Potential link between excess added sugar intake and ectopic fat: a systematic review of randomized controlled trials

Abstract

Context: The effect of added sugar intake on ectopic fat accumulation is a subject of debate. Objective: A systematic review and meta-analysis of randomized controlled trials (RCTs) was conducted to examine the potential effect of added sugar intake on ectopic fat depots. Data Sources: MEDLINE, CAB Abstracts, CAB Global Health, and EBM (Evidence-Based Medicine) Reviews – Cochrane Central Register of Controlled Trials databases were searched for studies published from 1973 to September 2014. Data Extraction: RCTs with a minimum of 6 days’ duration of added sugar exposure in the intervention group were selected. The dosage of added sugar intake as a percentage of total energy was extracted or calculated. Means and standard deviations of pre- and post-test measurements or changes in ectopic fat depots were collected. Data Synthesis: Fourteen RCTs were included. Most of the studies had a medium to high risk of bias. Meta-analysis showed that, compared with eucaloric controls, subjects who consumed added sugar under hypercaloric conditions likely increased ectopic fat, particularly in the liver (pooled standardized mean difference = 0.9 [95%CI, 0.6–1.2], n = 6) and muscles (pooled SMD = 0.6 [95%CI, 0.2–1.0], n = 4). No significant difference was observed in liver fat, visceral adipose tissue, or muscle fat when isocaloric intakes of different sources of added sugars were compared. Conclusions: Data from a limited number of RCTs suggest that excess added sugar intake under hypercaloric diet conditions likely increases ectopic fat depots, particularly in the liver and in muscle fat. There are insufficient data to compare the effect of different sources of added sugars on ectopic fat deposition or to compare intake of added sugar with intakes of other macronutrients. Future well-designed RCTs with sufficient power and duration are needed to address the role of sugars on ectopic fat deposition.

INTRODUCTION

Ectopic fat depots represent accumulation of fat in tissues or organs that typically do not accommodate fat, including the liver, abdominal cavity (visceral adipose tissue, or VAT), muscle, kidney, pancreas, heart, and blood vessels.¹ Emerging evidence suggests that ectopic fat accumulation is associated with increased cardiometabolic risk.² Ectopic fat depots may be classified on the basis of their postulated association with cardiometabolic risk: liver, VAT, and muscle depots are associated with systematic effects, whereas kidney, pancreas, heart, and blood vessel depots are associated with local effects.² Imaging techniques such as ultrasonography, computed tomography, magnetic resonance imaging, or magnetic resonance spectroscopy are the preferred approaches for accurately quantifying ectopic fat deposition in research studies.²

Several narrative reviews have proposed that excess intake of fructose induces fatty liver disease; however, much of the evidence is based on data from animal studies.^3–6 A recent meta-analysis of controlled feeding trials in human adults concluded that fructose increased liver fat only when consumed under hypercaloric conditions.⁷ Fructose is a major component of added sugars such as sucrose (50% fructose) and high-fructose corn syrup (42%–55% fructose), and added sugar consumption has been linked to greater cardiometabolic risk.⁸ Therefore, it is critical to study not only isolated fructose but also commonly consumed added sugars, i.e., sugars and syrups that are added to foods or beverages when they are processed or prepared.⁹ The purpose of this systematic review was to examine (1) whether added sugar intake affects fat accumulation in ectopic fat depots and (2) whether fat accumulation in ectopic fat depots, if any, differs by type of added sugars or when compared with the effects of other components of a diet.

METHODS

The present systematic review was conducted according to the methods outlined in the Cochrane Handbook for Systematic Reviews of Interventions¹⁰ and employed the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) for the presentation of results (see Table S1 in Supporting Information online).^11,12

Search strategy

A literature search was performed through the OvidSP platform by searching the following databases: MEDLINE, CAB Abstracts, CAB Global Health, and EBM (Evidence-Based Medicine) Reviews – Cochrane Central Register of Controlled Trials from 1973 to September 2014. An “exploded” version of relevant Medical Subject Heading (MeSH) terms and other relevant non-MeSH keywords was used. Details of search terms for intervention treatment, which included “sweetening agents,” “fructose,” “sucrose,” “sugar-sweetened beverages,” and other relevant keywords, are listed in Table S2 in the Supporting Information online. Search terms for ectopic fat depots included “fatty liver,” “intra-abdominal fat,” “intramuscular fat,” “myocardial fat,” “pericardial fat,” “perivascular fat,” “renal sinus fat,” “pancreatic fat,” and other relevant keywords. Reference lists in relevant studies were also searched for additional studies. Two investigators (J.M. and M.K.) independently screened abstracts. A full-text review for the accepted abstracts was conducted by at least 1 investigator. Discrepancies on eligibility for inclusion were resolved by discussions among investigators.

Selection criteria

The PICOS (Participants, Intervention, Comparators, Outcomes, Study Design) criteria are presented in Table 1. Only studies that were conducted in adults and children older than 4 years of age and published in English were included. During abstract and full-text screening, both observational studies and clinical trials were included. However, the present systematic review only synthesized original, peer-reviewed, and parallel- and crossover-designed randomized controlled trials (RCTs) with a minimum of 6 days’ duration of added sugar exposure in the intervention group. Only studies in which added sugars were consumed together with foods or beverages were considered. As shown in the Table S3 in the Supporting Information online, commonly consumed sugars have similar chemical composition, i.e., 50% fructose and 50% glucose. Thus, added sugars of interest in this systematic review were essentially fructose and fructose-containing sugars such as sucrose and high-fructose corn syrup. The present systematic review required the quantification of at least 1 site of ectopic fat deposition through the use of biopsy or imaging techniques (ultrasonography, computed tomography, magnetic resonance imaging, or spectroscopy). Hepatic steatosis related to alcohol consumption, medication use, or specific diseases such as hepatitis C or Wilson’s disease was excluded.

