Comparative efficacy of exercise training modes on systemic metabolic health in adults with overweight and obesity: a network meta-analysis of randomized controlled trials

Objective This network meta-analysis (NMA) was conducted to compare and rank the effects of training interventions including aerobic exercise (AE), resistance training (RT), combined aerobic and resistance training (CT), and high-intensity interval training (HIIT) on vital metabolic indicators in adults with overweight and obesity. Methods PubMed, Cochrane, Embase, and Web of Science were searched from 1990 to February 2023. Articles were included if they described randomized controlled trials (RCTs) examining the effects of exercise training on anthropometry parameters, lipid profiles, glucose metabolism, blood pressure, and cardiorespiratory fitness in adults with overweight and obesity. Weighted mean difference with 95% CI was calculated. Results A total of 28 studies with 1,620 patients were included. Results revealed that AE exerts best effects on weight loss (−2.35 [−4.05, −0.64]) and body mass index (−0.9 [−1.38, −0.42]), while HIIT is the most effective in reducing waist circumference (−5.93 [10.71, −1.15]), percentage body fat (−3.93 [−5.73, −2.12]), serum triglycerides (−20.55 [−37.20, −3.91]), and fasting blood glucose (−14.31 [−22.47, −6.16]) and improving VO2 max (7.41 [4.37, 10.45]). However, no significant benefit was observed in terms of total cholesterol and blood pressure. Conclusions AE is the optimal exercise type for reducing body weight and BMI, while HIIT exerts the most beneficial effects on improving body composition, cardiorespiratory fitness, and metabolic abnormalities in adults with overweight and obesity. Systematic review registration https://www.crd.york.ac.uk/prospero/display_record.php?ID=CRD42023444322, identifier CRD42023444322.


Introduction
Obesity is predicted to affect nearly 20% of people around the world and is causing a huge burden as a noncommunicable disease (1).Given the enormous personal and medical costs of overweight and obesity, it is urgent for public health policymakers to identify effective interventions.
Owing to the lack of safe and effective drugs for the treatment of obesity, the main interventions for obesity are still diet control and exercise.The main forms of exercise are aerobic exercise (AE), resistance training (RT), combined aerobic and resistance training (CT), and high-intensity interval training (HIIT) (2).According to the recommendation of guidelines of exercise for people with obesity, moderate-intensity AE for at least 300 min per week (3) was the preferred choice.However, the effects of different exercise modalities on body composition, metabolic health, and cardiovascular health in people with obesity were inconsistent.A meta-analysis reported that AE is beneficial for reducing body weight and fat mass, while RT is effective in reducing fat mass (4).Another meta-analysis showed that HIIT was slightly more effective than AE in improving VO 2 max (5).Moreover, a metaanalysis of adolescents with obesity has shown that compared with AE alone, CT resulted in greater reductions in LDL levels (6).
Despite the fact that it is well known that exercise is effective in reducing body weight and fat mass and promoting cardiovascular metabolism, a potential drawback is the large weekly time consumption, which totals approximately 300 to 400 min per week.Thus, in recent years, HIIT is being increasingly recommended as an alternative option mainly because it has high efficiency and is time-saving, as lack of time is the main obstacle that prevents people from exercising (7).There is increasingly strong evidence showing that HIIT has better effects on a series of health indicators in both healthy and chronically ill people than continuous moderate-intensity training (8)(9)(10).However, in terms of the effects of exercise training on people with overweight and obesity, most studies are pairwise meta-analyses that cannot rank the effectiveness of different trainings.
Different from a conventional meta-analysis, a network metaanalysis (NMA) is a powerful tool for comparing the effectiveness of different interventions as it can combine direct and indirect evidence and allow for the ranking of treatments (11).However, the current available NMA focused only on the effects among AE, RT, and CT, without considering the HIIT.More importantly, it is unclear which type of exercise is most effective for the different metabolic abnormalities in people with overweight and obesity.Therefore, the aims of this NMA were to comprehensively evaluate and rank the effectiveness of exercise across the vital metabolic indicators including anthropometry, glucose homeostasis, lipid profile, and cardiovascular fitness in adults with overweight and obesity.

Materials and methods
This systematic review and NMA follows the Preferred Reporting Items for Systematic Reviews and Meta-Analyses for Network Meta-Analyses (PRISMA-NMA) (12) and is registered in the PROSPERO database (Registration number: CRD42023444322).

