Impact of Different Exercise Modalities on Visceral Adipose Tissue in Normal Weight and Low-Overweight Individuals: A Systematic Review and Meta-Analysis

Abstract

The “thin outside, fat inside” (TOFI) phenotype, also termed normal weight obesity (NWO), describes individuals with normal body mass index (BMI) but elevated visceral adipose tissue (VAT). Individuals at the lower boundary of the overweight range may also present with excess VAT despite near-normal BMI. Although exercise is widely recommended to improve body composition, its effects on VAT in these populations remain unclear. This review evaluated the effects of different exercise modalities on VAT in adults with BMI ≤27 kg/m². A systematic search of PubMed, Scopus, and Web of Science was conducted for studies published between January 1999 and July 2024. Eligibility was determined at the study level based on reported mean BMI (≤27 kg/m²). Studies including aerobic, resistance, combined, or interval training interventions with VAT assessed via computed tomography, magnetic resonance imaging, or dual-energy X-ray absorptiometry were included. Between-group (controlled) and within-group (single-arm or pre–post) analyses were conducted separately using standardized mean differences (SMD). The protocol was registered on OSF (10.17605/OSF.IO/3DCXJ). Thirteen studies (n = 354 participants) were included. Between-group analysis demonstrated a significant reduction in VAT favoring exercise (SMD = −1.590; 95% CI: −2.283 to −0.898; p < 0.001), with substantial heterogeneity (I² = 86.4%). Aerobic training showed the most consistent effects, while interval training was also associated with significant reductions based on a limited number of studies. Longer intervention duration (>5 months) was associated with greater effects. Within-group analyses showed a similar direction of effect but did not reach statistical significance. Sensitivity analyses indicated that results were directionally robust but partially influenced by high-impact studies. Exercise is associated with reductions in VAT in adults with normal BMI and those at the lower boundary of the overweight range. However, the certainty of evidence is low due to methodological limitations, heterogeneity, and potential bias. Greater emphasis should be placed on findings from controlled studies. Further well-designed randomized controlled trials are needed to confirm these findings and clarify the roles of different exercise modalities and intervention characteristics.

Keywords

exercise body composition visceral fat body mass index normal weight

Introduction

Visceral adipose tissue (VAT) is increasingly recognized as a metabolically active fat depot with a disproportionate influence on cardiometabolic risk, independent of total body fat or body mass index (BMI). Unlike subcutaneous adipose tissue, VAT exhibits higher lipolytic activity, increased inflammatory cytokine secretion, and greater portal delivery of free fatty acids to the liver, thereby contributing directly to insulin resistance, dyslipidemia, and other components of metabolic syndrome (Conus et al., 2007; De Lorenzo et al., 2006; Kaur, 2014). These pathogenic characteristics position VAT as a critical therapeutic target, particularly in individuals who do not meet conventional BMI-based criteria for overweight or obesity.

The “thin outside, fat inside” (TOFI) phenotype, also referred to as normal-weight obesity (NWO) or metabolically obese normal weight (MONW), describes individuals with a normal BMI (18.5 – 24.9 kg/m²) who nonetheless exhibit excessive VAT accumulation and elevated metabolic risk. This phenotype is commonly identified using imaging-derived indices, including the intra-abdominal adipose tissue (De Lorenzo et al., 2006; Harrison et al., 2016; Kaur, 2014; Van Der Beek et al., 2017). The TOFI phenotype has been described using overlapping terms, including NWO and MONW. Although these constructs share common features, they are not fully interchangeable, as they may emphasize different aspects of metabolic dysfunction or body composition. In the present review, these terms are considered within a unified framework characterized by relatively normal BMI in the presence of disproportionate visceral adiposity, which serves as the primary feature of interest.

The distribution of visceral relative to subcutaneous adipose tissue, often expressed as the intra-abdominal to abdominal subcutaneous adipose tissue ratio (IAAT/ASAT), has also been proposed as an indicator of metabolic risk (Zdrojewicz et al., 2017). Higher ratios have been associated with adverse metabolic profiles, with some studies suggesting sex-specific differences in fat distribution patterns. However, universally accepted threshold values remain lacking, and reported cut-offs vary across populations and measurement techniques (Bennett et al., 2024). VAT areas exceed approximately 100 – 130 cm², assessed via computed tomography, have been associated with elevated cardiometabolic risk in both men and women. However, these thresholds are not universally standardized and may vary according to sex, age, and measurement modality (Bennett et al., 2024; Després, 2006; Nicklas et al., 2003). Visceral adipose tissue accumulation is also influenced by age, with evidence indicating a progressive increase across adulthood due to hormonal changes, reduced physical activity, and alterations in fat distribution (Kuk et al., 2005).

Exercise is a well-established non-pharmacological intervention for improving cardiometabolic health and body composition, with substantial evidence demonstrating preferential reductions in VAT relative to total body mass (Bellicha et al., 2021; Park et al., 2003; Verheggen et al., 2016). Aerobic training (AT) is consistently associated with significant VAT reductions, while high-intensity interval training appears to elicit comparable effects to moderate-intensity continuous training (MICT) when energy expenditure is matched (Gonzalez-Ruiz et al., 2017). Resistance training (RT), although less effective for VAT reduction in isolation, contributes to increases in lean mass and overall metabolic health (Khalafi et al., 2021). Combined AT and RT protocols do not consistently produce greater VAT reductions than AT alone, potentially due to insufficient aerobic training volume when modalities are integrated (Ismail et al., 2012; Vissers et al., 2013).

Despite this body of evidence, current exercise-VAT literature is dominated by studies conducted in mixed-BMI or overtly obese populations. Consequently, it remains unclear whether conclusions regarding exercise modality, intensity, and volume can be directly extrapolated to individuals with normal BMI or those at the lower boundary of the overweight category. Emerging evidence suggests that VAT accumulation and metabolic dysfunction may already be present in individuals near the upper limit of normal BMI or slightly above it, despite the absence of clinical obesity (Chen et al., 2024; Oldham & Robinson, 2018; Robinson, 2017). Including individuals with BMI values up to 27 kg/m² therefore represents a pragmatic strategy to capture early VAT-driven metabolic risk, despite not meeting conventional BMI-based definitions of overweight or obesity. As mentioned, while a BMI range of 18.5 – 24.9 kg/m² is conventionally defined as normal weight, evidence suggests that visceral adiposity and related metabolic risk may increase progressively (Conus et al., 2007) within the upper range of normal BMI and extend into the lower boundary of overweight. Individuals within this range may be misclassified as metabolically healthy based solely on BMI despite harboring excess VAT (Oldham & Robinson, 2018; Robinson, 2017). Therefore, a threshold of BMI ≤27 kg/m² was selected to capture individuals at the interface between normal weight and early overweight, where disproportionate visceral fat accumulation may already be present, while minimizing confounding effects associated with established obesity. This approach reflects a pragmatic strategy commonly adopted in aggregate-data meta-analyses, where individual participant data are not available to define phenotype boundaries more precisely (Higgins et al., 2024).

Therefore, the aim of this systematic review and meta-analysis was to evaluate the efficacy of different exercise modalities in reducing visceral adipose tissue in adults with normal BMI (18.5 – 24.9 kg/m²) and those at the lower boundary of the overweight range (25 – 27 kg/m²). Accordingly, the target population in this review is defined as adults with normal BMI or near-normal BMI who may exhibit elevated or disproportionate VAT, regardless of external anthropometric appearance. This review seeks to clarify whether established exercise prescriptions for VAT reduction are applicable to individuals who are not classified as obese by BMI.

Methods

Search Strategy and Study Identification

A systematic literature search was conducted in PubMed/MEDLINE, Scopus, and Web of Science from January 1999 to July 2024. Google Scholar was used as a supplemental source for citation tracking and identification of additional relevant studies not captured in the primary database searches. The search strategy combined controlled vocabulary and free-text terms related to visceral adipose tissue, body mass index, and exercise interventions. Database-specific search strategies were developed using Boolean operators and adapted to the syntax of each database. The full reproducible search strategies for all databases are provided in the Supplementary Material.

Inclusion criteria were: (1) human intervention studies, (2) adult participants (≥18 years), (3) BMI ≤27 kg/m², (4) chronic exercise interventions (≥4 weeks), (5) direct measurement of visceral adipose tissue using CT, MRI, or DXA. Exclusion criteria included observational designs, pharmacological interventions, pediatric populations, and studies assessing adiposity solely via anthropometry or bioelectrical impedance. The review protocol was prospectively registered on the Open Science Framework (OSF; DOI: 10.17605/OSF.IO/3DCXJ). The study selection process is summarized in a PRISMA flow diagram presented in the Results section (Figure 1), including reasons for exclusion at each stage.

Figure 1.

Flowchart of systematic review search strategy

Risk-Of-Bias Assessment

Risk of bias was assessed using the Cochrane Risk of Bias 2 (ROB2) tool (Chandler et al., 2019), which evaluates five domains: (1) bias arising from the randomization process, (2) bias due to deviations from intended interventions, (3) bias due to missing outcome data, (4) bias in measurement of the outcome, and (5) bias in selection of the reported result (Crocker et al., 2023). Two reviewers independently assessed each study. Disagreements were resolved through discussion or consultation with a third reviewer when necessary.