Table 1.

PICOS criteria for inclusion and exclusion of studies

Parameter	Inclusion criteria	Exclusion criteria
Participants	Adults and children older than 4 y of age
Intervention	Commonly consumed added sugars as presented in Table S2 in Supporting Information online. Added sugar consumed together with foods or beverages	Less than 6 d of added sugar exposure in the intervention group
Comparators	Comparison of a diet with increased or reduced added sugar vs a weight-maintaining diet (usual sugar intake). Comparison of different sources or types of added sugars. Comparison of added sugars and other dietary components
Outcomes	Fat deposition in ectopic depots (liver, visceral adipose tissue, muscle, kidney, pancreas, heart, and blood vessels). Ectopic fat deposition was quantified through the use of biopsy or imaging techniques (ultrasonography, computed tomography, and magnetic resonance imaging or spectroscopy)	Hepatic steatosis related to alcohol consumption, medication use, and specific diseases such as hepatitis C and Wilson’s disease
Study design	Randomized controlled trials (parallel or crossover design)

Data extraction and methodological quality assessment

The following information was extracted from each study selected: study characteristics (e.g., authors, publication year, study location, and funding sources); characteristics of the study sample (e.g., sample size, attrition, gender, age, and body mass index); dietary intake, including dietary assessment methods used and estimates of sugar intake; intervention strategy (e.g., randomization, dosage, intervention duration, and analysis strategy); and potential confounders and effect modifiers. The risk of biases (selection, performance, detection, attrition, reporting, and other biases) was assessed using the Cochrane risk-of-bias assessment tool.¹⁰ As shown in Figure S1 in the Supporting Information online, 10 criteria were evaluated. The risk of bias was rated as high, medium, or low for each criterion.

The dosage of added sugar intake as a percentage of total energy was extracted or calculated from all included RCTs. In most studies, dietary exposure was expressed as the percentage of energy from added sugar intake in treatment groups. In 2 studies,^13,14 the authors provided only the absolute intake in grams of sugar and, thus, intake was converted to percentage of energy intake using the estimated energy intake in the Dietary Reference Intake for the corresponding age groups.⁹ Means and standard deviations of pre- and post-test measurements or changes in ectopic fat depots were collected. When data were presented in figures, a Java program (Plot Digitizer, v.2.6.4) was used to convert the scanned figures to numerical data.

Data synthesis

Studies were synthesized by age groups (children vs adults) and by study design (comparisons between eucaloric diets and either high-sugar hypercaloric diets or low-sugar hypocaloric diets, between various sources or types of sugars, and between sugar and other dietary components). The terms hypercaloric, hypocaloric, and eucaloric, respectively, were used to represent whether the dietary regimen was designed to be positive, negative, or neutral in energy balance compared with baseline. The term isocaloric was also used when comparing 2 test dietary regimens that provided the same amount of calories. Data on levels of sugar intake, measures of ectopic fat deposition, and body weight were presented.

Meta-analysis

A meta-analysis was conducted when at least 2 studies were identified with the same exposure and outcome of interest. Because various techniques and scales were used to measure ectopic fat deposition, the standardized mean difference (SMD) (Hedges’s) was calculated as the effect size in the meta-analysis according to the methods described in the Cochrane handbook.¹⁰ For randomized crossover trials, a correlation coefficient of 0.5 was used to impute the standard error of the mean difference in ectopic fat deposition between intervention and control groups if it was not reported in the original manuscript. Sensitivity analysis was conducted using correlation coefficients of 0.2 and 0.8. For all meta-analyses, DerSimonian and Laird random-effects models were used to combine study effect sizes.¹⁵ Heterogeneity was assessed using the Cochran Q statistic (P < 0.1 was considered significant) and the I² index.¹⁰ Publication bias was assessed using Egger’s regression test. All analyses were conducted using Stata 13 software (Stata Corp., College Station, Texas, USA). A P value of <0.05 was considered statistically significant.

RESULTS

The initial literature search identified 6314 abstracts, of which 6273 were discarded during initial abstract screening (Figure 1). Of the 41 potentially relevant full-text articles, 27 were excluded on the basis of study eligibility criteria, including 4 single-arm trials,^16–19 5 case–control studies,^20–24 and 4 cross-sectional studies.^25–28 No prospective observational studies were identified. Two RCTs^29,30 comparing the potential effect of sucrose with that of isomaltulose (a disaccharide similar to sucrose but with a lower hydrolysis rate) on VAT fat deposition was excluded since isomaltulose is rarely consumed as added sugar anywhere in the world.

Figure 1.

Flow diagram of search and selection process

A total of 14 RCTs were included in this systematic review. Some studies examined the effect of added sugar on multiple ectopic depots. Liver fat, VAT, and muscle fat were outcomes reported in 13, 4, and 7 studies, respectively. The present systematic review found no RCTs examining the effects of added sugar intake on several local fat depots, such as myocardial, pericardial, perivascular, pancreatic, and renal sinus fat. Assessment of methodological quality showed more than half of included studies had medium to high risk of selection, performance, detection, and other biases (see Figure S1 in the Supporting Information online).