Inclusion and exclusion criteria
Studies were considered eligible for inclusion if the following criteria were met: (1) RCT must be published in English.( 2 The exclusion criteria were as follows: (1) The intervention was combined with diet control.(2) The subjects were reported to be taking medication or had other noncommunicable diseases, such as cancer, hypertension, and diabetes.(3) Articles that did not include the outcome of interest.(4) Conference abstracts, case reports, and dissertations were excluded.(5) Duplications of the searched studies.
Abstracts and full texts were screened by two authors independently, and any uncertainty was discussed among authors.

Assessment of study quality
We used the Cochrane Risk of Bias Tool (13) to assess the risk of bias (ROB) of the included studies independently.The following domains were considered: (1) randomization process, (2) treatment allocation concealment, (3) blinding of participants and researchers, (4) blinding of outcome assessment, (5) incomplete outcome data, (6) selective reporting, and (7) other bias.

Statistical analysis
We use weighted mean difference and 95% confidence interval (CI) to report continuous variables.Review Manager 5.3 was used for the traditional pairwise meta-analysis.Statistical heterogeneity was evaluated using the (I 2 ) statistic: I 2 > 50% was considered high heterogeneity and random-effects model was employed; otherwise, the fixed-effects model was adopted.For the NMA, Stata16.0.software package "mvmeta" and "network" was used to analyze indirect comparisons between different interventions.
The network diagram was used to present the relationship between exercise interventions.The node-splitting method and loop inconsistency test were performed to test inconsistency.p > 0.05 indicates that there is no significant inconsistency and the consistency model is chosen; otherwise, the inconsistency model is employed (14,15).
The effectiveness of different interventions was ranked by using the surface under the cumulative ranking probability diagram (SUCRA) (16).The value of SUCRA is higher, and the effects of interventions are better.League tables were used to present the pairto-pair comparisons between the exercise interventions.
Furthermore, a comparison-adjusted funnel plot was used to assess publication bias.To explore the heterogeneity, subgroup analyses were performed in pairwise meta-analyses, including different duration of exercise (4-8 weeks as short-term duration and >9 weeks as long-term duration) and mean baseline BMI (25-30 kg/m 2 as individuals with overweight and ≥30 kg/m 2 as individuals with obesity).

Literature selection
According to the search strategy, 4,381 studies were identified in those four databases.After 1,425 duplicates were removed, 2,956 articles were left to screen the titles and abstracts.After reading the full text of 105 studies, we included 28 RCTs for the NMA.The flowchart of the search and selection process is presented in Figure 1.

Characteristics of included studies
Table 1 shows the clinical and demographic characteristics of the 28 included studies (17-44).A total of 1,620 participants were enrolled in the NMA; sample size ranged from 21 to 349.As for gender, 15 studies recruited women only and 8 studies recruited men only, and 5 studies included both men and women.Of the total, 75% participants were women.With regard to exercise

Risk of bias and publication bias in the enrolled studies
The summary of the risk of bias assessment is shown in Supplementary Figure 1.In brief, 15 studies showed unclear ROB in random sequence generation, allocation concealment, and blinding of the participants and implementers.All studies showed low ROB in blinding of outcome assessment and selective reporting.Only one study showed risk of other biases.Overall, the quality of the included studies was considered moderate.

Network meta-analysis and subgroup analyses
Four measures of anthropometry and body composition (BW, BMI, WC, and %BF), three measures of lipid metabolism (TG, HDL, and LDL), three measures of glucose metabolism (FBG, FINS, and HOMA-IR), and one measure of cardiovascular function (VO 2 max) were included in the NMA.Table 2 illustrates the results of pair-to-pair comparisons between the exercise interventions.Figure 2 illustrates NMA maps of studies examining the efficacy of exercise modes on 11 indicators.Table 3 presents the rank of exercise interventions in order of effectiveness.The results of subgroup analyses are shown in Supplementary Table 3.The results of interval plot and cumulative ranking probability plots of NMA can be found in Supplementary Figures 2-23.The results of local inconsistency are displayed in Supplementary Figures 24-34.Funnel plot graphics are illustrated in Supplementary Figures 35-45.3).Subgroup analysis showed that short-term exercise could only reduce the %BF (−4.59 [−8.42, −0.76]) while long-term duration of exercise could efficiently reduce body weight, BMI, waist circumference, and %BF.In addition, all exercise could improve anthropometric outcomes for individuals with overweight and obesity (Supplementary Table 3).