Certainty of Evidence Assessment (GRADE)

The certainty of evidence for the primary outcome (change in VAT) was evaluated using the Grading of Recommendations Assessment, Development, and Evaluation (GRADE) approach. The body of evidence was assessed across five domains: risk of bias, inconsistency, indirectness, imprecision, and publication bias. Randomized controlled trials were initially rated as high-certainty evidence and were downgraded if concerns were identified in any of the domains. Risk of bias judgments were informed by the ROB2 assessment. Statistical heterogeneity (I²), confidence intervals, total sample size, and funnel plot asymmetry with Egger’s regression test were considered when evaluating inconsistency, imprecision, and publication bias. The overall certainty of evidence was categorized as high, moderate, low, or very low (Guyatt et al., 2011, 2016).

Comparators

Studies were eligible if they reported either: between-group comparisons (exercise vs. control or alternative intervention), or within-group pre-post changes in visceral adipose tissue in the absence of a control group.

Interventions

Included studies examined chronic exercise interventions. Participants were assigned to groups according to the study design, based on different exercise modalities: aerobic exercise, endurance training, resistance exercise, or combined aerobic and resistance exercise, conducted in either continuous or interval formats.

Participants

Studies involving adult participants aged 18 years or older with a normal BMI (18.5 – 24.9 kg/m²) and the lower portion of the overweight range (≤27 kg/m²) were included. The threshold of BMI ≤27 kg/m² was selected to capture individuals near the upper limit of normal weight and the lower range of overweight, where disproportionate visceral adiposity and early metabolic dysfunction have been reported, while minimizing confounding effects associated with established obesity.

Eligibility was determined at the study level based on reported mean BMI values, as individual participant data were not available across all studies. This approach is consistent with aggregate-data meta-analyses (Higgins et al., 2024) and allows for broader evidence inclusion while approximating the target population. However, it is acknowledged that within-cohort variability (e.g., BMI standard deviation) may include participants outside the intended BMI range, potentially reducing phenotype specificity. No restrictions were placed on sex, health status, or ethnic origin. BMI categories follow WHO definitions, but the cutoff was chosen for evidence capture and confounder control. These studies were retained to reflect the available evidence base; however, sensitivity analyses were conducted to evaluate their influence on the pooled estimates. In the present review, strict VAT cut-off values were not applied as eligibility criteria, as individual participant data were not available and reporting across studies was heterogeneous. Instead, the focus was placed on populations characterized by normal or near-normal BMI. This approach reflects a pragmatic strategy commonly used in aggregate-data meta-analyses, where phenotype definitions cannot be consistently operationalized at the individual level (Higgins et al., 2024).

Outcome Measures

Only studies that assessed VAT using computed tomography (CT), magnetic resonance imaging (MRI), or dual-energy X-ray absorptiometry (DXA) were included, while those relying on bioelectrical impedance or anthropometric evaluations were excluded. CT and MRI are widely regarded as reference-standard imaging modalities for the quantification of VAT due to their high spatial resolution and ability to directly differentiate visceral from subcutaneous compartments (Graffy & Pickhardt, 2016). DXA provides an indirect estimate of VAT based on regional attenuation algorithms (Katzmarzyk et al., 2013; Lin et al., 2013) and has demonstrated good agreement with CT and MRI, although some systematic differences may occur, particularly at higher levels of adiposity (Chan et al., 2023; Cheung et al., 2016; Micklesfield et al., 2012; Neeland et al., 2016). To account for potential variability across measurement techniques, standardized mean differences (SMD) were used in the meta-analysis, which minimizes the impact of differences in scale and measurement units across imaging modalities.

Data Synthesis and Grouping Strategy

Studies were grouped into two main categories for synthesis: within-group analyses, which included single-arm intervention studies reporting pre- and post-intervention changes in visceral adipose tissue (VAT), and between-group analyses, which included randomized or non-randomized controlled trials comparing exercise interventions with control or alternative treatment groups. Eligibility for each synthesis was determined based on the availability of comparable outcome data and sufficient statistical detail for effect size calculation.

Study Selection

The study selection process followed PRISMA guidelines. First, duplicate records were removed prior to screening. Titles and abstracts of the remaining studies were then screened for relevance according to the predefined eligibility criteria. Full-text articles were subsequently retrieved and assessed for eligibility. Studies that met the inclusion criteria were retained for the final analysis. Two investigators independently screened the studies, and any discrepancies were resolved through discussion. In cases where published articles contained incomplete or ambiguous data, the study authors were contacted for clarification. Extracted variables included study design, participant characteristics, intervention type and duration, comparator group (within- or between-group), visceral adipose tissue outcomes, and measurement methods (e.g., CT, MRI, DXA).

Statistical Analysis

Meta-analysis was conducted using Stata 17 (StataCorp, College Station, TX, USA) to calculate SMD for between group, standardized mean changes (SMC) for within group and confidence intervals (CI) to evaluate the impact of various exercise modalities on changes in visceral fat. A random-effects model was applied, and heterogeneity was assessed using the I² statistic. Values of I² greater than 50% indicated substantial heterogeneity, while values below 50% indicated low heterogeneity (Higgins et al., 2024). Data synthesis was conducted separately for controlled and single-arm studies. These analyses were conducted separately due to inherent methodological differences between study designs. Between-group comparisons provide stronger causal inference, whereas within-group analyses are more susceptible to bias and were therefore interpreted as supportive evidence. Where heterogeneity was present, subgroup analysis was used to explore potential sources. Effect sizes were interpreted following Cohen’s d guidelines, with d = 0.2 representing a small effect, d = 0.5 a medium effect, and d = 0.8 or greater a large effect (Cohen, 2013). Sensitivity analysis involved sequential exclusion of individual studies to assess the influence of each on overall results. Publication bias was assessed using a funnel plot and Egger’s test, with a p-value greater than 0.05 indicating no significant publication bias. For studies that did not report the standard deviation (SD) of change, were estimated using the following formula:

SD_change = \sqrt{(SD_{pre}^{2}} + SD_{post}^{2} - 2 \times r \times SD_pre \times SD_post)

and for mean following this formula: Mean change = Mean _post −Mean _pre

Assuming a correlation coefficient (r) of 0.8 between baseline and post-intervention values. A correlation coefficient of 0.8 was used, reflecting the strong relationship typically observed between repeated body composition measures in exercise trials. Sensitivity analyses were conducted using alternative values (r = 0.5 and r = 0.9) to assess the robustness of the results (Higgins et al., 2024; Lefebvre et al., 2024) forest plots were planned for effect sizes.

Results

Study Selection

A total of 341 original studies were identified through electronic database searches, supplemented by 13 additional manually identified articles. After initial screening and applying eligibility criteria, 344 articles were excluded, leaving 10 publications for analysis (see Figure 1). Several studies included multiple intervention arms or analytically distinct subgroups (different exercise modalities). Each arm was treated as a separate experimental cohort for meta-analytic purposes, resulting in 18 cohorts derived from 10 studies.

Subject Characteristics

The analysis included 18 experimental cohorts with a total of 472 participants (202 men and 270 women; see Table 1). Among the studies, two exclusively recruited men (Knauer et al., 2023; Poggi et al., 2024), five recruited only women (Benatti et al., 2012; Louise Thomas et al., 2000; Park et al., 2003; Poehlman et al., 2000; Zhang et al., 2021), and the remaining three recruited both genders. Participants were adults aged 18 – 75 years. Seven cohorts included individuals with a BMI in the normal range, while 11 cohorts included participants within the lower portion of the overweight range. Most studies involved inactive individuals, except one that included healthy, trained older men (Poggi et al., 2024). As detailed in Table 2, the included studies comprised a mix of randomized controlled trials (n = 4), non-randomized controlled trials (n = 3), and single-arm intervention studies (n = 3).

Table 1.