Effect of increased or reduced added sugar intake on fat depots

Table 2 ^13,31–36 summarizes the results from 7 randomized controlled intervention trials: 6 in adults^13,31–35 and 1 in children.³⁶ The 6 studies in adults were designed to compare sugar intake as part of a hypercaloric diet with intake of a eucaloric control diet that contained no excess sugars. As such, these studies were confounded by energy intake, and therefore one cannot separate out the effects of energy and sugar. In the studies of Koopman et al.³¹ and Maersk et al.,¹³ surplus energy was provided from the daily consumption of 1 L of sucrose-sweetened cola. In 4 trials, fructose or glucose or both were added to a weight-maintaining diet to create a hypercaloric diet that was then compared with a weight-maintaining diet alone.^32–35 In these 4 trials, fructose or glucose or both provided calories ≥35% of the energy requirement.

Table 2.

RCTs designed to compare high-sugar hypercaloric diets (or low-sugar hypocaloric diets) with eucaloric control diets in adults and children^a

Reference	Country	Study design	Participants	Intervention	Duration (weeks)	Energy intake (designed)	Results				Comments
							BW	Liver fat	VAT	Muscle fatb
Adults
Koopman et al. (2014)31	The Netherlands	Parallel	n = 8, 100% men; BMI 22.6 kg/m2; age, 22 y	Sucrose	6	Hypercaloric (ad libitum diet supplemented with SSB at1 L/d that accounted for 40% of energy)	Sucrose interventions increased BMI by 2.2% and 3.6%, respectively. Change in controls was not reported	Sucrose interventions increased liver fat by 108% (P = 0.07) and 16% (P = 0.92), respectively	NR	NR	Sucrose was given between meals
			n = 7, 100% men; BMI, 21.7 kg/m2; age, 22 y	Sucrose	6	Hypercaloric (ad libitum diet supplemented with SSB at1 L/d that accounted for 40% of energy)					Sucrose was given as part of meals
				Control	6	Eucaloric (ad libitum WM diet)
Lecoultre et al. (2014)32	Switzerland	Crossover	n = 10, 100% men; BMI 22.6 kg/m2; age, 23 y	Fructose	≈1	Hypercaloric (WM diet supplemented with fructose that accounted for 43% of energy)	BW was increased by 0.4% after fructose intervention and decreased by 1.3% after control diet. No significant difference between study arms (P < 0.05)	Fructose intervention increased liver fat significantly, by 85%, compared with control diet (P < 0.001)	NR	Fructose intervention did not significantly alter muscle fat (P > 0.05)
				Control	2
Maersk et al. (2012)13	Denmark	Parallel	n = 10, 60% men; BMI 31.3 kg/m2; age, 39 y	Sucrose-sweetened cola	26	Hypercaloric (ad libitum diet supplemented with SSB at1 L/d that accounted for ≈20% of energyc)	BW was increased 1.3% in SSB intervention; 0.1% in diet soda intervention; and 0.6% in water intervention. No significant difference between study arms (P = 0.8)	Liver fat was increased 134% in SSB intervention; decreased 4% in diet soda intervention; and increased 2% in water intervention. Compared with diet cola and water interventions, SSB intervention significantly increased liver fat accumulation (both P < 0.05)	VAT was increased 24% in SSB intervention; increased 1% in diet soda intervention; and decreased 1% in water intervention. No significant difference between study arms (P = 0.14 and P = 0.1, respectively)	Muscle fat was increased 198% in SSB intervention; decreased 31% in diet soda intervention; and increased 80% in water intervention. No significant difference between study arms (P = 0.08 and P > 0.05, respectively)	Change in ectopic fat depots observed may be due to small absolute increase in BW
			n = 12, 25% men; BMI 32.8 kg/m2; age, 39 y	Diet cola	26	Eucaloric (ad libitum diet supplemented with diet soda)
			n = 13, 38% men; BMI 32.2 kg/m2; age, 39 y	Water	26	Eucaloric (ad libitum diet supplemented with water)
Theytaz et al. (2012)33	Switzerland	Crossover	n = 9, 100% men; BMI 22.6 kg/m2; age, 23 y	Fructose	≈1	Hypercaloric (WM diet supplemented with fructose that accounted for 35% of energy)	Postintervention BW was 71 kg in the 2 fructose arms and 70 kg in controls. No significant difference between study arms (P > 0.05)	Fructose interventions increased liver fat 116% and 81% compared with control diet (both P < 0.05), respectively	NR	NR	Fructose intervention exceeded 3 times the mean intake in US diet58
				Fructose + essential amino acids	≈1
				Control	≈1	Eucaloric (WM diet)
Ngo Sock et al. (2010)34	Switzerland	Crossover	n = 11, 100% men; BMI 19–25 kg/m2; age, 25 y	Fructose	1	Hypercaloric (WM diets supplemented with either fructose or glucose that accounted for 35% of energy)	Compared with control diet, fructose intervention increased BW by 0.6 kg (P < 0.01), and glucose intervention increased BW by 1 kg (P < 0.05)	Compared with control diet, fructose intervention increased liver fat 52% (P < 0.05), and glucose intervention increased liver fat 58% (P = 0.06)	NR	Compared with control diet, fructose intervention increased muscle fat 49% (P > 0.05), and glucose intervention increased muscle fat 84% (P < 0.05)
				Glucose	1
				Control	1
Le et al. (2009)35	Switzerland	Crossover	n = 24 (n = 16 offspring of type 2 DM patients, n = 8 normal healthy adults), 100% men; BMI 19–25 kg/m2; age, 25 y	Fructose	1	Hypercaloric (WM diet supplemented with fructose that accounted for 35% of energy)	Compared with control diet, fructose intervention significantly increased BW (P < 0.05)	Compared with control diet, fructose intervention increased liver fat ≈50% (P < 0.05)	NR	Compared with control diet, fructose intervention increased muscle fat ≈20% (P < 0.05)
				Control	1	Eucaloric diet (WM diet)
Children
Hasson et al. (2012)36	USA	Parallel	n = 30; obese; age, 14–18 y	Reduced energy intake from added sugar	16	Hypocaloric diet (10% reduction in energy intake from added sugar)	No significant difference in BW change between study arms	No significant difference between study arms (pre–post test change was <1% in both arms)	No significant difference between study arms	NR	Added sugar intake was reduced by 26% in nutrition intervention and increased by 1% in controls. Absolute intake of fat content NR
			n = 39; obese; age, 14–18 y	Control	16	Eucaloric diet (no dietary intervention)