Discussion
This NMA comprehensively analyzed the current studies about exercise interventions for people with overweight and obesity.Our main findings indicate that AE is probably the best choice for reducing body weight and BMI, while HIIT may be the optimum strategy to improve glucose homeostasis, lipid metabolism, and cardiorespiratory fitness.

Anthropometric and body composition
BW and BMI are often the main endpoints in evaluating obesity.BMI could effectively predict overall mortality but limit the alterations in body composition (45,46).Therefore, %BF may be considered as a more significant indicator for reflecting the changes in body composition (47).Waist circumference represents the level of visceral fat, which is independently related to all-cause mortality and is considered as a better outcome for predicting the conditions of obesity (48,49).Our study showed that AE is probably the best in controlling BW and BMI, whereas HIIT could be the optimal exercise intervention for reducing WC and %BF in individuals with overweight and obesity.A recent NMA has shown that CT is the best intervention for reducing WC and %BF with a small reduction in weight loss, but this study did not assess the benefits of HIIT in people with obesity (50).A systematic review comparing HIIT with continuous moderate-intensity training on body composition in adults with overweight and obesity has found that HIIT has similar effects on %BF reduction, but requires ~40% less time commitment (51).Another meta-analysis has reported that HIIT provided nearly 30% more reductions in fat mass than continuous moderate-intensity training (52).The subgroup analysis has found that more than 9 weeks of exercise could effectively improve anthropometric outcomes, and individuals with overweight and obesity could obtain benefits from exercise.In general, our results demonstrated that compared with other types of exercise, AE and HIIT were superior in improving body composition.

Glucose metabolism
Individuals with overweight or obesity are often associated with poor glycemic control, which increase their risk of developing diabetes (53).Our study has shown that HIIT could be the most effective exercise mode in reducing fasting blood glucose, insulin, and HOMA-IR and exercise is more effective in people with obesity (BMI > 30).This may be because overweight participants had relatively healthy glucose metabolism than people with obesity.Additionally, although previous meta-analysis did not consider the effects of duration of exercise on glucose metabolism, our subgroup analysis has found that 4 to 8 weeks of exercise could lower fasting insulin level.

Lipid profile
Guidelines from the American College of Sports Medicine (ACSM) have recommended that more than 150 min of moderate-intensity exercise per week can effectively reduce blood lipid levels (3).The results of NMA showed that HIIT is the most effective mode in improving TG, LDL, and HDL levels.Previous studies have suggested that exercise could improve the TG and HDL levels, but seldom alter TC levels (54), which is consistent with our results.Additionally, a previous meta-analysis demonstrated that HIIT could improve the blood lipid metabolism for type 2 diabetes mellitus (55).Although patients with overweight and obesity included in our study had no complications of dyslipidemia, the results have shown that HIIT moderately reduced lipid levels within the normal range, which was beneficial for reducing the risk of cardiovascular disease.Subgroup analysis found that TG, LDL, and HDL levels were improved in short-term exercise, indicating that exercise could improve lipid metabolism before obvious body weight reduction.However, the reductions in TG and LDL levels were only observed in individuals with overweight, indicating that exercise alone has a limited effect on improving lipid metabolism in people with obesity.

Cardiovascular function
Blood pressure is a commonly assessed measure related to cardiovascular health (56).Many studies have shown that exercise can lower diastolic and systolic blood pressure in people with hypertension (57).For patients with overweight and obesity, we did not find significant changes in blood pressure in both pairwise meta-analysis and subgroup analysis.This may because the baseline of blood pressure in included participants is normal.
Cardiorespiratory fitness (CRF) has long been disregarded as an important moderator of the negative association between mortality and obesity (58).Although increasing studies demonstrated that CRF has a greater impact on morbidity and mortality than body fat percentage (59), most meta-analyses have continued to focus on anthropometric measurements more than CRF when evaluating the effects of exercise for patients with obesity, but have not included CRF in their analyses.Our study has found that all types of exercise can significantly improve CRF, of which HIIT has the highest probability of elevating CRF, followed by AE.One previous metaanalysis has reported that compared with moderate-intensity continuous training, HIIT leads to greater improvement in CRF in children (60).Moreover, another meta-analysis demonstrated that HIIT significantly increases CRF more than continuous moderate-intensity training in patients with chronic diseases (10).Our subgroup analysis found that short-term exercise is enough to increase the VO 2 max and the benefits existed in both individuals with overweight and obesity.Overall, HIIT may be the preferred option for improving CRF in patients with overweight and obesity.
Previous studies have mainly focused on the effectiveness of HIIT on CRF; beyond that, our research has found that HIIT is also the optimal choice for improving body composition and glucose and lipid metabolism compared with other exercise modalities.Although HIIT has favorable overall improvements, it may not be suitable for people without training foundation.Considering AE is the best for reducing body weight and BMI, it is necessary to take into account the patient's physical fitness and personal preferences when health professionals advise on exercise training to improve obesity.