Interventions and Participant Characteristics

Study	P.CH	N	Age (yr)	BMI (kg/m²)	Mode	Form	Fre (day/week)	Int	Sess du	Du
Benatti et al. (2012)	Women, inactive, underwent a small-volume liposuction surgery	18	20–35	23.2 ± 1.3	COM	RT: 8 exercises, one to three sets, 8 to 12 repetitions/AT on a treadmill	3	67–80% 1RM/approximately 75% VO₂ max	30 min/session/30–40 min/session	4-month
Carr et al. (2005)	17 men, 15 women with IGT	32	57.2 ± 1.8	26.6 ± 0.8	Stretching exercises	Nr	3	Nr	1 h	24 months
Carr et al. (2005)	12 men, 20 women with IGT	32	55.7 ± 1.8	25.7 ± 0.8	AT	Walking or jogging on a treadmill	3	70% of HRR	1 h	24 months
He et al. (2018)	Sedentary subjects Family study cohorts	395	Men = 36 ± 14 Women = 34 ± 14	Men = 26.6 ± 4.7 Women = 25.5 ± 5.4	AT	Duration and intensity were increased every 2 weeks	3	55% of baseline VO₂ max/75% of the HR max	30 min baseline/50 min	20-week
Knauer et al. (2023)	Dwelling old men	21	77.8 ± 3.6	25.0 ± 3.0	RT	5 Consecutive phases; all major and most of the minor muscle groups were exercised.	2	65 and 80% 1RM	Nr	18 months
Miyamoto-Mikami et al. (2015)	Healthy, young 16 men, 9 women sedentary or moderately active	25	21 ± 1	21.9 ± 1.7	AT	Cycling	3	60–70% VO₂ peak	55 min	8 weeks
Miyamoto-Mikami et al. (2015)	Healthy middle-aged/older adults 12 men, 16 women sedentary or moderately active	28	67 ± 8	23.1 ± 3.4	AT	Cycling	3	60–70% VO₂ peak	55 min	8 weeks
Park et al. (2003)	Middle-aged, healthy women	10	42.2 ± 1.91	25.3 ± 1.74	AT	A group of dynamic cardio exercises	6	60–70% of HRmax	60 min	24 weeks
Park et al. (2003)	Middle-aged women, healthy	10	43.4 ± 1.04	25.8 ± 1.43	COM	A group of dynamic cardio exercises plus resistant exercises	6 3:3 (AT:RT)	60–70% of HRmax/60%−70% of 1RM	60 min/60 min	24 weeks
Poehlman et al., (2000)	Women	14	29 ± 5	22 ± 2	AT	Jogging and add interval training sessions in second phase	3	90% of HR max/85% HR max	40 min/60 min	28 weeks
Poehlman (2000)	Women	17	28 ± 3	22 ± 2	RT	Full body exercises, 3 set, 10 repetitions	3	Approximately 80% of 1-RM	45 min	28 weeks
Poggi et al. (2024)	Well-trained older men	5	66.3 ± 3.0	23.9 ± 0.3	AT	2,558 km of cycling				16 days totally
Louise Thomas et al. (2000)	Non obese, premenopausal healthy women	17	32.6 ± 1.8	24.5 ± 0.8	AT	Treadmill	3	60–70% of VO₂ max	Minimum of 30-min	6 months
Zhang et al. (2021)	Women	11	20.9 ± 1.7	25.6 ± 2.4	SITall-out	Cycle ergometer	3 4	All-out effort		12 weeks
Zhang et al. (2021)	Women	12	19.7 ± 1.3	26.1 ± 3.2	SIT120	Cycle ergometer ex: 1 min rest: 1.5 min	3 4	120% VO₂ peak	1 min exercise, 1.5 min rest	12 weeks
Zhang et al. (2021)	Women	12	19.7 ± 1.1	26.0 ± 2.9	HIIT	Cycle ergometer ex: 4 min rest: 3 min	3 4	90% VO₂ peak	4 min exercise, 3 min rest	12 weeks
Zhang et al. (2021)	Women	11	21.0 ± 2.4	25.1 ± 3.0	MICT	Cycle ergometer	3 4	60% VO₂ peak		12 weeks

Note. In studies with multiple exercise groups, each intervention arm is presented separately; therefore, some studies appear more than once in the table; P.CH: participants characteristics, N: number of intervention group participants, BMI: Body mass index, Mode: Mode of exercise, Form: Form of exercise, Fre: frequency of exercise, Int: intensity of exercise, Sess du: every session duration, Du: study duration, COM: combination training, RT: resistance training, AT: aerobic training, 1RM: One-repetition maximum, VO₂ max: maximum rate of oxygen consumption, Min: Minute, IGT: impaired glucose tolerance, Nr: not reported, HRR: heart rate reserve, HRmax: maximum heart rate, VO₂ peak: Maximal oxygen uptake, SITall-out: sprint interval training all-out, HIIT: high-intensity interval training, MICT: moderate-intensity continuous training, ex: Exercise time.

Table 2.

Study Design Characteristics

Study	Groups N	Study design	Groups mode	Nutritional intervention	Fat evaluation	Control variable	Region
Benatti et al. (2012)	2	RCT	Trained = 18 Untrained = 18	Maintain their food intake pattern	CT	Age, sex, BMI 30 kg/m² or higher, smoking, CVD, musculoskeletal disturbances, metabolic disorders, unstable weight, previous liposuction surgery, current use of medications	Abdominal
Carr et al. (2005)	2	Single-arm	Stretching = 32 Endurance = 30	Stretching: Iso caloric and consisted of 30% of total calories as fat (10% as saturated fat), 50% as carbohydrate, and 20% as protein, giving <300 mg cholesterol daily Endurance: Isocaloric and consisted of <30% of total calories as fat (<7% saturated fat), 55% as carbohydrate, and the balance as protein, giving 200 mg cholesterol daily.	CT	Age, sex, nation, liver or renal disease, anemia, fasting triglyceride levels >300 mg/dl, a history of significant coronary artery disease, valvular heart disease, uncontrolled hypertension, or conditions that limited their ability to perform the lifestyle modifications, such as arthritis, pulmonary disease, neurological or psychiatric disease, or dietary restrictions.	Abdominal
He et al. (2018)	4	Single-arm	Black men = 77 White men = 193 Black women = 102 White women = 202	Nr	CT	Age, sex, nation,	Between the fourth (L4) and fifth (L5) lumbar vertebrae
Knauer et al. (2023)	2	RCT	Exercise group = 21 Control group = 22	Individual supplementation of whey protein was provided to all participants to generate a total protein intake 1.5–1.6 in the exercise group and 1.2–1.3 g/kg body mass per day in the control group.	MRI	Secondary osteoporosis, a history of hip fracture, (osteo) anabolic and antiresorptive pharmaceutical therapy, glucocorticoid therapy, participation in resistance type exercise, limitations or problems that inhibit intense, and alcohol abuse	Between 12^th (T12) thoracic to mid third lumbar vertebrae (L3)
Miyamoto-Mikami et al. (2015)	4	NRC	Young training = 14 Young control = 11 Middle-aged/older training = 14 Middle-aged/older control = 14	Maintain their usual levels of food intake	MRI	Age, taking medication such as anti-hyperlipidemic, anti-hypertensive, or anti hyperglycemic, history of stroke, diabetes, hypertension, hyperlipidemia, cardiac disease, chronic renal failure, mental disorder	The intervertebral level between the third and fourth lumbar (L3–L4) spaces
Park et al. (2003)	3	NRC	Control = 10 Aerobic training = 10 Combined training = 10	Nr	CT	Age, sex, medication	Abdominal
Poehlman et al. (2000)	3	RCT	Endurance training = 14 Resistance training = 17 Control = 20	Diets prepared by the metabolic kitchen at the general clinical research center containing 55% carbohydrate, 25% fat, and 20% protein.	CT	Age, diabetes, orthopedic limitations or history of pathological fractures, hypertension use of prescription or over the counter medications smoking, alcohol consumption	The L4–L5 vertebrae level
Poggi et al. (2024)	2	NRC	Cycling group = 5 Control group = 8	Nr	DXA	Nr	Nr
Louise Thomas et al. (2000)	1	Single-arm	Intervention = 17	Subjects were given a standard booklet.	MRI	Age, history of obesity, dieting, eating disorders, any form of medication including the contraceptive pill, smoking, a previous history of regular exercise.	Slices from the femoral heads to the slice containing the top of the liver or the base of the lungs (T10).
Zhang et al. (2021)	5	RCT	SITall-out = 11 SIT120 = 12 HIIT90 = 12 MICT = 11 CON = 13	Participants provided the dietary information, including type of food, portion size, and preparation method under the guidance of a dietician.	CT	Age, body fat percentage, constant body weight, physical activity, history of metabolic, hormonal, orthopedic, or cardiovascular diseases, current use of prescribed medication including oral contraception	Umbilicus level, approximately the L4-L5 intervertebral space

Note. Group N: Group number, Region: visceral fat analysis region, RCT: randomized controlled trial, CT: computed tomography, Single-Arm: Single-Arm Intervention Study, Nr: not reported, MRI: magnetic resonance imaging, NRC: Non-Randomized Controlled, DXA: dual-energy X-ray absorptiometry, SITall-out: sprint interval training all-out, HIIT: high-intensity interval training, MICT: moderate-intensity continuous training.

Across studies, mean participant age ranged from 19.7 ± 1.3 to 77.8 ± 3.6 years, with the BMI mean ranged from 23.1 ± 3.4 to 25.5 ± 5.4 kg/m², encompassing normal-weight and lower-overweight individuals. Across studies, BMI variability within cohorts was substantial, indicating that some participants may have exceeded the nominal BMI threshold despite meeting study-level eligibility criteria. Most cohorts involved otherwise healthy but physically inactive adults, with three cohorts including specific clinical or trained populations. Seven studies recruited healthy individuals, while others included participants who had recently undergone a minor liposuction procedure (n = 1; Benatti et al., 2012) individuals with impaired glucose tolerance (n = 1; Carr et al., 2005), and older men with osteo-sarcopenia (n = 1; Knauer et al., 2023).