Abbreviations: BMI, body mass index; BW, body weight; DM, diabetes mellitus; NR, not reported; SSB, sugar-sweetened beverage; VAT, visceral adipose tissue; WM diet, weight-maintaining diet.

^aHypercaloric, hypocaloric, or eucaloric reflects energy balance when interventions were compared with baseline.

^bTibialis anterior muscle.

^cEnergy from SSB was estimated based on dietary reference intake of similar age group.

In adults, liver fat was measured in 6 studies,^13,31–35 VAT in 1 study,¹³ and muscle fat in 2 studies.^33,35 Meta-analyses were performed to examine the pooled effect of high-sugar hypercaloric diets on liver and muscle fat deposition compared with the effect of a eucaloric control diet. As shown in Figure 2A, the pooled analyses demonstrate that high-sugar, hypercaloric diets increased liver fat compared with eucaloric control diets (pooled SMD = 0.93; 95% confidence interval [95%CI], 0.64–1.21). Figure 2B shows a significant pooled effect of high-sugar hypercaloric diets on muscle fat accumulation (pooled SMD = 0.63; 95%CI, 0.23–1.04). In addition, the significant pooled effect remained in both meta-analyses when only studies that examined isolated fructose were analyzed. In the study that examined fat accumulation in VAT, Maersk et al.¹³ showed that sucrose-sweetened cola tended to increase VAT compared with diet soda and water interventions, although this observed effect was not statistically significant.

Figure 2.

Effects of high-sugar hypercaloric diets vs eucaloric diets with no excess added sugars on fat accumulation in (A) liver and (B) lower-extremity muscle. Data were presented as standardized mean differences (SMD) with 95% confidence intervals (95%CI). Meta-analyses were conducted using DerSimonian and Laird’s random-effects models. Sucrose was consumed in between meals in Koopman et al.³¹* and with meals in Koopman et al.³¹**. Participants were offspring of type 2 diabetic patients in Le et al.³⁵* and were normal individuals in Le et al.³⁵**. (A) overall effect, z = 6.37; P < 0.001. (B) overall effect, z = 3.04; P = 0.002. In the Maersk et al.¹³ study, comparisons between sucrose-sweetened cola and diet soda were used in meta-analysis; however, results were the same as when sucrose-sweetened cola was compared with water.

The study by Hasson et al.³⁶ examined the potential effect of reducing added sugar intake on ectopic fat depots in the context of a hypocaloric diet in children. This study had 1 control and 2 intervention arms (nutrition education and nutrition education plus strength training). The nutrition education program was designed to reduce energy intake by decreasing added sugar intake and increasing dietary fiber intake, whereas the controls largely maintained the baseline diet. At the end of this 16-week study, the self-reported added sugar intake was reduced by 26% of energy in the nutrition education intervention group and was increased by 1% in the controls. No changes in liver fat or VAT depots were observed after the nutrition education intervention.

Comparison of the effects of different sources or types of added sugars on fat depots

In Table 3,^{14,34,37–41} data were extracted from 7 RCTs – 6 in adults and 1 in children – that compared the potential effects of various types of sugars. In the 6 studies in adults, liver fat was examined in 5,^{34,37,38,40,41} VAT in 2,^40,41 and muscle fat in 4.^34,37,38,40 Five of the 6 studies compared high-fructose diets with high-glucose diets,^34,38–41 and 1 study³⁷ compared sucrose with high-fructose corn syrup.

Table 3.

Comparisons of the potential effects of various types of sugars on ectopic fat deposition

