding conditions, rather than just changes in fat mass [93,95]. Importantly, there are other measures that may provide more accurate insights into the cellular mechanisms underlying the metabolic IF benefits. Examples include changes in biomarkers of cellular stress, oxidative damage or even autophagy [81,83,96]. Nevertheless, the measurement of these biomarkers is not well defined in the clinical context, neither included in the hospitals' routine analysis. Therefore, new approaches must be applied in order to overcome this drawback.
Despite the described limitations, one of the main findings of this study is that IF interventions have beneficial effects for most of individuals included in the studies, and independently of the IF protocol used. IF has been associated with improvements in weight loss, control of blood sugar and blood pressure and cholesterol levels [3,75,88,[97][98][99]. However, not all individuals may benefit from IF, since it is described that some individuals may experience negative side effects such as hunger, fatigue and irritability [84,99]. IF benefits are mostly dependent of mechanisms associated with a metabolic shift towards the predominant use of fatty acids as fuel for energy [3,100]. However, IF benefits are far more complex and not only restricted to the positive effects of fatty acids usage for energy generation. IF interventions involve alternated periods of fasting and feeding, which correspond to different metabolic homeostasis status. In the fasting time, cells adopt a stress-resistance mode through reduction in insulin signaling and overall protein synthesis [3,55,84,87,101,102]. This stress-resistance mode is associated with activation of signaling pathways, which improve mitochondrial function, stress resistance and antioxidant defenses, and also increase autophagy to remove damaged molecules and recycle their components [3,83,103]. In contrast, during the feeding, glucose levels and protein synthesis increase while ketone bodies levels drop, allowing cellular growth and repair [83,84,87,102,103]. Therefore, the maintenance of IF regimens lead to long-term adaptations, most likely through a hormesis effect that improve cellular homeostasis and increase disease resistance [3,84,[102][103][104][105][106].

Conclusions
Overall, we verified that the implementation of the different types of IF protocols has distinct effects. According to data herein presented, ADF and TRF protocols have major beneficial effects in the improvement of dysregulated metabolic conditions. In addition, IF protocols have a major beneficial impact for obese and MetS individuals, through the improvement of adiposity, lipid homeostasis and blood pressure. For T2D individuals, the IF beneficial effects were limited, but associated with their major metabolic dysfunctions. These individuals carefully require consideration of IF protocols, proper medication adjustment and self-monitoring of blood glucose levels to improve results (Graphical abstract).
One of the caveats of this work is related with the high heterogeneity seen, particularly in studies of individuals with obesity and MetS, which can be explained by diversity of IF interventions, as well as, the metabolic status of the individuals, strengthen the notion that IF should be tailored to the individual. Therefore, our data clearly show that IF impacts metabolic homeostasis differently depending on the individual's basal metabolic status.
In a forward-looking perspective, there is still much research to be carried out in this area, and more studies are needed to fully understand the potential benefits and risks of IF for different populations and under different conditions. Supplementary Materials: The following supporting information can be downloaded at: https:// www.mdpi.com/article/10.3390/jcm12113699/s1, Table S1: Overview of the main analysis of studies included in qualitative synthesis, Table S2: Assessment of studies quality, Table S3: Analysis of the impact of intermittent fasting on different outcomes and Table S4: Analysis of the impact of intermittent fasting, without the fasting Ramadan studies, on different outcomes. Funding: This work has been funded by National funds, through the Foundation for Science and Technology (FCT)-project UIDB/50026/2020 and UIDP/50026/2020.