Exercise Intervention Characteristics

Aerobic training was the most common intervention, typically involving treadmill running, while resistance training frequently utilized weight machines. For aerobic interventions, the majority (8 studies) implemented a frequency of 3 days per week, with one study progressively increasing to 4 (Zhang et al., 2021) days, and one study beginning with 6 days per week (Park et al., 2003). Resistance training was performed 2 (Knauer et al., 2023) and 3 days per week (Poehlman et al., 2000), while combined aerobic and resistance training was conducted 3 days per week (Benatti et al., 2012) or 6 days per week (3 days each for aerobic and resistance training; Park et al., 2003).

The intensity of AT was typically indicated as a percentage of maximum oxygen consumption, Maximal oxygen uptake, heart rate reserve, or maximum heart rate (HRmax), ranging from 55% to 120% with 60–70% HRmax being the most common range. Resistance exercise intensity was based on the one-repetition maximum (1-RM), with loads between 60 – 80% of 1-RM. Two studies employed interval training (IT) modalities, including sprint, supra-maximal, and high-intensity intervals (Zhang et al., 2021), or intervals as part of an endurance training program (Poehlman et al., 2000).

Dietary Intake

In five studies, participants were provided specific diet prescriptions outlining macronutrient and energy intake (Carr et al., 2005; Knauer et al., 2023; Louise Thomas et al., 2000; Poehlman et al., 2000; Zhang et al., 2021; see Table 2). Two other studies advised participants to maintain their usual dietary patterns (Benatti et al., 2012; Miyamoto-Mikami et al., 2015), while the remaining three studies did not report on dietary intake. Dietary control varied substantially across studies and represents a potential source of residual confounding, particularly given the sensitivity of VAT to energy balance and macronutrient intake.

Visceral Adipose Tissue Assessments

Most studies used CT to assess VAT. Three studies used MRI (Knauer et al., 2023; Louise Thomas et al., 2000; Miyamoto-Mikami et al., 2015), and one used DXA (Poggi et al., 2024). The majority of assessments focused on the abdominal region, particularly the lumbar interspace between L4 and L5. Other studies varied in measurement regions, such as L3–L4 (Miyamoto-Mikami et al., 2015), from the femoral heads to the liver or lung bases (T10; Louise Thomas et al., 2000) and T12 to mid-L3 (Knauer et al., 2023). Poggi et al. (2024) did not specify the VAT region assessed.

Significant Changes in Visceral Adipose Tissue

Table 3 illustrates the effect of exercise on VAT across studies. Most studies reported VAT in cm²; however, four studies used other units: cm³ (Knauer et al., 2023; Park et al., 2003), liters (Louise Thomas et al., 2000), or percentage of total body fat (Poggi et al., 2024). All studies, except He et al. (2018), reported post-intervention VAT values. Eight studies found significant reductions in VAT following intervention (Benatti et al., 2012; Carr et al., 2005; Knauer et al., 2023; Louise Thomas et al., 2000; Miyamoto-Mikami et al., 2015; Park et al., 2003; Poggi et al., 2024; Zhang et al., 2021). In Carr et al.'s study, only the endurance group exhibited significant reductions in VAT (Carr et al., 2005). AT was the primary mode of exercise associated with VAT reduction, though two studies utilized combined training (COM; Benatti et al., 2012; Park et al., 2003). One study found no significant differences in VAT from baseline to post-intervention, in protocols involving RT (Poehlman et al., 2000). Carr et al. also found no effect of stretching exercise on VAT (Carr et al., 2005).

Table 3.

Study Results

Study	Pre, mean (± SD)	Post, mean (± SD)	Change score	p
Benatti et al. (2012)	T: 42.9 ± 10.2 cm² C: 43.1 ± 14.9) cm²	T:38.1 (9.1) cm² C: 47.2 (14.2) cm²	−9.15 (−18.1 to −0.2) cm²	0.04
Carr et al. (2005)	Stretching: 112.3 ± 9.9 cm² AT: 87.2 ± 8.4 cm²	Stretching: 112.7 ± 10.4 cm² AT: 75.0 ± 7.9 cm²	−1.6 ± 6.9 cm² −10.6 ± 3.5 cm²	Not sig sig
He et al. (2018)	Men = 107.8 ± 64.2 cm² Women = 71.4 ± 51.8 cm²	Nr	−6.6 ± 17.2 cm² −1.7 ± 19.9 cm²	p < 0.0001
Knauer et al. (2023)	T = 1,067 ± 440 cm³ C = 1,095 ± 405 cm³	Nr	−82.4 ± 20.5 cm³ −14.7 ±19.5 cm³	p < 0.001
Miyamoto-Mikami et al. (2015)	Young T = 29.3 ± 8.4 cm² Young C = 26.0 ± 7.8 cm² Middle-aged/older T = 96.7 ± 66.8 cm² Middle-aged/older C = 73.1 ± 46.0 cm²	Young T = 26.9 ± 9.0 cm² Young C = 26.5 ± 7.4 cm² Middle-aged/older T = 81.5 ± 51.4 cm² Middle-aged/older C = 77.3 ± 50.2 cm²	−2.5 ± 7.2 cm² 0.5 ± 5.4 cm² −15.2 ± 22.2 cm² 4.2 ± 16.2 cm²	< 0.05 < 0.05
Park et al. (2003)	C = 182.9 ± 16.8 cm³ AT = 195.0 ± 12.55 cm³ COM = 201.6 ± 28.03 cm³	C = 190.4 ± 15.74 cm³ AT = 112.4 ± 10.50 cm³ COM = 108.6 ± 17.85 cm³	7.5 cm³ −82.6 cm³ −93.0 cm³	p < 0.01 p < 0.01
Poehlman et al. (2000)	AT = 40 ± 11 cm² RT = 36 ± 17 cm² C = 36 ± 13 cm²	AT = 41 ± 13 cm² RT = 36 ± 13 cm² C = 41 ± 15 cm²	Nr	p > 0.05 p > 0.05
Poggi et al. (2024)	21.4 ± 4.3%	20.8 ± 2.7%	Nr	Sig
Louise Thomas et al. (2000)	1.7 ± 0.2 L	1.3 ± 0.2 L	−25.0 %	p < 0.001
Zhang et al. (2021)	SITall-out = 79.3 ± 24.9 cm² SIT120 = 75.0 ±2 9.1 cm² HIIT90 = 78.2 ± 24.2 cm² MICT = 70.7 ± 29.6 cm² CON = 74.5 ± 18.3 cm²	SITall-out = 59.9 ± 24.5 cm² SIT120 = 57.1 ± 18.9 cm² HIIT90 = 62.9 ± 16.6 cm² MICT = 67.2 ± 28.2 cm² CON = 73.9 ± 19.2 cm²	−19.3 ± 12.7 cm² −17.9 ± 18.9 cm² −15.3 ± 14.1 cm² −3.5 ± 13.1 cm² −0.7 ± 7.5 cm²	p < .01 p < .01 p < .01

Note. T: training, C: control, AT: aerobic training, sig: significant, COM: combination training, RT: resistance training, Nr: not reported, SITall-out: sprint interval training all-out, HIIT: high-intensity interval training, MICT: moderate-intensity continuous training; p-values indicate statistical significance for either between-group comparisons (intervention vs. control) or pre–post changes in single-arm studies without a control group.

Risk of Bias

Ten studies were analyzed for risk of bias, with results presented in Table 4. The methodological quality of the included studies was variable. Out of the total studies, one study was rated as low risk of bias (Knauer et al., 2023), 4 as some concern (Benatti et al., 2012; Carr et al., 2005; Poehlman et al., 2000; Zhang et al., 2021), and 5 as high risk. Despite these concerns, the majority of studies were judged to be of acceptable quality for inclusion in the meta-analysis.

Table 4.

Risk-Of-Bias Assessment Using the ROB2 Tool

Study	Randomization process	Deviations from intended interventions	Missing outcome data	Measurement of outcome	Selection of reported result	Overall, ROB-2
Benatti et al. (2012)	Low	Some concerns	Low	Some concerns	Low	Some concerns
Carr et al. (2005)	Some concerns	Some concerns	Some concerns	Some concerns	Low	Some concerns
He et al. (2018)	High	Some concerns	Some concerns	Some concerns	Low	High
Knauer et al. (2023)	Low	Low	Low	Low	Low	Low
Miyamoto-Mikami et al. (2015)	High	Some concerns	Some concerns	Some concerns	Low	High
Park et al. (2003)	High	Some concerns	Some concerns	Some concerns	Low	High
Poehlman et al. (2000)	Some concerns	Some concerns	Some concerns	Some concerns	Low	Some concerns
Poggi et al. (2024)	High	Some concerns	Low	Some concerns	Low	High
Louise Thomas et al. (2000)	High	Some concerns	Low	Some concerns	Low	High
Zhang et al., (2021)	Some concerns	Some concerns	Low	Some concerns	Low	Some concerns

Meta-Analysis

The meta-analysis included 10 studies. Although 10 studies were analyzed, the meta-analysis included 18 data entries due to multiple study groups within some studies, which involved different interventions.