Reference	Country	Study design	Participants	Intervention	Duration (weeks)	Energy intake (designed)a	Results				Comments
							BW	Liver fat	VAT	Muscle fatb
Adults
Bravo et al. (2013)37	USA	Parallel	n = 8, 50% men; BMI, 27.1 kg/m2; age: 39 y	HFCS (55% fructose)	10	Eucaloric (WM diet with HFCS replacing 8% of energy)	Compared with baseline, BW was increased 2 kg and 1 kg in 30% HFCS and 30% sucrose arms, respectively (both P < 0.05). No significant changes in other groups	Compared with baseline, changes in liver fat were 1.7%, 2.4%, 1.4%, −3.1%, −1.9%, and −0.7% in 8% HFCS, 18% HFCS, 30% HFCS, 8% sucrose, 18% sucrose, and 30% sucrose interventions, respectively. Overall between-group difference was not significant (P for interaction = 0.21)	NR	No significant difference between groups in vastus lateralis muscle (P for interaction = 0.2). No significant difference between groups in gluteus maximus muscle (P for interaction = 0.6)	Sugar was consumed in reduced-fat (1%) milk
			n = 13, 77% men; BMI, 28.4 kg/m2; age 34 y	Sucrose	10	Eucaloric (WM diet with sucrose replacing 8% of energy)
			n = 12, 83% men, BMI: 27.5 kg/m2; age. 37 y	HFCS (55% fructose)	10	Eucaloric (WM diet with HFCS replacing 18% of energy)
			n = 10, 30% men; BMI, 26 kg/m2; age. 42 y	Sucrose	10	Eucaloric (WM diet with sucrose replacing 18% of energy)
			n = 11, 36% men; BMI, 28.6 kg/m2; age, 44 y	HFCS (55% fructose)	10	Eucaloric (WM diet with HFCS replacing 30% of energy)
			n = 10, 50% men; BMI, 25.6 kg/m2; age, 37 y	Sucrose	10	Eucaloric (WM diet with sucrose replacing 30% of energy)
Johnston et al. (2013)38	UK	Parallel	n = 15, 100% men; BMI, 30 kg/m2; age, 35 y	Fructose	2	Eucaloric (fructose or glucose replacing 25% of energy of ad libitum diet)	No change in either arm compared with baseline. No differences between groups	No change in either intervention compared with baseline. No difference between study arms	NR	No change in muscle fat in either arm compared with baseline. No differences between groups	In the eucaloric phase, energy from fat was likely replaced by sugars
			n = 17, 100% men; BMI, 28.9 kg/m2; age, 33 y	Glucose	2
Johnston et al. (2013)38	UK	Parallel	n = 15, 100% men; BMI, 30 kg/m2; age, 35 y	Fructose	2	Hypercaloric (ad libitum diet supplemented with fructose or glucose that accounted for 25% of energy)	Both arms increased BW (≈1 kg). No difference between study arms	Compared with baseline, liver fat was increased by ≈14% in fructose intervention and ≈24% in glucose invention (both P < 0.05). No significant difference between groups	NR	No change in soleus muscle fat in either intervention compared with baseline. No difference between study arms
			n = 17, 100% men; BMI, 28.9 kg/m2; age, 33 y	Glucose	2
Lecoultre et al. (2013)39	Switzerland	Parallel	n = 17, 100% men; BMI, 22.4 kg/m2; age, 23 y	Fructose	≈1	Hypercaloric (WM diet supplemented with either fructose or glucose that accounted for ≈35% of energy)	Both arms increased BMI by 1%. No significant difference between groups	Liver fat was increased by 213% and 129% after fructose and glucose interventions, respectively. No difference between groups (P > 0.05)	NR	NR
			n = 11, 100% men; BMI, 22.4 kg/m2; age, 23 y	Glucose	≈1
Silbernagel et al. (2011)40	Germany	Parallel	n = 10, 70% men; BMI, 25.5 kg/m2; age, 33 y (n = 8 had imaging data)	Fructose	4	Hypercaloric (WM diet supplemented with either fructose or glucose that accounted for 22% of energy)	No significant pre–post test change in BW in fructose intervention (0.2 kg, P = 0.4). Glucose intervention increased BW by 1.7 kg (P = 0.001). No statistical difference between study arms (P = 0.06)	Increase of liver fat was 0.45% and 0.52% in fructose and glucose interventions, respectively. No difference between groups (P = 0.98)	Fructose intervention increased VAT by 0.07 kg, while glucose intervention increased VAT by 0.07 kg. No difference between groups (P = 0.98)	Tibialis anterior muscle fat increased by 0.97 ± 0.61 (arbitrary units) with fructose and decreased by 0.26 ± 0.54 with glucose. No difference between groups (P = 0.17)
			n = 10, 50% men; BMI, 26.2 kg/m2; age, 28 y (n = 9 had muscle fat measures)	Glucose	4
Stanhope et al. (2009)41	USA	Parallel	n = 17, 53% men; BMI, 29.3 kg/m2; age, 53 y	Fructose	10	Hypercaloric (ad libitum diet with 25% of energy replaced by either fructose or glucose for 2 wk, and followed by 8 wk of an ad libitum diet supplemented with either fructose or glucose that accounted for 25% of energy)	Compared with baseline, BW was increased 1.4 kg in fructose intervention and 1.8 kg in glucose intervention. No difference between study arms (P = 0.47)	NR	VAT was increased 14% (P < 0.01) in fructose intervention and 3.2% (P > 0.05) in glucose intervention. No statistical difference between groups, P = 0.06 (a marginally significant fructose effect was observed in men, P = 0.048)	NR
			n = 15, 47% men; BMI, 29.4 kg/m2; age, 55 y	Glucose	10
Ngo Sock et al. (2010)34	Switzerland	Crossover	n = 11, 100% men; BMI, 19−25 kg/m2; age, 25 y	Fructose	1	Hypercaloric (WM diets supplemented with either fructose or glucose that accounted for 35% of energy)	BW was increased 0.6 kg in fructose intervention and 1 kg in glucose intervention. Difference between groups NR	Liver fat increased by 52% (P < 0.05) in fructose intervention and increased by 58% (P = 0.06) in glucose intervention. Difference between groups was not significant	NR	Tibialis anterior muscle fat increased by 49% (P > 0.05) in fructose intervention and increased by 84% (P < 0.05) in glucose intervention	Authors concluded that muscle fat was increased more with glucose than with fructose
				Glucose	1
Children
Jin et al. (2014)14	USA	Parallel	n = 9, 33% boys; BMI z−score, 2.25; age, 14 y	Fructose	4	Eucaloric diet (ad libitum diet with 3 servings of fructose- or glucose-sweetened drinks, ≈20% energyc)	BW was barely changed in both arms. No differences between groups	Hepatic fat was barely changed in both arms. No differences between groups	NR	NR	Overweight children with habitual intake of 3 servings of SSB per day
			n = 12, 67% boys; BMI z−score, 2.15; age, 13 y	Glucose	4

Abbreviations: BW, body weight; HFCS, high-fructose corn syrup; NR, not reported; SSB, sugar-sweetened beverages; WM diet, weight-maintaining diet.