Strengths and limitations
There are some strengths in this study.First, the included indicators are comprehensive and can represent the systemic metabolic status of patients effectively.Additionally, we used NMA to evaluate and rank the efficacy of interventions by combining direct and indirect evidence obtained from RCTs.Moreover, we performed subgroup analysis using the mean baseline BMI to explore the different efficacy of exercise on people with overweight and obesity and the duration of exercise to explore how long it takes to get metabolic benefits.However, there are several limitations to our study.First, most of the included RCT had high risks in the blinding of participants, which is unlikely to carry out in this kind of intervention, but because the outcomes were all measured objectively, this ROB does not seem to influence the objective outcomes.Second, owing to the limited number of included studies, we did not consider the sequence and intensity of exercise.Third, because of the limited direct comparisons for some interventions, the results should be interpreted cautiously.Fourth, of all patients included, 75% were women.Owing to the limited number of included studies, it is difficult to conduct subgroup analysis of gender, and considering the imbalance of sex ratios, this result may be more suitable for female patients.Fifth, although the majority (46%) of included patients used AE, there was no significant inconsistency in the effects of different exercises, indicating that the difference in number of people did not significantly affect our conclusions.

Conclusion
In summary, our study has demonstrated that different types of exercise have varying efficacy for different metabolic abnormalities.AE produces the best results in reducing BW and BMI, whereas HIIT is most likely the best exercise intervention for improving body composition and systemic metabolic status in adults with overweight and obesity.The findings display a deeper insight into the exercise prescription for individuals with overweight and obesity in the real world.

Funding
The author(s) declare financial support was received for the research, authorship, and/or publication of this article.This work was supported by grants from National Key R&D Program, China (2020YFA0803604; 2022YFA0806102), National Natural Science Foundation of China (91957113; 31871180) to FH.

FIGURE 1
FIGURE 1Schema of the search strategy.

2
FIGURE 2 Network diagram of network meta-analysis comparisons.(A) BW, body weight, (B) BMI, body mass index, (C) %BF, percentage body fat, (D) WC, waist circumference, (E) FBG, fasting blood glucose, (F) FINS, fasting blood insulin, (G) HOMA-IR, (H) TG, triglycerides, (I) HDL, high-density lipoprotein, (J) LDL, low-density lipoprotein, (K) VO 2 max.CON, control group; AE, aerobic exercise; RT, resistance training; CT, aerobic combined resistance training; HIIT, high-intensity interval training.The size of the nodes represents the number of participants in an intervention, and the thickness of lines between interventions represents the number of studies that compare them.

TABLE 1
Study characteristics.

TABLE 1 Continued
categories, participants were included in the AE (n = 757), RT (n = 108), CT (n = 107), and HIIT (n = 111) groups, respectively.The duration of the training ranged from 4 weeks to 48 weeks (mostly 12 weeks).The number of weekly sessions was mostly three times a week (n = 20).

TABLE 2
Matrix of the network meta-analysis results.

TABLE 2 Continued
The league table demonstrates the relative efficacy for each pair of comparison showing the effect of exercise type in each row compared with exercise type in each column.Data are presented as mean difference (95% CI).CON, control group; AE, aerobic exercise; RT, resistance training; CT, aerobic combined resistance training; HIIT, high-intensity interval training.BW, body weight; BMI, body mass index; WC, waist circumference; %BF, percentage body fat; FBG, fasting blood glucose; FINS, fasting blood insulin; TG, triglycerides; HDL, high-density lipoprotein; LDL, lowdensity lipoprotein.The most effective exercise for each indicator was highlighted in bolded form

TABLE 3
Ranking of exercise interventions in order of effective.