Between-Group Analysis

A meta-analysis of 13 controlled comparisons (n = 354) showed that exercise was associated with a significant reduction in visceral adipose tissue (SMD = −1.590, 95% CI: −2.283 to −0.898, p < 0.001) (Figure 2(a)). Substantial heterogeneity was observed (I² = 86.4%) (Table 5).

Figure 2.

Forest plot; (a) between-group subgroup analysis; showing the weighted mean differences (Weight %) and standardized mean differences (95% confidence limits) between exercise and control groups, (b) within-group subgroup analysis; presenting the pre-to-post intervention changes in VAT within exercise-only groups; SMD: Standardized Mean Difference

Table 5.

Meta-Analyses Results: Effect of Exercises on VAT for Between Groups Analysis

Overall effect	N	ES	SMD (95%CI)	p-value	Heterogeneity
Overall effect	N	ES	SMD (95%CI)	p-value	p	I²
	13	−1.590	−2.283, −0.898	<0.001	<0.001	86.4%

Note. N: number of study intervention, ES: effect size, SMD: Standardized Mean Difference, CI: Confidence Interval.

Within-Group Analysis

A within-group meta-analysis of five studies was conducted to evaluate changes in visceral adipose tissue following exercise interventions. The pooled analysis showed a reduction in visceral adipose tissue; however, the effect did not reach statistical significance (SMD = −1.247, 95% CI: −2.531 to 0.037, p = 0.057). Moderate heterogeneity was observed across studies (I² = 54.8%, p = 0.065), indicating variability in effect sizes (Figure 2(b)).

Sensitivity Analysis

Between-Group Analysis

Sensitivity analyses were conducted to evaluate the robustness of the pooled estimates. Leave-one-out influence analysis demonstrated that the overall effect size remained statistically significant regardless of the study omitted, with pooled SMDs ranging from −1.291 (when excluding Park et al., 2003) to −1.730 (when excluding Zhang MICT 2020). These findings indicate that no single study disproportionately altered the direction or statistical significance of the overall effect. Notably, the exclusion of studies with comparatively large effect sizes, particularly Park et al. (2003) and Knauer et al. (2023), resulted in a reduction in the magnitude of the pooled estimate, while the overall direction and statistical significance remained unchanged.

Additional sensitivity analyses excluding studies involving atypical or clinical populations (Benatti et al., 2012; Knauer et al., 2023; Poggi et al., 2024) yielded a pooled effect size of SMD = −1.563 (95% CI: −2.342 to −0.784) for between-group comparisons, which remained consistent in both direction and statistical significance. In contrast, the corresponding within-group (Carr et al., 2005) analysis (SMD = −1.250, 95% CI: −3.107 to 0.607) showed greater uncertainty, with confidence intervals crossing zero.

Leave-one-out sensitivity analysis for r = 0.5 demonstrated that the pooled effect remained stable (SMD range: −1.38 to −1.85), indicating that no single study substantially influenced the overall result. The pooled effect was statistically significant (SMD = −1.70, 95% CI: −2.41 to −0.99, p < 0.001), although substantial heterogeneity was observed (I² = 86.5%). In addition, influence analysis for r = 0.9 demonstrated that the pooled effect size remained negative across all leave-one-out iterations (SMD range: −0.97 to −1.35), indicating that no single study substantially influenced the overall estimate. The pooled effect remained statistically significant (SMD = −1.24, 95% CI: −1.87 to −0.60, p < 0.001), although substantial heterogeneity was observed (I² = 85.4%).

Overall, these analyses suggest that the observed reduction in visceral adipose tissue is robust across analytical assumptions, although partially influenced by a small number of high-impact studies.

Publication Bias

Between-Group Analysis

Publication bias was assessed using visual inspection of the funnel plot and Egger’s regression test. The funnel plot demonstrated asymmetry, with several studies showing large effect sizes and relatively high standard errors (Figure 3). Egger’s test indicated significant small-study effects (p = 0.001), suggesting the presence of potential publication bias or related biases (Table 6). These findings should be interpreted with caution, as asymmetry in funnel plots may also arise from substantial between-study heterogeneity, differences in study design, or the presence of outlier studies. Publication bias and sensitivity analyses were not performed for within-group estimates due to the limited number of included studies, which restricts the reliability of such assessments.

Figure 3.

Funnel plot; effects of exercise interventions on VAT

Table 6.

Egger’s Test

Std_Eff	Coefficient	Std. err	t	p > \|t\|	95% conf. interval
Slope	1.738,313	0.6,852,086	2.54	0.028	0.2,301,787 3.246,447
Bias	−6.673,917	1.535,939	−4.35	0.001	−10.0545 −3.293,338

Note. p = 0.001, Test of H₀ (no small-study effects), Number of Studies: 19. Std_Eff: Standardized Effect; Std. Err: Standard Error; t: t-statistic; 95% Conf. Interval: 95% Confidence Interval.

Certainty of Evidence

The certainty of evidence for the effect of exercise on VAT was evaluated using GRADE principles based on the pooled between-group analysis. As randomized controlled trials constituted the majority of included studies, the body of evidence was initially considered high certainty; however, several important limitations led to downgrading. The certainty was downgraded for risk of bias, as multiple studies were assessed as having high risk or some concerns using ROB2, and some analyses incorporated non-randomized or single-arm designs. A further downgrade was applied for inconsistency, given the substantial heterogeneity observed across studies (I² = 86.4%), indicating considerable variability in effect estimates. The evidence was also downgraded for imprecision, due to relatively small sample sizes in several studies and wide confidence intervals in subgroup analyses. Publication bias was considered serious, as visual inspection of the funnel plot suggested asymmetry and Egger’s regression test indicated small-study effects (p = 0.001). Indirectness was not considered serious, as the population, intervention, comparator, and outcome were directly aligned with the review question. Overall, the certainty of evidence was rated as very low, indicating that the true effect may be substantially different from the estimated effect and that further high-quality randomized controlled trials are required (Table 7).

Table 7.

GRADE Assessment; Overall Effect of Exercise on VAT

Domain	Judgment	Justification
Risk of bias	Serious (−1)	Several studies high risk or some concerns; inclusion of non-randomized/single-arm designs
Inconsistency	Serious (−1)	High heterogeneity (I² = 86.4%) across studies
Indirectness	Not serious	Population, intervention, and outcome directly relevant
Imprecision	Serious (−1)	Modest sample sizes and wide CIs in subgroups
Publication bias	Serious (−1)	Funnel asymmetry and Egger’s test significant (p = 0.001)
Overall certainty	Very low	Downgraded across multiple domains

Subgroup Analysis

Mode of Exercise

Between-Group Analysis

Subgroup analyses were conducted according to exercise modality, including AT, RT, COM, and IT. A statistically significant reduction in visceral adipose tissue was observed in the AT subgroup (SMD = −1.081, 95% CI: −2.064 to −0.099, p = 0.031) and the IT subgroup (SMD = −1.422, 95% CI: −1.939 to −0.905, p < 0.001). In contrast, the COM and RT subgroups did not reach statistical significance. Substantial heterogeneity was observed within the COM (I² = 95.1%), RT (I² = 95.8%), and AT (I² = 84.3%) subgroups, whereas heterogeneity was negligible in the IT subgroup (I² = 0.0%). The test for subgroup differences was not statistically significant (Q = 1.30, p = 0.730), indicating that the observed differences in effect size across exercise modalities should be interpreted with caution (Figure 4(a)).

Figure 4.

Subgroup forest plot for (a) Exercise mode, (b) Duration; (c) BMI (d) Age; showing weighted mean differences (Weight %) and SMD (95% confidence limits); SMD: Standardized Mean Difference, COM: combination training, RT: resistance training, AT: aerobic training, INT: interval training; DU: duration of study, x > 5: more than 5 months, x ≤ 5: less than 5 months, BMI: Body mass index, NW: normal-weight, LBO: lower boundary of overweight, Y: young, O: old, M: middle-aged

Duration

Between-Group Analysis

Subgroup analyses were conducted based on intervention duration (≤5 months vs. > 5 months). A statistically significant reduction in visceral adipose tissue was observed in both subgroups. Interventions with a duration of ≤5 months demonstrated a moderate effect size (SMD = −0.941, 95% CI: −1.298 to −0.583, p < 0.001), with relatively low heterogeneity (I² = 29.3%). In contrast, interventions lasting >5 months showed a substantially larger pooled effect (SMD = −3.659, 95% CI: −5.740 to −1.578, p = 0.001), but with considerable heterogeneity (I² = 94.7%). The test for subgroup differences was statistically significant (Q = 6.36, p = 0.012), indicating that intervention duration may act as a moderator of the effect of exercise on visceral adipose tissue (Figure 4(b)). The 5 months threshold was selected a priori as it corresponded to the median duration of the included interventions, providing a balanced basis for subgroup comparison.

BMI

Between-Group Analysis

Subgroup analyses were performed based on body mass index (BMI) categories, including normal weight (NW) and lower-bound overweight (LBO). A statistically significant reduction in visceral adipose tissue was observed in both subgroups. In the NW subgroup, the pooled effect size was SMD = −0.676 (95% CI: −0.987 to −0.365, p < 0.001), with no observed heterogeneity (I² = 0.0%). In the LBO subgroup, the pooled effect size was SMD = −2.921 (95% CI: −4.319 to −1.523, p < 0.001), with substantial heterogeneity (I² = 91.1%). The test for subgroup differences was statistically significant (Q = 9.44, p = 0.002), indicating a difference in effect sizes between BMI categories (Figure 4(c)).