^aHypercaloric or eucaloric reflects energy balance of interventions compared with baseline.

^bMuscle fat was measured in various locations in lower extremity.

^cEnergy from SSB was estimated on the basis of dietary reference intake of similar age group.

Bravo et al.³⁷ examined the effect of 3 levels of both sucrose and high-fructose corn syrup, accounting for 8%, 18%, and 30% of the energy requirement for maintaining body weight, on ectopic fat depositions in a 10-week parallel-designed RCT. In this study, participants were instructed to maintain their usual energy intake by replacing usual foods consumed with low-fat (1%) foods containing test sugars. In the meta-analysis, the contribution of high-fructose corn syrup was compared with the contribution of sucrose at the same energy intake, e.g., high-fructose corn syrup (30% energy) vs sucrose (30% energy). The pooled analysis (Figure 3A) showed no statistical difference between high-fructose corn syrup and sucrose for fat accumulation in liver (pooled SMD = 0.33; 95%CI, −0.17 to 0.83) or muscle (pooled SMD = 0.02; 95%CI, −0.48 to 0.52 in gluteus maximus, and pooled SMD = −0.14; 95%CI, −0.79 to 0.50 in vastus lateralis).

Figure 3.

Comparisons of potential effects on ectopic fat accumulation (A) between sucrose and HFCS (high-fructose corn syrup) and (B) between fructose and glucose. Data were presented as standardized mean differences (SMD) with 95% confidence intervals (95%CI). Meta-analyses were conducted using DerSimonian and Laird’s random-effects models. Energy intakes from each sugar and diet tested were the same in each comparison. (A) (1) in liver, z = 1.31, P = 0.19; (2) in vastus lateralis muscle, z = 0.43, P = 0.67; and (3) in gluteus maximus muscle, z = 0.08, P = 0.94. (B) (1) in liver fat, z = 0.56, P = 0.57; (2) in visceral adipose tissue, z = 1.18, P = 0.24; and (3) in muscle, z = 0.89, P = 0.37. All comparisons between HFCS and sucrose were extracted from the same trial. Two high-sugar eucaloric diets* and 2 high-sugar hypercaloric diets** were compared in the Johnston et al. study³⁸.

Johnston et al.,³⁸ Lecoultre et al.,³⁹ Silbernagel et al.,⁴⁰ and Ngo Sock et al.³⁴ compared fructose with glucose at contributions of 22%–35% of energy intake. In these 4 studies,^34,38–40 high fructose and glucose contributed to a hypercaloric diet. Johnston et al.³⁸ also examined the effect of fructose compared with glucose as part of a eucaloric diet. As shown in Figure 3B, meta-analysis showed no significant differential effects between high fructose and high glucose intake on fat accumulation in either liver (pooled SMD = 0.1; 95%CI, −0.2 to 0.4) or muscle (pooled SMD = −0.2; 95%CI, −0.7 to 0.3). Exclusion of the Johnston et al.³⁸ study in which fructose and glucose were part of a eucaloric diet did not change the observations.

Two studies were identified that compared the effects of high fructose and high glucose on VAT accumulation.^40,41 Similar levels of sugar intake were reported in both studies (22% and 25% energy). The pooled analysis (Figure 3B) showed no significant difference between fructose and glucose (pooled SMD = 0.3; 95%CI, −0.2 to 0.9). However, the Stanhope et al.⁴¹ study showed that, in VAT, fructose tended to promote greater fat accumulation than glucose after a 10-week intervention, and this difference reached statistical significance in men.

Comparison between the effects of added sugars and other dietary components on ectopic fat depots

As shown in Table 4,^13,39,42 the present systematic review identified 2 studies that compared the potential effects of sugar intake vs fat intake on the accumulation of liver fat.^39,42 Both studies showed that a high-fructose hypercaloric diet and a high-fat hypercaloric diet increased the accumulation of liver fat, but the difference between the 2 dietary regimens was not statistically significant. It should be noted that saturated fat was the major component in high-fat regimens in both studies.^39,42 Maersk et al.¹³ compared the effect of sucrose-sweetened cola intake with the effect of isocaloric consumption of milk. This study found that, compared with milk, sucrose-sweetened cola significantly increased liver fat and VAT. Intake of sucrose-sweetened cola also led to a greater increase in accumulation of muscle fat than intake of milk, although the observed difference was marginal and not statistically significant.

Table 4.

RCTs examining potential differences between the effects of sugars and other dietary components

Reference	Country	Study design	Participants	Intervention	Duration (weeks)	Energy intake (designed)a	Results
							BW	Liver fat	VAT	Muscle fat
Maersk et al. (2012)13	Denmark	Parallel	n = 10, 60% men; BMI 31.3 kg/m2; age, 39 y	Sucrose-sweetened cola	26	Hypercaloric (ad libitum diet supplemented with either cola or milk, 1 L/d, ≈20% of energyb)	BW increased 1.3 kg in SSB group and 1.4 kg in milk intervention; no difference between groups (P = 0.8)	Liver fat increased 134% in SSB group and decreased 9% after milk intervention; difference between groups was significant (P < 0.05)	VAT increased 24% in SSB group and decreased 8% in milk group; difference between groups was significant (P < 0.05)	Tibialis anterior muscle fat increased 198% in SSB group and decreased 25% in milk group; difference between groups was not significant (P = 0.09)
			n = 12, 25% men; BMI 31.9 kg/m2; age, 38 y	Milk	26
Lecoultre et al. (2013)39	Switzerland	Parallel	n = 17, 100% men; BMI 22.4 kg/m2; age, 23 y	Fructose	≈1	Hypercaloric diet: WM diet supplemented with fructose (32% of energy), glucose (37% of energy), or saturated fat (30% of energy)	BW barely changed after all 3 treatments; no differences between groups	Liver fat increased 213%, 159%, and 190% in fructose, glucose, and fat interventions, respectively. No significant difference between groups	NR	NR
			n = 11, 100% men; BMI 22.4 kg/m2; age, 23 y	Glucose	≈1
			n = 10, 100% men; BMI 22.4 kg/m2; age, 23 y	Fat	≈1
Sobrecases et al. (2010)42	Switzerland	Parallel	n = 12, 100% men; BMI 22.6 kg/m2; age, 24 y	Fructose	1	Hypercaloric diet: WM diet supplemented with fructose (35% of energy), fat (30% of energy), or fructose and fat (65% of energy)	BW increased in all study arms, by ≈0.3 kg. No significant difference between study arms (P > 0.05)	Liver fat increased by 16%, 86%, and 133% in fructose, fructose plus fat, and fat interventions. Fructose plus fat increased liver fat more than the other 2 interventions (P < 0.05)	NR	NR
			n = 8, 100% men; BMI 22.6 kg/m2; age, 24 y	Fructose and fat	4 d
			n = 10, 100% men; BMI 22.6 kg/m2; age, 24 y	Fat	4 d

Abbreviations: BW, body weight; NR, not reported; SSB, sugar-sweetened beverage; WM diet, weight-maintaining diet.

^aHypercaloric reflects energy balance when intervention intakes were compared with baseline intakes.

^bEnergy from SSBs was estimated on the basis of the dietary reference intake of a similar age group.

Publication biases and sensitivity analysis

All Egger’s tests were not significant (all P > 0.05). Sensitivity analyses utilizing correlation coefficients of 0.2 and 0.8 in place of a correlation coefficient of 0.5 for estimating standard error did not notably change the results in meta-analyses. Meta-analyses showed low or moderate heterogeneity (I² range, 0%–42%).

DISCUSSION

The present systematic review and meta-analysis observed an increased risk of fat accumulating in liver and muscle when excess added sugar intake was consumed as part of a hypercaloric diet compared with a eucaloric control diet. However, the differences in energy intake between the treatment and control groups in these studies complicate the interpretation of these findings. Only 3 studies have examined whether isocaloric intake from sugars may have a differential effect on fat accumulation relative to other attributes of diet,^13,39,42 and it is important to note that 1 study (out of 14 RCTs) observed that daily consumption of cola may differentially affect ectopic fat accumulation compared with an isocaloric intake of milk. This meta-analysis also showed no significant differential effect on fat accumulation in liver, muscle, or VAT between various types of sugars. In addition, this systematic review found that no RCTs have examined the effects of sugar intake on fat accumulation in local ectopic fat depots.

The primary observation, i.e., high-sugar hypercaloric diets increased ectopic fat accumulation compared with eucaloric control diets, is consistent with a recent meta-analysis in which a high-fructose hypercaloric diet increased liver fat accumulation.⁷ This finding is also consistent with the findings of a recent experiment conducted in monkeys. Monkeys on a high-fructose (24% energy) diet consumed more calories and had greater hepatic fat deposition than monkeys on a low-fructose (<1% energy) diet. Of the RCTs identified, only 1 study compared sucrose with high-fructose corn syrup, while 5 studies compared fructose with glucose. Both sucrose and high-fructose corn syrup are composed of 2 monosaccharides – fructose and glucose – and both sugars are able to be fully digested under normal conditions. As such, similar effects of various types of sugars on ectopic fat accumulation are not unexpected.

Interestingly, several review articles suggest that fructose tends to increase liver fat.^3–6 Under normal conditions, however, isolated fructose is rarely consumed in the absence of glucose in human diets. Moreover, as demonstrated recently by Lanaspa et al.,⁴³ glucose may be converted to fructose through the polyol pathway in the liver to promote fatty liver accumulation in a mouse model. As such, glucose may not be an appropriate comparator under a hypercaloric condition.

Other dietary components, particularly protein, may play a role in mitigating fat accumulation in ectopic depots. Maersk et al.¹³ observed that, compared with cola, isocaloric milk consumption may have a protective effect against ectopic fat depots, perhaps due to the protein. In line with this observation, Theytaz et al.³³ showed that consumption of 5 essential amino acids significantly attenuated the effect of a high-fructose hypercaloric diet on liver fat accumulation in a small group of healthy men. Similarly, in a short metabolic intervention study conducted in 10 healthy young men, Bortolotti et al.⁴⁴ observed that protein intake ameliorated liver fat accumulation induced by a high-fat hypercaloric diet. Thus, the interaction between sugar and protein on ectopic fat accumulation warrants further consideration.