Age

Between-Group Analysis

Subgroup analyses were conducted based on age categories (young [Y], older [O], and middle-aged [M]). A statistically significant reduction in visceral adipose tissue was observed in the Y subgroup (SMD = −0.849, 95% CI: −1.201 to −0.497, p < 0.001) and the M subgroup (SMD = −8.249, 95% CI: −11.097 to −5.402, p < 0.001). In contrast, the O subgroup did not reach statistical significance (SMD = −1.569, 95% CI: −3.342 to 0.203, p = 0.083). Moderate heterogeneity was observed in the Y subgroup (I² = 37.9%) and the M subgroup (I² = 47.8%), whereas substantial heterogeneity was present in the O subgroup (I² = 90.5%). The test for subgroup differences was statistically significant (Q = 25.99, p < 0.001), indicating differences in effect sizes across age categories (Figure 4(d)).

Subgroup analyses were exploratory and should be interpreted with caution due to the limited number of studies within some subgroups.

Discussion

This systematic review and meta-analysis evaluated the effects of different exercise modalities on VAT in adults with predominantly normal BMI and those at the lower boundary of the overweight range (BMI ≤27 kg/m²), based on data from 10 studies encompassing 18 cohorts and 472 participants. It should be noted that the included studies did not consistently report VAT thresholds corresponding to established metabolic risk levels, and individual participant data were not available. Therefore, it was not possible to determine whether participants met specific criteria for VAT-defined metabolic risk. The findings should thus be interpreted as reflecting changes in VAT within generally normal or near-normal BMI populations, rather than effects in clinically defined high-risk groups.

The relevance of these findings is particularly evident for individuals exhibiting the TOFI phenotype, also referred to as normal-weight obesity, who present with relatively normal BMI yet disproportionately elevated VAT and heightened cardiometabolic risk (Zdrojewicz et al., 2017). In addition, individuals near the lower boundary of the overweight range may underestimate their health risk due to weight-status misperception, potentially delaying preventive behaviors (Robinson, 2017).

The inclusion of both controlled and single-arm studies introduced methodological heterogeneity. To address this, analyses were conducted separately for between-group (controlled studies) and within-group estimates. While this approach allows for a broader synthesis of the available evidence, within-group analyses are inherently more susceptible to bias due to the absence of a comparator and may overestimate intervention effects. In the present analysis, the direction of effect was generally consistent across study designs; however, the magnitude and statistical robustness differed, with within-group estimates showing greater uncertainty.

While most prior research on VAT reduction has focused on populations with overweight or obesity, our findings extend existing evidence by demonstrating that VAT remains responsive to exercise even in individuals without overt obesity. This observation is consistent with broader evidence linking physical activity to reductions in VAT and metabolic disease risk across BMI categories (Powell-Wiley et al., 2021).

The present meta-analysis indicates that exercise interventions are associated with reductions in VAT in adults with normal BMI and those at the lower boundary of the overweight range. The between-group analysis demonstrated a statistically significant effect favoring exercise; however, the magnitude of this effect should be interpreted with caution due to substantial heterogeneity and evidence of small-study effects. Sensitivity analyses confirmed that the direction of the effect remained consistent, although the pooled estimate was partially influenced by a small number of studies reporting comparatively large effects.

In contrast, the within-group analysis showed a similar direction of effect but did not reach statistical significance and was characterized by greater uncertainty. This discrepancy highlights the methodological limitations of uncontrolled designs, which are more susceptible to bias and may overestimate intervention effects. Accordingly, greater weight should be placed on the between-group findings derived from controlled comparisons.

Intervention duration appeared to play a role in observed outcomes. Although established guidelines suggest that health-related benefits of exercise may emerge after approximately three months (Pescatello et al., 2021), the present findings indicate that longer intervention periods may be required to elicit detectable changes in VAT in normal-weight populations. This observation contrasts with findings in populations with obesity (Merlotti et al., 2017), suggesting that VAT responsiveness to exercise may differ across BMI.

With respect to exercise modality, AT demonstrated the most consistent association with VAT reduction. This is in line with previous meta-analyses (Ismail et al., 2012; Verheggen et al., 2016), although some discrepancies exist, potentially due to differences in exercise dose, intensity, and adherence. IT was also associated with a statistically significant reduction in VAT in the present analysis. However, this finding was based on a limited number of studies, largely derived from a single research group, which may limit the generalizability of the result. These findings should be interpreted in the context of prior evidence. For example, Maillard et al. (Maillard et al., 2018) reported significant reductions in visceral fat following high-intensity IT in individuals with overweight or obesity. The partial inconsistency with the present findings may reflect differences in baseline body composition, suggesting that responsiveness to interval-based training may be attenuated in normal-weight populations. RT and COM training did not demonstrate statistically significant effects. Overall, differences across exercise modalities may reflect variability in exercise dose, intensity, adherence, and participant characteristics, as well as underlying physiological differences related to adiposity status.

Subgroup analyses further indicated variability across BMI and age categories, although these findings should be interpreted cautiously due to the limited number of studies within certain subgroups and the potential for residual confounding. The observed differences may reflect underlying heterogeneity in study design, intervention characteristics, or participant profiles rather than true effect modification.

Dietary intake represents an important potential confounder. Only a subset of included studies implemented controlled dietary protocols, while others allowed habitual intake or did not adequately report dietary adherence. Given the sensitivity of VAT to energy balance and macronutrient composition, the absence of standardized dietary control limits the ability to isolate the independent effects of exercise.

An additional consideration is the predominance of female participants across the included studies, which may influence the generalizability of the findings. Sex-related differences in fat distribution are well established, with men typically exhibiting greater VAT accumulation, whereas women tend to have higher subcutaneous fat storage. These differences are partly mediated by sex hormones, particularly estrogen, which is associated with a more favorable fat distribution profile (Karastergiou et al., 2012). Exercise-induced reductions in VAT may also differ by sex, potentially due to variations in lipolytic activity (Blaak, 2001), hormonal responses, and baseline adiposity patterns. Although some evidence suggests that both men and women benefit from exercise interventions (Hands et al., 2016), the magnitude and mechanisms of VAT reduction may not be identical. Given the limited representation of male participants in the included studies, the present findings may be more reflective of responses in women, and caution is warranted when extrapolating results to male populations. Future studies should aim to include more balanced samples or conduct sex-specific analyses to better understand potential differences in responsiveness.

Age-related differences in visceral adipose tissue should also be considered when interpreting the findings. VAT is known to increase with advancing age, even in the absence of substantial changes in body mass index, with distinct patterns observed between men and women. These changes are thought to reflect alterations in hormonal milieu, fat distribution, and metabolic regulation across the lifespan. Accordingly, variation in age across the included studies may have influenced baseline VAT levels and responsiveness to exercise interventions (Swainson et al., 2020).

Several limitations should be acknowledged. First, substantial heterogeneity was observed across analyses, likely reflecting variability in study design, populations, intervention characteristics, and measurement approaches. Second, Dietary intake represents a potential confounding factor. Only a subset of the included studies implemented structured dietary control, while others allowed habitual intake or did not report dietary adherence. Given the sensitivity of visceral adipose tissue to energy balance and macronutrient composition, the absence of standardized dietary control across studies may have influenced the observed effects. Third, eligibility was determined at the study level based on reported mean BMI values, as individual participant data were not available, which may have resulted in the inclusion of participants outside the intended BMI range. Fourth, some standard deviations of change were estimated using assumed correlations, which introduces additional uncertainty. Fifth, the relatively small number of studies and the presence of small-study effects suggest that the magnitude of the pooled effect may be overestimated. Finally, although different imaging modalities (CT, MRI, DXA) were used to assess VAT, the application of standardized mean differences mitigates concerns related to scale heterogeneity but does not eliminate potential measurement variability.

Overall, while the findings support an association between exercise and reductions in visceral adipose tissue in individuals with normal or near-normal BMI, the certainty of evidence remains low. Further well-designed randomized controlled trials with standardized protocols, adequate sample sizes, and controlled dietary conditions are required to confirm these findings and clarify the relative effects of different exercise modalities and intervention durations.

Conclusions

This meta-analysis indicates that exercise is associated with reductions in VAT in adults with normal BMI and those at the lower boundary of the overweight range. Among exercise modalities, AT demonstrated the most consistent effects, while IT also showed significant reductions, albeit based on a limited number of studies. Although the pooled findings were statistically significant and directionally consistent, substantial heterogeneity and methodological limitations, including variability in study design and the inclusion of within-group analyses, may influence the magnitude of the observed effects. Accordingly, the overall estimate should be interpreted with caution, and greater weight should be placed on evidence derived from controlled comparisons. The certainty of evidence was rated as low due to risks of bias, imprecision, and potential small-study effects.

From a practical perspective, AT may represent a valuable and potentially underutilized strategy for managing visceral adiposity in individuals who may otherwise be overlooked due to normal body weight. However, further well-designed randomized controlled trials with larger sample sizes, standardized protocols, and appropriate dietary control are required to confirm these findings and to better define the roles of RT, COM, and IT modalities, as well as the interaction between exercise and dietary factors in optimizing VAT reduction.

Supplemental Material

Supplemental Material - Impact of Different Exercise Modalities on Visceral Adipose Tissue in Normal Weight and Low-Overweight Individuals: A Systematic Review and Meta-Analysis

Supplemental Material for Impact of Different Exercise Modalities on Visceral Adipose Tissue in Normal Weight and Low-Overweight Individuals: A Systematic Review and Meta-Analysis by Hamid Arazi, PhD, Fatemeh Bagherpour, Grant M. Tinsley in Biological Research For Nursing.

Footnotes

Hamid Arazi

Fatemeh Bagherpour

Grant M. Tinsley

Author Contributions

Arazi, H contributed to conception and design; contributed to acquisition, analysis, and interpretation; drafted manuscript; critically revised manuscript; gave final approval; and agrees to be accountable for all aspects of work ensuring integrity and accuracy. Bagherpour, F contributed to conception; contributed to acquisition, analysis, and interpretation; drafted manuscript; gave final approval; and agrees to be accountable for all aspects of work ensuring integrity and accuracy. Tinsley, G.M contributed to conception; contributed to analysis and interpretation; critically revised manuscript; gave final approval; and agrees to be accountable for all aspects of work ensuring integrity and accuracy.

Funding

The authors disclosed that this research was conducted independently of external funding.

Declaration of Conflicting Interests

The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Data Availability Statement

Further data supporting the findings of this study are available from the corresponding author upon reasonable request.*

Supplemental Material

Supplemental material for this article is available online. Supplementary material including searched keywords at the different databases is appended to the manuscript and can be accessed.

References

Bellicha

van Baak

M. A.

Battista

Beaulieu

Blundell

J. E.

Busetto

Carraça

E. V.

Dicker

Encantado

Ermolao

Farpour‐Lambert

Pramono

Woodward

Oppert

(2021). Effect of exercise training on weight loss, body composition changes, and weight maintenance in adults with overweight or obesity: An overview of 12 systematic reviews and 149 studies. Obesity Reviews, 22(S4), Article e13256. https://doi.org/10.1111/obr.13256

Benatti

Solis

Artioli

Montag

Painelli

Saito

Baptista

Costa

L. A.

Neves

Seelaender

Ferriolli

Pfrimer

Lima

Roschel

Gualano

Lancha

Jr. (2012). Liposuction induces a compensatory increase of visceral fat which is effectively counteracted by physical activity: A randomized trial. Journal of Clinical Endocrinology & Metabolism, 97(7), 2388–2395. https://doi.org/10.1210/jc.2012-1012

Bennett

J. P.

Prado

C. M.

Heymsfield

S. B.

Shepherd

J. A.

(2024). Evaluation of visceral adipose tissue thresholds for elevated metabolic syndrome risk across diverse populations: A systematic review. Obesity Reviews, 25(9), Article e13767. https://doi.org/10.1111/obr.13767

Blaak

(2001). Gender differences in fat metabolism. Current Opinion in Clinical Nutrition and Metabolic Care, 4(6), 499–502. https://doi.org/10.1097/00075197-200111000-00006

Carr

D. B.

Utzschneider

K. M.

Boyko

E. J.

Asberry

P. J.

Hull

R. L.

Kodama

Callahan

H. S.

Matthys

C. C.

Leonetti

D. L.

Schwartz

R. S.

Kahn

S. E.

Fujimoto

W. Y.

(2005). A reduced-fat diet and aerobic exercise in Japanese Americans with impaired glucose tolerance decreases intra-abdominal fat and improves insulin sensitivity but not β-Cell function. Diabetes, 54(2), 340–347. https://doi.org/10.2337/diabetes.54.2.340

Chan

Huang

Vardhanabhuti

(2023). Towards visceral fat estimation at population scale: correlation of visceral adipose tissue assessment using three-dimensional cross-sectional imaging with BIA, DXA, and single-slice CT. Frontiers in Endocrinology, 14, Article 1211696. https://doi.org/10.3389/fendo.2023.1211696

Chandler

Cumpston

Page

M. J.

Welch

(2019). Cochrane handbook for systematic reviews of interventions (Vol. 4, 1002, p. 14651858). Wiley.

Chen

Xie

Zhang

Cao

(2024). Effects of various exercise types on visceral adipose tissue in individuals with overweight and obesity: A systematic review and network meta-analysis of 84 randomized controlled trials. Obesity Reviews, 25(3), Article e13666. https://doi.org/10.1111/obr.13666

Cheung

A. S.

de Rooy

Hoermann

Gianatti

E. J.

Hamilton

E. J.

Roff

Zajac

J. D.

Grossmann

(2016). Correlation of visceral adipose tissue measured by Lunar Prodigy dual X-ray absorptiometry with MRI and CT in older men. International Journal of Obesity, 40(8), 1325–1328. https://doi.org/10.1038/ijo.2016.50

10.

Cohen

(2013). Statistical power analysis for the behavioral sciences. Routledge. https://doi.org/10.4324/9780203771587

11.

Conus

Rabasa-Lhoret

Péronnet

(2007). Characteristics of metabolically obese normal-weight (MONW) subjects. Applied Physiology Nutrition and Metabolism, 32(1), 4–12. https://doi.org/10.1139/h06-092

12.

Crocker

T. F.

Lam

Jordão

Brundle

Prescott

Forster

Ensor

Gladman

Clegg

(2023). Risk-of-bias assessment using Cochrane’s revised tool for randomized trials (RoB 2) was useful but challenging and resource-intensive: Observations from a systematic review. Journal of Clinical Epidemiology, 161, 39–45. https://doi.org/10.1016/j.jclinepi.2023.06.015

13.

De Lorenzo

Martinoli

Vaia

Di Renzo

(2006). Normal weight obese (NWO) women: An evaluation of a candidate new syndrome. Nutrition, Metabolism, and Cardiovascular Diseases, 16(8), 513–523. https://doi.org/10.1016/j.numecd.2005.10.010

14.

Després

J. P.

(2006). Is visceral obesity the cause of the metabolic syndrome? Annals of Medicine, 38(1), 52–63. https://doi.org/10.1080/07853890500383895

15.

Gonzalez-Ruiz

Ramirez-Velez

Correa-Bautista

J. E.

Peterson

M. D.

Garcia-Hermoso

(2017). The effects of exercise on abdominal fat and liver enzymes in pediatric obesity: A systematic review and meta-analysis. Childhood Obesity, 13(4), 272–282. https://doi.org/10.1089/chi.2017.0027

16.

Graffy

P. M.

Pickhardt

P. J.

(2016). Quantification of hepatic and visceral fat by CT and MR imaging: Relevance to the obesity epidemic, metabolic syndrome and NAFLD. British Journal of Radiology, 89(1062), Article 20151024. https://doi.org/10.1259/bjr.20151024

17.

Guyatt

G. H.

Alonso-Coello

Schünemann

H. J.

Djulbegovic

Nothacker

Lange

Murad

M. H.

Akl

E. A.

(2016). Guideline panels should seldom make good practice statements: Guidance from the GRADE working group. Journal of Clinical Epidemiology, 80, 3–7. https://doi.org/10.1016/j.jclinepi.2016.07.006

18.

Guyatt

G. H.

Oxman

A. D.

Schünemann

H. J.

Tugwell

Knottnerus

(2011). GRADE guidelines: A new series of articles in the journal of clinical epidemiology. Journal of Clinical Epidemiology, 64(4), 380–382. https://doi.org/10.1016/j.jclinepi.2010.09.011

19.

Hands

B. P.

Parker

Larkin

Cantell

Rose

(2016). Male and female differences in health benefits derived from physical activity: Implications for exercise prescription. Journal of Women’s Health, Issues and Care, 5(4), Article 1000238. https://doi.org/10.4172/2325-9795.1000238

20.

Harrison

Amirabdollahian

Davies

(2016). Nutritional status, dietary intake and adiposity of normal-weight individuals with clustered metabolic risk factors in the UK population. Proceedings of the Nutrition Society, 75(OCE3), E237. https://doi.org/10.1017/S0029665116002524

21.

Rankinen

Leon

A. S.

Skinner

J. S.

Tchernof

Bouchard

(2018). Plasma steroids, body composition, and fat distribution: Effects of age, sex, and exercise training. International Journal of Obesity, 42(7), 1366–1377. https://doi.org/10.1038/s41366-018-0033-1

22.

Higgins

Thomas

Chandler

Cumpston

(2024). In Page

M. J.

Welch

V. A.

(Eds.), Cochrane handbook for systematic reviews of interventions version 6.5 (updated August 2024). Cochrane.

23.

Ismail

Keating

Baker

Johnson

(2012). A systematic review and meta‐analysis of the effect of aerobic vs. resistance exercise training on visceral fat. Obesity Reviews, 13(1), 68–91. https://doi.org/10.1111/j.1467-789X.2011.00931.x

24.