The underlying mechanism for abnormal fat accumulation in ectopic depots is complex and remains to be elucidated. Excess added sugar intake is a contributing factor to greater energy intake. It is postulated that, under the condition of surplus energy intake, subcutaneous adipose tissue may become dysfunctional, and then a fraction of surplus energy is stored as fat in ectopic depots.¹ Dysfunction of subcutaneous adipose tissue may result from the inability of adipocytes to proliferate and differentiate normally.⁴⁵ Multiple factors may be involved in the accumulation of fat in ectopic depots, including genetic factors^46,47 and local hormone activity.⁴⁸ However, whether excess added sugar intake alone or whether the interaction between high sugar intake and a genetic predisposition or other external factors leads to dysfunctional subcutaneous adipose tissue remains largely unknown.

Alternatively, fructose, a sugar moiety in sucrose and high-fructose corn syrup, may play a crucial role in ectopic fat accumulation.³ It has been hypothesized that excess fructose consumption stimulates the synthesis of triglycerides in the liver, and, in the process of lipogenesis, intermediate products such as diacylglycerols may impair insulin signaling in the liver and peripheral tissues.⁴⁹ Subsequently, insulin resistance may facilitate excessive newly synthesized fat to be accumulated in ectopic fat depots.¹ This hypothesis is consistent with the findings of 1 study,⁵⁰ which demonstrated that individuals with higher liver fat had lower insulin sensitivity than individuals with lower liver fat, despite having similar VAT. Under normal conditions, triglyceride flux primarily leads to fat accumulation in subcutaneous adipose tissue because lipoprotein lipase in subcutaneous adipose tissue is more sensitive to insulin than lipoprotein lipase in VAT.⁵¹ However, when insulin resistance occurs, the less-insulin-sensitive lipoprotein lipase in VAT shunts triglycerides to deposit in VAT. Nevertheless, the molecular pathway channeling fat to be accumulated in ectopic sites is not completely understood.

This systematic review gives rise to several potential research questions. To date, no study in humans has examined whether added sugar intake may have different effects on ectopic fat depots relative to complex carbohydrates. Although lipid metabolism is different in animals than in humans, using animal models is useful to generate hypotheses. Several animal experiments in rats found that added sugar intake may induce adverse fat accumulation in the liver compared with starch intake under eucaloric conditions.^52–54 Experiments conducted by Pagliassotti et al.⁵³ and Roncal-Jimenez et al.⁵² showed that sucrose intake (68% and 40% of energy, respectively) promoted more fat accumulation in the liver in rats than the same intake levels of cornstarch. Sanchez-Lozada et al.⁵⁴ also showed that, compared with a control diet (starch as 46% of energy and maltodextrin as 20% of energy), a combination of fructose and glucose (60% of energy) increased liver fat and tended to increase intra-abdominal fat in rats. Therefore, studies designed to compare the effect of added sugar with the effect of complex carbohydrates on ectopic fat accumulation are warranted. Additionally, animal studies have observed that fructose intake may play a role in the development of local fat depots.^55–57 To date, however, no study in humans has examined the effects of sugar on local ectopic depots, including such sites as the kidney, pancreas, heart, and blood vessels.

The overall caveats of the present systematic review are that only limited numbers of studies were identified and that most of the included RCTs showed medium to low overall methodological quality. The overall risk of biases with respect to methodological quality of the studies included in the present systematic review was medium to high. For instance, several studies did not adequately describe the randomization procedure or state whether study participants were blinded to the study design.^23,30,37 A majority of included studies were conducted in men, thereby limiting the generalizability of the findings. Importantly, sample size was small in most of the trials, thereby limiting the power of these trials to detect a positive finding. Furthermore, the power calculations were often not reported, and the duration of intervention in most of these trials was short (<10 weeks). Thus, it is difficult to know whether studies observing long-term sugar consumption would result in different findings. In addition, many studies utilized a very high dose of fructose, which is far higher than the mean intake in the United States.⁵⁸

The present systematic review also considered whether observational studies, as complementary evidence to RCTs, may facilitate understanding of the relationship between long-term sugar consumption and ectopic fat accumulation. However, no prospective cohort studies examining sugar intake and ectopic fat deposition were identified through the literature search. Furthermore, because of the scarcity of literature, this systematic review could not examine factors that may modify the association between sugar intake and fat accumulation in ectopic depots such as liver and VAT. Last, but not least, the SMD method is intended to facilitate the synthesis of data rather than provide quantitative estimation.

CONCLUSION

This systematic review showed, on the basis of a limited number of RCTs, that excess intake of added sugars in combination with a hypercaloric diet would likely increase systemic ectopic fat depots, particularly in the liver and in muscle fat. There were too few studies that examined the potential differential effects of isocaloric intake of different sources or types of added sugars on ectopic fat accumulation. It remains unknown whether intake of sugar may have different metabolic effects on ectopic fat depots compared with intake of other macronutrients. Well-designed prospective cohort studies and RCTs with sufficient sample sizes and intervention durations are needed to further the understanding of the relationship between excess sugar intake and ectopic fat deposition.

SUPPORTING INFORMATION

The following Supporting Information is available through the online version of this article at the publisher’s website.

Figure S1: Risk of biases in randomized controlled trials

Table S1: PRISMA checklist

Table S2: Search strategy

Table S3: Sugar content and composition