Karastergiou

Smith

S. R.

Greenberg

A. S.

Fried

S. K.

(2012). Sex differences in human adipose tissues–the biology of pear shape. Biology of Sex Differences, 3(1), 13. https://doi.org/10.1186/2042-6410-3-13

25.

Katzmarzyk

P. T.

Greenway

F. L.

Heymsfield

S. B.

Bouchard

(2013). Clinical utility and reproducibility of visceral adipose tissue measurements derived from dual-energy X-ray absorptiometry in white and African American adults. Obesity, 21(11), 2221–2224. https://doi.org/10.1002/oby.20519

26.

Kaur

(2014). A comprehensive review on metabolic syndrome. Cardiology Research and Practice, 2014(1), Article 943162. https://doi.org/10.1155/2014/943162

27.

Khalafi

Malandish

Rosenkranz

S. K.

Ravasi

A. A.

(2021). Effect of resistance training with and without caloric restriction on visceral fat: A systemic review and meta‐analysis. Obesity Reviews, 22(9), Article e13275. https://doi.org/10.1111/obr.13275

28.

Knauer

Chaudry

Uder

Kohl

Kemmler

Bickelhaupt

Engelke

(2023). Effects of high-intensity resistance training on visceral adipose tissue and abdominal aortic calcifications in older men with osteosarcopenia–results from the FrOST study. Clinical Interventions in Aging, 18, 71–80. https://doi.org/10.2147/CIA.S388026

29.

Kuk

J. L.

Lee

Heymsfield

S. B.

Ross

(2005). Waist circumference and abdominal adipose tissue distribution: Influence of age and sex. American Journal of Clinical Nutrition, 81(6), 1330–1334. https://doi.org/10.1093/ajcn/81.6.1330

30.

Lefebvre

Glanville

Briscoe

Featherstone

Littlewood

(2024). In Higgins

J. P. T.

Thomas

Chandler

Cumpston

Page

M. J.

Welch

V. A.

(Eds.), Cochrane Handbook for Systematic Reviews of Interventions version 6.5. Available from. https://www.training.cochrane.org/handbook

31.

Lin

Yan

Rao

Xia

Zhou

Rothney

M. P.

Xia

Wacker

W. K.

Ergun

D. L.

Zeng

Gao

(2013). Quantification of visceral adipose tissue using lunar dual-energy X-ray absorptiometry in Asian Chinese. Obesity, 21(10), 2112–2117. https://doi.org/10.1002/oby.20325

32.

Louise Thomas

Brynes

A. E.

McCarthy

Goldstone

A. P.

Hajnal

J. V.

Saeed

Frost

Bell

J. D.

(2000). Preferential loss of visceral fat following aerobic exercise, measured by magnetic resonance imaging. Lipids, 35(7), 769–776. https://doi.org/10.1007/s11745-000-0584-0

33.

Maillard

Pereira

Boisseau

(2018). Effect of high-intensity interval training on total, abdominal and visceral fat mass: A meta-analysis. Sports Medicine, 48(2), 269–288. https://doi.org/10.1007/s40279-017-0807-y

34.

Merlotti

Ceriani

Morabito

Pontiroli

(2017). Subcutaneous fat loss is greater than visceral fat loss with diet and exercise, weight-loss promoting drugs and bariatric surgery: A critical review and meta-analysis. International Journal of Obesity, 41(5), 672–682. https://doi.org/10.1038/ijo.2017.31

35.

Micklesfield

L. K.

Goedecke

J. H.

Punyanitya

Wilson

K. E.

Kelly

T. L.

(2012). Dual-energy X-Ray performs as well as clinical computed tomography for the measurement of visceral fat. Obesity, 20(5), 1109–1114. https://doi.org/10.1038/oby.2011.367

36.

Miyamoto-Mikami

Sato

Kurihara

Hasegawa

Fujie

Fujita

Sanada

Hamaoka

Tabata

Iemitsu

(2015). Endurance training-induced increase in circulating irisin levels is associated with reduction of abdominal visceral fat in middle-aged and older adults. PLoS One, 10(3), Article e0120354. https://doi.org/10.1371/journal.pone.0120354

37.

Neeland

I. J.

Grundy

S. M.

Adams-Huet

Vega

G. L.

(2016). Comparison of visceral fat mass measurement by dual-X-ray absorptiometry and magnetic resonance imaging in a multiethnic cohort: The Dallas heart study. Nutrition & Diabetes, 6(7), e221. https://doi.org/10.1038/nutd.2016.28

38.

Nicklas

B. J.

Penninx

B. W.

Ryan

A. S.

Berman

D. M.

Lynch

N. A.

Dennis

K. E.

(2003). Visceral adipose tissue cutoffs associated with metabolic risk factors for coronary heart disease in women. Diabetes Care, 26(5), 1413–1420. https://doi.org/10.2337/diacare.26.5.1413

39.

Oldham

Robinson

(2018). Visual body size norms and the under‐detection of overweight and obesity. Obesity Science & Practice, 4(1), 29–40. https://doi.org/10.1002/osp4.143

40.

Park

S.-K.

Park

J.-H.

Kwon

Y.-C.

Kim

H.-S.

Yoon

M.-S.

Park

H.-T.

(2003). The effect of combined aerobic and resistance exercise training on abdominal fat in obese middle-aged women. Journal of Physiological Anthropology and Applied Human Science, 22(3), 129–135. https://doi.org/10.2114/jpa.22.129

41.

Pescatello

L. S.

Panza

G. A.

Zaleski

Guidry

(2021). Development of a novel clinical decision support system for exercise prescription among patients with multiple cardiovascular disease risk factors. Mayo Clinic Proceedings: Innovations, Quality & Outcomes, 5(1), 193–203. https://doi.org/10.1016/j.mayocpiqo.2020.08.005

42.

Poehlman

E. T.

Dvorak

R. V.

DeNino

W. F.

Brochu

Ades

P. A.

(2000). Effects of resistance training and endurance training on insulin sensitivity in nonobese, young women: A controlled randomized trial. Journal of Clinical Endocrinology & Metabolism, 85(7), 2463–2468. https://doi.org/10.1210/jcem.85.7.6692

43.

Poggi

A. I.

Andreasen

M. W.

Sahl

R. E.

Hansen

M. T.

Rømer

Frandsen

Linden

F. H.

Høgh‐Schmidt

Boesen

M. P.

Ostrowski

S. R.

Larsen

Dela

Helge

J. W.

(2024). Effect of repeated prolonged exercise on liver fat content and visceral adipose tissue in well‐trained older men. Scandinavian Journal of Medicine & Science in Sports, 34(4), Article e14612. https://doi.org/10.1111/sms.14612

44.

Powell-Wiley

T. M.

Poirier

Burke

L. E.

Després

J.-P.

Gordon-Larsen

Lavie

C. J.

Lear

S. A.

Ndumele

C. E.

Neeland

I. J.

Sanders

St-Onge

M. P.

(2021). Obesity and cardiovascular disease: A scientific statement from the American Heart Association. Circulation, 143(21), e984–e1010. https://doi.org/10.1161/CIR.0000000000000973

45.

Robinson

(2017). Overweight but unseen: A review of the underestimation of weight status and a visual normalization theory. Obesity Reviews, 18(10), 1200–1209. https://doi.org/10.1111/obr.12570

46.

Swainson

M. G.

Batterham

A. M.

Hind

(2020). Age- and sex-specific reference intervals for visceral fat mass in adults. International Journal of Obesity, 44(2), 289–296. https://doi.org/10.1038/s41366-019-0393-1

47.

Van Der Beek

E. M.

Abrahamse-Berkeveld

Oosting

Alles

M. S.

(2017). Metabolic imprinting effects of specifically designed lipid component. Google Patents.

48.

Verheggen

Maessen

Green

D. J.

Hermus

Hopman

Thijssen

(2016). A systematic review and meta‐analysis on the effects of exercise training versus hypocaloric diet: Distinct effects on body weight and visceral adipose tissue. Obesity Reviews, 17(8), 664–690. https://doi.org/10.1111/obr.12406

49.

Vissers

Hens

Taeymans

Baeyens

J.-P.

Poortmans

Van Gaal

(2013). The effect of exercise on visceral adipose tissue in overweight adults: A systematic review and meta-analysis. PLoS One, 8(2), Article e56415. https://doi.org/10.1371/journal.pone.0056415

50.

Zdrojewicz

Popowicz

Szyca

Michalik

Śmieszniak

(2017). TOFI phenotype–its effect on the occurrence of diabetes. Pediatric Endocrinology, Diabetes and Metabolism, 23(2), 96–100. https://doi.org/10.18544/pedm-23.02.0079

51.

Zhang

Tong

T. K.

Kong

Shi

Liu

Nie

(2021). Exercise training‐induced visceral fat loss in obese women: The role of training intensity and modality. Scandinavian Journal of Medicine & Science in Sports, 31(1), 30–43. https://doi.org/10.1111/sms.13803

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.06 MB

0.00 MB